Highly effective Fe–N–C electrocatalysts toward oxygen reduction reaction originated from 2,6-diaminopyridine

Fe–N–C electrocatalysts have been intensively studied due to their extraordinary catalytic activity toward oxygen reduction reaction (ORR). Here we prepare a Fe–N–C electrocatalyst through cost-effective and nontoxic precursors of 2,6-diaminopyridine (DAP) and FeCl3, where iron ions react with DAP to formed Fe-Nx species first, followed by polymerization and pyrolysis. X-ray diffraction patterns display no obvious Fe2O3 peaks observed in the catalyst as the nominal content of iron addition is less than 10 wt%. X-ray photoelectron spectroscopy spectra indicate that the catalyst has rich pyridinic nitrogen, graphitic nitrogen and Fe-Nx species, which are considered as active sites for ORR. Therefore the catalyst demonstrates an excellent catalytic activity with an onset potential of about 0.96 V, half-wave potential of about 0.84 V, and a limiting current density of 5.8 mA cm−2, better than commercial Pt/C catalyst in an alkaline medium. Furthermore its stability is also much more excellent than that of Pt/C. This work provides a strategy to synthesize universal M–N–C catalysts.


Introduction
With increasing concerns on pollution of environment and shortage of fossil energy [1][2][3], clean hydrogen energy has attracted great attention worldwide. H 2 -O 2 fuel cell is a device to convert chemical energy of hydrogen and oxygen directly into electrical energy by electrochemical reactions. Among all reactions, oxygen reduction reaction (ORR) occurring at the cathode determines the whole process due to its sluggish kinetics. Therefore, to resolve the cathode problem, ORR catalysts with excellent catalytic activity are imperative [4][5][6][7]. Nowadays, the most widely used and efficient ORR catalysts are Pt-based electrocatalysts in an acidic medium. Due to high price and small reserve of precious metal Pt, large-scale applications of fuel celldriven vehicles are limited. Therefore, scientists are devoted to developing various non-precious metal catalysts to replace Pt/C catalysts [8][9][10]. The nonprecious metal catalysts currently studied are including transition metal nitrogen-carbon catalysts, nitrogen-doped carbon catalysts, metal oxide catalysts, metal nitride catalysts, metal carbide catalysts, and so on. It is found that transition metal nitrogencarbon catalyst is the most promising alternative for the noble metal catalyst, especially the Fe-N-C catalyst. The Fe-N-C catalysts usually possess various types of N-C active sites and Fe-N x , thus the catalyst delivers outstanding performances [11][12][13][14]. Bao et al. [15] developed a novel method for one-pot direct synthesis of N-doped graphene via the reaction of tetrachloromethane with lithium nitride under mild conditions. The distinct electronic structure perturbation induced by the incorporation of nitrogen in the graphene network was observed for the first time by scanning tunneling microscopy. The nitrogen content varied in the range of 4.5-16.4% and enhanced catalytic activity was demonstrated in a fuel cell cathode oxygen reduction reaction. Raymond [16] discovered that cobalt phthalocyanine has the ability to catalyze the reduction of oxygen in alkaline media. This discovery opened up a new field for the research of fuel cell catalysts. Since then, non-precious metal macrocyclic molecular complexes have received widespread attention as oxygen reduction catalysts, and a large number of non-precious metal macrocyclic molecular complex catalysts have been developed [17][18][19][20][21]. Scientists have studied the performances of catalysts prepared with polypyrrole (PPy) [22][23][24], formamide [25], polythiophene (PTh) [26], poly-3methylthiophene (PMeT) [27] and dicyandiamide (DCD) [28] for oxygen reduction reaction. Zenelay's research group [29] attached PPy on a carbon support, and then deposited metal cobalt on the resulting carbon-supported PPy by the sodium borohydride reduction method, and obtained a transition metalcontaining carbon-supported conductive polymer oxygen reduction catalyst. They found that after introducing metal cobalt, the oxygen reduction catalytic activity and stability of the carbon-supported conductive polymer catalyst have been greatly improved. In the H 2 -O 2 single cell test, the power density can reach 240 mW cm -2 in the 30-h H 2 -O 2 cell stability test, and its performance is stable. Lv et al. [30] synthesized hydrogen-substituted graphdiyne (HsGDY) using tris[(trimethylsilyl)ethynyl] benzene, and then heated to 900°C in an ammonia atmosphere for one hour to prepare pyridinic-N selectively doped catalyst. The performance of the catalyst in an alkaline medium is better than that of commercial Pt/C catalyst. This study fully proved that pyridine nitrogen demonstrated ORR activity. Researchers such as Scherson [31][32][33] found that these transition metal macrocyclic complexes heattreated at high temperature can greatly improve the catalytic activity and stability of these catalysts for oxygen reduction. Therefore, high temperature heat treatment provides a practical method for preparing high activity and high stability carbon-based oxygen reduction catalyst. Dodelet et al. [34] prepared ORR catalysts by pyrolyzing Fe acetate or Fe porphyrin adsorbed on a synthetic carbon, which is made from the pyrolysis of perylene tetracarboxylic dianhydride between 400 and 1000°C in a H 2 /NH 3 /Ar atmosphere. They discovered that when Fe acetate was the Fe precursor, FeN 2 /C may represent up to 80% of the catalytic sites, while this fell to a maximum of about 50% when Fe porphyrin was the precursor, and FeN 2 /C was more electrocatalytically active than FeN 4 /C.
In this manuscript, a high-performance Fe-N-C catalyst will be synthesized by pyrolysis of a polymer containing Fe-Nx complex. Cost-effective and environmental-friendly 2,6-diaminopyridine (DAP) is used as monomer, FeCl 3 is added to preform Fe-N complex with DAP, later DAP is polymerized. Eventually the polymer will be treated at elevated temperature, and the resulting catalyst will be characterized and tested.

Experimental
2.1 Preparation of Fe-N-C catalysts 1 mmol of DAP and specified amount of FeCl 3 were dissolved in 30 mL of DI water, then 1 mmol of APS solution was added drop by drop, thus the polymer PDAP containing Fe 3? was synthesized, the polymerization reaction of Fe-DAP is shown in Fig. 1. Later the Fe-PDAP polymer will be filtrated and washed by DI water, till most of soluble small molecules, Cl -, and free Fe 3? will be removed, and then the filter cake is dried and pulverized. After that, the polymer was heated to 900°C at a rate of 5°C Eventually the Fe-N-C catalyst can be obtained. Fe contents of 2.5, 5, 7.5, 10 wt% (defined as mass ratio of Fe element to DAP) and pyrolysis times of 0.5, 1, 2 h were investigated, and the catalysts were denoted as Fe-N-C-900-mwt% and Fe-N-C-900-nh, respectively. For comparison, Ni-N-C and Co-N-C catalysts were also prepared with the same parameters of synthesizing the Fe-N-C catalyst by using NiCl 2 and CoCl 2 as precursors, respectively.

Physicochemical characterization
Thermogravimetric analysis (TGA) of Fe-PDAP mixtures was carried out from room temperature to 1000°C in a N 2 atmosphere at a heating rate of 10°C min -1 on SDTA851e instrument. Transmission electron microscopy (TEM) was performed using a Libra 200FE microscope (Carl Zeiss SMT Pte Ltd) under an operating voltage of 200 keV. Powdered X-ray diffraction pattern (XRD) was recorded on an X'Pert Pro MPD diffractometer at a scan rate of 5°m in -1 . X-ray photoelectron spectroscopy (XPS) analysis was carried out on Thermo Scientific ESCALAB 250Xi X-ray photoelectron spectrometer with Al-Ka source. The Brunauer-Emmett-Teller (BET) surface area was analyzed through N 2 adsorption-desorption isotherms measured at 77 K on a Micromeritics ASAP-2460. Pore size distributions and average pore volume were calculated by BJH method.

Electrochemical characterization
All the electrochemical measurements were performed on a rotating disk electrode in a typical threeelectrode cell using electrochemical workstation (AutoLab Potentiostat 302 N, Metrohm, Holland) at room temperature. The cell was composed of a glassy carbon as a working electrode (GC, 5 mm in diameter), HgO/Hg (0.1 M KOH) as a reference electrode and a platinum wire as a counter electrode. The catalyst ink was prepared by mixing 5 mg of catalysts, 48 lL of 5 wt% NafionÒ solution (Du Pont), and 1 mL of ethanol after sonication for 30 min. Then the catalyst ink was dropped onto the glassy carbon electrode and the loading was controlled at 1.2 mg cm -2 . For comparison, 40 wt% Pt/C (Johnson Matthey) was made same as above with a Pt loading of 25 lg cm -2 . Both cyclic voltammetry (CV) and linear sweep voltammetry (LSV) were performed in 0.1 M KOH solution. All potentials related to electrochemical tests in the context are referred to a reversible hydrogen electrode (RHE). The Koutecky-Levich equation can be used to express the current density generated by the reaction at a certain potential. In the RDE test, when the reaction reaches the limit diffusion control on the electrode surface, the concentration of the reactant on the electrode surface is zero and does not change with time. The expression of the rotating disk electrode is obtained as follows: where j is the measured current density(mA/cm 2 ), j K is the kinetic current density(mA/cm 2 ), j L is the limiting current density(mA/cm 2 ), x is the angular velocity of the electrode rotation(rpm), n is the number of electrons reduced by an oxygen molecule. F is Faraday's constant (96,487 C mol -1 ), CO is concentration of oxygen in the electrolyte (1.117 E -6mol mL -1 ), D is diffusion coefficient for oxygen in the electrolyte (1.9 E -5 cm 2 s -1 ), t is the kinetic viscosity of the electrolyte (0.01 cm 2 s -1 for 0.1 M KOH). The K-L diagram (J -1 vs. x -1/2 ) can be drawn from the current density measured at different speeds, and the electron transfer number n is calculated according to the above formula to further explore the mechanism of the oxygen reduction reaction. In order to investigate the ORR activities of the catalysts, the polarization plots are recorded in 0.1 M KOH oxygen-saturated solution at a rotation rate of 1600 rotation per minute (rpm) and the stability test was carried out at 0.6 V vs RHE for 3 h by chronoamperometry method.

Results and discussion
As schematically illustrated in Fig. 2, when DAP and FeCl 3 solution are dissolved in DI water, concurrently Fe and DAP may form a complex to obtain Fe-DAP containing Fe-N species, then Fe-DAP complex will be polymerized to form Fe-PDAP. Later Fe-PDA will be filtrated and washed by DI water, thus most of soluble small molecules, Cl -, and free Fe 3? will be removed. After the filter cake is dried and pulverized, it will be treated at elevated temperature in nitrogen flow. Eventually the Fe-N-C catalyst can be prepared. Because only trace of Fe 3? remains in the power, so the carbonization may be not very complete during the pyrolysis process. Figure 3 displays the mass loss of Fe-PDAP tested by thermogravimetric analysis (TG) and corresponding differential thermogravimetric analysis (DTG) in N 2 flow. We can see that the pyrolysis process of Fe-PDAP can be mainly divided into three stages. The first starts at 40°C and reaches a plateau at about 180°C, with an associated mass loss of 19%, which can be assigned to the volatilization of absorbed water [35,36]. The second step, from 180 to 420°C, shows a mass loss of 20%, which might be related to the cleavage of C-N bond of pyridine C-NH 2 group in PDAP as well as some polymers with low molecular weight. With temperature further increasing, the third stage shows that the PDAP is gradually carbonized and transformed. For example, all of hydrogen and the rest of polymers with low molecular weight will get removed, and the polymers with medium and high molecular weights will be transformed into active sites such as pyridinic-N, graphitic-N, and Fe-Nx. From 500 to 1000°C, we can see that the weight loss is occurring without a plateau, this may be ascribed to no additional Fe species existing for the carbonization procedure, because except for the Fe 3? forming the Fe-Nx complex with DAP, the rest of Fe 3? has been washed after the polymerization process. One Fe 3? could form Fe-Nx complex by combining with pyridine nitrogen in the DAP, because the nitrogen atom has lone-pair electron which can enter the outer orbital of Fe ion. From here we can deduce that the iron will be uniformly distributed in the following polymer and carbonized product due to the preformed Fe-Nx complex; furthermore the Fe element could also exist in a separate single atom state in the Fe-N-C catalyst.
The XRD patterns of Fe-N-C catalysts with various Fe contents, standard patterns of graphite and Fe 2 O 3 are shown in Fig. 4. The diffraction peaks at 26°and 43°are characteristic planes of (002) and (101) of graphitized carbon. The strong peak (002) indicates that the carbon material in the Fe-N-C catalyst has relatively high degree of graphitization [37]. With increasing Fe content, an obviously characteristic peak corresponding to the crystal plane (110) of Fe 2 O 3 [38,39] appears at 35.6°for the Fe-N-C-10wt% sample, this indicates that there are still considerably free Fe 3? encapsulated in the polymer after washing by DI water, the Fe 3? ions convert into Fe 2 O 3 particles in the following carbonization procedure. As shown in Fig. 5, it can be seen that the nitrogen adsorption-desorption curve is a typical type IV isotherm with an H3 hysteresis loop, so Fe-N-C-5wt% is a mesoporous carbon material with specific surface area as high as 1279 m 2 g -1 , measured by Brunauer-Emmett-teller (BET) method [40]. The curve of pore size distribution confirms the mesoporous state, that the catalyst mainly shows an average pore diameter of about 2.5 nm, with BJH Adsorption cumulative volume of 0.342 cm 3 g -1 . The large specific surface area and pore volume will provide more sufficient reaction sites for the gasliquid-solid three-phase reaction and facilitate the mass transportation process.
TEM images of Fe-N-C-5wt% are shown in Fig. 6. We can see that the catalyst is composed of small and uniform particles with graphitized carbon fringes, and no obvious Fe 2 O 3 and other metal particles are observed. Combined with above XRD pattern, we don't see obvious diffraction peaks of Fe and its compounds, indicating that Fe element may be presented in a form of single atom. The graphitized stripes of carbon at the edge of the catalyst are in consistent with the graphitized carbon peak in the XRD pattern.
The composition and binding energy of the Fe-N-C catalyst are analyzed by XPS, as shown in Fig. 7. The Fe-N-C catalyst is mainly composed of elements of C, O, and N, only trace of Fe is observed. The contents of carbon, oxygen, nitrogen, and iron are 85. 7, 9.4, 4.5, and 0.4 wt%, respectively. N can be mainly divided into four categories, as well-known pyridinic N (398.5 eV), pyrrolic N or Fe-N (399.2 eV), graphitic N (401.2 eV), and oxidized N (402.2 eV) [41,42]. From Fig. 7c, the N 1s spectrum in the Fe-N-C catalyst consists of graphite-N, pyridine-N, Fe-Nx, and oxidized N, with respective mass proportions of 39%, 26%, 10%, and 25%. According to literature [43][44][45], graphite-N, pyridine-N, and Fe-Nx are considered as the active sites for an oxygen reduction reaction, so we can predict that the catalyst could possess excellent ORR activity. XPS spectrum of Fe 2p in Fig. 7c demonstrates that Fe exists in the forms of Fe 2? and Fe 3? [46][47][48], correlating the N information, we can deduce that Fe might be in the form of Fe-Nx.
The ORR performances of different catalysts are evaluated on a rotating disk electrode (RDE) in 0.1 mol L -1 KOH O 2 -saturated solution at a rotation rate of 1600 rpm. Firstly different pyrolysis temperatures are investigated, for example 800°C, 850°C, 900°C and 950°C, the LSV curves are shown in Fig. 8a. We can see that pyrolysis temperature has great important influence on the ORR activity; the Fe-N-C-900-5wt% exhibits much better catalytic activity than Fe-N-C-800-5wt%. However, when the temperature increases to 1000°C, the polymer is decomposed completely, later we will try the pyrolysis temperature of 950°C. Later the effect of pyrolysis time on catalyst performance is tested and shown in Fig. 8b. As the pyrolysis time increases, the ORR performance increases and then decreases, one hour is the optimum time. This may be due to shorter time leading insufficient carbonization, and longer time decreasing the specific surface. Different mass ratios of nominal Fe to DAP (2.5:100, 5:100, 7.5:100, and 10:100) are further investigated; the catalysts are denoted as Fe-N-C-900-2.5 wt%, Fe-N-C-900-5 wt%, Fe-N-C-900-7.5 wt%, and Fe-N-C-900-10, respectively. Their ORR activities are shown in Fig. 8c, the ORR performance increases with iron content firstly, and the Fe-N-C-900-5 wt% catalyst demonstrates the most positive on-set potential, half-wave potential and biggest limiting current density; then decreasing with Fe content. This reason is that lower Fe content cannot form enough Fe-N x active sites, but more Fe content may influence the DAP polymerization and following carbonization process. We know that Fe, Co, and Ni are all carbonization catalysts and Fe-Nx, Co-Nx, and Ni-Nx species are active sites for ORR. Therefore, the ORR activity of Fe-N-C-900-5wt%, Co-N-C-900-5wt%, and Ni-N-C-900-5 wt% are investigated and compared with commercial Pt/C catalysts (40 wt%, from Johnson Matthey). As shown in Fig. 8d, the sequence in ORR activity is Fe-N-C-900-5 wt% % Pt/C (JM) [ Co-N-C-900-5 wt% [ Ni-N-C-900-5 wt%. We can see that the ORR activity of Fe-N-C catalyst is comparable to that of Pt/C catalyst, and is much better than those of Co-and Ni-based catalysts.
In order to study the reaction mechanism, we carried out linear sweep tests at different rotation rates. As shown in Fig. 8e, the limiting current density increases with rotation rates, while the onset potential remains unchanged. Figure 8f is the Koutecky-Levich (K-L) plots of Fe-N-C, which shows very good linearity at different potentials. According to the K-L plots, the numbers of electron transfer per oxygen molecule during the oxygen reduction reaction at different potentials (0.6-0.3 V) is calculated to be 3.6-3.99. This indicates that the Fe-N-C catalyst mainly undergoes a four-electron oxygen reduction process in alkaline media. Cyclic voltammetry and Chronoamperometry methods are used to evaluate the stability of Fe-N-C-900-5 wt% and commercial Pt/C catalysts, respectively. From Fig. 9a, b, after scanning at sweep rate of 50 mV s -1 for 1000 cycles, the intrinsic catalytic activity of Fe-N-C-900-5 wt% exhibits almost no change, only the limiting current density reduces slightly, indicating mass transportation becomes less easy; However, for Pt/C catalyst, the catalytic activity and limiting current density decrease, and the halfwave potential drops nearly 30 mV. If Chronoamperometry is running at 0.6 V (vs RHE) for 3 h, the current retention of Fe-N-C-900-5 wt% is 94%, while only 52% for Pt/C, as shown in Fig. 9c. We can draw a conclusion that the stability of Fe-N-C-900-5 wt% catalyst is much better than Pt/C in an alkaline medium. Furthermore, whether the ORR activity or the stability, the Fe-N-C-900-5 wt% catalyst shows better performance than Pt/C catalyst. It can be seen from Table 1 that under alkaline conditions, the catalytic performance of Fe-N-C-900-5 wt% catalyst is comparable to the similar Fe-N-C catalysts.

Conclusions
Fe-N-C catalyst with single-atom iron distributed has been successfully synthesized by pyrolysis of PDAP containing preformed Fe-Nx complex. XRD pattern and TEM images confirm that iron atoms are The Fe-N-C catalyst demonstrates best ORR activity and stability when control the pyrolysis temperature at 900°C and last for 1 h with nominal Fe content of 5 wt%. The Fe-N-C-900-5wt% catalyst exhibits comparable ORR activity and much better stability with respect to commercial Pt/C catalyst in an alkaline medium. This can provide a strategy to synthesize single-atom distributed catalysts and apply cheap and nontoxic C and N precursor to prepare highefficient ORR catalyst for alkaline fuel cells.