Selective hydrogenation of nitroarenes is a promising approach to produce aromatic amines and aminocyclohexanes [1, 2]. Aromatic amines are important bulk chemicals in many fields, including production of fine chemicals, commercial products, and polymers [3, 4]. Along with the products of hydrogenation of the nitro groups, some byproducts are also produced, such as nitroso and azo compounds. For substituted nitroarenes with reducible groups (e.g., –Cl, –C=O, and –CHO groups), selective hydrogenation of the nitro group is a great challenge. Noble-metal-based catalysts have been used for the chemical transformation of nitrobenzene to aniline, and excellent selectivities and activities have been achieved under mild conditions [5-13]. However, noble-metal-free catalysts are preferred in industry because of their low cost and abundance. In this case, transition-metal-[14-22] and carbon-based [23-28] catalysts show potential for selective hydrogenation of nitrobenzene, but they suffer from low activity and require harsh reaction conditions. To achieve high activity and selectivity under mild conditions, both the H₂ molecules and nitro groups must be effectively activated, which is difficult to achieve with a catalyst containing only a single type of active site.

For bifunctional catalysts composed of a carbon-based material loaded with a transition-metal-based catalyst, the support material can not only confine the metal atoms, but it can also modify the electronic environment of the metal nanoparticles (in a way similar to homogeneous catalysts with organic ligands), which increases the catalytic activity. Recently, earth-abundant metals (e.g., Co) have been used for hydrogenation of nitrobenzene [14, 20]. Reduced cobalt can act as a hydrogen acceptor owing to the electron-deficient state of its surface, which promotes cleavage of H–H bonds. However, it suffers from low chemoselectivity because of coadsorption of two or more nitroarenes. In this case, catalytic coupling of azo compounds can occur. Heteroatom-doped carbon with an adjustable electron structure and chemical stability can stabilize metal nanoparticles [29-33]. Doping N atoms into the carbon matrix can cause charge delocalization [34-36] and a high positive charge density on the carbon atoms next to the doped N atoms, which are beneficial for adsorption of compounds containing a nitro group. N-doped carbon has been used as a catalyst for hydrogenation of nitrobenzene because of the strong interaction between nitrobenzene and carbon catalysts [37], but its activity is still very low owing to its poor ability for H₂ activation. Metal catalysts can also be activated by carbon-nitride-based materials [38-40], where the electron density of the metal centers changes because in-lattice N atoms act as electron-accepting groups. Therefore, using N-doped carbon as a support for metal-based nanocatalysts is expected to enhance their catalytic activity.

Inspired by the above findings, we developed a synergetic catalyst composed of Co nanoparticles encapsulated in N-doped carbon (Co@NC) for room-temperature selective hydrogenation of nitroarenes. The N-doped carbon and cobalt center play roles in preferential adsorption of nitroarene and H₂ activation, so their synergetic effect leads to excellent activity and selectivity. The Co@NC catalysts were synthesized by calcining a mixture of a cobalt salt, an inexpensive organic molecule, and carbon nitride. The optimized Co@NC-1 catalyst shows excellent activity and selectivity at room temperature, which cannot be achieved using bare N-doped carbon or pristine-carbon-supported Co nanoparticle catalysts.

To fabricate carbon nitride (g-C₃N₄), melamine was placed in a covered crucible, heated to 550 ℃ for 4 h under an inert environment, and then maintained at 550 ℃ for another 4 h. The as-synthesized yellow solid was ground into a powder, which was used to produce N-doped carbon. The homogeneous cobalt-containing precursor solution was prepared by dissolving cobalt nitrate hexahydrate (1.168 g) and triethylene diamine (1.92 g) in 40 mL of deionized water. g-C₃N₄ (1.0, 2.0, or 4.0 g for Co@NC-0.5, Co@NC-1, or Co@NC-2, respectively) was added to the aqueous solution under magnetic stirring. Owing to the effect of the Co–N bond, Co²⁺ ions can embed in the holes of carbon nitride and triethylene diamine molecules prevent excessive growth of Co particles during the following pyrolysis treatment. After the solvent was removed by vacuum freeze-drying, the solid powder was transferred to a covered crucible, heated to 900 ℃ at a rate of 2.5 ℃/min under a N₂ atmosphere, and maintained at this temperature for 2 h. The as-obtained black solid samples were used for characterization and the catalytic reactions. To further investigate the role of N-doped carbon, Co@C was prepared by the conventional impregnation method using commercial carbon (VX720R) as the support. Pure N-doped carbon was also synthesized by pyrolysis of g-C₃N₄.

Transmission electron microscopy (TEM) was performed with a Tecnai G2 F20 transmission electron microscope at 200 kV. The crystal structures were determined by a RigaKu D/max-2500 X-ray diffractometer (XRD) equipped with a Cu K_α irradiation source. Elemental analysis was performed with a Vario EL Cube equipped with a METTLER x86 instrument. The elemental composition and bonding information were obtained by X-ray photoelectron spectroscopy (XPS) at a pass energy of 187.85 eV (Physical Electronics PHI 1600 ESCA XPS system using a monochromated Al K_α X-ray source) using the C 1s peak at 284.6 eV as the internal standard. The nitrogen adsorption–desorption isotherms were determined with a ASAP 2020 physisorption analyzer at –196 ℃, and the specific surface areas were calculated by the conventional Brunauer–Emmett– Teller (BET) method. The pore size distribution in the mesopore range was determined by the Barrett–Joyner–Halenda method. Before performing the measurements, all the samples were outgassed under vacuum at 200 ℃ for 12 h until the pressure was less than 0.66 Pa. The Fourier transform infrared spectroscopy (FTIR) spectra were recorded with a BioRad FTS 6000 spectrometer. Raman spectroscopy was performed with a Raman spectrometer (DXR Microscope) using a green semiconductor laser (532 nm) as the excitation source.

Hydrogenation was performed in a 50-mL autoclave with a polytetrafluoroethylene lining using 0.5 mmol of the nitroarene, 5 mL of the solvent, and 30 mg of the catalyst. The autoclave was purged with N₂ and H₂ (15 bar) three times and then pressurized to 10 bar. After the reaction, the catalyst was separated by centrifugation, followed by analysis of the sample by gas chromatography (GC)–mass spectroscopy (MS) (Agilent 5975 equipped with a HP-5 capillary column) and GC (Agilent 7820 equipped with a flame ionization detector and AT-SE-54 capillary column).

To enable large-scale production of Co@NC for possible use as a heterogeneous catalyst, we developed a freeze-drying method using cobalt nitrate as the metal salt and carbon nitride as the N precursor. The solid mixture was carbonized at 900 ℃ under an inert atmosphere (Fig. 1(a)). The X-ray diffraction (XRD) patterns indicate formation of metallic cobalt [41] (the peaks at about 44° and 51° are assigned to the (111) and (200) diffractions of Co nanoparticles) and graphitic carbon (the peak at about 26°) [42, 43] (Fig. 1(b)), and also exclude coexistence of detectable amounts of cobalt oxides, nitrides, and carbides. The carbon peak becomes more intensive with addition of graphitic carbon nitride. All of the TEM images (Fig. 1(c)–(f)) show that the Co nanoparticles in Co@NC have a similar average size of ~10 nm, indicating that the amount of carbon nitride does not influence the size of the Co nanoparticles. Furthermore, the high-resolution TEM (HRTEM) image of Co@NC shows a highly integrated nanostructure of layered carbon-coated Co nanoparticles (Fig. 1(d)). The lattice spacing of all of the catalysts is 0.205 nm (Fig. 1(d)–(f)), which is attributed to the (111) plane of metallic cobalt [41]. The carbon layers can act as a shell to protect the metallic Co nanoparticles from aggregation and oxidation during pyrolysis and the following reaction processes. The porosity of the synthesized Co@NC catalysts was evaluated by nitrogen adsorption–desorption (Fig. 2(a)). With increasing amount of carbon nitride, the BET surface area of the catalyst does not significant change (80.78, 81.46, and 86.31 m² g⁻¹ for Co@NC-0.5, Co@NC-1, and Co@NC-2, respectively). XPS and elemental analysis were performed to obtain further information about the chemical compositions of the catalysts (Fig. 2(b) and Table 1). The results indicate that the N concentration in the bulk is very close to that on the surface and the N concentration increases with increasing carbon nitride addition (1.75 and 3.62 at% for Co@NC-0.5 and Co@NC-1, respectively). Further increasing the amount of carbon nitride powder does not significantly increase the nitrogen concentration (3.73 at% for Co@NC-2). The Raman spectra (Fig. 2(c)) show two distinct signals at 1350 and 1590 cm⁻¹, which are assigned to defective/disordered sp³ hybridized carbon (D band) and crystallized graphitic sp² carbon (G band), respectively. The ratio of the intensity of the D band to the intensity of the G band I_D/I_G increases from 0.908 (Co@NC-0.5) to 1.0 (Co@NC-1) upon increasing the added amount of carbon nitride.

The Co@NC catalysts were applied to hydrogenation of 4-nitrophenol to test their catalytic activity. The hydrogenation reactions were performed at room temperature under a H₂ pressure of 10 bar. When N-doped carbon was used as the catalyst, no catalytic hydrogenation product was obtained even after reacting for 12 h, demonstrating the inert nature of N-doped carbon for hydrogenation of 4-nitrophenol under mild reaction conditions, which is because of its poor ability for hydrogen activation. The catalytic behavior of the Co@NC catalysts is shown in Fig. 3(a), where conversion of 4-nitrophenol is plotted as a function of the reaction time. Co@NC-0.5 can initiate hydrogenation of 4-nitrophenol, with total conversion of 4-nitrophenol in 5 h but with poor chemoselectivity for 4-aminophenol (78.1%). Co@NC-1 exhibits the best performance among the catalysts tested, giving complete conversion of 4-nitrophenol in 3 h and a turnover frequency (TOF) based on the amount of Co at 30 min of 12.3 h⁻¹. More importantly, the chemoselectivity for 4-aminophenol is > 99.9% and the amounts of the byproducts are below the detection limitation of GC–MS and GC (Fig. 3(b) and (c)). Co@NC-2 does not perform better than Co@NC-1 because a further increase in the amount of g-C₃N₄ does not further increase the N doping concentration. To understand the important role of N-doped carbon, commercial carbon-supported Co nanoparticles (Co@C) were synthesized and applied for hydrogenation of 4- nitrophenol under the same conditions. After 3 h, conversion of 4-nitrophenol was only 5.6% and it did not significantly increase even for a prolonged reaction period. Furthermore, large amounts of byproducts (i.e., azoxy compounds) were produced and the selectivity for 4-aminophenol was only 67.1%.

Different solvents (water, tetrahydrofuran (THF), and cyclohexane) were tested using Co@NC-1 as the catalyst (Table 2). The activity improves with increasing polarity of the solvent. The polarity of the solvent has a significant effect on many reactions [41], especially hydrogenation of nitrobenzene [14, 42]. One important factor is the interaction between the nitro group and the polar solvent by OH…N and OH…O hydrogen bonds, which accelerates polarization of N=O and N–O and weakens the bond strength [46, 47]. Another important factor is the easy of adsorption and dissociation of hydrogen on the catalyst surface in the polar solvent [48]. As shown in Table 2, higher reaction temperature and H₂ pressure accelerate the reaction.

The control experiments indicate that Co@NC-1 has very high catalytic activity and excellent chemoselectivity for hydrogenation of 4-nitrophenol. To determine how N-doped carbon and Co nanoparticles synergistically catalyze selective hydrogenation of 4-nitrophenol, further characterizations of Co@NC-1 were performed. The similar morphologies of the Co@NC catalysts mean that it is not the cobalt content, particle size, or nanostructure of the Co nanoparticles that affects the catalytic activity. The surface area can also be excluded because Co@NC-1, Co@NC-0.5, and Co@NC-2 have similar surface areas, as shown in Fig. 2(a). As shown in the N 1s spectra (Fig. 4(a) and (b)), the doped N atoms are mainly graphitic and pyridinic N [30] (the peaks centered at about 401.5 and 399.0 eV, respectively). The shifts of the typical Co 2p XPS peaks to higher energy and N 1s XPS peaks to lower energy indicate a decrease in the electron density of the Co nanoparticles in Co@NC-1 compared with Co@NC-0.5, which confirms electron transfer at the interface of Co and N-doped carbon (Fig. 4(c)).

Theoretical investigation suggests that introduction of dopants into the carbon framework can make the electronic structure negative [34-36]. The nitro group, which is a strong electron-withdrawing group, will have a strong interaction with the electron-rich sites. Consequently, N-doped carbon has been used for catalytic hydrogenation of nitrobenzene [37]. For Co@NC-1, the relatively high N doping concentration results in electron redistribution at the interface of cobalt and N-doped carbon, which enriches the positive charges on the metallic Co nanoparticle side and the negative charges on the N-doped carbon side, as indicated by the XPS results (Fig. 4). Although both carbon with a high concentration of N atoms and metallic Co can act as adsorption sites for nitro groups, adsorption to electron-rich N-doped carbon is preferred. In addition, the metallic phase provides active sites for H₂ activation and promotes cleavage of H–H bonds owing to its good electrophilicity. Therefore, for Co@NC-1, the electron-transfer effect accelerates H₂ activation over the Co nanoparticles and deprotonation of the H–metal intermediates, which generates catalytically active sites to promote the whole reaction process. Here, we presume that 4-nitrophenol adsorbs to N-doped carbon and the absorbed nitro groups are then reduced by the surface hydrogen species, which is described as the synergistic effect of Co nanoparticles and N-doped carbon.

To better understand the excellent selectivity of Co@NC-1, the adsorption mode of 4-nitrophenol on different catalyst surfaces was investigated by FTIR spectroscopy (Fig. 5). For N-doped carbon, there are three characteristic peaks at 1210, 1460 and 1540 cm⁻¹ after adsorption of 4-nitrophenol, which are assigned to the stretching vibration of C–OH, framework vibration of benzene, and asymmetrical stretching of –NO₂, respectively. In contrast, there are no characteristic peaks for the carbon support before and after adsorption. This confirms that 4-nitrophenol adsorbs on N-doped carbon but not on pure carbon, which is because the pure carbon support is nearly inert for adsorption of both nitro and hydroxyl groups. However, Co@NC-1 shows similar signals to N-doped carbon, indicating that 4-nitrophenol molecules preferentially adsorb on N-doped carbon.

The catalytic stability is also a very important criterion to evaluate the potential of a catalyst. Co-based catalysts can be easily recycled using an external magnetic field and reused up to ten times. As shown in Fig. 6(a), both the conversion of 4-nitrophenol and selectivity for 4-aminophenol are unchanged for ten cycles. The unchanged structural features further demonstrate the excellent catalytic stability of Co@NC-1 (Fig. 6(b)).

We also investigated hydrogenation of eight other substituted nitroarenes using the Co@NC-1 catalyst. The corresponding anilines were obtained in excellent yields with high selectivities (Table 3). Co@NC-1 selectively hydrogenates the nitro group while other functional groups, such as –CHO, –COOH, and –C≡C, are unaffected. For different substituted nitroarenes, different times are required to achieve full conversion. 4-Nitrostyrene takes 0.5 h more than 4-nitrophenol because the π–π conjugative effect of the –C=C group and benzene ring makes a relatively difficult for nitrostyrene to adsorb to the catalyst. For 3-nitrophenol, the nitro group is difficult to polarize because of the electron-donating effect of the –OH group at the meta position, which leads to a longer reaction time. In conclusion, this catalyst shows high activity and selectivity for hydrogenation of challenging substrates, suggesting that Co@NC-1 is a versatile catalyst for selective hydrogenation of substituted nitroarenes.

We have designed a heterogeneous synergetic catalyst for highly active and selective room-temperature hydrogenation of nitroarenes. The excellent catalytic performance can be attributed to the synergistic effect of N-doped carbon for preferential adsorption of the nitro group and Co nanoparticles for activation of H₂. Electron transfer from the Co nanoparticles to the N-doped carbon atoms at the interface results in electron-rich N-doped carbon and electron-deficient Co nanoparticles, which further enhances the activity of N-doped carbon for nitro group adsorption and Co for H₂ activation. The optimized catalyst (Co@NC-1) exhibits excellent catalytic performance with a TOF of 12.3 h⁻¹ and selectivity for aminophenol of > 99.9% at room temperature. This catalyst also exhibits high activity for hydrogenation of nitroarenes with various substituents.