Cheng
Han
a,
Xiaodeng
Zhang
a,
Qiuhong
Sun
a,
Dandan
Chen
a,
Tingting
Miao
a,
Kongzhao
Su
b,
Qipeng
Li
bc,
Shaoming
Huang
d and
Jinjie
Qian
*ab
aKey Laboratory of Carbon Materials of Zhejiang Province, College of Chemistry and Materials Engineering, Wenzhou University, Wenzhou, Zhejiang, P. R. China. E-mail: jinjieqian@wzu.edu.cn
bState Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou, Fujian, P. R. China
cCollege of Chemistry and Chemical Engineering, Zhaotong University, Zhaotong, Yunnan, P. R. China
dSchool of Materials and Energy, Guangzhou Key Laboratory of Low-Dimensional Materials and Energy Storage Devices, Guangdong University of Technology, Guangzhou, P. R. China
First published on 4th April 2022
The oxygen reduction reaction (ORR) on the cathode is of great significance in fuel cells and metal–O2 batteries, which is well demonstrated and highly efficient in nature. Herein, iron phthalocyanine (FePc) is encapsulated into the nanosized pores of a new porous metal–organic framework (MOF) of BMM-14 to give FePc@BMM-14. It shows a series of atomically dispersed Fe-based active species embedded into N-doped Fe@NC(x+y) samples after calcination, where x and y represent the FePc concentration and temperature, respectively. During pyrolysis, the rapid volatilization of Zn contents can further expand the framework to form more loose carbon nanomaterials with abundant active components, including FeNx and ultrafine Fe-based nanoparticles. Compared to the state-of-the-art Pt/C, the obtained Fe@NC(10+900/1000) exhibits satisfactory ORR performance to give a large limiting current of −5.22/−5.16 mA cm−2 (0.3 V vs. RHE), small Tafel slope of −51.1/−38.9 mV dec−1, and excellent stability. Meanwhile, the assembled batteries based on the two above catalysts show high activity, indicative of being potential substitutes to those noble metals consolidated by the theoretical calculations. This work demonstrates phthalocyanine-induced Fe-based active species in MOF-derived graphitic carbon as efficient ORR electrocatalysts for zinc–air batteries, which can be extended into the reasonable design and preparation of cost-effective but highly efficient catalysts for various energy-related applications.
Microporous metal–organic frameworks (MOFs) have been fully demonstrated in gas storage and separation, drug delivery and release, heterogeneous catalysis, and so on.14–19 As one subclass of crystalline MOFs, zeolite imidazole frameworks (ZIFs) have also been widely investigated on account of their high porosity, adjustable structure, and rich carbon and nitrogen sources.20,21 MOF materials are structurally rich in inorganic metal units and have a high content of organic ligands that make them potential precursors for multifunctional metal carbon composites. Due to the fast volatilization of Zn species, the formation of porous N-containing carbons calcined from Zn-based ZIFs can be achieved with a great possibility.22,23 Therefore, a large number of ZIF-derived carbon nanomaterials such as carbon nanotubes and carbon nanosheets have been widely utilized in different energy-related applications, because of their large specific surface area, excellent electrical conductivity, high stability and other significant advantages.24–27 In this context, the rational design and facile preparation of porous MOF-derived carbon-based catalysts in energy conversion and storage is highly important and desired.
On the other hand, the phthalocyanine compounds show a large conjugated system of 18 electrons structurally similar to the porphyrins that exist widely in nature.28,29 Among them, the structure of iron phthalocyanine (FePc) resembles the core of hemoglobin, which is highly selective for adsorption and transport of oxygen in human blood.30–32 In addition, the atomically dispersive iron active centers have been considered to be an efficient ORR catalyst in those recent reports.33–35 For example, Dai et al. have synthesized FeNx species well confined within micropores of the hierarchically porous carbon by the direct pyrolysis of the FePc complex, which outperforms the commercial Pt/C in terms of ORR activity.36 Meanwhile, these carbon-coated iron nanoparticles (NPs) as active catalytic species could be another type of non-precious metal oxygen reduction catalysts previously demonstrated by the Gewirth group.37 Thus, the rational combination between MOF and FePc will lead to the formation of Fe-based species after calcination, namely FeNx sites as well as Fe NPs, which deserves an in-depth investigation for efficient ORR performance.
Herein, a new porous Zn-based MOF (BMM-14) is solvothermally synthesized by three different imidazole ligands; for a more detailed experimental procedure please see the ESI.† Through a simple two-solvent diffusion method, discrete FePc can be successfully dispersed into this MOF to yield FePc@BMM-14(x), where x (0.1, 1, 10 g L−1) represents the concentration of FePc within 50 mg of the MOF precursor. After being pyrolyzed at different temperatures, the morphology of FePc@BMM-14(10) exhibits a flake-like shape to prove the influence of the introduced phthalocyanine on the obtained carbon. It also shows that the volatilization of Zn species can further expand the carbon layer, which is beneficial for the formation of a more loose N-containing framework, especially for Fe@NC(10+900/1000). These as-calcined Fe,N-doped carbon nanomaterials are endowed with the advantages of high porosity and a large specific surface area, while the low content of FePc results in generating atomically dispersive FeNx centers and ultrafine Fe nanoclusters. In this case, the optimal products of Fe@NC(10+900/1000) present a satisfactory ORR activity, showing the limit current of −5.22/−5.16 mA cm−2 at 0.3 V vs. RHE, low Tafel slope of −51.1/−38.9 mV dec−1, and excellent stability compared to the state-of-the-art Pt/C. Meanwhile, the assembled zinc–air batteries (ZABs) based on the Fe@NC catalysts give a superior performance with robust long-term stability, indicative of the potential substitutes to those noble metal based catalysts that are further supported by the following theoretical models. The present work confirms the reasonable design and synthesis of transition metal based ORR electrocatalysts by low-concentration filling of phthalocyanine compounds, which would pave a new pathway to prepare novel catalysts for efficient energy storage and conversion.
The electron transfer number (n) and yield of the H2O2 intermediate were calculated according to the Koutecky–Levich (K–L) eqn (1) and (2) and RRDE technologies eqn (3) and (4), as shown below:
(1) |
(2) |
(3) |
(4) |
In these formulae, J is the experimental current density, JL is the diffusion-limited current density, and JK is the kinetic current density, respectively; ω is the rotation speed in rpm (round per minute), F is the Faraday constant (96485 C mol−1), C0 is the bulk concentration of oxygen (1.2 × 10−6 mol cm−3), D0 is the diffusion coefficient of oxygen in 0.1 M KOH (1.9 × 10−5 cm2 s−1), and υ is the kinetic viscosity (0.01 cm2 s−1). When the speed of rotation is also represented as rpm, 0.2 is a constant. The n can be extracted from the slope of the K–L plot. Ir is the ring current, Id is the disk current and N expresses the collection efficiency of the ring electrode (0.37).
The aforesaid obtained electrocatalyst ink as the cathode is dropped onto carbon paper until the loading mass is 1.0 mg cm−2 and dried in air. Besides, one Zn foil becomes the anode and the paper is used to absorb 6.0 M KOH electrolytes and separate the electrodes. Finally, these materials are assembled in perforated button-cell shells for testing. All Zn–air button battery tests are performed under ambient atmosphere at room temperature. Discharge power density is calculated from the discharge polarized profiles using the following equation:
P = Ud × Jd |
Ea = EO2+sub − EO2 − Esub |
Due to the large porosity and abundant channels, a simple double-solvent dispersion method is used to encapsulate different amounts of FePc molecules.39 In Fig. 2a, the powder X-ray diffraction (PXRD) verifies that the as-made BMM-14 crystals are in good agreement with the simulated peaks, in which four strong diffraction signals at 2-theta values of 6.2°, 6.7°, 9.7°, and 10.2° are well matched with the (2, −1, 1), (3, 0, 0) (0, 2, 2) and (2, −1, 2) planes, respectively. After that, a total of 50 mg of pristine MOF is dispersed in 4 mL of n-hexane, while x (0.1, 1, 10) g L−1 of FePc in dimethyl sulfoxide is also prepared, then the above two solvents are mixed to obtain the FePc@BMM-14(x) sample. From FT-IR spectra, there are no new characteristic peaks detected after loading with different concentrations of FePc into the crystalline MOF (Fig. 2b). However, the optical color of the exchanged samples exhibits a significant change confirmed by the UV-vis spectra before and after the dispersion treatment in Fig. 2c and S5.† The PXRD patterns of FePc-encapsulated MOFs show that the obtained FePc@BMM-14(10) with the highest concentration owns a typical characteristic peak at 27.1° validating the successful loading of FePc (Fig. 2d). Meanwhile, the composite presents no obvious difference in the crystal form, of which the overall framework is almost retained. In Fig. 2e and f, these collected N2 isotherms exhibit the type-I curves, and a trend of reduced gas capacity with the filling of FePc is also observed. Furthermore, the fitted pore size distribution (PSD) curves can further confirm that all the pores and channels are centered in the microporous regions, while the number of micropores of FePc@BMM-14(x) is significantly reduced after the loading of FePc. For more specific surface area and PSD data please see Table S2.† It is indicative that the small FePc molecule has been evenly distributed in the MOF cavity that will be conducive to the formation of dispersive Fe atoms after the subsequent thermal treatment.
The morphology change of the FePc@BMM-14 series is first studied by SEM images. After continuous stirring, the bulk single crystals are cracked into micro-sized particles with an increased FePc concentration to achieve an adequate dispersion. For the control sample of FePc@BMM-14(0), there is no difference compared with the broken particles in Fig. 3a1, a2 and S6.† However, the surface of FePc@BMM-14(x) becomes significantly rough (Fig. 3b–d and S7–S9†), especially for FePc@BMM-14(10), which is reasonably selected to carry out the following thermal treatments. After calcination, these original particles will be almost destroyed which can be reflected in the products of Fe@NC(10+900/1000) with the most obvious sheet-like structure. This can be well explained by the bubbling process induced by the volatilization of Zn species in the carbon layer, thereby achieving the pore-forming effect.40 The subsequent gas sorption and PSD data verify that the obtained Fe@NC(10+900/1000) possess a higher adsorption capacity with distinct hysteresis loops in Fig. 3g, h, S13 and S14.† However, the total N2 uptake of Fe@NC(10+700/800) does not change significantly, which stemmed from the lower carbonization process in Fig. 3e, f, S11 and S12.† Meanwhile, the increased BET surface areas are in the following order Fe@NC(10+700) (60.32 m2 g−1), Fe@NC(10+800) (104.00 m2 g−1), Fe@NC(10+900) (334.75 m2 g−1), and Fe@NC(10+1000) (420.30 m2 g−1), respectively. Furthermore, the difference in PSD further proves that the carbon at higher temperature presents the larger total pore volume induced by the Zn evaporation. The enhanced micropore content of Fe@NC(10+900/1000) is indicative of the superiority of both carbon nanomaterials which are endowed with abundant active sites that would accelerate the electron transfer and mass diffusion during the electrochemical performance.
Fig. 3 SEM images of (a–d) FePc@BMM-14(x); (e–h) SEM images, sorption isotherms and PSD curves for the calcined Fe@NC(10+y) series. |
Transmission electron microscopy (TEM) and high resolution TEM (HR-TEM) further reveal the nanostructure and morphological features of these carbon materials. In Fig. 4a1–d1, the Fe@NC(10+y) series present the microstructure of bulky body and sheet-shaped edge, but the nanosheet structures of Fe@NC(10+700/800) are not as obvious as Fe@NC(10+900/1000), which verifies the incompleteness of low-temperature carbonization and the characteristics of amorphous carbon. Each sample shows the tendency of the carbon layer to wrap the metal sources where the distribution of metals is relatively compact and poorly dispersive, especially for Fe@NC(10+700/800) (Fig. 4a2–d2). In this case, the metal aggregation in the graphitic carbon comes from a large amount of Zn from the MOF precursor as well as Fe from the FePc molecules. On the other hand, the ultrafine NPs are evenly distributed and atomically dispersive metals wrapped in the N-doped carbons are detected in Fe@NC(10+900/1000). Meanwhile, low-temperature carbonization leads to the formation of some inconspicuous lattice stripes of carbon (0.33 nm) in Fig. 4a3 and b3. An ordered nanostructure on the edge corresponds to the (0 0 2) plane of graphite for Fe@NC(10+800) confirming the existence of graphitized carbon, but the inner carbon maintains a disordered structure (Fig. 4b3). However, the inner/outer carbon reaches an ordered structure for Fe@NC(10+900) in Fig. 4c2 and c3, and a small number of cavities are in situ formed during pyrolysis. The same phenomenon can also be detected in Fe@NC(10+1000) where cavities in the middle of the carbon layer become more obvious. Meanwhile, the hollow carbon structure coated with Fe atoms shows orderly structure characteristics in Fig. 4d2 and d3, and for more TEM images please refer to Fig. S15–S18.† Finally, the uniform distribution of Fe@NC(x+y) can be clearly examined in the EDX spectra and elemental mappings (Fig. 4a4–d4). This is indicative of the successful formation of the thermodynamically stable FeNx species because of the low content of FePc in the microporous channels. The obtained hierarchically porous and graphitic carbon nanomaterials ensure the fast transfer of the electrolytes and electrons, and the active sites are thoroughly exposed to achieve the enhanced electrochemical performance under an extremely high atomic utilization.
Fig. 4 TEM/HR-TEM images, EDX spectra of (a1–a4) Fe@NC(10+700), (b1–b4) Fe@NC(10+800), (c1–c4) Fe@NC(10+900) and (d1–d4) Fe@NC(10+1000). (e) Elemental mappings of Fe@NC(10+1000). |
To characterize the content and coordination environment, X-ray photoelectron spectroscopy (XPS) is performed to show the coexistence of Zn, Fe, C, N, and O (Fig. 5a and S19†). Almost invisible Fe 2p peaks in the full spectra of Fe@NC(10+900/1000) confirm the presence of trace FePc introduced in the MOF pores. Meanwhile, the decrease or even disappearance of the Zn 2p signal also proves the evaporation of Zn after high temperature carbonization.41 In Fig. 5b, S20 and Table S3,† the deconvoluted O 1s peak shapes in Fe@NC(0/10+900) are almost the same where three peaks at 530.1, 531.2 and 532.2 eV are considered as the M–O, C–O and the CO bonds, respectively. The deconvoluted C 1s spectra in Fig. 5c reveal the existence of four different carbon coordination environments (C–C sp3, CC sp2, C–O/C–N and CO/CN) and there is no change in the positions of the four peaks, which confirms that the coordination environment of carbon atoms in the five material has not changed during the change of pyrolysis temperature. In this case, element N mainly comes from imidazoles in the microporous MOF in the high-resolution spectra of N 1s in Fig. 5d and Table S5.† The fitting peak positions of N in Fe@NC(10+900/1000) exhibit almost no deviations, indicating that the high-temperature carbonization poses no influence on the coordination environment of Fe and N. On the other hand, the changes in the coordination environment of Zn and Fe have been further studied in Fig. 5e and f. There is a slight negative shift trend of the 2p1/2 and 2p3/2 peak positions of Zn(II) in Fe@NC(10+700/800) compared to those in Fe@NC(10+900/1000), while the binding energies of 2p1/2 and 2p3/2 of Fe(0, II, III) tend to shift in the positive direction compared to those of Fe@NC(10+900/1000). The deconvoluted Fe 2p spectra show three pairs of weak peaks at 706/720, 709/721, and 712/724 eV for Fe(0), Fe(II), and Fe(III), respectively, implying the Fe–Nx sites in Fe@NC(10+y) samples. At higher temperatures, the binding energies of 2p1/2 and 2p3/2 of Zn(II) for Fe@NC(10+900/1000) suddenly rise as a result of the reduction and evaporation of Zn(II) species, while the binding energies of 2p1/2 and 2p3/2 of Fe(0, II, III) decrease, especially in Fe(II) and Fe(0). For more XPS data and peak changes, please refer to Fig. S19–S22 and Tables S3–S7.†
Fig. 5 (a) Full XPS survey spectra; the deconvoluted (b) O 1s, (c) C 1s and (d) N 1s of Fe@NC(0+900) and Fe@NC(10+900/1000); the deconvoluted (e) Zn 2p and (f) Fe 2p of Fe@NC(0+900) and Fe@NC(10+y). |
Inspired by the nanostructure as well as abundant FeNx species in Fe@NC(x+y), the electrochemical ORR performance is further evaluated in 0.1 M KOH, and all applied potentials have been calibrated to the reversible hydrogen electrode. In Fig. 6a, the cyclic voltammetry (CV) curves in N2-/O2-saturated solution at 5 mV s−1 show that the obtained Fe@NC(0+900) owns the inferior oxygen reduction peak of 0.798 V. It reveals that the position of the reduction peak has moved significantly to 0.854/0.912/0.915/0.921 V for Fe@NC(10+700/800/900/1000), respectively, but it is still lower than Pt/C (0.940 V). Meanwhile, the linear sweep voltammograms (LSV) at 1600 rpm in Fig. 6b for Fe@NC(10+900/1000) exhibit a higher onset potential and a larger diffusion limited current density (JL) of −5.22/−5.16 mV cm−2 at 0.3 V, higher than Fe@NC(0+900) (−2.49 mV cm−2), Fe@NC(10+700/800) (−3.99/−4.24 mV cm−2), and Pt/C (−4.95 mV cm−2). When compared to previous reports, the Fe@NC(10+900/1000) catalyst demonstrates a competitive ORR performance in Table S8.† The LSV curves from 100 to 2500 rpm are summarized, and the Koutecky–Levich curves of Fe@NC(10+900/1000) present almost linear and parallel fitting lines from 0.3–0.7 V (Fig. S23 and S24†). In Fig. 6c, all samples possess relatively small Tafel slopes for Fe@NC(10+700/800/900/1000) (−48.6/−47.4/−51.1/−38.9 mV dec−1) lower than Pt/C (−54.2 mV dec−1), indicating the excellent reaction kinetics with enhanced mass transfer ability. Furthermore, the calculated number of transferred electrons (n) and the HO2− yields validate the high O2 selectivity and effective 4-electron process (Fig. 6d). The electrochemical impedance spectrum (EIS) in Fig. 6e shows that the obtained Fe@NC(10+1000) is endowed with the low charge transfer resistance of 7.2 Ω in the high-frequency region, lower than for Fe@NC(0+900) (9.4 Ω), Fe@NC(10+700/800/900) (14.5/7.6/12.0 Ω) and Pt/C (18.8 Ω). On the other hand, the calculated double-layer capacitance (Cdl) shows that Fe@NC(10+800/900) (37.66/36.71 mF cm−2) owns a larger electrochemical surface area than Fe@NC(10+700/1000) (13.17/15.25 mF cm−2), Fe@NC(0+900) (16.95 mF cm−2) and Pt/C (9.32 mF cm−2), proving that both high temperature and multi-element doping can improve the ability of electron/ion transport (Fig. 6f and S25†). By inserting methanol, the excellent tolerance of Fe@NC(10+900/1000) with current retention of 98.60%/98.99% is observed compared to Pt/C of 72.43% with an obvious current attenuation in Fig. 6g. Finally, a constant current at 0.6 V is also performed at a rotation rate of 1600 rpm to evaluate the long-term ability in Fig. 6h. The loss is calculated to be 94.05% for Fe@NC(10+900) and 90.49% for Fe@NC(10+1000) better than that of 20 wt% Pt/C (85.62%), verifying the robust stability. Above all, these porous carbon nanomaterials not only behave as a rigid supporting substrate to mitigate the structural change of catalysts, but also improve the electrochemical property with well dispersed Fe-based active centers.
In order to better understand the improvement of ORR performance in the Fe@NC(x+y) series, the theoretical calculations are used to clarify the adsorption behavior of the oxygen molecule on different types of Fe/N-doped graphitic carbon (FeNxCy). There are a total of six differentiated situations for Fe and N coordination environments, namely FeC4, FeNC3, FeN2C2-a, FeN2C2-b, FeN3C and FeN4. First of all, the standard model of the O2 unit cell is established, and its calculated energy is about −867.91 eV (Fig. S26†). In Fig. 7a, S27 and S28,† it shows that FeNxCy sites doped with N have lowered the adsorption energy of O2 loaded on the active sites as FeC4 (−1.94 eV), FeNC3 (−1.79 eV), FeN2C2-a (−2.33 eV), FeN2C2-b (−2.04 eV), FeN3C (−2.81 eV) and FeN4 (−2.82 eV). In this case, FeNC3 exhibits a higher adsorption energy than the pristine FeC4 caused by an inevitable breakdown of structural symmetry.42 With greater N-doping, the overall adsorption energy can be further reduced with more π electrons, which are experimentally supported by the high-efficiency ORR activity of FeNx sites in our work. In Fig. 7b, the partial density of states (PDOS) for FeNxCy at approximately 0, −6.42, and −13.87 eV are significantly broadened by valence hybridization compared to the isolated O2 molecule. Furthermore, FeNC3, FeN2C2-a/b, FeN3C and FeN4 are also endowed with obviously lower energies than that of FeC4, indicating the transfer of the partial charge from oxygen to the metal atom in the N-coordinated centers.43 In addition, we have further applied the optimal samples to assemble Zn–air batteries (ZABs). In Fig. 7c, the Fe@NC(10+1000) provides the largest open circuit voltage (OCV) of 1.53 V, higher than Pt/C (1.47 V) and Fe@NC(10+900) (1.45 V). Meanwhile, the maximum current density and peak power density of Fe@NC(10+900) in Fig. 7d are calculated to be 233.6 mA cm−2 and 107.1 mW cm−2, which is inferior to that of Fe@NC(10+1000) (291.0 mA cm−2, 133.0 mW cm−2), while the Pt/C based ZAB exhibits the best performance of 319.7 mA cm−2 and 146.8 mW cm−2. Furthermore, the specific capacity of the Pt/C-based battery is 734.0 mA h g−1, lower than that of Fe@NC(10+1000) at 777.6 mA h g−1 (Fig. 7e). These above results confirm the high catalytic ORR performance and high stability in alkaline media for MOF-derived porous carbons with abundant FeNx sites as well as Fe-based NPs. And the long-term charge and discharge curves of the ZABs with Fe@NC(10+1000) and Pt/C show a high round-trip efficiency of 52% and 54% within 100 cycles to confirm their robust stability for possible practical devices (Fig. S29†).
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2130792. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d2qi00394e |
This journal is © the Partner Organisations 2022 |