As shown in Scheme S1, melamine foam was used for the first time as a carbon precursor for the direct pyrolysis synthesis of 3D-NDCS. Different from powdered melamine, foamed melamine displays a porous network structure, which is conducive to the formation of 3D porous carbon skeleton during pyrolysis. Finally, the carbon skeleton gradually collapses and polymerizes when heated, forming a 3D porous sphere. In addition, the nitrogen content of melamine accounts for a high proportion of 66.7%, which is expected to bring high nitrogen doping. Figure S1 shows the scanning electron microscope (SEM) image of melamine foam after pyrolysis at 1000°C. The overall shape formed by the arrangement of rows of spheres, specifically, each column presents a candied haws-like string morphology composed of nano-sized porous spheres with a diameter of 40–60 nm, which helps to improve the transmission kinetics of active ions and ensure the adequate infiltration of electrolyte. The broad typical peak (002) with low peak intensity indicates the imperfect alignment of 3D-NDCS with large layer spacing (Fig. 1a), which is caused by the formation of amorphous structure due to the release of pyrolysis gas during the carbonization process of melamine foam [33, 34]. The R value reflecting the crystallinity of amorphous carbon introduced by Dahn is 3.36, indicating a low crystallinity with a high disorder of 3D-NDCS [48]. It can also be proved from the Raman spectrum that the G band at ~ 1580 cm− 1 is quite weak, and the value of ID/IG is 1.01, shows the large amount of defects and high amorphous state of 3D-NDCS (Fig. 1b) [49]. The results of EDX mapping represent an ultra-high nitrogen doping content of 36.46% (Figs. 1c and S2), which is significantly higher than most reported nitrogen doping levels. It can be proved from the FT-IR analysis in Figure S3 that the broad absorption peak around 1130 cm− 1 is attributed to the stretching vibration of C-N bond, and the absorption peaks near 1560 and 1655 cm− 1 are associated with C = N and N-H bond [50, 51], respectively. Figure 1c suggests that C, N, and O elements are uniformly distributed. The relative ratios of pyridine nitrogen (N6), pyrrole nitrogen (N5), graphitized nitrogen (NQ), and nitrogen oxide (NO) obtained via fitting high-resolution N1s XPS spectra are 42.3%, 39.5%, 14.0% and 4.2%, respectively, with a high proportion of 81.8% for edge-nitrogen (N6 and N5), which is of great significance to improve the Li+ storage capacity of 3D-NDCS (Fig. 1d) [32, 34, 52]. Remarkably, the N2 adsorption/desorption curves show a hierarchical pore distribution of micropores, mesopores and macropores for 3D-NDCS, which helps to store more active ions (Fig. 1e). Besides, the low BET specific surface area of 16.731 m2/g is expected to ensure sufficient contact with the electrolyte and inhibit the occurrence of parasitic reactions (Fig. 1f), thus forming a steady SEI film [53, 54].
The Li+ storage capability and electrochemical performances of the 3D-NDCS electrode was investigated by cyclic voltammetry scanned between 2.5-5.0 V, Fig. 2a shows the CV curves at a scan rate of 0.3 mV/s. The reduction peaks near − 2.8 and − 3.2 V in the first circle can be attributed to the decomposition of electrolyte for the formation of SEI [55, 56]. Notably, they disappear in the subsequent cycles, indicating the stability of SEI. Two pairs of redox peaks near − 4.8 and − 4.2 V correspond to the electrochemical reaction process. Moreover, all the curves except the first cycle coincide, indicating that the storage of Li+ is a stable and reversible process [49, 57]. The galvanostatic charge-discharge (GCD) curve of NDCS//NG-DIB at 1 C (1 C = 100 mA g− 1) displays typical active ion (de)intercalation characteristics with two pairs of charge/discharge plateaus (Fig. 2b), matching with the redox peaks in CV curves (Fig. 2a). Figure 2c exhibits the corresponding dQ/dV differential profile that each peak is consistent with the platform in the GCD curve. The charging process can be identified as three regions of 3.63–4.39 V (stage I), 4.39–4.83 V (stage II), and 4.83-5.0 V (stage III), representing different stages of Li+ intercalation into 3D-NDCS [58–60]. Similarly, the discharge process is also divided into three periods of 4.95–4.66 V, 4.66–4.27 V, and 4.27–3.63 V, implying different stages of Li+ deintercalation from 3D-NDCS. Furthermore, the NDCS//NG-DIB displays an operating voltage range of 4.95–3.63 V with a medium discharge voltage (Vm) of 4.28 V, which is much higher than commercial LiBs. To explore the Li+ storage behaviour and kinetics of the 3D-NDCS electrode, CV curves based on different scan rates have been obtained (Fig. 2d). Specifically, these curves present a rectangular-like shape with clear redox peaks, indicating the synergistic effect of diffusion and capacitive behaviour [61]. This synergistic effect can be quantitatively analyzed based on the power-law relationship between i (peak current) and v (scan rate): i = avb, where the value of b between 0.5-1 can be determined by plotting log(i) vs log(v), the b-value close to 0.5 or 1 meaning a diffusion-controlled or capacitive-controlled process, respectively [62, 63]. Figure 2e shows the b-values of the anodic and cathodic peaks are calculated to be 0.8226 and 0.8108, respectively, indicating that the kinetics of 3D-NDCS is mainly attributed to the capacitive-controlled process. Quantitatively, the mixed mechanisms can be divided into two separate mechanisms at a fixed potential by i(V) = k1v + k2v1/2, where k1 and k2 are constants, and k1v and k2v1/2 represent capacitance contribution and diffusion contribution, respectively [64]. As shown in Fig. 2f, correspondingly, the capacitance contribution to Li+ storage at 0.1 mV/s is 0.738. In addition, it increases to 0.829, 0.863 and 0.882 as the scan rates increase to 0.3, 0.5, and 0.7 mV/s, respectively.
To further investigate the rate capability and cyclic performance of NDCS//NG-DIB, a proof-of-concept full battery based on a 4 M concentrated electrolyte and a 3D-NDCS anode is constructed. Figure 3a demonstrates the typical GCD curves at different current densities from 1 to 15 C with voltage range of 2.5-5.0 V. As the current density increases, a slight charge-discharge plateau separation is observed, indicating the occurrence of electrochemical polarization. It can be proved from the dQ/dV differential curves in Figure S4 that as the rate increases, the oxidation/reduction peaks shift to a high/low potential with the peak intensities decreasing [49, 65]. Nevertheless, the separation is slow with the voltage platforms corresponding to the (de)intercalation of active ions still being clearly examined, indicating the weak polarization and fast kinetics behaviour. Moreover, specific discharge capacities (SDC) of 335, 278, 242, 223, 185, and 156 mAh g− 1 can be delivered at the rate scope of 1, 2, 3, 5, 10, and 15 C (Fig. 3b), the SDC can be restored to its initial value and cycled stably when the rate returns to 2 C, demonstrating the excellent reversibility and rate capability of NDCS//NG-DIB. For comparison, we also assembled a full DIB based on diluted electrolyte of 1 M, which displays unsatisfactory performance compared to concentrated electrolyte. As well as the increase of current density, the polarization becomes more serious and the charge/discharge plateaus tilt (Figure S5). Only 163, 142, 119, 88, 63, and 49 mAh g− 1 achieved at 1, 2, 3, 5, 10, and 15 C, respectively (Figure S6). Figure 3c shows the cycling performance of NDCS//NG-DIB based on concentrated/diluted electrolyte at 1 C. On the one hand, the initial specific discharge capacity (ISDC) based on concentrated electrolyte is as high as 351 mAh g− 1, and the capacity retention rate (CRR) of 95% after 80 cycles. Furthermore, the GCD curves of various cycles basically overlap (Fig. 3d), indicating outstanding cyclic stability, which showcases the best performance of DIBs reported so far. On the other hand, a ISDC of 148 mAh g− 1 based on diluted electrolyte is released, then decreases continuously in subsequent cycles with a limited SDC of 66 mAh g− 1 obtained after 80 cycles. The GCD curves display poor cyclic stability and serious polarization with disappearing charge/discharge plateaus as shown in Figure S7. The enormous discrepancy mainly benefits from the formation of sturdy SEI via the use of concentrated electrolyte. Researches by Atsuo Yamada et al. have shown that concentrated electrolytes are not only beneficial to the rapid transport of Li+ and steady circulation, but also inhibit the further decomposition of electrolyte and occurrence of side reactions, resulting in better electrochemical performance [42, 43, 66, 67]. This can be inferred from the Nyquist plots based on electrochemical impedance spectroscopy (EIS) in Fig. 3e and Figure S8, the intermediate frequency semi-circle represents the charge transfer resistance (Rct) at the electrode/electrolyte interface. The initial Rct based on concentrated electrolyte is slightly larger than that of diluted electrolyte, which is owing to the higher viscosity of high-concentration electrolyte. Interestingly, the Rct of diluted electrolyte is significantly higher than that of concentrated electrolyte in subsequent cycles. It is caused by the utilization of diluted electrolyte resulting in a fragile SEI, which will lead to the continuous decomposition of electrolyte and the formation of SEI. Besides, the NDCS//NG-DIB based on concentrated electrolyte exhibits a quite stable Vm up to 4.2 V during the long-cycling process (Fig. 3f). As the rate up to 4 C, the ISDC based on concentrated electrolyte is 255 mAh g− 1 with no capacity fade after 200 cycles, while the ISDC based on diluted electrolyte is only 110 mAh g− 1 with a CRR of 71% (Figure S9). Surprisingly, the battery still shows superior electrochemical performance even at a high rate of 15 C. The ISDC is 166 mAh g− 1, and it can be continuously cycled for 1300 cycles without capacity degradation with the CE maintains at ~ 99.5%. While the ISDC of 46 mAh g− 1 based on diluted electrolyte is much less than that of concentrated electrolyte and decays continuously to 36 mAh g− 1 after 1300 cycles, signifying a poor cycling performance.
Severe self-discharge will limit further practical application, especially in dual-graphite or dual-carbon systems, where the batteries usually deliver a serious self-discharge rate [68–70]. Therefore, the self-discharge performance was conducted. First, charging the full battery to the upper cut-off voltage of 5.0 V, and then discharging it after resting for 24 h. Figures 4a-b show the voltage-time curves of (un)resting, the battery still maintains a voltage of 4.41 V after resting. Consequently, the self-discharge rate of the system is calculated as 2.46% h− 1, which is considerably lower than that of the reported DIBs. In addition, stable fast charging and slow discharging can not only shorten the charging time, but also increase the usage time of batteries. Therefore, the performance was tested by fast charging at an ultra-high rate of 20 C, then slowly discharged at a low rate of 1 C to investigate the cyclic stability. The first 10 cycles of fast charging-slow discharging shows that the battery is fully charged within 5 minutes, while the discharging time exceeds 100 minutes (Fig. 4c). Besides, the ISDC reaches 176 mAh g− 1 and steadily increases to 186 mAh g− 1 after 300 cycles (Fig. 4d). This results in excellent fast charging-slow discharging performance, indicating that this dual-ion full battery system shows excellent stability with great potential for further practical application.
To elucidate the working mechanism of NDCS//NG-DIB. X-ray photo-electron derivative spectroscopy (XPS) was performed to characterize and analyze the elements of electrodes during the GCD processes, and the corresponding results are presented in Fig. 5. Figure 5a shows the high-resolution Li 1s spectrum. A sharp Li 1s characteristic peak appears at the fully charged stage, indicating the intercalation of Li+ [55]. After being fully discharged, the characteristic peak disappears completely, which means the reversible (de)intercalation of Li+. Moreover, a distinct F 1s characteristic peak appears after being fully charged (Fig. 5b). According to the previous work, which is due to the sharp decrease of free solvent molecules in the concentrated electrolyte, the anion (TFSI−) preferentially decomposed to form a robust SEI [42, 43, 66, 67, 71–73]. As shown in the high-resolution C 1s spectrum (Fig. 5c), the C-N characteristic peak significantly strengthened [52, 74]. Besides, a new characteristic peak corresponding to the decomposition of TFSI− appears [75]. After being fully discharged, the characteristic peaks of F 1s and C-N remains basically unchanged, while the decomposition peak of TFSI− weakened, indicating the formation of a robust SEI that no longer consumed electrolyte. As for NG cathode, the high-resolution S 2p spectra appears after being fully charged because of the intercalation of TFSI− (Fig. 5d), which becomes weaker with the deintercalation of TFSI− after being fully discharged. Furthermore, the (de)intercalation of TFSI− at NG cathode can be demonstrated from the FT-IR spectra based on pristine and fully charged/discharged stages. As shown in Fig. 5e, stretching vibration peaks corresponding to C-S, S = O, C-F and N-H appears at ~ 1047, 1127, 1188 and 1324 cm− 1, respectively after being fully charged, implying the intercalation of TFSI− [61, 65, 76]. Then all the characteristic peaks are significantly weakened due to the deintercalation of TFSI− after being fully discharged. Combined with the XRD results (explained in the following section), the remaining signal peaks are mainly due to the residual electrolyte and the decomposition of TFSI− and ES to form a CEI on the surface of cathode, just as there are still strong N-S (Figure S10), Li-F (Figure S11) and TFSI− decomposition peaks (Figure S12) after being fully discharged.
Combined with the above analysis, the working mechanism of the full battery is shown in Fig. 6, and the electrode reactions involved are as follows:
Anode: xCA + Li+ + e− ↔ CAx(Li) where A represents Anode
Cathode: yCC + TFSI− ↔ CCy(TFSI) + e− where C represents Cathode
Overall: xCA + Li+ + yCC + TFSI− ↔ CCy(TFSI) + CAx(Li)
XRD and Raman characterizations were performed to further explore the (de)intercalation behaviour of Li+/TFSI− with the concentrated electrolyte at the selected charged/discharged states. Figure 7a shows the corresponding time-voltage curve of the full battery for XRD measurement. As for NG cathode, the intensity of 002 characteristic peak at 26.5° gradually decreases during the charging process (Fig. 7b), and a new characteristic peak is split after being fully charged, which is caused by the formation of graphite intercalation compounds (GICs), corresponding to a layer spacing of 3.9 Å with intercalation stage of 1 [77–79]. The split peak disappears during the subsequent discharging process, while the 002 peak recovers to its original intensity after being fully discharged. Moreover, it remains a comparable intensity to the pristine peak even after 200 cycles, indicating an outstanding reversibility of the (de)intercalation process with excellent structural stability of NG. The Raman spectra of cathode in Fig. 7e can also corroborate the reversible (de)intercalation of TFSI−. The D band and G band located at ~ 1360 cm− 1 and ~ 1580 cm− 1 represent the defects and sp2 carbon atomic plane vibrations of the graphitic layer, respectively, and the ratio of ID/IG is usually used to measure the disorder degree of graphite [49, 69, 73]. The ID/IG ratio gradually increases during the charging process, representing a reduction of the crystallinity due to the intercalation of TFSI−. During the following discharging process, the ID/IG ratio gradually decreases and the crystallinity of graphite recovers with the deintercalation of TFSI−. For 3D-NDCS anode, as shown in Fig. 7c, the 002 peak gradually decreases with the continuous intercalation of Li+ during the charging process. In the meantime, the D band in the Raman spectrum continuously enhances, accompanied by an increase in the ID/IG ratio (Fig. 7f), indicating the enhancement of layer spacing and disorder of anode. During the discharging process, the 002 peak returns to original strength with the continuous deintercalation of Li+, which is manifested in the decrease of D band and ID/IG ratio, and the crystal structure of anode gradually recovers. Remarkably, all characteristic peaks do not shift during the entire process, and still maintain the comparative intensity to pristine peak after 200 cycles, proving the excellent reversibility of (de)intercalation and splendid structural stability.
The structure and morphology evolution of electrodes after 200 cycles were investigated by scanning electron microscope (SEM). On one hand, Figures S13a-b show the SEM images of NG cathode after 200 cycles. The lamellar structure of graphite can be clearly seen without any volume expansion and structural exfoliation, which proves the structural stability of NG. On the other hand, the surface morphology and structure of anode after 200 cycles are shown in Figures S13c-d, the complete 3D spherical structure is still displayed, indicating the splendid cyclic stability, which is consistent with the conclusion of XRD and Raman in Fig. 7.