Structure and Morphology of N, F-MCFs
The overall preparation process for N, F-MCFs is illustrated in Scheme 1. As shown in Scheme 1, the PAN/PVP/PVDF tricomponent nanofibers are first obtained by electrospinning a mixture of PAN, PVP, and PVDF in the DMF solution. Further, they are hydrothermally treated under 110 °C for 6 h, and then preoxidized and carbonized at 220 and 1000 °C, respectively, thereby resulting in N, F-MCFs. There are two mass ratios of PAN, PVP, and PVDF for the prepared N, F-MCFs, 1/1/1 and 1/1/1.5, which are separately named as N, F-MCFs-A and N, F-MCFs-B, respectively. To investigate the effects of pores and F atoms on N, F-MCFs, we synthesized N-MCFs-C without F atom doping from PAN/PVP bicomponent polymers and N, F-CFs-D without micropores from PAN/PVDF bicomponent polymers for comparison.
The morphology of electrospun PAN/PVP/PVDF nanofibers for the entire preparation process was characterized by SEM. As shown in Fig. 1a, smooth and uniform electrospun PAN/PVP/PVDF nanofibers exhibited a mean diameter of about 180 nm with random orientation. After removing PVP of the hybrid fibers by hydrothermal treatment, the surface of the nanofibers becomes rough and uneven, as shown in Fig. 1b. We can observe that some pores appear on the fibers. However, the continuous and uniform fiber structure is maintained after the hydrothermal treatment as well. The heat treatment of the porous nanofibers includes two processes, pre-oxidation and carbonization. During pre-oxidation in air at 220 °C, PAN undergoes cyclization and partial dehydrogenation, which make fibers more stable during subsequent high-temperature carbonization and create more defects for doping heterogeneous atoms . As shown in Fig. 1c, d, the obtained N, F-MCFs became thinner and denser after carbonization, and cracks formed on their surface. The nanofiber structure with pores and the amorphous carbon microcrystalline structure can be observed in the TEM images of N, F-MCFs-A (Fig. 1e, f).
The crystal structure and degree of graphitization are further characterized by XRD analyses and Raman spectrum. Two broad peaks at around 2θ = 25° and 44°, corresponding to the (002) and (100) planes of carbon, respectively, validate the amorphous carbon structure in Fig. 2a. Figure 2b shows three typical D, G, and 2D bands at about 1344, 1598, and 2798 cm−1, respectively. The D-band represents the defects and disordered structure of carbon lattice, while the G-band is a characteristic feature of in-plane vibration of sp2 bonded carbon atoms, which indicates the ordered structure of the carbon materials. It is known that the ratio of D-band and G-band (ID/IG) is attributed to determine the degree of graphitization or the defect density of carbon materials. The ID/IG ratios of N, F-MCFs-A and N, F-MCFs-B are as high as 2.98 and 2.31, respectively, which suggests that many defect sites and disordered structures are caused by doping N and F atoms. Moreover, they are in accordance with the broad peak of (002) obtained from the XRD results.
The pore structure of N, F-MCFs was investigated by adsorption–desorption isotherms of N2 at − 196 °C. As shown in Fig. 2c, d, both isotherms are between type I isotherms with type H4 hysteresis loops following IUPAC classification, thereby indicating the existence of micropores. The BET surface areas of N, F-MCFs-A and N, F-MCFs-B are 709.78 and 621.46 m2 g−1, respectively. The corresponding pore size distributions of the two samples are calculated by Barrett–Joyner–Halenda (BJH) desorption, and their average pore widths are 2.37 and 2.77 nm. The large BET surface area and the existence of micropores are advantageous for high ORR activities.
The elemental composition and the contents of N and F heteroatoms onto the catalysts surface were characterized by XPS measurements. The XPS survey spectra for N, F-MCFs-A and N, F-MCFs-B showed the existence of the C 1s, O 1s, N 1s, and F 1s peaks. The XPS quantitative result demonstrated that the relative surface mass ratios of C, O, N, and F are 89.01, 6.12, 2.06, and 2.81% in N, F-MCFs-A, and 87.48, 8.26, 1.79, and 2.48% in N, F-MCFs-B, respectively. The high-resolution N 1s spectrum can be further deconvoluted into four peaks centered at 398.5 ± 0.2 eV, 400.1 ± 0.2 eV, 401.1 ± 0.2 eV, and 404 ± 0.2 eV, corresponding to pyridinic N, pyrrolic N, graphitic N, and quaternary N, respectively. The relative content of pyridinic N and graphitic N is much higher than that of pyrrolic N and quaternary N, which may contribute to the high ORR activity of the catalyst [26, 27, 34]. The method of doping F atoms via PVDF can obtain higher content (> 2%) than those methods discussed in the earlier literature, which could be beneficial to the ORR electrocatalytic activity. The high-resolution F 1s spectrum was usually deconvoluted into semi-ionic F (688.8 ± 0.2 eV) and ionic F (685.4 ± 0.2 eV). These two peaks can be observed in the F 1s spectrum of N, F-MCFs-B in Fig. 3f, while the F 1s spectrum of N, F-MCFs-A can be deconvoluted into three peaks in Fig. 3e, which include ionic F (685.4 ± 0.2 eV) and two kinds of semi-ionic F, CH2–CF2 (689 ± 0.2 eV) and CHF–CHF (687.2 ± 0.2 eV). We compared the high-resolution F 1s spectrum of F-monodoped catalyst (prepared by PVDF carbon fibers catalysts) with N, F-MCFs in Fig. S4. The binding energy of the semi-ionic F in F-doped catalysts is lower than that in N, F-codoped catalysts, and the ionic F content of F-monodoped catalyst is much lower than N, F-codoped catalysts, which are probably because of the synergistic interactions between the F and N atoms. It is observed that ionic F can result in higher electrical conductivity and modification of electronic structures of carbon frameworks. Compared to the mass ratio of ionic F in N, F-MCFs-B (7.18%), the higher mass ratio of the ionic F in N, F-MCFs-A (14.83%) can provide more active sites for ORR to enhance the activity of the catalysts.
Electrocatalytic Activities Toward ORR of N, F-MCFs
The electrocatalytic activities of N, F-MCFs were first evaluated by cyclic voltammetry (CV) measurements in N2- and O2-saturated 0.1 M KOH solution. The CV curves of N, F-MCFs-A and N, F-MCFs-B in N2- and O2-saturated 0.1 M KOH solution are shown in Fig. S1. Figure 4a shows that the CV curves of N, F-MCFs-A and N, F-MCFs-B present two peak potentials at 0.881 and 0.846 V, respectively, which are higher than those of N-MCFs-C (prepared from PAN/PVP bicomponent polymers, without F-doped atoms) and N, F-CFs-D (prepared from PAN/PVDF bicomponent polymers without micropores). The more positive peak potentials indicate that more active sites for ORR are created by the high BET surface area and synergistic effect of the codoped heteroatoms. As shown in Fig. S2, the onset potential, half-wave potential, and limiting current density of N, F-MCFs-A are all higher than those of N-MCFs-C, which can also prove that F doping improves the ORR activity. To further investigate the high ORR catalytic activity of N, F-MCFs-A and N, F-MCFs-B, the linear sweep voltammetry (LSV) measurements were performed via a rotating disk electrode (RDE) in O2-saturated 0.1 M KOH solution. As depicted in Fig. 4b, N, F-MCFs-A present more positive onset potential (0.94 V vs. RHE) than that of N, F-MCFs-B (0.87 V vs. RHE) and it is also more approaching to that of the commercial Pt/C (0.95 V vs. RHE) because of its larger BET surface area and more doped content of N and F. The high ORR electrocatalytic activity of N, F-MCFs-A can also be gleaned from its higher half-wave potential (0.81 V vs. RHE) than that of N, F-MCFs-B (0.71 V vs. RHE) and close to commercial Pt/C (JM20, 0.83 V vs. RHE). However, N, F-MCFs-A exhibit a lower limiting current density (4.9 mA cm−2) than Pt/C (6.1 mA cm−2). In addition, because of concerning about the impact of glass corrosion , we measured the LSV curves of N, F-MCFs-A and Pt/C in O2-saturated 0.1 M KOH solution with Teflon container (Fig. S5) and confirmed that in the short test time poison of Pt/C catalyst by impurities released from the glass cell in alkaline medium should be very little and not detectable.
To further estimate the ORR reaction kinetics of N, F-MCFs, a series of LSV tests were carried out with various rotation speeds from 400 to 2000 rpm in an O2-saturated 0.1 M KOH electrolyte (Fig. 4c, e). On the basis of LSV curves at different rotations, the Koutecky–Levich plots and the electron transfer numbers (n) for ORR were obtained from the K–L equations. The great linearity and parallelism of the K–L plots (Fig. 4d, f) suggest a direct four-electron pathway for better ORR efficiency. The average n value of N, F-MCFs-A and N, F-MCFs-B is 4.0 and 4.1, which suggested that the complete reduction of O2 to OH− over N, F-MCFs by a four-electron transfer process in 0.1 M KOH. For further confirmation of 4e− selectivity, the yield of H2O2 was measured via rotating ring-disk electrode (RRDE) in Fig. S3. The percentage of H2O2 of N, F-MCFs-A is below 10%, and n is about 3.75 in the potential range from 0 to 0.8 V versus RHE (Fig. 4g) in 0.1 M KOH, which indicates the low peroxide formation and promising ORR activity. The poisoning experiment with CN−, which can strongly bond to active metal sites, was performed to identify the active sites of the prepared catalysts . As shown in Fig. 4h, although the limiting current density drops a little, the onset potential and half-wave potential of N, F-MCFs-A almost unchanged with and without CN− in 0.1 M KOH. That means the catalytic active sites in N, F-MCFs-A are primarily derived from the F-doped and N-doped carbon sites rather than from other metallic coordination sites.
The ORR electrocatalytic performance of N, F-MCFs was also tested in acid media. In Fig. 5a, CV measurements in O2-saturated 0.5 M H2SO4 show that the curve peak of N, F-MCFs-A is at 0.649 V (vs. RHE), at which the potential is negative as compared to that of commercial Pt/C (0.748 V vs. RHE). Similarly, LSV measurements (Fig. 5b) in O2-saturated 0.5 M H2SO4 with various rotating speeds from 400 to 2000 rpm and a scan rate of 10 mV s−1 also exhibit slightly poor onset potential (0.635 V vs. RHE) and half-wave potential (0.257 V vs. RHE). However, the electron transfer number is calculated to be approximately 4.0 for N, F-MCFs in 0.5 M H2SO4 from the corresponding K–L plots (Fig. 5c), which indicates that N, F-MCFs have a probable application prospect for ORR electrocatalytic activity in acid media.
Furthermore, the durability and tolerance to methanol oxidation of N, F-MCFs-A in 0.1 M KOH were also investigated. The current–time (i–t) chronoamperometric responses of N, F-MCFs-A and commercial Pt/C were measured under 0.6 V (vs. RHE) in O2-saturated 0.1 M KOH at a rotating rate of 1600 rpm for 10,000 s. As shown in Fig. 6a, the relative current density of N, F-MCFs-A decreased more slowly than commercial Pt/C with continuous reaction. It retained a superior higher relative current of 95.7%, while commercial Pt/C only retained 87.0%. The CV curves of methanol tolerance test (Fig. 6b, c) showed that methanol has almost no effect on N, F-MCFs-A; however, a typical methanol oxidation/reduction curve can be observed for Pt/C. Thus, N, F-MCFs-A exhibit not only higher stability but also better methanol resistance than commercial Pt/C, which indicates that N, F-MCFs-A can be a practical metal-free ORR catalyst in fuel cells.