The fabrication schematic of D-PCS is shown in Fig. S1, the polymer spheres with high N contents were first in-situ converted into reductive carbon spheres (CS-NH3) through the carbonization process under high-purity ammonia flow. Then the D-PCS with amorphous structure (Fig. S2-3) and abundant N-derived vacancy defects (Fig. S4-5) can be thus constructed during the rich pore formation process under KOH treatment. D-PCS exhibits monodisperse spheres with an average diameter of 500 nm (Fig. 1a). The symmetrical geometry is good at ions’ transportation during electrochemical tests. The electrochemical tests were carried out to evaluate the capacitance of D-PCS in common IL, 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIMBF4), as electrolyte within the high stable potential window of electrolyte from − 1.6 V to 2.4 V vs. Ag/Ag+ (Fig. S6)24, 25. The D-PCS express unique pseudo capacitive behavior in a 3-electrode cell with an Ag/Ag+ reference electrode, which is resulted from a narrower EDL based on the strong interactions between ions and vacancy defects that shifting ions from outer Helmholtz plane (OHP) to inner Helmholtz plane (IHP) (Fig. 1a).
The cyclic voltammetry (CV) profile exhibits a rectangular shape including four typical processes alongside additional broad anodic peak at 0.4 V and cathodic peak at -1.2 V, implying the occurrence of charge transfer process on the electrode (Fig. 1b). While the nitrogen doped porous carbon spheres (N-PCS), as the controlled sample, exhibited a rectangular shape CV indicative of typical EDL behavior (Fig. S7). The galvanostatic charge and discharge (GCD) curve of D-PCS shows a distinct change of slope at the potential of -0.2 V (Fig. 1c), similar with faradaic behavior in biredox ILs. There are also four distinguished electrochemical processes, which is corresponding well with the CV profile. Specifically, the Process I and III display high capacitance of 481 F g− 1 and 370 F g− 1 benefited from surface charge transfer reaction, respectively. The Process II and IV show relatively low capacitance of 203 F g− 1 and 277 F g− 1 mainly resulting from EDLC contribution, respectively. To the best of our knowledge, it is the first time pure porous carbon materials could achieve efficient pseudo capacitance in pure IL-based EC.
From the GCD curves at different densities, D-PCS can release reversible capacitance of 340, 308, 280, 262, and 250 F g− 1 at 5, 10, 20, 30, and even 50 A g− 1, respectively (Fig. 2a). Along with a tenfold increase in current density from 5 A g− 1 to 50 A g− 1, the capacity retention rate remains at 74%, demonstrating the outstanding rapid charge-discharge capability of D-PCS. In contrast, The N-PCS only shows straight triangle GCD curves and low EDL capacitance (Fig. S8). To provide deep insight into the charging/discharging kinetics, the relationship between capacity (Q) and GCD time (T) is discussed26, 27. In a plot of Q versus T1/2, linear regions represent capacity limited by semi-infinite linear diffusion whereas capacitive contributions are rate-independent. By taking a linear extrapolation of this region to T = 0, the capacity free from diffusion control can be estimated, which corresponds to the outer charge of the material, that is easily accessible by the electrolyte.
Q = Q outer + mT 1/2 (1)
1/Q = 1/Q total + nT − 1/2 (2)
In Eq. 1, the mT1/2 term represents the long-T data in linear region, which extrapolate y-intercept yields to Qouter of D-PCS (1332 C g− 1) in Fig. 2b, approaching the theoretical capacity of carbon (1339 C g− 1). Similarly, according to Eq. 2, a plot of 1/Q versus T− 1/2 (Fig. 2c) can be extrapolated to T = ∞ to estimate the total charge, which includes extra capacity shielded from diffusion inside the electrode. Here, the total charge of D-PCS is estimated at 1518 C g− 1, representing rate-independent capacity dominated ratio of 88% in total charge storage. This ratio indicates that the fast kinetics in faradaic reactions of D-PCS is comparable with EDL charge behavior, originating from the widespread distribution of vacancy defects in rich porous structure with high specific surface area (SSA).
It should be mentioned that a sealed 3-electrode testing system with an Ag/Ag+ reference electrode for electrochemical measurements is crucial to achieve more accurate and deep observation of possible pseudocapacitive processes than a traditional 2-electrode setup. The resulting CV curve scanned between 0 V and 4 V in a 2-electrode configuration is symmetrically rectangular and lacks any reversible peaks as shown in Fig. S9. This is understandable because the pair of reversible peaks shown in Fig. 1b spans the equilibrium potential of -0.2 V. In contrast, the two electrodes in the 2-electrode system are located on opposite sides of the equilibrium potential, with the potential (-0.2 ― 1.43 V) of positive electrode always above − 0.2 V and that (-1.57 ― -0.2 V) of negative electrode always below − 0.2 V (Fig. 2d). D-PCS express specific EDL capacitance of 215 F g− 1 in 2-electrode configuration, which is comparable with that of N-PCS (206 F g− 1). Hence, the asymmetric faradic capacitance of D-PCS can be detected in 3-electrode testing system but it fails in symmetric capacitor with 2-electrode configuration.
The additional CV tests in Fig. S10 show that the redox peaks of D-PCS express highly reversible in the smallest potential range from − 1.6 V to 1.0 V. Considering only the EDL capacitance can be measured in a 2-electrode test due to the limited potential ranges, the faradaic proportion of D-PCS can be obtained from the difference between capacitance tested under 2-electrode configuration (215 F g− 1) and 3-electrode configuration (340 F g− 1). Hence, it is concluded that the faradaic capacitance of D-PCS accounts for as much as 37% of the total capacitance, which is crucial for capacitance improvement. Thus, the radar map in Fig. 2e summarizes the Qouter of D-PCS achieves high ratio of 88% Qtotal, which is higher than 74% rate retention and also surpass the 63% EDLC contribution, indicating that the surface charge transfer reaction will not sacrifice its potential rate performance.
From the electrochemical impedance spectroscopy (EIS) plots in Fig. S11, all D-PCS electrodes show additional Warburg diffusions at intermediate frequencies compared to N-PCS, further validating faradic reaction occurred on the D-PCS electrode. To clarify the impact of testing temperature and electrode thickness on electrochemical kinetics, three D-PCS electrodes were used for EIS tests (D-PCS1 is labeled for default loading of 3.3 mg cm− 2 at 20°C, D-PCS2 and D-PCS3 are labeled for 3.3 and 7.3 mg cm− 2 at 60°C, respectively). The smaller semi-circles show that D-PCS2 and D-PCS3 have a lower equivalent series resistance compared to D-PCS1, originating from the lower viscosity and higher conductivity of the IL electrolyte at higher temperature. The obtained phase angle versus frequency results (Fig. S12) also indicate that the induced surface charge transfer reaction does not weaken rate capability. Furthermore, according to the GCD tests at 5 A g− 1, D-PCS2 and D-PCS3 exhibited high capacitance of 427 and 303 F g− 1, respectively (Fig. S13).
To assess the intrinsic capacitive storage ability of D-PCS electrode, the normalized area capacitance (CA) are calculated from the specific capacitance divided by SSA. D-PCS2 achieved a CA of 16 µF cm− 2, surpassing the theoretical EDL CA (~ 11 µF cm− 2) in neat IL electrolytes28. Hence, D-PCS electrode can achieve a reversible max capacitance over 400 F g− 1, which was much higher than that of reported organic ECs (Fig. 2f). In addition, even at 20°C, D-PCS1 achieved a higher CA than that of N-PCS (12.7 vs 9.8 µF cm− 2), demonstrating the efficient energy storage capability of D-PCS. The full comparative data are presented in Table S1. The long cycling capacitance of 262 F g− 1 at 30 A g− 1 for D-PCS presents no obvious decay within 10000 cycles (Fig. 2g). This indicates that the strategy of introducing vacancy defects not only implements excellent faradic capacitance, but also achieves outstanding cycling stability.
Since capacitive charge storage mainly occurs at the interface between electrode and electrolyte, the surface and porosity of carbon materials will play critical roles in the interactions and transportations of ions during the charging and discharging process. Here, the porosity of as-fabricated D-PCS and N-PCS is evaluated by nitrogen adsorption-desorption isotherms (Fig. 3a). It shows that D-PCS has a type IV isotherm and N-PCS has a type I isotherm, revealing that D-PCS exhibits more pronounced mesopores. The pore size distribution in Fig. 3b further confirms D-PCS displays larger mesopore diameter (2.73 nm) than that of N-PCS (2.16 nm). The mesopore size difference (0.57 nm) between D-PCS and N-PCS is approximately twice the diameter (0.28 nm) of the carbon hexagon, originating from the N vacancies that derived from the removal of N atoms from the layered carbon structure. Due to the widespread surface N vacancies, D-PCS also exhibits higher SSA (2676 vs 2109 m2 g− 1) and larger pore volume (2.36 vs 1.23 cm3 g− 1) than that of N-PCS.
To further identify defective carbon structure, electron energy loss spectroscopy (EELS) and electron paramagnetic resonance (EPR) are both conducted. In typical carbon K-edge EELS spectrum, the characteristic narrow π* peak (287 eV) and broad σ* peak (293 eV) are related to the electronic transition from 1s to the unoccupied anti-bonding π*-states of sp2 and σ*-states sp3 hybridized carbon, respectively29. Figure 3c demonstrates D-PCS’s lower π*/σ* ratio than that of N-PCS, indicating the lower sp2 hybridization in D-PCS. It is contributed to the formation of massive vacancy defects due to N atoms removing completely in D-PCS. That is also evidenced by the Lorentzian lines of D-PCS and N-PCS in Room-temperature EPR spectra (Fig. S14) 30, 31.
Based on the above characterization, the presence of vacancy defects is likely to have an impact on the charge distribution of surface carbon atoms, which could result in a potential effect on the adsorption of cation and anion in IL electrolyte during electrochemical process. The similar phenomenon also exists in a paper published by Gleb Yushin. Their CV profile of assembled supercapacitor at potential window from − 2.3 V to 2.3 V under 10 mV s− 1 displayed a pair of redox peaks in EMIMBF4 based ILs electrolyte. However, since the quite dilute nitrogen vacancy density (nitrogen content decreases from 2.52% in precursor to 0.72% in end product), compared with the rich nitrogen vacancy in D-PCS (N content below 0.5%) derived from CS-NH3 (high N content of 12.69%), the weak peak current provides limited capacitance improvement32. In addition, the O removed D-PCS(H) sample (see details in SI) also expressed evident reversible peaks (Fig. S15), which implies the faradic reactions not resulted from O surface groups but from vacancy defects. This is consistent with the typical EDLC behavior of N-PCS, which contain O functional groups as well (Fig. S16).
Ex-situ XPS C1s characterization was carried out on D-PCS and N-PCS at potentials of 2.4 V and − 1.6 V (vs. Ag/Ag+), to elucidate the inner mechanism of high capacitance for D-PCS. As shown in Fig. 3d, the XPS peak of D-PCS at 286.5 eV rose dramatically in comparison with that of N-PCS, which corresponds to the C-N band in CnEMIM. This can be attributed to the N atom of EMIM cations occupying the N-derived vacancy defects in D-PCS at the − 1.6 V potential. Furthermore, ex-situ XPS F1s spectra of D-PCS were obtained at potentials of -1.6 V and 2.4 V, as shown in Fig. 3e. Upon reaching 2.4 V, the F1s signals of CnBF4 at 690 eV and 685 eV were detected alongside the F1s signal of the PTFE binder at 686 eV. However, only the F1s signal of PTFE remained at the − 1.6 V potential, indicating interactions between BF4 anions and D-PCS occurred at the 2.4 V potential. Therefore, it can be inferred that the charge transfer reaction of D-PCS in pure EMIMBF4 is attributed to vacancy defects on the porous carbon surface of D-PCS.
In general, the GCD process of N-PCS demonstrates typical EDL capacitance behavior of adsorbed ions in the OHP, as shown in Fig. S17. However, in the case of D-PCS, a narrower EDL formed in IHP triggers faradaic reaction 22. Considering the distinct characteristic (e.g. the naked nonmetal cation and anion groups with non-solvation shell) of pure ILs from traditional dilute electrolytes33, vacancy defects in carbon structure would enhance the interactions between carbon and IL, which will impact the thickness of EDL. It is further confirmed by the adsorption energies between ions in IL electrolyte and electrodes calculated using density functional theory (DFT) methods (Fig. 3f). D-PCS showed stronger interactions with EMIM cations and BF4 anions, as indicated with the binding energy (Eb−EMIM = -0.98 eV, Eb− BF4 = -1.35 eV) compared to N-PCS (Eb−EMIM = -0.48 eV, Eb−BF4 = -0.93 eV). It may contributed to the existing CH-π intermolecular hydrogen bond with ILs electrolyte since more exposed π*(C = C) layer in D-PCS than N-PCS 34. In addition, alkyl imidazolium tetrafluoroborate ILs could cause n-type doping of carbon materials, originating from the changes in the electrostatic potential at the carbon/IL interface, which may also enable the faradic reaction to obtain high specific capacitance in the IL based ECs 35, 36.
Except for the neat IL electrolyte, the fundamental charge mechanism related to ions was explored by using mix solution with various concentration of EMIMBF4 in ACN solvent as electrolytes. CV profiles (Fig. 4a) found the peak potentials shifted to lower value along with the increase of ion concentration from 1.9 M to 5.8 M, which confirms EMIMBF4 participate in the charge transfer reaction. The data also allow us to construct the Tafel plot in Fig. 4b, to obtain the energetic of faradaic reaction. The potential at peak current (0.93 V, 0.95 V, and 0.98 V) lies well in a Tafel line with a slope of -102 (± 5) mV/decade, close to double of the theoretical slope of 2.3*RT/F (-59.2 mV/decade)10. Based on the fact that the Tafel line indicates a 2-electron reaction, it is possible that the concentration dependent potential shift arises from the simultaneous incorporation of both EMIM cations and BF4 anions (Eq. 3). To elaborate on this, we suggest that the electrochemical mechanism of the charge and discharge process can be illustrated as follows:
CnEMIM+BF4− – 2e = CnBF4+EMIM+ (3)
To fully leverage the pseudo-capacitive advantage of D-PCS and obtain high average voltage, we choose asymmetric capacitor for efficient energy storage. Also given the inability of BF4 encountering Li metal, 5M LiFSI in EMIMFSI was used as electrolyte to assemble Lithium hybrid capacitor (LIC) with D-PCS1 cathode and Li metal anode for potential practical application (Fig. 4c). The high reversible capacity of 192 mAh g− 1 and coulombic efficiency of 97% are achieved at the initial cycle in voltage range of 1.5 V-4.3 V under current density of 5 A g− 1, as shown in Fig. 4d. After 900 cycles, an increased capacity of 232 mAh g− 1 and coulombic efficiency of surpassing 99% are both remained (Fig. 4e). Based on the mass of cathode and anode, the device could achieve high energy density of 635 Wh kg− 1 and power density of 14.5 kW kg− 1. Considering the active material only make up 30% mass of the whole device, a practical energy density of 190 Wh kg− 1 can be estimated, still comparable with current Lithium ion batteries. The presented prototype of an asymmetric LIC demonstrates that pseudo capacitance from D-PCS can achieve practical high capacity in non-flammable electrolytes.