Transition metal catalysts such as Co, Fe, Ni, and Mn have attracted intensive research because of their abundance and economical utilization [1, 2, 3, 4]. These transition metals have been utilized in electrocatalysis and synthesized to replace Pt, which is limited by its rareness, high price, and low methanol resistance [5]. Due to their superior properties, nitrogen-doped carbon (N-C) supported non-precious catalysts have been extensively studied [6, 7, 8]. M-N-Cs can be made from a variety of different metals, nitrogen and carbon precursors such as ammonia, acetonitrile and chelating agents and macrocycles (porphyrins or phthalocyanines) by high heat treatment [9].

Wang et al. [10] synthesized ultrafine dispersed iron oxide-based nanoparticles (NPs) embedded in nitrogen-doped carbon by a one-pot method with D-glucosamine hydrochloride as the carbon and nitrogen source, melamine as the soft template, and Fe(NO₃)₃·6H₂O as the metal source. The hybrid catalyst exhibited high activity and excellent durability for the hydrogen evolution reaction (HER) in alkaline solution. They pointed out that the high HER activity of the hybrid catalyst was due to the strong interaction between Fe-based NPs and N-doped graphitic carbon, especially the synergistic effect of Fe and Fe₂O₃. Herrmann et al. [11] modified Co-based TMPP by adding various metal oxalates where the oxalates behaved as a structure forming agent to provide a nano-scaled template for the carbonization of the CoTMPP. After removing the oxalates by an etching step, catalysts with enhanced specific surface area were obtained.

Metalloporphyrins with a unique carbon-rich macrocycle and metal-nitrogen coordination, especially cobalt porphyrin, iron porphyrin, and manganese porphyrin, which are porphyrins with transition metal elements, are attractive precursors for the one-step synthesis of M-N-C catalysts [12]. The catalytic performance of M-N-Cs in the selective oxidation of arylalkanes has been explored [13, 14].

Here, we synthesized transition metal catalysts embedded in nitrogen-doped carbon M-N-C (M = Co, Fe, Mn) by a template-free method by heating different meso-tetraphenyl porphyrins (CoTPP, FeTPPCl, MnTPPCl) precursors. These catalysts were characterized by N₂ adsorption-desorption, thermogravimetry (TG), transmission electron microscopy (TEM), high-resolution TEM (HRTEM), and Raman and X-ray photoelectron spectroscopy (XPS). The selective oxidation of ethylbenzene with molecular oxygen under solvent-free conditions was carried out to explore the catalytic performance of the M-N-Cs.

CoCl₂·6H₂O, FeCl₂·4H₂O, MnCl₂·4H₂O, benzaldehyde, p- formylbenzoic acid, propanoic acid, dichloromethane, anhydrous N,N′-dimethyl formamide (DMF), ethanol, ethylbenzene, bromobenzene, and 1, 4-dichlorobenzene were obtained commercially and used without further purification. Pyrrole was distilled before utilization.

Typically [15], distilled pyrrole (4.69 g, 70.0 mmol) was added dropwise into a three-neck flask containing a mixture of propanoic acid (250 mL), benzaldehyde (5.56 g, 52.5 mmol), and 4-carboxy benzaldehyde (2.62 g, 17.5 mmol), and then refluxed for 1 h. The product was cooled overnight, then filtered and purified. This was denoted as TPP.

The TPP sample (1.0 g, 1.6 mmol) was dissolved in 100 mL DMF. After loading 10.5 mmol of CoCl₂·6H₂O in batches, the mixture was heated to reflux under stirring until the meso-triphenyl porphyrin was exhausted. After cooling overnight, the mixture was filtered and washed repeatedly with deionized water. The product Co(Ⅱ) meso-tetraphenyl porphyrin was denoted as CoTPP.

For the synthesis of Fe(Ⅲ) meso-tetraphenyl porphyrin (FeTPPCl), FeCl₂·4H₂O was introduced to the synthesis. After the reflux, when the mixture was cooled to 100℃, 100 mL HCl solution (pH = 1) was added. Then the mixture was stirred for another 30 min before cooling to ambient temperature. The post-treatment was the same as that for CoTPP.

The synthesis of Mn(Ⅲ) meso-tetraphenyl porphyrin MnTPPCl was similar to CoTPP, but the added metal salt was MnCl₂·4H₂O instead of CoCl₂·6H₂O.

A portion of metal porphyrin was ground and placed in the porcelain boat and heated directly in N₂ flow with a heating rate of 10℃/min and kept at 800℃ for 1 h. The catalyst was ground well and denoted as M-N-C (M = Co, Fe, Mn).

The specific surface area was measured by N₂ adsorption-desorption at -196℃ on a NOVA 1000e apparatus from Quantachrome Instruments. The sample was degassed at 300℃ for 3 h prior to the experiment. The Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) methods were used to determine the specific surface areas and pore sizes of the samples, respectively. TEM and HRTEM studies were obtained on a JEM-3010 high-resolution transmission electron microscope operating at 200 kV. TG was analyzed on a NETZSCH apparatus. Raman measurements were performed at ambient conditions using a 632 nm laser on a Labram-010 apparatus. XPS measurements were evaluated on a RBD upgraded PHI-5000C ESCA system (PerkinElmer) with Mg K_α radiation (hν = 1253.6 eV) or Al K_α radiation (hν = 1486.6 eV). The binding energies were calibrated using the C 1s peak at 284.6 eV as reference.

The selective oxidation of ethylbenzene with molecular oxygen as oxidant was conducted in a 50 mL autoclave. Ethylbenzene (10 mL) and catalyst (30 mg) were loaded in the reactor and then sealed and the pressure raised to 0.8 MPa with O₂. The temperature was increased to 120℃ and kept for 5 h.

The products were analyzed by a gas chromatograph (Shimadzu GC-2014 equipped with a capillary column (RTX-5)) using a flame ionization detector with the internal standard method using bromobenzene and 1, 4-dichlorobenzene as reference. The recovered catalyst was obtained by centrifugation, then washed with ethanol and dried at 80℃ in air.

N₂ adsorption-desorption isotherms and the pore size distribution curves of the M-N-Cs are shown in Fig. 1. It showed that the amount of adsorbed N₂ increased at a high p/p₀ (> 0.9) without an adsorption limit for Co-N-C and Fe-N-C. According to the IUPAC classification, the N₂ adsorption isotherms were Type Ⅳ and indicated the presence of mesopores. There was a relatively large adsorption-desorption hysteresis loop of the H3 type at p/p₀ > 0.4, which was due to the capillary condensation of N₂ in the slit-shaped nanopores [16]. However, the isotherm of Mn-N-C was not smooth. From Table 1, it can be seen that the specific surface areas of the M-N-Cs exhibited the order Co-N-C > Fe-N-C > Mn-N-C, as did the total pore volume, D_p, and dV(d). In our earlier report, we found that the specific surface area of Co-N-C increased with elevated temperature, which was caused by the different calcination degree [17]. Thus, it can be concluded that the M-N-Cs (M = Co, Fe, Mn) had different calcination degree, which may be attributed to the different catalytic performance of the transition metal in the formation of the N-C.

Fig. 1. N2 adsorption-desorption isotherms (a) and pore size distribution (b) of Co-N-C (1), Fe-N-C (2), and Mn-N-C (3).

Table 1 Physical properties of the M-N-Cs (M = Co, Fe, Mn). Catalyst ABET a (m2/g) Vtot b (cm3/g) Dp c (nm) dV(d) (cm3 nm-1 g-1) Co-N-C 193.8 0.21 3.73 0.28 Fe-N-C 182.5 0.19 3.73 0.13 Mn-N-C 147.0 0.08 3.68 0.02 a Surface areas calculated using the BET method. b Total pore volumes estimated from the N2 adsorption isotherm at p/p0 = 0.98. c Mesopore diameters calculated from the N2 desorption branch using the BJH method.

Table 1 Physical properties of the M-N-Cs (M = Co, Fe, Mn).

TG analysis was conducted to further investigate the structure evolution of the transition metal porphytins with increasing temperature (Fig. 2). Taking CoTPP as an example, there was a loss of water of crystallization at 95℃ with a mass loss of 6%. The total mass loss was 81% from the TG curve. The curve showed decomposition temperatures at 410, 513, and 746℃, which indicated that CoTPP decomposed by steps and the different stages are associated with the sequential thermal degradation of the porphyrin macrocycles, that is, detachment of phenyl rings from pyrole rings, rupture and detachment of cobalt from pyrole rings, cleavage of phenyl rings, and decomposition of the remaining organic fragment [18, 19]. The TG curve of FeTPPCl was similar to that of CoTPP but had less total mass loss (72%). As for MnTPPCl, the stage of the loss of crystallization water was almost similar to that of CoTPP and FeTPPCl. However, after that, the decomposition temperatures were 331, 497, and 688℃, which were lower. The total mass loss was only 34% until the end stage. This indicated that the thermal stability of MnTPPCl was better than that of CoTPP and FeTPPCl.

Fig. 2. TG curves of CoTPP (1) , FeTPPCl (2), and MnTPPCl (3).

The morphology and structure of the Co-N-C, Fe-N-C, and Mn-N-C composites were examined by TEM and HRTEM. The results are presented in Fig. 3. As shown in Fig. 3(a), the TEM image of Co-N-C showed that Co nanoparticles were embedded in graphite carbon and amorphous carbon (Fig. 3(d)). It can be clearly seen that the graphitic spiral structures were twisted and spiral, which showed ringed graphitic layers had formed. However, the fringes of graphitic carbon were rarely seen in Fe-N-C (Fig. 3(b) and (e)) and Mn-N-C (Fig. 3(c) and (f)). In Mn-N-C, particularly, the Mn nanoparticles were embedded in a flat sheet-like structure composed of amorphous carbon (Fig. 3(c)). It was clearly observed that the M-N-C composites (Co-N-C, Fe-N-C and Mn-N-C) possessed different graphitization degree. Besides, the metal particle size in Co-N-C is about 43 nm. In Fe-N-C, the metal compound particles are about 114 nm, while that it is ~20 nm for Mn-N-C. This observation was in agreement with the N₂ adsorption-desorption and TG results.

Fig. 3. TEM and HRTEM images of M-N-C. (a, d) Co-N-C; (b, e) Fe-N-C; (c, f) Mn-N-C.

Raman spectroscopy was used to study the graphitization process and chemical bonding state of the pyrolyzed metal porphyrins (Fig. 4). All these M-N-C materials have an obvious D band at 1328 cm^-1, which is associated with defects, and a G band at 1593 cm^-1, which is due to graphitic carbon (Fig. 4) [20]. The intensity ratio of D and G bands, Ⅰ(D)/Ⅰ(G), is used to evaluate the disorder in the materials. The higher Ⅰ(D)/Ⅰ(G) implies more defects in carbon materials [21]. The Ⅰ(D)/Ⅰ(G) ratio of the M-N-Cs (M = Co, Fe, Mn) followed the order Co-N-C (1.48) > Fe-N-C (1.19) > Mn-N-C (1.18). This indicated that Co-N-C possessed the highest graphitization degree, which could simultaneously produce more defects than the other samples. The trends of the Raman spectra were consistent with the TG results and TEM images.

Fig. 4. Raman spectra of M-N-C. (1) Co-N-C; (2) Fe-N-C; (3) Mn-N-C.

To get the more detailed elemental composition and surface chemical states of the M-N-Cs, XPS analysis was performed. The results are shown in Fig. 5. The full scan of the M-N-Cs presented in Fig. 5(a) revealed the existence of C, N, and O and Co, Fe, and Mn. The ratio analysis of the M-N-Cs is presented in Table 2. It can be seen that the atom percent of C decreased in the order Co-N-C > Fe-N-C > Mn-N-C, while the content of O followed the opposite trend. The metal amounts of Co-N-C and Fe-N-C were similar and larger than that of Mn in Mn-N-C. Besides, the amount of N was almost the same in Co-N-C and Fe-N-C and a little lower than that in Mn-N-C.

Fig. 5. XPS spectra of M-N-C (M = Co, Fe, Mn) composites. (a) Survey; (b) C 1s; (c) N 1s; (d) Co 2p; (e) Fe 2p; (f) Mn 2p.

Table 2 Ratio analysis of the peaks in the XPS spectra of the M-N-Cs. Catalyst Atomic composition (at%) N content (%) C O M N Pyridinic N Pyrrolic N Graphitic N Co-N-C 90.7 5.8 1.0 2.5 40.3 30.7 29.0 Fe-N-C 87.8 8.9 1.1 2.2 37.5 36.7 25.8 Mn-N-C 85.1 11.1 0.6 3.2 33.2 41.0 25.8

Table 2 Ratio analysis of the peaks in the XPS spectra of the M-N-Cs.

The C 1s spectrum was fitted using the following carbon bonding environments: 284.6 eV (C=C), 285.7 eV (C=N), and 286.6 eV (C-N) (Fig. 5(b)) [22], which suggested the successful doping of N atoms in the M-N-Cs. The high resolution N 1s peaks in M-N-C are shown in Fig. 5(c). The N 1s XPS spectrum was deconvoluted into three peaks, namely pyridinic N (398.6 eV), pyrrolic N (399.8 eV), and graphitic N (401.3 eV) [23]. This indicated that the transition metal type has no obvious effect on the kind of N species present. However, as shown in Table 2, the ratios of the three types of N in the M-N-Cs were different. The ratios of pyridinic N and graphitic N in Co-N-C were higher than in the others while the pyrrolic N content was the lowest. Mn-N-C contains the most pyrrolic N and lowest amount of pyridinic N and graphitic N in the M-N-C. This showed that the transition metal influences the amount of the different N species [24].

In the cobalt region (Fig. 5(d)), the band at 778.5 eV was attributed to Co⁰ [5]. Peaks characteristic of oxidic Co with the typical binding energy of 780.2 eV for the Co 2p_3/2 and 796.2 eV of the Co 2p_1/2 electrons were also found. The shake-up satellite peaks at 784.9 and 802.0 eV suggested that the oxidic Co was mainly composed of Co₃O₄ [25]. The low energy bands can be deconvoluted into two peaks with the binding energies of 781.2 and 779.6 eV, which represented Co²⁺ and Co³⁺, respectively [26]. The deconvolution of the Fe spectrum is shown in Fig. 5(e). The peaks with the binding energy of 723.5 and 710.0 eV were attributed to Fe 2p_1/2 and Fe 2p_3/2 [27, 28]. The low energy bands can be deconvoluted into two peaks with the binding energies of 712.8 and 710.6 eV, which represented Fe²⁺ and Fe³⁺ [29, 30]. In the Mn 2p spectrum (Fig. 5(f)), there were two main peaks, 652.9 eV of Mn 2p_1/2 and 640.8 eV of Mn 2p_3/2, and a satellite peak at 646.9 eV [2]. The peak of Mn 2p_3/2 can be deconvoluted into three peaks, i.e., 640.3 eV for Mn²⁺, 641.3 eV for Mn³⁺, and 642.7 eV for Mn⁴⁺ [2].

All these observations showed that the M-N-C composites with different transition metals possessed different graphitization degree, and Co-N-C had the most defects and graphitization degree, which are beneficial to the catalytic activity.

The selective oxidation of ethylbenzene was used as the probe reaction to measure the catalytic performance of the M-N-Cs (Table 3). Co-N-C showed a higher conversion of ethylbenzene (14.1%) than Fe-N-C (12.4%) and Mn-N-C (10.7%). This may be because of the higher content of Co and the Ⅰ(D)/Ⅰ(G) ratio that indicated more defects caused by graphitization. It has been reported that pyridinic N and graphitic N are the active sites [31]. Thus it is reasonable to suggest that it was the higher content of pyridinic N and graphitic N that was beneficial and improved the catalytic performance of Co-N-C. In addition, three consecutive oxidation experiments were performed to demonstrate the stability and reusability of the catalyst. Only slight changes of the catalytic performance in ethylbenzene oxidation was found, which indicated that the M-N-Cs have good recyclability in the heterogeneous catalysis system.

Table 3 Catalytic performance of M-N-C for ethylbenzene selective oxidation. Catalyst Recycle times Conversion of ethylbenzene (%) Selectivity (%) Acetophenone Phenethyl alcohol Benzaldehyde Co-N-C 1 14.1 73.2 22.1 4.7 2 12.2 72.2 21.8 6.0 3 11.5 71.7 22.4 5.9 Fe-N-C 1 12.4 72.2 23.1 4.7 2 12.0 67.5 26.5 6.0 3 11.7 68.4 25.5 6.1 Mn-N-C 1 10.7 73.7 18.3 8.0 2 10.8 74.0 17.5 8.5 3 10.2 74.2 17.5 8.3 Reaction conditions: catalyst 30 mg, ethylbenzene 10 mL, O2 0.8 MPa, 120℃, 5 h.

Table 3 Catalytic performance of M-N-C for ethylbenzene selective oxidation.

Transition metal M-N-C (M = Co, Fe, Mn) catalysts have been synthesized by a template-free method by heating different meso-tetraphenyl porphyrins (CoTPP, FeTPPCl, and MnTPPCl) precursors. The M-N-Cs exhibited different catalytic performance for ethylbenzene oxidation. This was ascribed to the difference in M (Co, Fe, Mn) and their catalytic behavior during M-N-C formation. A higher cobalt content, higher Ⅰ(D)/Ⅰ(G) ratio (meaning more defects caused by graphitization), and more pyridinic N and graphitic N species were beneficial for improving the catalytic performance of Co-N-C. All the M-N-C composites had remarkable recyclability in the selective oxidation of ethylbenzene.