1 Introduction

Potassium-based energy storage devices are promising for the substitution of lithium-based energy storage devices in large-scale energy storage systems, because of several advantages such as naturally abundant potassium resources, low manufacturing costs, and comparable redox potential of potassium with lithium (K/K+: -2.93 V vs. Li/Li+: -3.04 V) (Ding et al. 2021; Zhang et al. 2021a). With the integrated merits of the high energy density of batteries and the high power density of supercapacitors, potassium-ion hybrid capacitors (PIHCs) are attracting increasing attention (Hu et al. 2020). Carbonaceous anode materials are one of the key components in PIHCs. Graphite, the typical anode material for commercial lithium-ion batteries, has been facing three thorny issues as follows when used as the anode material for PHICs: (1) a relatively low potassium-ion storage capacity (KC8, 278 mAh g−1) compared with lithium-ion storage capacity (LiC6, 372 mAh g−1) (Chen et al. 2019; Jian et al. 2015); (2) a relatively larger volume variation of approximately 60% due to a larger ionic size of potassium-ion (1.38 Å) than lithium-ion (0.68 Å) (Zhong et al. 2023), resulting in the structural collapse during repeated intercalation and deintercalation processes; (3) sluggish diffusion kinetics in (de-)intercalation processes caused by large ionic size of potassium-ion (Qiu et al. 2019), enabling low capacity at large current densities and short cycle life. Therefore, it is necessary to develop high-performance carbonaceous materials for PHICs.

Amorphous carbons, with short-range ordered graphene structure, enlarged interlayer spacing, and adjustable active sites, show great potential for enhanced potassium-ion storage (Guo et al. 2020; Wang et al. 2021, 2018; Zhao et al. 2023). In general, amorphous carbons are irregular monoliths with randomly distributed graphene nanodomains, in which potassium ions suffer from insufficient storage sites and inferior diffusion kinetics (Lu et al. 2018). Therefore, engineering strategies are necessary for amorphous carbon to construct accessible active sites for enhanced potassium-ion storage capacity and short diffusion distances to accelerate kinetics for improved rate capability and cycling stability.

Heteroatom doping is an effective way to engineer surface active sites by introducing electronegative atoms (such as nitrogen, sulfur, and oxygen) into carbon structure to strengthen the adsorption interactions with potassium ions. Edge-nitrogen dopant is demonstrated to be an effective site for the efficient adsorption of potassium ions (Li et al. 2018; Liu et al. 2020a; Share et al. 2016). There are three common methods for nitrogen dopant, including self-nitrogen dopant of carbon sources, co-pyrolysis of carbon sources and nitrogen sources, and pyrolysis of chemically-bonded macromolecules or supermolecules from carbon sources and nitrogen sources. Due to the trial and error approaches of nitrogen atoms into carbon structure in the pyrolysis of carbon sources and physically mixed carbon sources and nitrogen sources (Qiu et al. 2019; Zhang et al. 2020b), the nitrogen doping content is lower than 5% and nitrogen configuration is uncontrollable (Huang et al. 2022; Romero-Cano et al. 2019). The pyrolysis of macromolecules or supermolecules could achieve high nitrogen content (> 20%) and elaborate edge-nitrogen configuration (> 18%) (Zhang et al. 2020b, 2021b, 2020c). However, it is only suitable for the minority of supermolecules by pyrolyzing specific structural units. The nitrogen content and the edge-nitrogen configuration are commonly uncontrollable due to the thermal instability of nitrogen-containing units for the majority of macromolecules or supermolecules (Zhang et al. 2020b; 2020c). Graphitic carbon nitride (g-C3N4) has many nitrogen atoms located at the edge sites of the tri-s-triazine units, which shows potential for the synthesis of edge-nitrogen-rich structure. It is reported that g-C3N4 could be converted into an edge-nitrogen-rich framework during the pyrolysis process in the presence of transition metals or transition metal oxides (Jia et al. 2021; Wang et al. 2022a; Zhang et al. 2021d; Zhao et al. 2022). Therefore, introducing an edge-nitrogen-rich framework on the surface of carbon materials would be undoubtedly efficient for enhanced potassium-ion storage capacity.

Morphology and mesopore engineering are two effective ways to shorten diffusion distances. Compared with mesopore engineering, morphology engineering causes a lesser increase in specific surface area (Huang et al. 2021; Zhang et al. 2020a), which induces relatively slight interfacial side reactions. Two-dimensional (2D) morphology can shorten the diffusion distances for potassium ions (Zhang et al. 2021c). Three-dimensional (3D) morphology can provide a continuous conductive network for fast transportation of electrons (Wang et al. 2022a). Therefore, the 3D carbon nanosheet framework integrated with 2D carbon nanosheets possesses fast kinetics for the transportation of potassium ions and electrons, which facilitates the enhancement of rate capability and cycling stability. In addition, the 3D carbon nanosheet framework could provide sufficient surface space for the integration of an edge-nitrogen-rich framework.

Lignin, a hyperbranched aromatic macromolecule with a carbon content higher than 60%, a high annual production, and a low cost, is an ideal carbon source for the preparation of amorphous carbon (Zhang et al. 2022; Zhong et al. 2023). However, the severe agglomeration of lignin in the carbonization process creates great difficulty in the regulation of specific morphology. Our research group has innovatively found that the carbon nanosheet framework can be constructed through the gas release inhibiting the lignin agglomeration by the demethylation of methoxyl groups and cleavage of carboxyl groups in the pyrolysis process (Wang et al. 2023). The organic integration of edge-nitrogen-rich framework and carbon nanosheet framework is the key challenge to be conquered. Whether added before or after pyrolysis, the edge-nitrogen-rich framework cannot be perfectly integrated on the surface of carbon materials due to the lack of covalent linkages with the carbon nanosheet framework. Therefore, it is necessary to introduce an edge-nitrogen-rich framework into the carbon nanosheet framework with the assistance of other intermediates.

In this work, an elaborate “twice-cooking” strategy was developed for the integration of an edge-nitrogen-rich framework into a carbon nanosheet framework to successfully prepare edge-nitrogen-rich lignin-derived carbon nanosheet framework (EN-LCNF). The “twice-cooking” strategy involves two-step carbonization processes. In the first-step carbonization process, the staged release of CO and CO2 from CaC2O4 decomposition exfoliates the lignin-derived carbon matrix into lignin-derived carbon nanosheet framework (LCNF) and the CaO from CaC2O4 decomposition is decorated on the surface of LCNF. In the second-step carbonization process, melamine is first pyrolyzed into g-C3N4, followed by conversion into an edge-nitrogen-rich framework by reacting with the generated CaO, which is then in situ integrated on the surface of LCNF through sp3-hybridized C–N bonds. The prepared EN-LCNF has a high nitrogen content of 9.8 at.% and a considerable edge-nitrogen ratio of 72.3%, which delivers an excellent reversible capacity of 310.3 mAh g−1 at 50 mA g−1, a robust rate performance (126.4 mAh g−1 at 5000 mA g−1) and a long cycle life. The dual-carbon PIHCs assembled by commercial activated carbon cathode and EN-LCNF display the highest energy density of 110.8 Wh kg−1 at a power density of 100 W kg−1 and durable cycle stability. This work demonstrates a general and novel strategy to develop edge-nitrogen-rich carbon materials for advanced potassium-ion storage.

2 Materials and methods

2.1 Materials

Sodium lignosulphonate (LS) was purchased from Chempack Co., Ltd. in Russia. Calcium oxalate (CaC2O4·H2O) and calcium oxide (CaO) were purchased from Shanghai Macklin Biochemical Co., Ltd. Melamine was purchased from Shanghai Aladdin Biochemical Co., Ltd. Hydrochloric acid (HCl) was purchased from Guangzhou Chemical Reagent Co., Ltd. All raw materials were analytical grade and used without further treatment.

2.2 Preparation of edge-nitrogen-rich carbon materials

Typically, 1 g of LS was dissolved into deionized water by stirring for 1 h. Then, 2 g of CaC2O4·H2O or 0.77 g of CaO was added into the LS solution to obtain the uniformly mixed solution by stirring and ultrasonic treatment three times. In this process, CaO occurred in a hydration reaction to generate calcium hydroxide (Ca(OH)2). After that, the LS/CaC2O4 and LS/Ca(OH)2 precursors were synthesized by the drying of the mixed solutions at 105 °C under magnetic stirring.

The edge-nitrogen-rich carbon materials were prepared by two-step carbonization. In the first-step carbonization process, the precursors were carbonized using a tube furnace reactor (OTF-1200X-II, Hefei Kejing Instrument, China) by heating from room temperature to a target temperature of 700 °C for 2 h at 10 °C min−1 under a nitrogen atmosphere. The lignin-derived carbon nanosheet framework (LCNF) with CaO and lignin-derived honeycomb-like carbon monolith (LHCM) with CaO were obtained from the carbonization of LS/CaC2O4 and LS/Ca(OH)2 precursors, which are labeled as LCNF/CaO and LHCM/CaO, respectively.

Before the second-step carbonization process, 1 g of melamine was dissolved into the deionized water at 80 °C under continuous stirring. Then, the carbonized products were added to the melamine solutions to obtain the mixed solutions. In this process, CaO was converted into Ca(OH)2 due to a hydration reaction. The mixed solutions were dried at 80 °C to obtain the LCNF/Ca(OH)2/melamine and LHCM/Ca(OH)2/melamine precursors, respectively. Afterward, the precursors were carbonized by heating from room temperature to a target temperature of 700 °C for 2 h at 10 °C min−1 under a nitrogen atmosphere. During the carbonization process, Ca(OH)2 was in situ decomposed into CaO. After removing the impurities with HCl solution (1 mol L−1), the edge-nitrogen-rich LCNF (EN-LCNF) and edge-nitrogen-rich LHCM (EN-LHCM) were obtained from the carbonization of LCNF/Ca(OH)2/melamine and LHCM/Ca(OH)2/melamine precursors, respectively. Meanwhile, the pure LCNF was prepared from the re-carbonization of LCNF/Ca(OH)2 precursor without melamine.

2.3 Characterization

The microstructure and morphology were characterized using field emission scanning electron microscopy (SU8220, Hitachi, Japan) and transmission electron microscopy (JEM-2100F, JEOL, Japan). The phase composition was detected using an X-ray diffractometer (D8 Advanced, Bruker, Germany) with a Cu Kα radiation wavenumber of 0.1541 nm at 40 mA and 40 kV. The surface composition was determined using an X-ray photoelectron spectroscopy (K-Alpha, Thermo Scientific, America) with a 300 W Al Kα radiation as the excitation source. The FTIR spectra were recorded using a Fourier transform infrared spectrophotometer (Vertex70, Bruker, Germany) by the KBr pellet method. The pyrolysis behavior was analyzed using a thermalgravimetric analyzer (SDT 650, TA instruments, America) coupled with a mass spectroscopy detector (Discovery MS, TA instruments, America). The pyrolysis characteristics were carried out by heating from room temperature to 800 °C at a heating rate of 10 °C min−1 under a nitrogen atmosphere using a thermogravimetry analyzer (TG209F3, Netzsch, Germany). The Raman spectra were recorded using a Raman spectrometer (LabRAM ARAMIS, HORIBA, France) with a 532 nm laser with a power of 1 mW. The room-temperature electron paramagnetic resonance (EPR) spectra were measured using a Bruker spectrometer (E500 9.5–12, Bruker, Germany). The nitrogen adsorption/desorption analysis was carried out at 77 K using an analyzer (Tristar II 3flex, Micrometrics, America). The Brunauer Emmett Teller (BET) method and density functional theory (DFT) model were performed to calculate the specific surface area and pore-size distribution. The element analyses of C, H, N, and S were performed using an elemental analyzer (Vario EL Cube, Elementar, Germany).

2.4 Electrochemical measurements

A certain weight of polyvinylidene fluoride (PVDF) was added to the N-methyl pyrrolidone (NMP) to form a uniform solution under continuous stirring. Certain weights of active materials (EN-LCNF, EN-LHCM, and pure LCNF) and super carbon-conducting black were added into the above solutions to form uniform black slurries with a stirring process for 8 h. The mass ratio of active material, super carbon black, and PVDF was 8:1:1. The above black slurries were coated on copper foils and dried at 120 °C for 12 h in a vacuum oven. The working electrodes were obtained by cutting the obtained foils into circular discs with a diameter of 12 mm. The mass loading of the working electrodes was 0.6 mg cm−2.

The 2032-type coin half cells were assembled using a glove box filled with argon gas under concentrations of oxygen and water below 0.01 ppm. The potassium metal and glass fiber filter (Whatman, GF/F) were used as a reference electrode and a separator, respectively. 0.8 M KPF6 dissolved in ethylene carbonate/diethyl carbonate (EC/DEC, 1:1 v/v) was used as the electrolyte.

The galvanostatic charge/discharge (GCD) tests were performed under current densities of 50–5000 mA g−1 in the voltage window range of 0.01–3.0 V using a Neware battery test system (CT-4008 T, Neware Electronics, China). The cyclic voltammetry (CV) at scan rates of 0.1–1.0 mV s−1 and electrochemical impedance spectra (EIS) in a frequency range from 0.01 Hz to 100 kHz were measured using an electrochemical workstation (CHI660E, Chenhua Instruments, China).

The b value and capacitive contribution were calculated using Eq. (1) and (2):

$$i={av}^{b}$$
(1)
$$i={{k}_{1}v+{k}_{2}v}^{1/2}$$
(2)

where, a and b are variables, respectively; i and v are the current densities and scan rates, respectively; k1v and k2v1/2 are the capacitive-controlled and diffusion-controlled contributions, respectively. The slope (b) is calculated by the ratio of log(v) versus log(i). The b value close to 1 or 0.5 indicates that the electrochemical process is dominated by surface capacitive-controlled or diffusion-controlled processes, respectively.

The normalized surface capacitance (Cs) and diffusion coefficient of potassium ions (DK+) were calculated using Eq. (3) and (4):

$${C}_{s}=\frac{-1}{2\pi fZm}$$
(3)
$${D}_{{{\text{K}}}^{+}}=\frac{{0.5{\text{R}}}^{2}{{\text{T}}}^{2}}{{{\text{S}}}^{2}{{\text{n}}}^{4}{{\text{F}}}^{4}{{\text{C}}}^{2}{\upsigma }^{2}}$$
(4)

where, Cs (F g−1) is the gravimetric capacitance; Z (Ω) is the imaginary impedance measured at the lowest frequency; f (Hz) is the frequency; m (g) is the mass of active materials. DK+ (cm2 s−1) is the diffusion coefficient, R (J mol−1 K−1) is the ideal gas constant of 8.314, T (K) is the Kelvin temperature, S (cm2) is the surface area of the electrode, n is the number of electrons, F (C mol−1) is the Faraday constant of 9.6 × 104, C (mol L−1) is the concentration of potassium ions, and σ is the Warburg factor.

The diffusion coefficients at different voltages were calculated based on galvanostatic intermittent titration technique (GITT) measurement using Eq. (5):

$${D}_{{{\text{K}}}^{+}}=\frac{4}{\pi \tau }{(\frac{{m}_{{\text{b}}}{V}_{{\text{M}}}}{{M}_{{\text{B}}}S})}^{2}{(\frac{{\Delta E}_{s}}{{\Delta E}_{\tau }})}^{2}$$
(5)

where, τ is the time duration of the current pulse; mb is the mass loading of the active material; VM is the molar volume of the electrode material; MB is the molar mass of the electrode; S is the area of the electrode–electrolyte interface; ΔEs is the steady-state voltage change between before and after the current pulse; ΔEτ is the voltage change during the current pulse.

2.5 Assembly of potassium-ion hybrid capacitors (PIHCs)

The PIHCs were assembled in 2032-type coin cells. Similar to the preparation of working electrodes, the working cathode consisted of PVDF (10 wt.%), superconducting carbon black (10 wt.%), and commercial activated carbon (AC) (YP-80F, Kuraray, Japan) as active materials (80 wt.%). The YP-80F AC, superconducting carbon black, and PVDF were uniformly mixed in NMP, and then coated on aluminum foil. The working cathode was obtained by drying coated aluminum foil at 120 °C for 12 h in a vacuum oven and punched into circular discs (12 mm). The PIHCs were assembled by using EN-LCNF as an anode, YP-80F AC as a cathode, glass fiber (Whatman, GF/F) as the separator, and 0.8 M KPF6 in EC/DEC as the electrolyte. Before assembling, the EN-LCNF and YP-80F AC were cycled for 5 cycles at 50 mA g−1 in half cells to eliminate the irreversible capacities, respectively. To balance between cathode and anode, a series of PIHCs with cathode-to-anode mass ratios of 1:2, 1:3, 1:4, and 1:5 were assembled to achieve a proper kinetic.

The GCD and CV measurements of the PIHCs were tested using a CHI660E electrochemical workstation system. The long-term cycling performance of the PIHCs was tested on a Neware CT-4008 T test system. The tested voltage window was set from 0.01 V to 4.0 V. The energy density and power density of the PIHCs were calculated based on the total mass of the anode and cathode using Eqs. (6–8):

$$V=\frac{{V}_{{\text{max}}}+{V}_{{\text{min}}}}{2}$$
(6)
$$E={\int }_{{t}_{1}}^{{t}_{2}}iV{\text{d}}t\approx \frac{iV\Delta t}{3600}$$
(7)
$$P=\frac{E}{\Delta t/3600}=iV$$
(8)

where, E (Wh kg−1) is the energy density, and P (W kg−1) is the power density; t1 (s) and t2 (s) are the start and end of discharge time, and Δt (s) is the time for a full discharge; i (A g−1) is the charge current based on the total mass of the active materials in both anode and cathode; Vmax is the voltage at the beginning of the discharge, and Vmin is the voltage at the end of the discharge.

3 Results and discussion

The edge-nitrogen-rich lignin-derived carbon nanosheet framework (EN-LCNF) was prepared through a “twice-cooking” strategy, as illustrated in Fig. 1. Sodium lignosulphonate (LS) was first dissolved into a water solution and then mixed with calcium oxalate (CaC2O4) powders to form uniformly-mixed LS/CaC2O4 precursor under stirring and solution evaporation (Fig. S1). There are two-step carbonization processes at 700 °C in a “twice-cooking” strategy. In the first-step carbonization process of LS/CaC2O4 precursor, the staged gas releasing of CO and CO2 from CaC2O4 decomposition (Fig. S2) exfoliates lignin-derived carbon matrix into lignin-derived carbon nanosheet framework (LCNF) and calcium oxide (CaO) is decorated on the surface of LCNF (Fig. S3). Then, the LCNF/CaO product is placed into a melamine solution to prepare LCNF/Ca(OH)2/melamine precursor, where CaO occurs in a hydration reaction to generate Ca(OH)2 (Fig. S4).

Fig. 1
figure 1

Schematic diagram for the preparation of edge-nitrogen-rich lignin-derived carbon nanosheet framework (EN-LCNF) through a “twice-cooking” strategy

Full size image

In the second-step carbonization process of well-mixed LCNF/Ca(OH)2/melamine precursor, the melamine is pyrolyzed into graphitic carbon nitride (g-C3N4) and the Ca(OH)2 is decomposed into CaO, followed by the interactive reaction between g-C3N4 and CaO to form an edge-nitrogen-rich framework, which is in situ covalent-bonded with LCNF to form EN-LCNF. For comparison, the pure LCNF is prepared from the second-step carbonization process of LCNF/Ca(OH)2 precursor without melamine (Fig. 2a). In addition, the effect without gas release in the first-step carbonization process on the structure-performance relationship is further studied through the same “twice-cooking” strategy for LS/Ca(OH)2 precursor.

Fig. 2
figure 2

SEM images of (a) pure LCNF, (b, c) EN-LCNF, and (d) EN-LHCM. e TEM image of EN-LCNF. f HRTEM image of EN-LCNF. The golden arrows show the edge defects. g High-angle annular dark-field (HAADF) image of EN-LCNF, and EDS mapping of carbon (C), nitrogen (N), oxygen (O), and sulfur (S)

Full size image

The scanning electron microscopy (SEM) images show that pure LCNF remains the morphology of the 3D nanosheet framework after two-step carbonization and the surface is smooth (Fig. 2a). The edge-nitrogen-rich framework shows a fluffy and porous morphology (Fig. S5a). EN-LCNF exhibits a rough 3D nanosheet framework (Fig. 2b and c), suggesting the introduction of an edge-nitrogen-rich framework. In the first-step carbonization of LS/Ca(OH)2 precursor without gas releasing, a lignin-derived honeycomb-like carbon monolith (LHCM) is obtained (Fig. S5b and c). After the second-step carbonization with melamine, the edge-nitrogen-rich LHCM (EN-LHCM) shows a rough surface (Fig. 2d). This phenomenon is similar to EN-LCNF, indicating that the edge-nitrogen-rich framework is introduced into EN-LHCM. In addition, the edge-nitrogen-rich framework on the surface of EN-LHCM shows severe agglomeration compared with that of EN-LCNF. This phenomenon demonstrates that a 3D nanosheet framework facilitates the uniform integration of an edge-nitrogen-rich framework compared with a 3D honeycomb-like structure. The transmission electron microscopy (TEM) image of EN-LCNF shows the complete morphology of the 3D nanosheet framework (Fig. 2e). The high-resolution TEM (HRTEM) image shows that EN-LCNF has a turbostratic structure with defective graphene layers, of which the interlayer spacing ranges from 0.416–0.444 nm (Fig. 2f), which are the typical structural characteristics of amorphous carbon. The defects could adsorb potassium ions to enhance the reversible capacity. The larger interlayer spacing of EN-LCNF than the (002) lattice spacing of graphite (0.334 nm) could facilitate the fast transportation of potassium ions to accelerate kinetics. The uniform elemental distribution of nitrogen (N) detected by energy-dispersive X-ray spectroscopy (EDS) mapping (Fig. 2g) demonstrates the successful introduction of an edge-nitrogen-rich framework into EN-LCNF.

The formation mechanism of the edge-nitrogen-rich framework is first explored. The g-C3N4 is generated from the melamine pyrolysis at 500 °C, proved by the X-ray diffraction (XRD) peaks at 13.1° and 27.4° (Fig. 3a; Wang et al. 2009). The diffraction peak intensities of g-C3N4 at 600 °C become weaker, indicating the gradual decomposition of g-C3N4 in a temperature range of 500–600 °C, where the color changes from light yellow to dark yellow (Fig. S6). The decomposition of g-C3N4 could be demonstrated by the decrease in carbon content with increasing temperature (Table S4). When the temperature increased to 700 °C, g-C3N4 was completely decomposed without any pyrolysis product. Interestingly, the brownish-black pyrolysis products can be obtained from the pyrolysis of melamine with CaO at 700 °C (Fig. S7). The high-resolution N 1 s X-ray photoelectron spectroscopy (XPS) of the brownish-black pyrolysis product exhibits an edge-N ratio of 85.5%, while there is no edge-N configuration for g-C3N4 (Fig. 3b). Therefore, the brownish-black pyrolysis products are edge-nitrogen-rich framework. As shown in Fig. 3c, in the pyrolysis of Ca(OH)2/melamine precursor at 500–800 °C, the pyrolysis products at 500 °C mainly show the characteristic diffraction peak of CaO, indicating that Ca(OH)2 has decomposed into CaO at 500 °C (Fig. S8) and CaO did not react with g-C3N4 at 500 °C. The pyrolysis products at 600 °C show new diffraction peaks of calcium cyanamide (CaNCN) and the diffraction peak of CaO disappears. The diffraction peak intensity of CaNCN becomes stronger as temperature increases to 700 °C and 800 °C. The results indicate that CaO started to react with g-C3N4 to form CaNCN (Fig. S9) and an edge-nitrogen-rich framework at 600 °C. As shown in Fig. 3d, the Fourier-transform infrared (FTIR) spectra of the edge-nitrogen-rich framework show a significant difference in the stretching vibration of CN heterocycles at 1000–1700 cm−1 compared with g-C3N4 (Zhu et al. 2017), indicating that the edge-nitrogen-rich framework is formed by CaO reacting with the CN heterocycles, mainly the units containing C–N and C = N bonds (Zhang et al. 2021c). This phenomenon could also be proven by the lower total N content of edge-nitrogen-rich framework (33.0 at.%) than that of g-C3N4 (40.6–43.3 at.%) (Table S1).

Fig. 3
figure 3

a XRD patterns of the pyrolysis products derived from the melamine pyrolysis at 500–600 °C. b High-resolution N 1 s XPS of g-C3N4 and edge-nitrogen-rich framework. c XRD patterns of the pyrolysis products derived from the Ca(OH)2/melamine pyrolysis at 500–800 °C. d FTIR spectra of g-C3N4 and edge-nitrogen-rich framework. e XRD patterns of the pyrolysis products from the LCNF/Ca(OH)2/melamine pyrolysis at 500–800 °C. f TG-MS profiles of LCNF/Ca(OH)2/melamine precursor. g Schematic illustration of the integration of edge-nitrogen-rich framework into LCNF

Full size image

The mechanism for the integration of edge-nitrogen-rich framework into LCNF is further revealed. As shown in Fig. 3e, in the pyrolysis of LCNF/Ca(OH)2/melamine precursor at 500–800 °C, the pyrolysis products at 500 °C show the co-existence of the characteristic peaks of CaNCN and CaO, indicating that nanosheet framework could make CaO react with g-C3N4 at 500 °C by the full contact. Besides CaNCN, the weak diffraction peaks of CaCO3 and CaS are found, which is attributed to the multi-stage decomposition of CaC2O4 and the reaction between CaO and sulfur in LCNF (Wang et al. 2022b). When the temperature increases above 500 °C, the strongest diffraction peak intensity of CaNCN for the pyrolysis products at 700 °C indicates the complete reaction of CaO with g-C3N4 to produce the edge-nitrogen-rich framework. Then, the edge-nitrogen-rich framework is integrated into LCNF to form black EN-LCNF (Fig. S10).

The structural evolution was further tracked by TGA coupled with mass spectroscopy (MS) (TG-MS). The typical products such as H2O (atomic mass unit [a.m.u.] 18), CO2 (a.m.u. 44), CH4 (a.m.u. 16), CO2− (a.m.u. 30) are detected in the pyrolysis of LCNF/Ca(OH)2 precursor (Fig. S11). Compared with LCNF/Ca(OH)2 precursor, a large amount of small-molecular-weight fragment of CNH3 (a.m.u. 30) and a smaller amount of cyanamide unit of NCNH4 (a.m.u. 44) are identified in the pyrolysis of LCNF/Ca(OH)2/melamine precursor (Fig. 3f). The results demonstrate that the edge-nitrogen-rich framework is formed by CaO reacting with the cyanamide units in g-C3N4. In addition, the signal peak of H2O (a.m.u. 18) demonstrates that Ca(OH)2 was decomposed into CaO and H2O approximately at 400 °C.

Therefore, the integration of edge-nitrogen-rich skeleton into LCNF is schematically illustrated in Fig. 3g. First, the melamine was pyrolyzed into g-C3N4 at a temperature of about 500 °C. Second, with increasing temperature to 600–800 °C, CaO reacted with the cyanamide units in g-C3N4 to generate the edge-nitrogen-rich framework. The formed edge-nitrogen-rich skeleton was integrated into the carbon nanosheet framework maybe through covalent bonds.

The XRD patterns show that the average interlayer spacing are 0.366, 0.366, and 0.349 nm for EN-LCNF, EN-LHCM, and pure LCNF according to Bragg’s law, respectively (Fig. 4a). The interlayer spacing of EN-LCNF and EN-LHCM significantly increases after the introduction of edge-nitrogen-rich framework, which could accelerate the potassium-ion storage kinetics. The R-value can be used to evaluate the order degree of amorphous carbon (Zhang et al. 2020b, 2020c). The R-values of EN-LCNF (2.89) and EN-LHCM (2.77) are higher than that of pure LCNF (2.21) (Fig. S12), which demonstrates low-level lattice defects (Zhang et al. 2020b), maybe due to the integration of edge-nitrogen-rich framework at defects. EN-LCNF shows a higher R-value than EN-LHCM, indicating that the nanosheet framework could facilitate the integration of an edge-nitrogen-rich framework at defects compared with the honeycomb-like structure. The Raman spectra show the higher intensity ratio of the D and G peaks (ID/IG) for EN-LCNF (1.21) and EN-LHCM (1.24) compared with pure LCNF (1.13) (Fig. 4b), demonstrating the high level of sp3-hybridized structure after the introduction of edge-nitrogen-rich framework. The relatively lower ID/IG value of EN-LCNF than EN-LHCM demonstrates the fewer defects in EN-LCNF. The electron paramagnetic resonance (EPR) spectroscopy was further used to distinguish the microstructure (Fig. 4c). The higher g-value of EN-LCNF than those of EN-LHCM and pure LCNF indicates the strong nitrogen doping effect (Zhang et al. 2020c). The intensities of Lorentzian lines are highest for EN-LHCM and lowest for EN-LCNF, demonstrating the highest defect concentration of EN-LHCM and the lowest defect concentration of EN-LCNF. Given that the edge-nitrogen framework enriches defects (Zhang et al. 2021c), their severe agglomeration results in the existence of abundant defects for EN-LHCM. However, the defect level of EN-LCNF is lower than that of pure LCNF. The results suggest that the defect-rich edge-nitrogen framework was uniformly integrated into the defects of the carbon nanosheet framework mainly by sp3-hybridized bonds. The FTIR spectra of EN-LCNF and EN-LHCM show a characteristic C–N bond centered on a wavenumber of 1082.1 cm−1, while that of pure-LCNF is located at 1231.5 cm−1 (Fig. S13). The result demonstrates that the edge-nitrogen-rich framework was integrated into the carbon structure through sp3-hybridized C–N bonds.

Fig. 4
figure 4

a XRD patterns. b Raman spectra. c EPR spectra. d N2 adsorption–desorption isotherms. e Pore size distribution. f XPS survey spectra. High-resolution N 1 s XPS of (g) EN-LCNF, (h) EN-LHCM, and (i) pure LCNF

Full size image

The adsorption quantity at a relative pressure range of 0.6–1.0 P/P0 for EN-LCNF is lower than that for pure LCNF, suggesting the less mesoporous/macroporous structure (Fig. 4d), which is proven by the significant decrease in microporous/mesoporous volumes of EN-LCNF compared with pure LCNF (Table S2). The results demonstrate that the integrated edge-nitrogen-rich framework is located in mesopores/macropores, which could be intuitively reflected by the decreased pore volume at 4–100 nm (Fig. 4e). Because the edge-nitrogen-rich framework is integrated in microporous/mesoporous volumes, the specific surface area of EN-LCNF (1012 m2 g−1) slightly decreases compared with pure LCNF (1036 m2 g−1). EN-LHCM shows a lower specific surface area of 409 m2 g−1 and a higher mesoporosity of 58.1% than EN-LCNF (Table S2). The results demonstrate that the carbon nanosheet framework constructed by gas release in the first-step carbonization process could provide a higher specific surface area than the honeycomb-like structure constructed without gas release. The high surface area could expose more active edge-N sites for potassium-ion storage.

The surface chemical configurations of EN-LCNF, EN-LHCM, and pure LCNF are detected as shown in Fig. 4f. The XPS measurements show that the main elements of carbon (C), nitrogen (N), and oxygen (O) and a less amount of sulfur (S) are determined. Compared with pure LCNF (2.5 at.%), the higher N content of EN-LCNF (9.7 at.%) and EN-LHCM (7.7 at.%) demonstrate the successful integration of edge-nitrogen-rich framework. The elemental contents of the EN-LCNF, EN-LHCM, and pure LCNF determined by XPS measurements are similar to that of elemental analyses (Table S3). The high-resolution N 1 s XPS can be fitted into three configurations of pyridinic N (N–6, 398.4 eV), pyrrolic N (N–5, 400.1 eV), and graphitic N (N–Q, 401.4 eV) (Fig. 4g–i). The edge-N ratios for EN-LCNF, EN-LHCM, and Pure LCNF are 72.3%, 69.7%, and 29.1% and the corresponding edge-N contents are 7.0 at.%, 5.2 at.%, and 0.7 at.%, respectively (Table S4). The higher N and edge-N content of EN-LCNF than EN-LHCM demonstrate that the nanosheet framework is beneficial for the integration of edge-nitrogen-rich framework and the exposure of active edge-N sites. The high-level edge-N could contribute to high potassium-ion storage capacity (Zhang et al. 2020c). The high-resolution C 1 s XPS can be deconvoluted into C = C/C–C (284.6 eV), C = N/C–O (286.3 eV), C = O/C–N (288.3 eV), and π–π* (290.8) (Fig. S14; Fu et al. 2020; Wang et al. 2022a). The high-resolution O 1 s XPS can be deconvoluted into C = O (531.1 eV), C–O (532.3 eV), and COOH (533.5 eV) (Fig. S15; Jian et al. 2022; Wang et al. 2022d). The total contents of C = O and COOH are 4.7%, 4.6%, and 6.1% for EN-LCNF, EN-LHCM, and pure LCNF, respectively (Table S5). The active oxygen species of C = O and COOH can boost the potassium storage capability as well (Cheng et al. 2022).

Therefore, an edge-nitrogen-rich carbon nanosheet framework could be prepared by CaO reacting with the produced g-C3N4 from the melamine pyrolysis to form the edge-nitrogen-rich framework, followed by the in situ integration into carbon nanosheet framework through sp3-hybridized C–N bonds. This strategy can be extended to other nitrogen sources, such as dicyandiamide (Fig. S16). The edge-nitrogen-rich carbon nanosheet framework prepared using dicyandiamide as nitrogen source shows an N content of 8.8 at.% and an edge-N ratio of 66.3%, where the edge-N content is 5.8 at.%. This strategy is a general way to prepare edge-nitrogen-rich carbon materials.

The potassium storage performances of EN-LCNF, EN-LHCM, and pure LCNF are tested in half cells as shown in Fig. 5. The cyclic voltammetry (CV) curves of EN-LCNF show a cathodic peak approximately at 0.65 V in the first cycle (Fig. 5a), which corresponds to the formation of solid electrolyte interphase (SEI). The cathodic peak approximately at 0 V and the hump anodic peak at 0.45 V are related to the intercalation/deintercalation of potassium ions into/from defective graphene layers. The weak anodic peaks centered at 1.45 V and 2.12 V are attributed to the redox reactions of the active oxygen species and sulfur species with potassium ions (Cheng et al. 2022; Qiu et al. 2022). After the first cycle, the CV curves overlap well in the subsequent cycles, indicating the high reversibility of EN-LCNF. EN-LHCM and pure LCNF show similar CV profiles with EN-LCNF (Fig. S17). The smaller area of enclosed CV curves for EN-LHCM and pure LCNF compared with EN-LCNF suggests lower reversible capacities.

Fig. 5
figure 5

a The first three CV curves of EN-LCNF. b GCD curves at a current density of 50 mA g−1. c Rate capability at current densities of 50–5000 mA g−1. d GCD curves at various current densities. e Cycling stability at a current density of 1000 mA g−1 after initial 3 cycles at 50 mA g−1. f Performance comparison

Full size image

The initial Coulombic efficiencies (ICEs) are 23.5%, 29.0%, and 37.3% for EN-LCNF, pure LCNF, and EN-LHCM at a current density of 50 mA g−1, respectively (Fig. S18). The high specific surface area can induce intensified interfacial side reactions to form the SEI layer (Wang et al. 2022c), resulting in lower ICEs of EN-LCNF and pure LCNF than EN-LCNF. Compared with pure LCNF, the low ICEs of EN-LCNF are attributed to the inactive C–O groups irreversibly bonding with potassium ions (Wang et al. 2023). The galvanostatic charge–discharge (GCD) curves (Fig. 5b) show the higher initial reversible capacities of EN-LCNF (467.1 mAh g−1) and EN-LHCM (342.4 mAh g−1) than pure LCNF (258.9 mAh g−1) at a current density of 50 mA g−1, ascribed to their high edge-N content (Zhang et al. 2021c). Therefore, EN-LCNF with the highest edge-N content exhibits the highest initial reversible capacities. After 100 cycles at 50 mA g−1, EN-LCNF can maintain a reversible capacity of 310.3 mAh g−1, while the reversible capacity of EN-LHCM significantly decreases to 162.1 mAh g−1. The result can be attributed to the faster kinetics of the nanosheet framework than that of the honeycomb-like structure with high mesoporosity. Pure LCNF can display a reversible capacity of 200.7 mAh g−1 after 100 cycles at 50 mA g−1, relying on the active C = O/COOH groups interacting with potassium ions (Cheng et al. 2022).

The rate performance tests are performed at the current densities of 50, 100, 200, 500, 1000, 2000, and 5000 mA g−1 as illustrated in Fig. 5c. The EN-LCNF delivers the high capacities of 328.5, 279.8, 249.9, 219.3, 195.8, 169.1, and 126.4 mAh g−1 at corresponding current densities. The corresponding capacities for EN-LHCM are 234.2, 200.5, 164.3, 128.4, 96.4, 68.4, and 33.3 mAh g−1, while those for pure LCNF are 197.1, 174.6, 157.5, 134.9, 113.7, 93.3, and 62.5 mAh g−1. The excellent rate capability of EN-LCNF is attributed to the nanosheet framework accelerating diffusion kinetics and accessible edge-N sites. At large current densities of 500–5000 mA g−1, the inferior rate capability of EN-LHCM than that of pure LCNF indicates that the nanosheet framework is more beneficial to accelerate kinetics than the honeycomb-like structure with high mesoporosity. As shown in Fig. 5d, all profiles of GCD curves at different current densities for EN-LCNF show slope characteristics without the obvious platform, suggesting that EN-LCNF is more suitable for potassium-ion hybrid capacitors (PIHCs).

The long-term cycling stability of EN-LCNF shows remarkable durability at a current density of 1000 mA g−1 compared with those of EN-LHCM and pure LCNF (Fig. 5e). The EN-LCNF shows a high reversible capacity of 195.9 mAh g−1 after 3600 cycles with a capacity retention of 88.1%, which is attributed the fast kinetics by nanosheet framework and accessible edge-N sites. Due to the nanosheet framework accelerating kinetics, pure LCNF exhibits higher reversible capacity (111.3 mAh g−1) and capacity retention (75.5%) than EN-LHCM (58.9 mAh g−1 and 39.6%) after 1950 cycles.

The potassium-ion storage capability of EN-LCNF is superior to the most of previously reported N-doped carbon materials, especially at high current density (Fig. 5f and Table S6). The structure-performance relationships between N or O doping and potassium storage capacity are further explored from the perspective of surface characteristics. With the increase in the N concentration, the reversible capacities of carbon materials at 0.1 A g−1 generally increase (Fig. S19a). However, these carbon materials with ultrahigh N concentration would enable the fast capacity fading at larger current density (Zhang et al. 2021b). Interestingly, the carbon materials with a high N/O ratio could obtain high reversible capacities at 5 A g−1 (Fig. S19b). Therefore, a high N doping level can effectively enhance potassium-ion storage capacity, while the relatively higher level of O doping versus N doping is beneficial for enhancing the rate capability for potassium-ion storage.

The electrochemical impedance spectroscopy (EIS) shows the intercept of electrolyte resistance (Rs) at high-frequency regions, the charge transfer resistance (Rct) and double-layer capacitance (Cdl) at medium/high-frequency regions, the ion diffusion impedance (Zw) at medium/low-frequency regions, and the surface capacitance (Cs) at low-frequency region (Fig. 6a). The Rct values are 679.9, 706.5, and 807.2 Ω for EN-LCNF, EN-LHCM, and pure LCNF, respectively. The result indicates that edge-N could enable fast charge transfer. As expected, EN-LCNF shows the highest diffusion coefficient of potassium ions and surface capacitance (Fig. 6b), demonstrating that the carbon nanosheet framework could accelerate kinetics and the accessible edge-N sites could adsorb potassium ions. The galvanostatic intermittent titration technique (GITT) results (Fig. S20) demonstrate that EN-LCNF has the highest diffusion coefficient of potassium ions, followed by pure LCNF and EN-LHCM (Fig. S21). The higher diffusion coefficient of pure LCNF than EN-LHCM demonstrates the faster kinetics of pure LCNF by nanosheet framework than EN-LHCM by honeycomb-like structure with high mesoporosity. The fast kinetics could result in superior rate capability and cycling stability. EN-LHCM shows a higher surface capacitance than pure LCNF, demonstrating a higher storage capacity of EN-LHCM by edge-N adsorbing potassium ions than pure LCNF by C = O/COOH interacting with potassium ions.

Fig. 6
figure 6

a Electrochemical impedance spectroscopy (EIS) analyses. b Diffusion coefficient of potassium ions and normalized surface capacitance. c Dependence of b values on voltages. Capacitive contribution percentages of (d) pure LCNF, (e) EN-LHCM, and (f) EN-LCNF

Full size image

The CV curves at scan rates of 0.1–1.0 mV s−1 are measured to elaborate the charge storage mechanism (Fig. S22). All b values of EN-LCNF, pure LCNF, and EN-LHCM at different voltages are close to 1 (Fig. 6c), demonstrating the capacitive-dominated storage mechanism. In particular, the highest b values of EN-LCNF indicate the strong capacitive adsorption behavior, due to its highest edge-N content. The capacitive contribution was further quantified using Dunn’s method (Fig. S23). EN-LCNF shows a high capacitive contribution of 83.3% at a low scan rate of 0.4 mV s−1 (Fig. 6d), which is attributed to a large number of accessible edge-N sites and fast kinetics. However, EN-LHCM shows a capacitive contribution of 70.4% (Fig. 6e), which is lower than pure LCNF (79.4%) (Fig. 6f), ascribed to the fast kinetics of pure LCNF.

The PIHCs were assembled to evaluate the practicability of EN-LCNF, where the EN-LCNF was used as anode, and the commercial activated carbon (YP-80F, Kuraray, Japan) was used as cathode as illustrated in Fig. 7a. The anode/cathode mass ratios of 1:2, 1:3, 1:4, and 1:5 were used to optimize the matched kinetics between anode and cathode (Figs. S24–S27). The EN-LCNF//YP-80F PIHCs can achieve the optimal energy densities under different power densities in the voltage window range of 0.01–4.0 V (based on the total mass of the anode and cathode) when the mass ratio is 1:3 (Fig. 7b). The near-linear shape GCD profiles (Fig. 7c) and near-rectangular shape CV curves (Fig. 7d) result from the combinatorial charge storage mechanism of the non-Faradaic and Faradaic reactions. The EN-LCNF//YP-80F PIHC assembled at an anode/cathode mass ratio of 1:3 delivers excellent cycling stability at 1 A g−1, which shows a capacity retention of 98.7% after 6000 cycles (Fig. 7e). The EN-LCNF//YP-80F PIHCs can exhibit the highest energy density of 110.8 Wh kg−1 at a power density of 100 W kg−1, and retain an energy density of 60.1 Wh kg−1 even at a power density of 4000 W kg−1 (Fig. 7f). The energy-power performance of EN-LCNF//YP-80F PIHCs is superior to the most of previously reported PIHCs (Table S7), such as SHPNC//AC (Luo et al. 2020), C3N4@NCNF//AC (Shen et al. 2021), NCP//AC (Liu et al. 2020b), NCNT//AC (Li et al. 2020), P/O-PCS//AC (Zhao et al. 2021), and NPCF//AC (Sun et al. 2022). The fully-charged EN-LCNF//YP-80F PIHCs can power a thermometer operating well as illustrated in the insert of Fig. 7f, demonstrating the promising practicability of EN-LCNF.

Fig. 7
figure 7

a Schematic illustration for the working principle of PIHCs. b Optimization for PIHCs assembled with different anode/cathode mass ratios. c GCD curves. d CV curves. e Long-cycling performance. f Regone plots with practical application exhibition

Full size image

4 Conclusion

An edge-nitrogen-rich lignin-derived carbon nanosheet framework (EN-LCNF) is successfully engineered through a “twice-cooking” strategy including two-step carbonization processes at 700 °C. In the first-step carbonization process, the carbon nanosheet framework was constructed by the gas release of CO and CO2 from CaC2O4 decomposition, and the produced CaO was decorated on the surface of the lignin-derived carbon nanosheet framework (LCNF). In the second-step carbonization process, the edge-nitrogen-rich framework was first formed by the generated CaO reacting with the cyanamide units of g-C3N4 from the pyrolysis of melamine and then integrated into the meso-/macropores of LCNF through sp3-hybridized C–N bonds to obtain EN-LCNF. EN-LCNF delivers a high reversible capacity of 310.3 mAh g−1 at 50 mA g−1, an excellent rate capability of 126.4 mAh g−1 at 5000 mA g−1, and a superior cycling stability with capacity retention of 88.1% at 1000 mA g−1 after 3600 cycles. The robust potassium-ion storage capability of EN-LCNF can be attributed to the high-level edge-nitrogen configuration (7.0 at.%) reversibly adsorbing potassium ions and the nanosheet framework accelerating kinetics. The potassium-ion hybrid capacitors (PIHCs) assembled with EN-LCNF anode and commercial YP-80F activated carbon cathode deliver a high energy density of 110.8 Wh kg−1 at a power density of 100 W kg−1, and a robust cycling stability with a capacity retention of 98.7% after 6000 cycles. This work provides a general and novel way to exploit edge-nitrogen-rich carbon materials for advanced potassium-ion storage.