Introduction

Since the shortage of the traditional fossil fuels and CO2 emission issues1,2, it become an urgent task to develop the renewable and clean energy sources. Toward this end, advanced technologies for both energy conversion (e.g., solar cells3,4 and fuel cells5) and storage (e.g., supercapacitors6 and batteries7,8) are being studied extensively. Actually, the catalysts for oxygen reduction play a key role in regulating the performance of fuel cells. However, it is still a great challenge to develop highly efficient and stable oxygen reduction reaction (ORR) electrocatalysts at a low cost since the scarcity, high cost and poor long-term stability of the Pt-based ORR catalysts are main obstacles for commercialization of the fuel cell technology.

The porous carbon family has been demonstrated to be promising candidates for high-performance energy-conversion and storage devices9. Recently, N-doping has been widely considered as one of the most feasible approaches to modulate the electronic and catalytic properties of graphene and its derivatives5,9,10,11,12,13,14,15,16. A large variety of heteroatom-doped carbon materials, including vertically-aligned nitrogen-containing carbon nanotubes (VA-NCNTs)12, N-doped graphene17, nitrogen carbide and N-doped ordered mesoporous graphitic arrays18, have been demonstrated to be efficient metal-free ORR electrocatalysts free from CO “poisoning” with a comparable even higher electrocatalytic activity and better long-term operation stability than the commercially available Pt-based electrodes (e.g., C2-20, 20% platinum on Vulcan XC-72R; E-TEK) in alkaline electrolytes. Moreover, N-doping technology has also been extensively used in highly efficient supercapacitors5,9,10,19,20.

Recently, porous covalent organic materials (COMs)21,22,23,24,25,26,27,28 with high porosity and controllable structural characteristics as well as multi-chemical functionality have been considered as self-sacrificed precursor to fabricate novel nanoporous carbons including graphene analogues. The COM-derived porous carbons show a lot of advantages. For example, their inherent optimized pore size allows ion migration and high surface area provides the possibility of formation of an electrostatic charge-separation layer and a highly ordered structure enables the formation of conductive paths. All these characteristics suggest that COM-derived porous carbon is a class of promising electrocatalyst materials. Pachfule et al. employed N-rich porous-organic-frameworks as template to prepare porous nitrogen-rich carbon under 1000°C and obtained ~40 mV enhancement in the onset potential compared to that of a commercial Pt/C catalyst29. Zhuang et al. prepared the S-/N-doped 2D porous carbons by the combination of graphene with 2D conjugated microporous polymers as metal-free catalysts for ORR and as supercapacitors30. Kou et al. synthesized a Aza-fused p-conjugated microporous framework (Aza-CMP) via phenazing ring fusion reaction and Aza-CMP@450 shows large capacitance of ~530 F g−1 at 0.2 A g−1 31.

3D structures provide an ideal platform towards the maximization of high specific surface areas, strong mechanical strengths and fast mass and electron transport kinetics for the development of high-performance catalysts, electromechanical and electrochemical devices10,32. Actually, the composition and structure of COM materials would determine the electrochemical properties of the resulting porous carbons. Here, to synthesize 3D carbon and reduce the effects of other elements in COM materials, we select porous aromatic framework (PAF-1, in which there are only C and H elements)33,34 with dia topology and ultrahigh BET surface area of 5640 m2 g−1 as self-sacrificed precursor to fabricate high efficient PAF-derived N-doped 3D carbon materials (marked as N-PAF-Carbon) under NH3 media and further investigate the electrochemical properties of the N-PAF-Carbon.

Results

Material characterizations

The FT-IR (Figure S1) and solid-state 13C/MAS NMR (Figure S2) measurements reveal the success of C-C coupling in synthesis processes, which is consistent with the previous report33. The pore size distribution derived from the N2 adsorption at 77 K shows that the pore size of as-synthesized PAF-1 is ~1.27 nm (Figure 1b), which also consists with the reported one35. Our thermo gravimetric (TG) results suggest the carbonization should be carried out with two steps, i.e., first preheated at 350°C for a period of time and subsequently carbonized at a given high temperature. As shown in Figure 1c, the sample heated directly under N2 atmosphere without preheating at 350°C almost loss all the weight before 900°C, while the sample with preheating at 350°C for 2 hours can reserve up 70% weight even under 900°C. The 1.54 wt% weight (the inset in Figure 1c) during this process can be ascribed to the adsorbed guests (e.g. O2, CO2) in the air due to ultrahigh surface area and the as-synthesized PAF-1 sample can be almost burned with the only adsorbed O2 molecules in the air.

Figure 1
figure 1

(a) The scheme for synthesis of PAF-1 materials. The cyan sphere in the right structure refers to the pore in PAF-1 and the purple lines refer to the framework of diamond structure. The actual structures of the statistical PAF-1 materials will be more complex than the one represented. (b) Pore size distribution and cumulative pore volume of PAF-1 obtained by non-local density functional theory modelling on the N2 adsorption curve. (b) TG analysis of as synthesized PAF-1. The inset refers to the enlargement of the TG analysis in blue circle region.

Full size image

For comparison, we also prepared undoped 3D PAF derived carbon material. For simplicity, we marked the carbonized PAF carbon materials as PAF-Carbon-T (T = 600, 800 and 1000°C and T is omitted when T = 800°C). As shown in Figure 2a, the sample PAF-Carbon-800 show the best graphitization degree (IG/ID = 1.07, Figure 2a) among the three PAF-derived carbon samples. The intensity of D band becomes higher than that of G band when the temperature is increased to 1000°C, suggesting the structure of PAF-1 may be completely destroyed and the disordered carbons dominate under 1000°C. The powder X-ray diffraction (XRD) spectra with two pronounced (0 0 2) graphitic peaks at 26° and (1 0 0) at 43° (Figure S3), indicate the formation of graphitic structures during carbonization. Combining the above results with the following ORR activities (vide infra), we employed 800°C as the optimal carbonization temperature for preparing N-PAF-Carbon. The Raman results of N-PAF-Carbon show a high ratio of IG/ID (1.23, Figure 2a), suggesting good graphitization degree. The X-ray photoelectron spectro-scopic (XPS) measurements show that the atomic percentage of N in N-PAF-Carbon reaches 4.1% (Figure 2b). The XPS N 1s spectra of N-PAF-Carbon (Figure 2c) could be deconvoluted into three sub-peaks: “graphitic” N at 401.1 eV, “pyrrolic” N at 400.1 eV and “pyridinic” N at 398.6 eV, in which the “graphitic” and “pyridinic” N dominate (Figure 2d). Transmission electron microscopic (TEM) images show porous texture in N-PAF-Carbon (Figure 2e). The XRD spectra of N-PAF-Carbon also shows two pronounced (0 0 2) graphitic peaks at 26° and (1 0 0) at 43° (Figure S3), suggesting that the N-PAF-Carbon has good crystalline quality.

Figure 2
figure 2

(a) Raman spectra of PAF-Carbons under different temperature and N-PAF-Carbon. (b) and (c) XPS survey spectrum and high-resolution N 1s spectrum of N-PAF-Carbon. The presence of oxygen can be ascribed to atmospheric O2, H2O, or CO2 adsorbed onto the surface of samples. (d) The percent content of three types of nitrogen in N-PAF-Carbon and PAF-Carbon. (e) TEM images for N-PAF-Carbon.

Full size image

Oxygen reduction reaction (ORR) evaluation

The electrocatalytic activities of the non- and N-doped PAF derived carbon materials were explored through cyclic voltammetry (CV), rotating disk electrode (RDE), rotating ring disk electrode (RRDE) and chronoamperometric measurements in 0.1 M KOH aqueous solution. Pt/C was also measured for comparation. The RDE measurements show the optimized annealing temperature is to be 800°C since the sample PAF-Carbon-800 possess the highest positive onset and half-wave potential (Figure 3b). Therefore, combining with the above characterization (vide supra), we employ 800°C as the carbonization temperature to obtain N-PAF-Carbon. As shown as in Figure 3a, both PAF-Carbon and N-PAF-Carbon show a well-defined cathodic peak centered in the CV curve when the electrolyte was saturated with O2-saturated 0.1 M KOH and present a featureless slope for the cathodic current with N2-saturated 0.1 M KOH within the entire potential range (Figure S4a and S5a). N-PAF-Carbon possesses ~20 mV positive cathodic peak than that of PAF-Carbon, suggesting a more pronounced catalytic activity of N-PAF-Carbon compared with non-doped PAF carbon.

Figure 3
figure 3

(a) CV curves of PAF-Carbon and N-PAF-Carbon in O2-saturated 0.1 M KOH at a sweep rate of 50 mV s−1. (b) LSV curves for PAF-Carbon-600, PAF-Carbon-800, PAF-Carbon-1000, N-PAF-Carbon and Pt/C in O2-saturated 0.1 M KOH at 1600 rpm at a sweep rate of 5 mV s−1. (c) RRDE curves for PAF-Carbon and N-PAF-Carbon at 1600 rpm. The calculated H2O2 yield derived from RRDE is also inserted. (d)–(f), Methanol and CO-poison effects and durability evaluation on i-t chronoamperometric responses for ORR at Pt/C (red) and N-PAF-Carbon (black) electrodes. In figure d, the arrow indicates the addition of 3 mL methanol into the O2-saturated electrochemical cell after about 400 s. In figure e, the addition of 55 mL min−1 CO gas into the 550 mL min−1 O2 flow saturated electrochemical cell after about 250 s.

Full size image

To gain further insight into the ORR process of PAF-Carbon and N-PAF-Carbon, RDE measurements were performed in O2-saturated 0.1 M KOH electrolyte at a scan rate of 5 mV s−1. As can be seen in Figure 3b, the N-doped sample exhibits much better catalytic activity compared with the non-doped PAF-Carbon. N-PAF-Carbon obtains ~70 mV enhancement for the half-wave potential compared with PAF-Carbon (0.75 vs. 0.68 V) and the limiting current is also improved as much as 80% (from 2.39 mA cm−2 to 4.30 mA cm−2) after nitrogen doping. In particular, the N-doped PAF sample shows only 44 mV less for the half-wave potential than the commercial Pt/C. Subsequently, RRDE measurements were carried out to determine their catalytic reaction pathway (Figure 4c). On the basis of the ring and disk currents, the electron transfer number (n) of N-PAF-Carbon was calculated to be 3.50–3.55 at 0 to 0.8 V (vs. reversible hydrogen electrode: RHE), suggesting a mainly four-electron oxygen reduction process. These results consist with those derived from RDE with K-L equation (Figures S5 and S6). In contrast, the electron transfer number for PAF-Carbon was less than 3 under similar condition. Particularly, the peroxide percentage for N-PAF-Carbon was remarkably reduced to <30% from 80% of non-doped PAF-Carbon sample. Basing on the above result, we can infer that the N-doping remarkably enhances the electrocatalystic activity of the PAF-Carbon for ORR. Furthermore, the chronoamperometric measurements show the N-PAF-Carbon is almost free from the CO and methanol crossover effects and holds better long-term durability compared with the commercial Pt/C. These results indicate that N-PAF-Carbon material is promising metal-free ORR electrocatalyst for fuel cells.

Figure 4
figure 4

(a) and (c) CV curves for PAF-Carbon and N-PAF-Carbon at scan rates of 10, 25, 50, 100, 200 and 500 mV s−1. (b) and (d), Charge–discharge profiles for PAF-Carbon and N-PAF-Carbon at current densities of 10, 5, 2, 1, 0.5 and 0.2 A g−1, respectively. (e), Specific capacitance of PAF-Carbon and N-PAF-Carbon at different current densities.

Full size image

Supercapacitance performance

Apart from enhancement for ORR activity of the N-doped technique, we also investigated the effect of N-doping on the supercapacitive behavior. As shown in Figure 4a and 4c, one significant feature in the I–V profiles is that no redox peaks are observable, which indicates that PAF derived carbon do not involve redox-related faradic processes during the charge-discharge processes. The I–V curves of both PAF-Carbon and N-PAF-Carbon at the different scan rates retain good symmetry, which confirms good electrochemical stability and capacitance. The current of the N-PAF-Carbon is more than 40% larger than those of non-doped PAF-Carbon. The specific capacitance was evaluated from the discharge curves (Figure. 4b and 4d). Under higher current densities than 0.5 A g−1, the capacitances of PAF-Carbon were largely enhanced after N-doping. In particular, the capacitance of PAF-Carbon is increased from 75 to 135 F g−1 (with 80% increment) after N-doping at 10 A g−1. When the current density was reduced to 0.2 A g−1, the charge-discharge profiles of PAF-Carbon show remarkable asymmetry character and the hysteresis appears in the discharge process, which may arise from the relatively poor electronic conductivity of PAF-Carbon. In a sharp contrast, the charge-discharge curves of N-PAF-Carbon sample exhibits good symmetry under low current density, even at 0.2 A g−1, which confirms the electronic conductivity of N-PAF-Carbon is largely enhanced after N-doping. The hysteresis results in the ‘inflated’ supercapacitance of PAF-Carbon (Figure 4c) compared with that of N-PAF-Carbon. Particularly, the supercapacitance of N-PAF-Carbon at 0.1 A g−1 reaches 385 F g−1, which is close to the previously reported ‘ultrahigh’ supercapacitance of Aza-CMPs (93–549 F g−1),31 N-doped graphene foam (484 F g−1)19 but superior to the state-of-the-art nanostructured carbon6,36, e.g., activated carbon materials (<270 F g−1)37,38,39, N-rich porous carbon materials (50–330 F g−1)6, carbon nanotubes (50–120 F g−1)40, graphene (135 F g−1)41, N-doped graphene film (280 F g−1)42 and carbon fibers (120–370 F g−1)43 and remarkably outperform those carbons derived from metal organic frameworks (MOFs)44,45,46. (See more results in Table S1 in the Supporting Information).

Discussion

To gain insight into the kinetic behavior of N-doping enhancement for supercapacitance, we further carried out electrochemical impedance spectroscopic (EIS) measurements. In Figure 5a, N-PAF-Carbon exhibits an approximately vertical line, which suggests that N-PAF-Carbon possesses lower diffusive resistivity for the electrolyte ions within the pores. N-PAF-Carbon show a near vertical line that was almost parallel to the Z″ axis (imaginary impedance axis), which indicated ideal capacitive behavior. The impedance spectra are fitted by using the electrical equivalent circuit illustrated in the insert of Figure 5a. In this circuit, Rs is the combined resistance of ionic resistance of electrolyte, intrinsic resistance of substrate and contact resistance at the active material/current collector interface. Rct is the contact interface resistance due to adsorption/desorption of ions on electrode materials. Zw is the Warburg impedance. CPE represents constant phase elements which are substituted for the capacitors to fit the high-frequency capacitive loop and Q is the limit capacitance. All fitted results are shown in Table S2. As shown in Table 1, the Rct was reduced from 1.02 to 0.20 Ω (81.6% decrease). The sum of Rs and Rct is the main contributor to equivalent series resistance (ESR), limiting the specific power of supercapacitors. The PAF derived carbon exhibits 52% shortage of the ESR (0.96 vs. 2.00 Ω) after N-doping. Moreover, we also evaluate the quantitative frequency response by plotting of imaginary capacitance versus frequency derived from the EIS profiles the quantitative. The maximum point of the resultant convex curves represents the frequency f0, at which purely resistive behavior is transformed into purely capacitive behavior. The reciprocal of f0 provides a time constant t0 (Figure 5b, inset), which is a quantitative measure of how fast the device can be reversibly charged and discharged. N-PAF-Carbon possesses much shorter reversible charge-discharge time (4 vs. 7s). All these results show the residence of N-PAF-carbon can be reduced and the electrolyte ions can be moved with faster way within in pores and the charge-discharge process can be finished in a shorter time, compared to PAF-Carbon without N-doping.

Table 1 Summary of porosities and supercapacitive behaviour of PAF-Carbon and N-PAF-Carbon
Full size table
Figure 5
figure 5

(a) Nyquist plots of PAF-Carbon and N-PAF-Carbon. Inset shows the impedance in the high and medium-frequency regions and an equivalent circuit as electrode materials is also inserted. (b) Plot of imaginary capacitance versus frequency. The inset shows a plot of the time constant τ0 of PAF-Carbon and N-PAF-Carbon. (c) N2 adsorption isotherms of PAF-Carbon and N-PAF-Carbon at 77K. Solid and open symbols represent adsorption and desorption, respectively. (d) NLDFT pore size distributions of PAF-Carbon and N-PAF-Carbon by incremental pore volume. (e) Charge–discharge curves of N-PAF-Carbon during a test of 9000 cycles. The profiles of three time zones of the initial, middle and last 700 seconds are selected to confirm that they are identical to each other in shape and symmetry. (f) Capacitance percentage of N-PAF-Carbon for a 9000-cycle charge–discharge test at a current density of 5 A g−1.

Full size image

The structure effects also co-operatively contribute to the superior capacitive performance of N-PAF-Carbon. First, the “graphitic” and “pyridinic” N species in N-PAF-Carbon increase the density of the electrolyte ions accumulated on the pore walls through electrostatic interactions with protons31 and facilitate the formation of charge layers. Second, a large surface area is necessary to provide an interface for the formation of electrostatic charge-separation layers in the pores. N-PAF-Carbon possesses a larger specific surface area (SSA) (764 vs. 418 m2 g−1 for SBET, see Figure 5c and Table 1) compared to PAF-Carbon. Moreover, the pore size has a significant effect on the performance of N-PAF-Carbon. pore size is perfectly adapted to the ion size and ion adsorption is achieved in the most efficient way, by minimizing the free space available47. It appears a smaller pore 1.09 nm in N-PAF-Carbon (Figure 5d), which is much closer to the size of electrolyte ions (sulfuric acid electrolyte (proton): 0.42 nm, sulfate: 0.53 nm) than 1.29 nm of PAF-Carbon. At this point, more suitable pore size in N-PAF-Carbon improves the movement of ion species, compared to PAF-Carbon. All these characters indicate that N-PAF-Carbon has outstanding supercapacitive performance.

Furthermore, N-PAF-Carbon also exhibits excellent stability without any loss in capacitance after 9000 charge–discharge cycles at a current density of 5 A g−1 and the charge–discharge profiles retain the linearity and symmetry well (Figures 5e and 5f). In short, N-PAF-Carbon possesses not only large capacitance but also an excellent cycle life and is very promising capacitor electrode materials.

In summary, we firstly developed the physical exposure to nitrogen-containing moieties doping approach to introduce N heteroatom into covalent organic materials. Our results suggest the N-doping technique enhance not only the ORR electronic catalysis but also supercapacitive behavior. The N-doped PAF derived carbon obtains ~70 mV enhancement as to half-wave potential and 80% increase as to the limiting current for ORR catalysis. Moreover, the N-doped PAF derived carbon displays free from the CO and methanol crossover effect and better long-term durability compared with the commercial Pt/C benchmark. Furthermore, N-PAF-Carbon based capacitor device is only 0.96 Ω equivalent series resistance low diffusive resistivity of the electrolyte ions within the pores and N-PAF-Carbon possesses not only large capacitance (385 F g−1) but also an excellent performance stability without any loss in capacitance after 9000 charge–discharge cycles. These results clearly indicate that N-PAF-Carbon material is promising metal-free ORR catalyst for fuel cells and promising capacitor electrode materials.

Methods

PAF-1 was prepared according with the previous reported method33. By heating the bis(1,5-cyclooctadiene) nickel(0) ([Ni(cod)2] and 2,2′-bipyridyl in dehydrated DMF with Tetrakis(4-bromophenyl)methane at 85°C for overnight, we obtained the PAF-1 powder after drying in vacuum. The scheme for synthesis of PAF-1 was shown in Figure 1a. The activated PAF-1 sample was put into the temperature programming-furnace for degassing at 350°C for 3 hours and then carbonized at a given temperature. The resulting sample was marked as PAF-Carbon. To prepare nitrogen doped 3D carbons (marked as N-PAF-Carbon), we employ the above similar carbonization method, i.e., the activated PAF-1 was put into the temperature programming-furnace for degassing at 350°C for 3 hours and then further carbonized under NH3 media at a given temperature to produce the final products. The detailed synthetic methods can be found in the supporting information.