The Stöber process with a high concentration of tetraethoxyorthosilicate (TEOS) is applied to prepare mesoporous siliceous materials. Conventional rod type SBA-15 is obtained by dissolving nonionic surfactant, i.e., Pluronic P123 triblock copolymer in 2M HCl solution at 35°C, followed by adding a silica source. In this study, platelet SBA-15 was prepared by adding a small amount of Zr(IV) salt (ZrOCl₂) in the rod type SBA-15 starting mixture. Since platelet SBA-15 has relatively shorter mesochannels, it has an advantage in terms of molecular diffusion. The molar composition of the starting gel is 0.017 P123/1 TEOS/0.05 ZrOCl₂·8H₂O /7.94 HCl/221 H₂O [17]. The mixture was left under stirring with 300 rates per minute (rpm) at 35°C for 24 h, and subsequently hydrothermally heated at 100°C for another 24 h under static conditions. The precipitated product was isolated by filtration and dried at 80°C for 24 h. After that, the product was calcined in air at 550°C for 3 h [28–31].

As for the KIT-6, it is obtained by adding BuOH as a co-solute. While silica species polymerize along with non-ionic surfactant, the presence of butanol under low HCl concentrations enables the phase controlling and micelle-folding. The molar composition of the starting mixture is 0.017 P123/1.2 TEOS/1.31 BuOH/1.83 HCl/195 H₂O [32–36]. Except for adding the n-butanol before adding TEOS, other processes were the same as those of SBA-15.

Lastly, SSP is obtained by using n-hexadecylamine as a non-ionic surfactant. The starting molecular composition for synthesis of SSP is 1 TEOS/0.162 n-hexadecylamine/50 IPA/191 H₂O [24]. Deionized water, IPA and n-hexadecylamine were blended for 3 h, and then a certain amount of TEOS was added dropwise at 30°C. The mixture was left under stirring (300 rpm) for 24 h. The precipitated product was isolated by filtration and dried at 80°C for 24 h. After the drying product was corrected by filtration and washed with ethanol (EtOH), the product was calcined in air at 550°C for 3 h.

The preparation of Fe-N/C catalysts was conducted by using the incipient wetness impregnation method [37–40]. Iron (III) chloride hexahydrate and 1,10-phenanthroline were dissolved in sequence in a mixture of EtOH and H₂SO₄ solution. Afterward, an equal volume of silica hard template was rubbing vigorously. After evaporation of the solvent, the catalyst precursor powder was dried in an oven at 80°C for 4 h and subsequently dried at 160°C for 12 h. Dried catalyst powder was pyrolyzed at 900°C in Ar for 3 h. In the case of silica host, washing with HF is required to remove the sacrificial template [37–40]. HF filtration should be handled with extreme care. Pyrolyzed powder was filtered and washed with HF, and dried for 24h. The three catalysts are designated as follows: S-Fe-N/C from platelet SBA-15, K-Fe-N/C from KIT-6 and SP-Fe-N/C from SSP, respectively.

The electrochemical properties were measured with a reference 600 potentiostat in a three-electrode cell equipped with a Pt-wire rod as a counter electrode and a saturated Hg/HgO as a reference electrode. The catalyst ink of Fe-N/C was prepared by dispersing 15 mg of fabricated catalyst in 1.2 mL of Nafion^® solution (0.1 mL water, 1.07 mL EtOH and 30 μL of 10 wt.% Nafion^® resin). The working electrode was prepared by pipetting 4.75 μL of prepared ink onto the embedded glassy carbon disk of the rotating disk electrode (RDE). Generally, the ORR performance of the catalyst is estimated with an RDE in a half cell setup. The RDE technique allows determining the half-wave potential and the limiting current density, which provides an assessment of the electrochemical catalyst activity [41]. Since the surface area of the glassy carbon disk is 0.196 cm², 302.93 μg_catalyst/cm² catalyst was loaded. The ORR polarization curves were recorded with 5 mV/s and 1600 rpm rotation speed in oxygen saturated 0.1 M KOH electrolyte solution. For comparison, 4 mg of 20 wt.% Pt/C (TKK) was dispersed in 2.4 mL of Nafion^® solution (1.99 mL water, 0.5 mL IPA and 10 μL of 10 wt.% Nafion^® resin). The prepared ink (29.1 μL) was coated on the glassy carbon surface with 250 μg/cm² of catalyst loading amount. In the durability test, coating conditions Fe-N-C was the same as ORR test, and Pt/C was coated with 100 μg/cm² and 300 μg/cm² respectively for glassy carbon by controlling the drop in the same ink composition. Cycling was performed at 200 mV/s from 0.6 V to 1.0 V under the condition of 0.1 M KOH purged with oxygen.

The small-angle XRD patterns confirm a structure of template and catalyst. It is widely recognized that the small-angle XRD patterns for SBA-15 structure feature three distinct peaks which are indexable as (100), (110), (200) planes, respectively [42]. The small-angle XRD patterns for both the resultant SBA-15 and S-Fe-N/C shows 3 distinct diffraction peaks as shown in Fig. 1A. Also, small-angle XRD pattern for Ia3d cubic symmetry of KIT-6 features has three distinct peaks which are corresponding to (211), (220), (332) planes [33]. However, the resultant K-Fe-N/C from KIT-6 only shows only 1 distinct diffraction peak and a blunt peak 2θ = 1.2–1.8° (Fig. 1B). Since the presence of mesoporous structures is checked out through nitrogen adsorption-desorption isotherms (Fig. 3B), the absence of 2 distinct peaks may result from amorphous carbon shells generated on the surface [42]. Lastly, SSP is characterized as disorderly pore structures, they give rise to a broad peak in the small-angle XRD as shown in Fig. 1C. In comparison with the silica template, overall shifts of all peaks of Fe-N/C catalyst to higher values of the diffraction angles are observed. This is because that Fe-N/C has the inverse crystalline phase of the mesoporous silica template.

SEM images (Fig. 2) reveal that both the resultant SBA-15 and S-Fe-N/C consist of many platelet type samples of which average width and thickness are about 1 μm and 300 nm, respectively. S-Fe-N/C maintained their template shape, which indicates that hard template synthesis works well. As for the resultant KIT-6 and K-Fe-N/C, they consist of the irregular shape samples of which sizes are much larger (> 10 μm) than that (< 1 μm) of platelet SBA-15. Then, SEM images reveal that the SSP and SP-Fe-N/C consist of many spherical types of domains of which the average diameter is about 1μm.

According to Fig. 3A, the nitrogen adsorption isotherms of the SBA-15 show the classical type IV isotherms with H1 hysteresis loops appeared at which P/P₀ is around 0.6–0.8. These types of isotherms are given by mesoporous adsorbents of which pore sizes are between 2 and 50 nm [43]. The average pore size calculated by using the Barrett-Joyner-Halenda (BJH) analysis is 9.36 nm. Hysteresis loops of type H1 are observed in materials that exhibit a narrow range of uniform cylindrical mesopores [28,31]. Also, the specific surface area and pore volume of this material is 719.55 m²/g and 1.14 cm³/g, respectively. As for the S-Fe-N/C, hysteresis loops appeared at which P/P₀ is around 0.5–0.9, which indicates that Fe-N/C consists of mesoporous adsorbents with relatively broad pore size distribution centered at 3.12 nm. The catalyst surface area and pore volume of this material are 1084.7 m²/g and 0.85 cm³/g.

According to Fig. 3B, the nitrogen adsorption-desorption isotherms of the KIT-6 and K-Fe-N/C show the classical type IV isotherms with H1 and H4 hysteresis loops appeared at which P/P₀ is around 0.4–0.9 (Fig. 3B), which implies the existence of mesopore structure in the samples [28,31]. The average pore size calculated by using the BJH method is 9.36 nm. Hysteresis loops of type H1 are given by materials, which exhibit a narrow range of uniform cylindrical mesopores and hysteresis loops of type H4 are associated with slit-like mesopores. Also, the specific surface area and pore volume of KIT-6 is 802.51 m²/g and 0.85 cm³/g, respectively. As for the resultant K-Fe-N/C, hysteresis loops appeared at which P/P₀ is around 0.4–1.0, which means that K-Fe-N/C consists of mesoporous adsorbents with relatively broad pore size distribution centered at 3.54 nm. The specific surface area and pore volume of K-Fe-N/C are 1369.7 m²/g and 1.21 cm³/g, respectively.

The nitrogen adsorption-desorption isotherms of the resultant SSP show classical type IV isotherms with H2 hysteresis loops appeared at which P/P₀ is around 0.4–0.8 as shown in Fig. 3C. The average pore size calculated by using the BJH method is 4.27nm. Hysteresis loops of type H2 are given by samples that feature the ink-bottle shape of pores. Also, the specific surface area and pore volume of SSP is 609.94 m²/g and 0.71 cm³/g, respectively. As for the SP-Fe-N/C, hysteresis loops appeared at which P/P₀ is around 0.4–1.0, which indicates that SP-Fe-N/C consists of mesopores with relatively broad pore size distribution centered at 4.02 nm. In addition, the specific surface area and pore volume of SP-Fe-N/C is 1389.4 m²/g and 1.40 cm³/g, respectively. Nitrogen isotherms data of the three catalysts are summarized in the Table 1. There are some differences between the physical properties of silica templates and those of Fe-N/C. This is because their phases are inversed by the nano-replication method and the constituent materials are different, like silica and carbon [44].

One of the significant ORR activity indicators is a half-wave potential. A half-wave potential is a potential at which current is equal to one half of the diffusion-controlled current. Fig. 4 shows ORR polarization curves of three types of synthesized Fe-N/C catalysts and platinum on carbon (Pt/C, 20wt.% Pt) the catalyst which is a state-of-the-art TKK’s commercial catalyst for ORR. All three Fe-N/C catalysts have superior ORR performances than the commercial Pt catalyst in the alkaline condition. However, it seems that there is no morphological preference towards ORR performance in contrast to the previous literature, which suggested the different ORR activity in acidic conditions [26,27]. The half-wave potential of S-Fe-N/C, K-Fe-N/C, and SP-Fe-N/C catalysts is 0.867 V, 0.875 V, and 0.875 V, respectively. Since these values are a little different from each other, it is difficult to judge which catalyst has a better activity towards ORR. Moreover, the recent study suggests that Fe-N/C catalysts prepared by using the KIT-6 template tend to have the worst ORR performance in the acidic condition because of its enormous particle size [26]. However, in this study, K-Fe-N/C features almost the same half-wave potential to the SP-Fe-N/C, but slightly higher potential than the S-Fe-N/C, which implies that K-Fe-N/C catalyst overcame its morphological disadvantages. In the half-cell, the synthesized S-Fe-N/C and the commercial Pt/C were tested for their durability (Fig. S1). The durability of Pt/C was tested with two different loading amounts. One was 100 μg/cm² and the other was 300 μg/cm². When measuring 100 μg/cm² and 300 μg/cm² of Pt/C, respectively, superior durability at 300 μg/cm² Pt/C was confirmed. As a result, S-Fe-N/C shifted about 30 mV, whereas only 18 mV shift occurred at 300 μg/cm² Pt/C in terms of half-wave potential after the test. This result was less than that of a 26 mV shift of 100 μg/cm² Pt/C, indicating the influence of the loading amount on the catalyst’s stability. The durability and performance difference of catalysts in the half-cell test depending on the loading amount was remarkable in the alkaline medium as precedent studies [45,46]. The decrease in the performance of Pt/C was assumed to be due to the change in surface chemistry, and in the case of Fe-N/C, it was predicted that carbon corrosion occurred due to the relatively low pyrolysis temperature compared to graphitic carbon [47].

To elucidate this result, XPS analysis was implemented in order to analyze the surface atomic states of N and Fe in the catalysts. Actually, the active sites of the Fe-N/C toward the ORR activity are still in controversy [48–52]. So it is expected to get information about active sites if it is possible to investigate what components K-Fe-N/C mostly contains. According to Fig. 5A, a rough outline of the XPS spectrum for K-Fe-N/C is a bit different from those of the others. The distribution of the N component on the right side of Fig. 5A indicates that the three kinds of catalysts have mainly graphitic N (400.8eV) and pyridinic N (398.4eV). Also, both N components are major controversial sites for the ORR activity, respectively. However, there is no obvious clue for the direct relationship of ORR activity from the different portions of two N species. Instead of N species, we concentrate on “Fe-N” bonding and “Fe species” because the amount of Fe component appears to be related to the main active sites in the Fe containing catalysts for ORR [9,51,53,54]. In Fig. 5B, K-Fe-N/C has the most abundant Fe elements, which contains 7 to 10 at.% more than other catalysts of this study. Also, the state of Fe elements with Fe²⁺ and Fe³⁺ components may affect the performance of the catalyst toward ORR under alkaline conditions. It is expected that the Fe-N₄ site would be considered as the main active site in the alkaline non-precious catalyst. This is because the position of the Fe-N₄ sites corresponds to K-Fe-N/C Fe species positions, which is the middle of Fe²⁺ and Fe³⁺ positions [10]. Thus, it is possible to conclude that large numbers of Fe species and Fe-N₄ sites, especially Fe²⁺ and Fe³⁺ 2p_1/2, in K-Fe-N/C as shown in Fig. 5B help to overcome morphological disadvantage in terms of the ORR activity in the half-cell configuration. Thus, it would be suggested that the activity towards ORR of the Fe-N/C catalysts mainly depends on the portion of the Fe species in the catalyst, which can mitigate the size-effect of the Fe containing catalysts in the half-cell configuration.