Introduction

Direct methanol fuel cells (DMFCs) and proton-exchange membrane fuel cells (PEMFCs) as clean and high efficient power sources have been attracting great attention for transportable and stationary applications due to their high power density, low operating temperature, and low environmental pollution. Nonetheless, the large-scale commercialization of fuel cells is limited by using expensive Pt-based catalysts as cathodic oxygen reduction reaction (ORR) [1–3]. Therefore, extensive studies [4–6] have been made on exploring new catalysts with low cost, high efficiency, and stability. In the last few decades, a variety of nonnoble metal ORR catalysts such as transition metal macrocyclic compounds [7], chalcogenides [8], conducting polymers [9], nitrogen-containing catalysts [10], and so forth have been investigated. Meanwhile, heat-treating transition metal-based nitrogen-containing complexes such as bipyridine, dibenzotetraazaannulenes, phthalocyanines, porphyrins, phenanthroline, polypyrrole, and tripyridyl triazine, etc. [11–19] has been the most promising ORR catalysts. However, the catalytic ORR performance of the abovementioned nonnoble electrocatalysts is still inferior compared to those of the Pt-based catalysts until now. Thus, the development of cost-effective method to obtain an active and durable nonnoble electrocatalysts is urgent in order to accomplish the practical and industrial implementations of DMFCs/PEMFCs.

In our previous work [20], a facile method was reported to synthesize a variety of carbon-incorporated FeN x (FeN x C y ) species on carbon blacks. It is noteworthy that we use a cheap and simple nitrogen-rich compound (pentaethylenehexamine) as a starting material to replace expensive heterocyclic nitrogen compounds (for instance, porphyrins and phthalocyanines) upon synthesizing carbon-incorporated FeN x catalysts which may have potentially and effectively active sites for uses in cost-effective fuel cells. In the present study, various FeN x electrocatalysts were synthesized by carbonizing a common nitrogen-containing compound (ethylenediamine) and two different iron precursors (iron(III) chloride and iron(II) ammonium sulfate) on carbon blacks. Consequently, a series of Fe(Z)N/C-2 and Fe(Z)N/C-8 (where Z = III and II represent different iron precursors) was obtained and microstructural properties as well as ORR performance have been investigated in order to explore the influence of iron precursors upon the synthetic process. In terms of catalysts’ properties, X-ray-based spectroscopies including X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and X-ray absorption spectroscopy (XAS) were conducted. By using XAS studies, the chemical structures of active sites and alloyed extents of target metals (PtM [21–25] and PtRu bifunctional [26, 27] electrocatalysts) have been examined thoroughly during the electrochemical processes. For studying electrocatalytic ORR performance, the linear scanning voltammetry (LSV), rotating disk electrode (RDE), and chronoamperometric techniques were employed to compare the results.

Experimental

Catalysts preparation

In a typical run, weighted amounts of FeCl3 (Acros) or (NH4)2Fe(SO4)2·6H2O (Acros) were dissolved in ethanol solution. To prepare the FeN x complex solution, nitrogen-containing precursors (ethylenediamine (EDA, Acros)) were added as chelated compounds into the above solutions, followed by the introduction of carbon blacks (Vulcan XC-72, Cabot Corp.) under stirring for 48 h. Then, the solvents were removed by using a vacuum oven at 363 K. Finally, the obtained powders were pyrolyzed with a heating rate of 5 K min−1 to the temperature of 1073 K in a nitrogen atmosphere and maintained at the same carbonized temperature for 2 h. The resultant samples were denoted as Fe(Z)N/C-2 and Fe(Z)N/C-8 (where Z = III and II represent the use of Fe(III)Cl3 and (NH4)2Fe(II)(SO4)2·6H2O, respectively, as iron precursors).

Characterizations

To obtain the bulk amounts of nitrogen in the Fe(Z)N/C-2 and Fe(Z)N/C-8 samples, elemental analyses (EA) were conducted with a CHN elemental analyzer (Heraeus varioIII). The bulk contents of iron in the samples were obtained by using an atomic absorption spectroscopy (AAS, ICE-3300). The crystallographic structures of samples were attained by using an XRD (X’Pert PRO diffractometer) under a CuKα radiation source (λ = 0.1541 nm). The surface chemical environments of samples were determined by using XPS (Kratos Axis Ultra DLD) under a monochromatic Al-Kα radiation (1486.6 eV). Nitrogen adsorption isotherms were measured at 77 K on a Micromeritics ASAP 2020 analyzer. The Fe K-edge (7112 eV) XAS was obtained in a transmission mode under room temperature at the Wiggler 17C beamline facility of the National Synchrotron Radiation Research Center (NSRRC) in Taiwan. The energy resolution (ΔE/E), estimated from the rocking curves of the double-crystal Si (111) monochromator, was ca. 2 × 10−4. The intensity of the incident beam intensity (I o) was measured by the ion chambers filled with a mixture of nitrogen and helium gases. Additionally, argon and nitrogen gases were mixed in an ion chamber to determine the transmitted beam intensity (I t). In terms of XAS data analysis, a standard procedure including energy calibration, pre-edge and post-edge background subtraction, normalization with respect to edge height, and Fourier transformation was performed. In order to isolate the X-ray absorption fine structure (EXAFS) contributions from different coordination shells, the k 3-weighted χ(k) spectra in the k-space range of 3.5–12.0 Å−1 were Fourier-transformed (FT) to the r-space. In the present study, all the XAS data analysis was performed by using the ATHENA and ARTEMIS software [28].

Electrochemical testing

A conventional three-electrode cell was constructed to measure the electrocatalytic performance of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts. An Ag/AgCl (saturated KCl) reference electrode and a Pt wire counter electrode were used for the half-cell system. Catalyst inks were prepared by dispersing ca. 5 mg of sample uniformly in 2.5 mL deionized water via the ultrasonic treatment. For the preparation of the working electrode, ca. 20 μL of the above ink was deposited onto the polished glassy carbon (GC, area = 0.1962 cm2) electrode, followed by drying in air at 333 K for 1 h. At last, ca. 20 μL of 5 % Nafion® (DuPont) solution was applied on the catalyst-coated GC electrode to form a thin film. To clean and activate the surface of working electrode, the cyclic voltammogram (CV) was conducted by scanning the potentials between −0.2 and 1.0 V for 10 cycles in a 0.5-M H2SO4 solution under high-purity N2 (99.9 %). The ORR catalytic activities of the electrocatalysts were evaluated via a LSV technique by scanning from 0.8 to −0.1 V with the scan rate of 5 mV s−1 in oxygen-saturated 0.5 M H2SO4 electrolyte. All current-voltage (I–V) curves were measured by a potentiostat (CHI Instruments, 727D) with an RDE system (Pine Instrument, AFMSRCE) that was used to handle the rotating speed during the electrochemical tests. In addition, the rotating ring disk electrode (RRDE) was employed to measure the H2O2 yield by using the same electrochemical analyzer.

Results and discussion

To study the influence of iron precursors on the properties of synthesized samples, a variety of X-ray-based spectroscopic analyses were performed. Figure 1 shows the XRD patterns of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples. The symbols of the square, triangle, and star in Fig. 1 represent the characteristic diffraction peaks of metallic α-Fe (JCPDS, No. 87-0722), Fe3C (JCPDS, No. 89-2867), and Fe2N (JCPDS, No. 89-3939), respectively. The diffraction peaks at ca. 26° observed for the samples are attributed to the (002) plane of graphitic carbon. Obviously, no noticeable diffraction peaks of Fe species were found for Fe(III)N/C-2 and Fe(II)N/C-2 samples. This is possibly because of their low contents (ca. 2 wt%) or amorphous states of Fe species. As the amount of Fe loading increased to 8 wt%, one diffraction peak at ca. 43.7° was observed for Fe(II)N/C-8 samples, suggesting the possible formation of the crystalline planes of metallic Fe. By contrast, less apparent feature of metallic Fe was found in the Fe(III)N/C-8. It can be seen that a mixture of Fe3C, Fe2N, and metallic Fe phases may coexist in the Fe(III)N/C-8 and Fe(II)N/C-8 samples by judging from XRD patterns. Nevertheless, it is noteworthy that chemical structures of Fe are hardly identified because the peak positions of metallic Fe, Fe3C, and Fe2N are rather overlaid in the XRD patterns. In addition, it is likely that some Fe species are unable to be observed by means of XRD due to their presence of amorphous phases. Therefore, to further investigate the information including the chemical environment of Fe, the coordination geometry, and the bonding of its local environment around the Fe atoms, additional X-ray based spectroscopies (XPS and XAS) were carried out.

Fig. 1
figure 1

XRD patterns of various Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts

Full size image

XPS analysis was employed to reveal the surface chemical compositions of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples. From the full-range XPS survey (not shown), all the samples mostly contain four species including carbon (C 1 s, 279–293 eV), nitrogen (N 1 s, 395–407 eV), oxygen (O 1 s, 525–537 eV), and iron (Fe 2p, 700–740 eV). As a result, their corresponding surface atomic contents (at.%) are summarized and compared in Table 1. As we can see, Fe(III)N/C-2 and Fe(III)N/C-8 samples synthesized using FeCl3 as iron precursors possess higher surface atomic nitrogen contents (1.0 and 1.5 at.%, respectively) than the bulk nitrogen contents (0.9 and 0.8 %, respectively). Nonetheless, the Fe(II)N/C-2 and Fe(II)N/C-8 samples have lower surface atomic (0.7 and 0.6 at.%, respectively) and bulk nitrogen contents (0.7 and 0.5 %, respectively) when compared to Fe(III)N/C-2 and Fe(III)N/C-8 samples. These results imply the preferential formation of nitrogen atoms on the sample surface upon using FeCl3 precursors in the synthesis process. As reported earlier [15], the better ORR activity can be achieved as the higher amounts of nitrogen were doped onto the catalysts surface. To further realize the structural states of these nitrogen species on the surfaces of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples, high-resolution XPS N 1 s spectra are attained and shown in Fig. 2. Three Lorentzian-Gaussian multipeaks at ca. 398.5 (pyridinic-N), 400.9 (graphitic-N), and 401.5 (pyridinic-N-H) were obtained by deconvoluting the broad peaks of all samples. The first peak at ca. 398.5 eV (pyridinic-N) [29] represents nitrogen atoms doped at the edges of the graphitic carbon layer. The second feature at ca. 400.9 eV (graphitic-N/quaternary-N) [29] is defined as nitrogen atoms doped inside the graphitic carbon plane. In the previous report [30], pyridinic-N and graphitic-N are the dominant species influencing the ORR activities, whereas the third peak at ca. 401.5 eV may be ascribed to pyridinic-N-H which is reported to be ORR inactive species [31]. To quantitatively obtain the relative fractions of the nitrogen species on the surfaces of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples, further spectra analyses by Lorentzian-Gaussian deconvolution were conducted and listed in Table 2. It can be observed that Fe(III)N/C-2 samples exhibit two major nitrogen-doping types of pyridinic-N (71 %) and pyridinic-N-H (29 %). Interestingly, pyridinic-N-H may be converted to graphitic-N in the Fe(III)N/C-8 samples while the content of Fe is increased to 8 wt%. In addition, while the Fe precursors were changed to (NH4)2Fe(II)(SO4)2·6H2O during the synthesis process, pyridinic-N and graphitic-N are the main nitrogen-doping types in the Fe(II)N/C-2 and Fe(II)N/C-8 samples. In the latest studies [32–34], graphitic-N is considered to be the superior sites of nitrogen-doped carbons during ORR. On the contrary, pyridinic-N is responsible for significant enhancement of the ORR activity [35] according to previous reports. Additionally, the pyridinic-N is not able to assist direct four-electron pathways for ORR without the help of graphitic-N [35, 36]. It seems that the role of actual electrocatalytic active sites is still contentious. However, the coexistence of pyridinic-N and graphitic-N species in this work is recognized as the most active sites for catalyzing ORR. In particular for Fe(III)N/C-8 samples, the higher concentrations of pyridinic-N (68 %) and the assistance of graphitic-N (32 %) as well as an increase in the surface nitrogen content (1.5 at.%) should be responsible for the observed enhancement of ORR activity (discussed below).

Table 1 Physical properties of various Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts
Full size table
Fig. 2
figure 2

XPS spectra of various Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts at the N 1 s region

Full size image
Table 2 Surface atomic concentrations of different types of N species in various Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts obtained from XPS results
Full size table

Table 1 also shows the surface iron contents of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples. It can be seen that the contents of Fe are only 0.8–1.4 at.%, indicating their rather low loadings on the surface of samples. Among them, Fe(III)N/C-8 samples possess the highest surface iron content (1.4 at.%) and atomic Fe/C ratio (0.017), suggesting that the Fe atoms present more on the surface of Fe(III)N/C-8 than those of other samples. In addition, the higher specific surface areas (see Table 1) can be found in the Fe(III)N/C-2 and Fe(III)N/C-8 catalysts than those in the Fe(II)N/C-2 and Fe(II)N/C-8, suggesting that iron precursors significantly affect the surface areas of micropores during the synthesis process. As displayed in Fig. 3, moreover, the high-resolution Fe 2p spectra show that the binding energies of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples in the Fe 2p3/2 region are all located at ca. 711.4 eV, indicating that the presence of oxidative surface Fe(III) state is based on the peak positions of the Fe (III) 2p3/2 binding energy (710.8–711.8 eV) and Fe(II) 2p3/2 binding energy (707.1–708.7 eV) [37]. Interestingly, no evident peak was observed at ca. 706.7 eV, implying the nonexistence of the iron nitride species [38] on the surface of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples.

Fig. 3
figure 3

XPS spectra of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts at the Fe 2p region

Full size image

To more realize the nature of iron active sites in the Fe(Z)N/C-2 and Fe(Z)N/C-8 samples during ORR, further XAS analyses were performed. Figure 4a shows the normalized K-edge EXAFS spectra of Fe(Z)N/C-2, Fe(Z)N/C-8 samples, and model compounds (Fe foil, Fe2O3, FeCl3, and (NH4)3Fe(SO4)2). Obviously, it can be seen that the shapes of the Fe(Z)N/C-2 and Fe(Z)N/C-8 samples in the XANES regions are diverse as compared to those of the model compounds. This result shows that the iron precursors (FeCl3 and (NH4)3Fe(SO4)2) have been converted to different forms of iron upon chelating and heat-treated carbonization processes. To recognize the corresponding positions of the edge caused by the excitation of a core electron to quasi-bound states [39] which are related to the valence of the iron atom, the high-resolution Fe K-edge XANES is shown in Fig. 4b. The curves of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples in the XANES regions are located between those of Fe foil and Fe2O3, indicating the existence of the oxidative valence of iron. In addition, the white line intensities (the magnitude of absorption hump at ca. 7128 eV) of samples are between those from Fe foil and Fe2O3, suggesting that iron species are present as oxidative states which are consistent with the above results. Additionally, the edges of Fe(III)N/C-2 and Fe(III)N/C-8 catalysts shift toward higher values at the absorption energy edges, suggesting their existence of large valences. Fourier transformations of the k 3-weighted EXAFS functions observed for Fe(Z)N/C-2 and Fe(Z)N/C-8 samples and model compounds are shown in Fig. 5. It can be seen that the radial distance distributions of Fe(Z)N/C-2 and Fe(Z)N/C-8 samples are different from those of model compounds, implying that these samples may be formed in the diverse local structures. The first peaks observed for Fe(Z)N/C-2 and Fe(Z)N/C-8 samples in the FT spectra are due to Fe-N or Fe-O bonds in the first shells. The second features are ascribed to the Fe-Fe bonds in the second coordination shells. The other peaks may be the Fe-Fe bonds in the higher coordination shells. Obviously, Fe(II)N/C-2 and Fe(II)N/C-8 samples possess mainly Fe-O bonds in the first shells, while large amounts of Fe-N bonding structure are found in the Fe(III)N/C-2 and Fe(III)N/C-8 samples. In addition, Fe(II)N/C-2 and Fe(III)N/C-2 samples have apparent Fe-Fe bonds in the second (or higher) coordination shells, implying that the aggregation of iron metals may occur. As a result, the Fe-N structure can be significantly formed in the samples synthesized from FeCl3 precursors. Interestingly, however, the increased intensities in the second shell coordination of Fe-Fe moieties observed for Fe(II)N/C-2 and Fe(III)N/C-2 samples indicate that the sintering reactions between metallic iron species may happen during high temperature carbonization process.

Fig. 4
figure 4

a Fe K-edge XANES spectra and b their corresponding magnification spectra of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts and Fe standard compounds

Full size image
Fig. 5
figure 5

Radial distribution functions of Fourier-transformed k 3-weighted χ(k) Fe K-edge EXAFS spectrum for Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts and Fe standard compounds

Full size image

As displayed in Fig. 6, CVs of the glassy carbon electrode surfaces coated by Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts measured in a N2-purging 0.5 M H2SO4 solution show weak features of redox waves in the potential range of −0.2 to 1.0 V. By contrast, the cathodic currents with the potentials ranging from 0.2 to 0.7 V can be evidently observed for Fe(III)N/C-2 and Fe(III)N/C-8 in a O2-purging 0.5 M H2SO4 solution, indicating the pronounced occurrence of ORR on these catalysts. Among these electrocatalysts, Fe(III)N/C-8 catalysts possess the largest capacitance over the measured range of potentials, suggesting that it has the largest catalytic surface area. The higher capacitance values can result in the higher electrochemical surface and faster electron transfer which could lead to the surpassing of ORR activity.

Fig. 6
figure 6

CV curves of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts on a glassy carbon electrode surface measured under N2-saturated (solid lines) and O2-saturated (dot lines) 0.5 M H2SO4 solution at a scan rate of 10 mV s−1

Full size image

The ORR performance of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts was assessed by LSV in oxygen-saturated 0.5 M H2SO4 solution at room temperature. Figure 7 displays the polarization curves of these catalysts and their corresponding onset and half-wave potentials are also listed in Table 3. The Fe(III)N/C-2 electrocatalysts showed ORR activities at the onset and half-wave potentials of ca. 0.57 and 0.33 V vs. Ag/AgCl, respectively. Upon increasing the iron contents to 8 wt%, the onset and half-wave potentials were further increased to 0.61 and 0.40 V vs. Ag/AgCl. While the (NH4)2Fe(SO4)2·6H2O was used as iron precursors in the synthesis, the ORR activities of Fe(II)N/C-2 and Fe(II)N/C-8 were found to be remarkably decreased. It is clear that the iron precursors have an essential impact on the ORR activities of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts. The ORR onset potentials of the Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts follows the order: Fe(III)N/C-8 > Fe(III)N/C-2 > Fe(II)N/C-8 ≅ Fe(II)N/C-2. In order to further explore the ORR mechanism on the catalysts prepared from different iron precursors, the RDE technique was carried out in O2-saturated 0.5 M H2SO4 solution at room temperature under different rotating rates. Consequently, the number of transferred electrons (n) during ORR can be attained from the slope by using the Koutecky-Levich first-order reaction equation [40]:

$$ {j}^{-1}={j_k}^{-1}+{j_{\mathrm{d}}}^{-1} $$

where j is the measured current density, j k is the kinetic current density for the ORR, and j d is the diffusion-limiting current density that can be expressed by the following equations:

$$ \begin{array}{l}{j}_{\mathrm{d}}=B{\omega}^{1/2}\hfill \\ {}B=0.62 nF{C}_{\mathrm{O}2}{D_{\mathrm{O}2}}^{2/3}{\nu}^{-1/6}\hfill \end{array} $$

where B represents the Levich slope, ω denotes the rotating angular frequency, n is the number of the exchange electrons during the ORR process, F is Faraday constant (96,487 C mol−1), C O2 and D O2, respectively, represent the concentration and diffusion coefficient of oxygen dissolved in the electrolyte, and v is the kinematic viscosity of the electrolyte (0.01 cm2 s−1). The RDE voltammograms of various Fe(Z)N/C-2 and Fe(Z)N/C-8 electrocatalysts in O2-saturated 0.5 M H2SO4 solution at room temperature under different rotation rates are displayed in Fig. 8 and their corresponding Koutecky-Levich plots at different potentials are shown in Fig. 9. As shown in Table 3, the n values of the catalysts prepared from (NH4)2Fe(SO4)2·6H2O precursors are obviously lower than 4, indicating the catalyzed ORR with both two-electron-transferred reduction and four-electron-transferred reduction, while the former is dominant. On the other hand, the Fe(III)N/C-8 catalysts proceed more four-electron-transferred reaction as compared to the Fe(II)N/C-2 and Fe(II)N/C-8 catalysts. Thus, the iron precursors have a critical influence on ORR mechanism which is in accordance with the order of catalytic activities discussed above.

Fig. 7
figure 7

Polarization curves during oxygen reduction for various Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts in O2-saturated 0.5 M H2SO4 solution at room temperature

Full size image
Table 3 Kinetic parameters of various Fe(Z)N/C-2 and Fe(Z)N/C-8 electrocatalysts during ORR
Full size table
Fig. 8
figure 8

RDE voltammograms for the Fe(Z)N/C-2 and Fe(Z)N/C-8 in O2-saturated 0.5 M H2SO4 with various rotation rates

Full size image
Fig. 9
figure 9

Koutecky-Levich plots for the Fe(Z)N/C-2 and Fe(Z)N/C-8 electrocatalysts at different electrode potentials

Full size image

The RRDE voltammetry curves which can be used to evaluate the generated amount of hydrogen peroxide during the ORR process of Fe(Z)N/C-2 and Fe(Z)N/C-8 electrocatalysts are also shown in Fig. 10a. The yields of hydrogen peroxide were determined by using the equation [41]:

$$ \%{\mathrm{H}}_2{\mathrm{O}}_2=\frac{2{I}_{\mathrm{r}}}{N\times {I}_{\mathrm{d}}+{I}_{\mathrm{r}}}\times 100\% $$

where I d (see in Fig. 7), I r (see in Fig. 10a), and N (0.24 in this work) are the disk current, ring current, and collection efficiency, respectively. As a result, the fractional yields of hydrogen peroxide for the synthesized electrocatalysts were obtained and illustrated in Fig. 10b. It can be seen that Fe(III)N/C-2 and Fe(III)N/C-8 electrocatalysts have superior ORR activity and generated a less amount of hydrogen peroxide during the reaction as compared to Fe(II)N/C-2 and Fe(II)N/C-8.

Fig. 10
figure 10

a RRDE data of hydrogen peroxide yield and b fractional yield of hydrogen peroxide for Fe(Z)N/C-2 and Fe(Z)N/C-8 electrocatalysts

Full size image

More importantly, tolerance to methanol crossover and electrocatalytic durability at the cathode have been two of the vital requirements for practical applications of fuel cells [42–44]. To further investigate the methanol-tolerant behaviors of the synthesized electrocatalysts, LSV tests were also carried out to realize the effect of various concentrations of methanol (0.5–2.0 M) on electrocatalytic ORR activities of Fe(III)N/C-8 catalysts. As shown in Fig. 11, it can be seen that the overpotential of Fe(III)N/C-8 electrocatalysts is intact upon the presence of 0.5 M methanol. A slight decrease with a current density loss of ca. 10 % measured at 0.5 V vs. Ag/AgCl was observed even in the existence of 2.0 M methanol while compared to ORR in pure H2SO4 solution. The significant resistance to methanol during ORR on the Fe(III)N/C-8 catalysts is due to the weak competition reaction between oxygen reduction and methanol oxidation. In addition, the stability of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts also can be evaluated by investigating the current loss after prolonged operation (ca. 13.9 h) at a constant voltage of 0.4 V vs. Ag/AgCl in a 0.5-M H2SO4 electrolyte saturated with O2 at a rotating rate of 1600 rpm. As shown in Fig. 12, the order of current-time (i–t) chronoamperometric response of Fe(III)N/C-2 > Fe(III)N/C-8 > Fe(II)N/C-2 > Fe(II)N/C-8 was observed after 13.9 h operation. These results of durability testing indicate that ORR performance of synthesized catalysts using FeCl3 as iron precursors is surpassing than those using (NH4)2Fe(SO4)2·6H2O as iron precursors. Based on the combined results from X-ray-based spectroscopies (XRD, XPS, and XAS), the better ORR performance (both activity and durability) of Fe(III)N/C-8 catalysts is possibly attributed to the more atomic ratios of surface N/C and Fe/C with the formation of FeN x species that iron may be coordinated with pyridinic-N and graphitic-N active sites.

Fig. 11
figure 11

Effect of methanol concentrations on ORR activities of Fe(III)N/C-8 at 0.5 V vs. Ag/AgCl in O2-saturated 0.5 M H2SO4 solution

Full size image
Fig. 12
figure 12

Chronoamperometric analysis of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts at 0.4 V vs. Ag/AgCl in O2-saturated 0.5 M H2SO4 solution at room temperature

Full size image

Conclusions

In this study, various nonnoble FeN x electrocatalysts (Fe(Z)N/C-2 and Fe(Z)N/C-8, Z = III and II) were fabricated by heat-treating two iron precursors (Fe(III)Cl3 and (NH4)2Fe(II)(SO4)2·6H2O) with a simple nitrogen-containing precursor (ethylenediamine) on carbons in a nitrogen atmosphere. The influence of iron precursors on the catalysts properties and ORR performance has been comparatively examined with a series of electrochemical analysis techniques including cyclic voltammogram, rotating disk electrode, and chronoamperometric methods. In addition, the physiochemical properties of Fe(Z)N/C-2 and Fe(Z)N/C-8 catalysts were analyzed and discussed by using various X-ray-based techniques. As a result, we found that iron precursors play a critical role on the microstructures of the synthesized catalysts and their corresponding ORR activities. Among all the catalysts, those prepared with FeCl3 precursors have the surpassing ORR performance. By employing XPS analysis, Fe(III)N/C-8 catalysts were found to have the notable increases in the atomic surface Fe/C and N/C ratios with coexistence of pyridinic-N (68 %) and graphitic-N (32 %) sites. Additionally, the FeN x clusters, which may be the active centers for the superior ORR activity and durability, can be observed for the Fe(III)N/C-8 catalysts in the analysis of XAS. Consequently, the developed nonnoble nanocatalysts may be possible electrocatalysts for practical implementations in DMFCs and PEMFCs.