Synthesis and characterization of Fe2P NC catalysts
The Fe2P NCs were synthesized from triphenylphosphite, hexadecylamine, and Fe(CO)5 in black powder form. Subsequently, the prepared Fe2P NCs were immobilized on TiO2, SiO2, and carbon (C) supports, denoted Fe2P NC/TiO2, Fe2P NC/SiO2, and Fe2P NC/C, respectively. For comparison, Fe NP/TiO2 was also prepared as a model of a conventional Fe NP catalyst37, and was used in the hydrogenation reaction without being exposed to air.
X-ray diffraction (XRD) of the prepared Fe2P NCs revealed three characteristic peaks at 2θ = 40.1, 52.1, and 54.2°, which were assigned to the (2\(\stackrel{-}{11}\)1), (0002), and (30\(\stackrel{-}{3}\)0) crystalline planes of hexagonal Fe2P, respectively (Fig. 1a). Transmission electron microscopy (TEM) of the Fe2P NCs revealed regular nanorods with a mean size of 26.5 × 8.7 nm (Figs. 1b, 1e, and Supplementary Fig. 1). Side and top view high-resolution TEM (HRTEM) of the Fe2P NCs (Figs. 1c and 1f, respectively) revealed lattice spacings of 0.17 and 0.51 nm, corresponding to the (0002) and (10\(\stackrel{-}{1}\)0) hexagonal Fe2P planes, respectively. The selected area electron diffraction (SAED) patterns from the side and top views (Figs. 1d and 1g) also showed the spot patterns indexed to the [10\(\stackrel{-}{1}\)0] and [0001] of the Fe2P hexagonal crystal, proving the formation of single crystalline Fe2P. Moreover, high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDX) of the Fe2P NCs recorded from side and top views (Figs. 1h–1k and Figs. 1l–1o, respectively) depicted uniform distributions of Fe and P within the Fe2P NCs. EDX elemental analysis of the Fe2P NCs (Supplementary Fig. 2) revealed that the molar ratio of Fe to P was close to 2:1, indicating that the nanorods formed an ideal composition. These results indicate that the prepared Fe2P NCs are composed of crystalline Fe2P (Fig. 1p) and have a hexagonal prism structure consisting of the (10\(\stackrel{-}{1}\)0) and (0001) surfaces, as represented in Fig. 1q.
The TEM image (Fig. 1r) and EDX maps (Figs. 1s–1w) confirmed that Fe2P NCs were dispersed uniformly on the TiO2 support without structural degradation. Subsequently, X-ray photoelectron spectroscopy (XPS) was used to obtain information regarding the electronic states of the bare and supported Fe2P NCs. As shown in Fig. 1x, the Fe 2p spectrum of the Fe2P NCs after exposure to air consisted of two peaks at 707.2 and 720.1 eV assigned to metallic Fe 2p3/2 (706.8 eV) and Fe 2p1/2 (720.0 eV), respectively38. The metallic nature of the Fe2P NCs was further supported by electron energy loss spectroscopy (EELS) (Fig. 1y)39. These results indicate that the Fe2P NCs contain air-stable low-valent Fe (i.e., Fe0) species. The most intense peaks in the XPS spectra of the supported Fe2P NC catalysts reflected the presence of the metallic Fe species, and the minor peaks observed for both Fe2P NC/TiO2 and Fe2P NC/SiO2 around 710 eV corresponded to ionic Fe (i.e., FeO) species. The peaks attributed to FeO may represent the formation of Fe–O–Ti or Fe–O–Si bonds through the strong metal–support interactions. The Fe 2p peaks observed for the supported Fe2P NCs appear at lower energies than those of the non-supported species. Importantly, the largest shift was observed when loading the Fe2P NCs onto TiO2 due to electron donation from TiO2 to the Fe2P NCs (inset of Fig. 1x)40. In addition, the P 2p spectrum of the Fe2P NCs revealed an asymmetric peak split into two P0 peaks at 129.9 eV (P 2p3/2) and 130.6 eV (P 2p1/2), and an additional peak at 132.1 eV corresponds to the phosphate species formed through surface oxidation (Supplementary Fig. 3)41.
Evaluation of catalytic activity
The catalytic activity of the Fe2P NCs in the hydrogenation of nitriles, which is important for the highly atom-efficient synthesis of primary amines, was then examined. Although Ni- and Co-based sponge metals (Raney catalysts) are used for the hydrogenation of nitriles in industry, these catalysts are prone to significant deactivation during storage and require harsh reaction conditions due to their low activities. Thus far, various metal NP catalysts based on noble metals (i.e., Pt, Pd, Ru, Rh, Re, and Ir) and non-precious metals (i.e., Co and Ni) have been developed as alternatives to Raney catalysts42–45. In contrast, Fe NP catalysts are extremely rare, with only one recent report discussing the use of Fe/FeOx core–shell NPs supported on SiO2 in the hydrogenation of nitriles25. However, the active surface Fe NPs prepared by high-temperature H2 pretreatment are encapsulated with an FeOx shell, resulting in low activity, and a significant loss of its activity in the reuse experiments was caused by the low stability. Moreover, the use of Al foil or Al(i-OPr)3 as an additive is required to activate the Fe NPs. Therefore, the development of new class of Fe-based heterogeneous catalysts with high activity and stability for the hydrogenation of nitriles remains in a considerable challenge.
Figure 2a shows the results of the hydrogenation of benzonitrile (1a) as a model substrate using Fe2P NCs and supported Fe2P NC catalysts without pretreatment under 3.8 MPa of H2 and 0.2 MPa of NH3 at 453 K for 2 h. Notably, Fe2P NCs promoted the hydrogenation of 1a to give benzylamine (2a) in a 20% yield. In addition, the use of a TiO2 support considerably improved the activity of the Fe2P NCs, and 2a was afforded in 78% yield, whereas the carbon and SiO2 supports did not lead to any significant increase in yield. These results suggest that the type of support affects the hydrogenation efficiency, thereby indicating the importance of the metal–support interactions. Fe2P NC/TiO2 also gave 2a in excellent yield (95%) when the reaction time was extended to 3 h; this represents the first example of a heterogeneous iron catalyst promoting an efficient nitrile hydrogenation reaction without additives. Furthermore, Fe2P NC/TiO2 performed well at a lower H2 pressure of 0.5 MPa, although a longer reaction time was required (i.e., 24 h, 93% yield). By contrast, a conventionally pre-reduced Fe NP/TiO2 catalyst and the commercially available bulk Fe2P were inactive in this hydrogenation reaction. These results clearly demonstrate that nanosized iron phosphide species exhibit unique and excellent catalytic properties for the hydrogenation of nitriles. The effects of adding bases to the Fe2P NC/TiO2 reaction system were investigated because bases are effective at activating nitrile hydrogenation catalysts46,47. As outlined in Fig. 2b, MgO was the most effective base examined, increasing the yield of 2a from 78 to 96%, whereas other bases gave comparable yields to those obtained in the absence of an additive.
Figure 2c shows the result of the recycling experiments performed on the spent Fe2P NC/TiO2 catalyst, which was easily recovered from the reaction solution by centrifugation under air and was used in the subsequent run without treatment. Interestingly, after the fifth recycle, the catalyst continued to afford 2a in excellent yield. The fresh and reused Fe2P NC/TiO2 exhibited similar reaction rates over a short reaction time (i.e., 1 h) during the recycling experiments (white diamonds in Fig. 2c), demonstrating the durability of this catalyst. This is the first example of high reusability in the iron-catalyzed nitrile hydrogenation. In addition, the hot filtration of the Fe2P NC/TiO2 catalyst was carried out to separate the catalyst from the reaction mixture when the yield of 2a reached ~ 50% (i.e., 80 min, Fig. 2d). The resulting filtrate was further treated under the same reaction conditions in the absence of the filtered catalyst, and no increase in the yield of 2a was observed, indicating that no leaching of the solid catalyst into the reaction solution occurred.
The structure and electronic states of the used Fe2P NC/TiO2 were then investigated. More specifically, elemental analysis by inductively coupled plasma-atomic emission spectrometry (ICP-AES) revealed that the quantities of Fe and P in the Fe2P NC/TiO2 before and after the reaction were comparable (Supplementary Table 1), while representative TEM images showed no significant changes in the morphology of the Fe2P NCs (Supplementary Fig. 4). In addition, the XPS spectra of Fe2P NC/TiO2 before and after the reaction were similar (Supplementary Fig. 5). Thus, overall, these observations confirm the excellent reusability of the Fe2P NC/TiO2 catalyst.
Importantly, the high catalyst durability allowed the reaction to be carried out on the gram scale. In this case, 1.2 g of 1a was converted into 2a in 96% yield (Fig. 2e), with a turnover number (TON) and turnover frequency (TOF) of 295 and 12.3 h− 1, respectively. These TON and TOF values are 26 times greater than those of the recently reported heterogeneous iron catalyst (Supplementary Scheme 1).
Substrate scope
The substrate scope of the hydrogenation of various nitriles was explored using the optimized Fe2P NC/TiO2 catalyst and reaction conditions (Fig. 3). Benzonitrile derivatives bearing electron-withdrawing (i.e., halogen and trifluoromethyl) or electron-donating groups (i.e., methyl, tert-butyl, methoxy, amino, dimethylamino, methyl sulfide, phenoxy, and methylenedioxy) were converted into the corresponding benzylamines in high yields (2a–2u). In addition, nitriles bearing aromatic moieties, including phenylacetonitrile, biphenylcarbonitrile, and naphthonitrile, were also successfully hydrogenated to afford the corresponding primary amines (2v–2x). Furthermore, the Fe2P NC/TiO2 system was suitable for use with heteroaromatic compounds, such as pyridine, indole, furan, and thiophene, without dearomatization (2y–2cc). Importantly, the less active aliphatic and alicyclic nitriles were also hydrogenated to afford their corresponding primary amines in high yields (2dd–2ll). Notably, Fe2P NC/TiO2 was applicable to the hydrogenation of dinitriles, which are industrially important for the synthesis of polymer precursors. For example, isophthalonitrile, terephthalonitrile, suberonitrile, and sebaconitrile were facilely converted to their corresponding diamines (2mm–2pp). Overall, these results demonstrate the high generality and applicability of Fe2P NC/TiO2 for the hydrogenation of nitriles.
Kinetic study
As outlined in Supplementary Fig. 6, kinetic studies revealed that the initial reaction rates of the hydrogenation of 1a using Fe2P NC/TiO2 increased upon raising the reaction temperature from 432 to 453 K. The corresponding Arrhenius plot showed a good degree of linearity, and the apparent activation energy (Ea) was determined to be 59.8 kJ·mol− 1. Investigation of the dependency of the reaction rate on the substrate concentration and the H2 pressure shows that the initial rates were directly proportional to both parameters (Supplementary Figs. 7a and 7b), thereby suggesting that the rate-determining step involves the reaction of the adsorbed hydrogen species with the nitrile substrate. A hydrogen–deuterium (H2–D2) exchange reaction was also carried out (Supplementary Fig. 8). The Fe2P NC catalysts promoted the H2–D2 exchange reaction at 453 K, confirming the H2 activation ability of this catalyst. Notably, the H2–D2 exchange activity of Fe2P NC/TiO2 was significantly higher than that of the other Fe2P NC catalysts, which is consistent with the superiority of Fe2P NC/TiO2 over the other Fe2P NC catalysts during the nitrile hydrogenation reaction, as shown in Fig. 2a. Furthermore, as mentioned above, XPS revealed the generation of electron-rich Fe species in Fe2P NC/TiO2, which can be attributed to the donation of electrons from TiO2 to the Fe2P NCs. Such electron-rich metal species are well-known to promote the activation of H248, thereby confirming the pivotal role of the TiO2 support in improving the H2-activation ability of the Fe2P NCs through the formation of electron-rich Fe2P NC; ultimately, this led to the high catalytic efficiency of this system in the nitrile hydrogenation reaction.
XAFS and data analyses
To gain additional insight into the origin of the hydrogenation catalysis of the Fe2P NC/TiO2 system, the atomic-scale structure of the Fe species in the Fe2P NCs was investigated using Fe K-edge X-ray absorption fine structure (XAFS) analysis under an air atmosphere. Figure 4a shows the X-ray absorption near-edge structure (XANES) spectra of Fe2P NCs and Fe2P NC/TiO2 along with those of the Fe foil and FeO. The absorption edge energies of the Fe2P NCs (yellow line) and Fe2P NC/TiO2 (orange line) were considerably lower than that of FeO (red line), and very close to that of the Fe foil (blue line), thereby suggesting that the Fe species in the Fe2P NCs retain a metal-like state, which is consistent with the XPS result shown in Fig. 1x. In addition, a Fourier transform of the extended XAFS (FT-EXAFS) spectrum of the Fe2P NCs revealed two peaks at ~ 1.8 and 2.3 Å, which were assigned to the Fe–P and Fe–Fe bonds, respectively (Fig. 4b). Previous reports on metal phosphide catalysts revealed that metal–metal sites play a key role in the hydrogenation reaction31, indicating that the Fe–Fe sites in the Fe2P NCs can function as active sites for the nitrile hydrogenation reaction. Furthermore, the wavelet transformation (WT) results (Figs. 4c–4f) showed that the Fe2P NCs and Fe2P NC/TiO2 produced similar patterns, which allowed the Fe–Fe and Fe–P bonds to be observed. However, their patterns were different in the region of r = 1–1.5 Å and k = 9–12 Å−1; specifically, the pattern of Fe2P NC/TiO2 was similar to that of FeO. This can be attributed to the Fe–O–Ti bond formation from an interfacial interaction within Fe2P NC/TiO2. This is also consistent with the XPS results shown in Fig. 1x. The Fe–Fe species were further examined by EXAFS curve-fitting analysis (Supplementary Fig. 9 and Supplementary Table 2). The Fe‒Fe bond length of the Fe2P NCs was 2.65 ± 0.03 Å, which is slightly longer than that of the Fe foil (i.e., 2.48 ± 0.02 Å). Notably, the coordination number ratio of Fe–Fe to Fe–P (CNFe‒Fe/CNFe‒P) was 1.68, which is considerably smaller than the value in bulk Fe2P (5.20), revealing the formation of coordinatively unsaturated Fe atoms in the Fe2P NCs. Based on these XAFS results, coordinatively unsaturated Fe sites that can adsorb nitrile and H2 are formed in the Fe2P NC, which accounts for the high activity of this catalyst.
Finally, to investigate their structural and catalytic stability in air, the Fe2P NCs were heated from 300 to 873 K under an air atmosphere, and their XAFS spectra and catalytic activities were evaluated at each temperature. As shown in Fig. 4g, the Fe K-edge XANES spectra has an isosbestic point at 7123 eV, and the absorption edge shifted to higher energies upon increasing the temperature from 423 K. The FT-EXAFS results also show a decrease in the peak intensity of the Fe–Fe bond with increasing temperature above 423 K, whereas an increase in the peak intensity of the Fe–O bond was observed due to the oxidation of the Fe2P NCs (Fig. 4h). When these heat-treated catalysts were used in the hydrogenation of 1a, the activity was maintained up to 423 K, and decreased gradually with further increases in temperature (Fig. 4i). These results demonstrate that both the structure and the catalytic activity of the Fe2P NCs are highly stable in air.