Introduction

Iron is the most abundant transition metal in the Earth’s crust and is considered an ideal metal for catalysts owing to its extremely low cost, low-toxicity, and unique catalytic properties1,2,3,4,5. In particular, iron-based heterogeneous catalysts play a key role in two reactions vital for society, namely the Haber–Bosch process for ammonia synthesis6,7,8 and the Fischer–Tropsch process for synthesizing gasoline-range iso-paraffins from syngas9,10. In these hydrogenation reactions, the iron oxide species deposited on metal oxide supports are reduced by H2 at high temperatures to form zero-valent iron nanoparticles (Fe NPs). Thus, low-valent Fe NPs produced in situ are the true active species for the aforementioned gas-phase reactions11,12. However, as conventional Fe NP catalysts require harsh reaction conditions to be active, they exhibit extremely low activity toward liquid-phase hydrogenation reactions that are conducted under mild reaction conditions (i.e., temperatures: < 473 K); such reactions are usually involved in the synthesis of bulk and fine chemicals and biomass transformation. Moreover, the Fe NPs are inherently unstable and are easily oxidized to inactive FeOx even in the presence of small amounts of oxygen13. Hence, the handling of such Fe NP catalysts requires strict oxygen-free conditions during all catalytic manipulation steps, including the preparation, reaction, separation, and recycling stages14. Furthermore, the difficulties of using these unstable Fe NPs have restricted attempts to improve their catalytic performance, such as through particle size control, support alterations, and heteroatom doping15,16,17. Therefore, the preparation of Fe NP catalysts has been limited to the classical in situ reduction of Fe ions supported on metal oxides that can withstand high temperatures and H2 pressures18,19. To address the limitation associated with the instability of Fe NPs in air, various strategies have been employed, such as coating Fe NPs with metal oxides or N-doped carbon layers20,21,22,23,24,25,26. However, this approach often results in improved air stability at the expense of decreased catalytic activity due to the shielding of active surface sites. A recent example illustrating this trade-off is the use of Fe/FeOx core–shell NPs supported on SiO2 as a heterogeneous catalyst for the hydrogenation of nitriles25. Unfortunately, this catalyst exhibited insufficient activity and stability, representing a significant limitation of conventional iron catalysts. It is therefore vital to overcome this trade-off for the development of highly active and stable iron-based heterogeneous catalysts suitable for liquid-phase reactions.

The rapid advancement in nanotechnology over the past decade has led to the development of various synthetic techniques for nanosized metal non-oxides, including metal nitrides, phosphides, and sulfides. These nanomaterials frequently exhibit unique catalytic properties distinct from those of conventional metal oxides and metals27,28,29. In this context, metal phosphide nanomaterials have recently emerged as highly promising catalysts, demonstrating superior activity and stability compared to conventional metal catalysts in gas-phase desulfurization30 and liquid-phase reactions31,32,33, as well as in the electrocatalytic hydrogen-evolution reaction34. However, the catalytic potential of iron phosphide in liquid-phase reactions remains largely unexplored, despite the high demand for iron-based heterogeneous catalysts in the field of fine and bulk chemical syntheses.

To overcome the limitation of conventional iron catalysts, our focus was on developing iron phosphide nanocrystals (Fe2P NCs). The Fe2P NC catalyst exhibited excellent stability and high activity for the liquid-phase hydrogenation of nitriles, a process of significant industrial importance in the production of primary amines35,36,37. The Fe2P NC catalyst demonstrated high reusability without a significant loss of activity. This study represents the successful demonstration of an air-stable and reusable iron-based heterogeneous catalyst for nitrile hydrogenation.

Results and discussion

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 catalyst38, 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\(\bar{11}\)1), (0002), and (30\(\bar{3}\)0) crystalline planes of hexagonal Fe2P, respectively (Supplementary Fig. 1). Transmission electron microscopy (TEM) of the Fe2P NCs revealed regular nanorods with a mean size of 26.5 × 8.7 nm (Fig. 1a, d, and Supplementary Fig. 2). Side and top view high-resolution TEM of the Fe2P NCs (Fig. 1b, e, respectively) revealed lattice spacings of 0.17 and 0.51 nm, corresponding to the (0002) and (10\(\bar{1}\)0) hexagonal Fe2P planes, respectively. The selected area electron diffraction (SAED) patterns from the side and top views (Fig. 1c, f, respectively) also showed the spot patterns indexed to the [10\(\bar{1}\)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 (Fig. 1g–j and Fig. 1k–n, respectively) depicted uniform distributions of Fe and P within the Fe2P NCs. EDX elemental analysis of the Fe2P NCs (Supplementary Fig. 3) 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. 1o) and have a hexagonal prism structure consisting of the (10\(\bar{1}\)0) and (0001) surfaces, as represented in Fig. 1p. The TEM image and EDX maps confirmed that Fe2P NCs were dispersed uniformly on the TiO2 support without structural degradation (Supplementary Fig. 4).

Fig. 1: Structural characterization of Fe2P NCs.
figure 1

a, b Side view TEM images of the Fe2P NCs. c SAED pattern of the Fe2P NCs indicated by the blue square in part (a). d, e Top view TEM images of the Fe2P NCs. f SAED pattern of the Fe2P NCs indicated by the red square in part (d). g Side view HAADF-STEM image of the Fe2P NCs with elemental mapping images of (h), P and (i), Fe. j Composite overlay of parts (h) and (i). k Top view HAADF-STEM image of the Fe2P NCs with elemental mapping images of (l), P and (m), Fe. n Composite overlay of parts (l) and (m). o Unit cell of Fe2P. p Proposed crystal structure of the Fe2P NCs.

Full size image

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. 2a, 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), respectively39. The metallic nature of the Fe2P NCs was further supported by electron energy loss spectroscopy (EELS) (Supplementary Fig. 5)40. 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 at ~710 eV corresponded to ionic Fe (i.e., FeO) species (Fig. 2b, c). The peaks attributed to FeO may represent the formation of Fe–O–Ti or Fe–O–Si bonds through the metal–support interactions (see details in Supplementary discussion). 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. 2e)41. In addition, the P 2p spectrum of the Fe2P NCs revealed an asymmetric peak split into two P0 peaks at 129.5 eV (P 2p3/2) and 130.3 eV (P 2p1/2), and an additional peak at 132.1 eV corresponds to the phosphate species formed through surface oxidation (Supplementary Fig. 6)42.

Fig. 2: Fe 2p XPS spectra of Fe2P NCs and supported Fe2P NC.
figure 2

a Fe2P NCs. b Fe2P NC/TiO2. c Fe2P NC/SiO2, d Fe2P NC/C. e Composite overlay of the spectra shown in parts (a–d). The inset shows an enlarged view of the Fe 2p3/2 peaks.

Full size image

Evaluation of catalytic property of Fe2P NC in nitrile hydrogenation

The catalytic activity of the Fe2P NCs was then evaluated in the hydrogenation of nitriles, which is an important reaction for the synthesis of primary amines. 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 catalysts43,44,45,46,47. 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 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 nitrile hydrogenation remains in a considerable challenge.

Figure 3a 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, SiO2, and other metal oxide supports did not lead to any significant increase in yield (Supplementary Table 1). 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. 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). In 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 then investigated due to the fact that bases are effective in activating nitrile hydrogenation catalysts48,49. As outlined in Fig. 3b, MgO was the most effective base examined, increasing the yield of 2a from 78 to 96%, whereas other bases did not lead to any significant improvement in yield. The effect of NH3 was also confirmed50, wherein the addition of NH3 enhanced the selectivity toward primary amine formation (Supplementary Figs. 7–10). In addition, we conducted the hydrogenation of 1a using deuterium-labeled 2-propanol (2-propanol-d8) as a solvent. The resulting 2a did not show any deuterium incorporation, confirming that hydrogen source is not 2-propanol but H2 in this reaction (Supplementary Figs. 11 and 12).

Fig. 3: Catalytic performance of Fe2P NCs in the hydrogenation of 1a.
figure 3

a Hydrogenation of 1a using various Fe catalysts. Reaction conditions: Fe catalyst (Fe: 7.6 mol%), 1a (0.5 mmol), 2-propanol (3 mL), H2 (3.8 MPa), NH3 (0.2 MPa), 2 h. Yield was determined by gas chromatography (GC) using the internal standard technique. a3 h. bH2 (0.5 MPa), 24 h. b Effect of bases on the hydrogenation of 1a. Reaction conditions: Fe2P NC/TiO2 (0.1 g), 1a (0.5 mmol), 2-propanol (3 mL), base (0.1 mmol), H2 (3.8 MPa), NH3 (0.2 MPa), 2 h. TEA: triethylamine. DBU: 1,8-diazabicyclo[5.4.0]undec-7-ene. DIPEA: N,N-diisopropylethylamine. c Hydrogenation of 1a under a high substrate/Fe ratio. Reaction conditions: Fe2P NC/TiO2 (0.1 g, Fe: 0.33 mol%), 1a (11.7 mmol), 2-propanol (20 mL). d Reuse experiments. Reaction conditions: Fe2P NC/TiO2 (0.1 g), 1a (0.5 mmol), 2-propanol (3 mL), H2 (3.8 MPa), NH3 (0.2 MPa). Reaction time: 3 h (blue columns), 1 h (white diamonds). e Hot filtration experiments. Reaction conditions: Fe2P NC/TiO2 (0.1 g), 1a (0.5 mmol), 2-propanol (3 mL), H2 (3.8 MPa), NH3 (0.2 MPa). Blue circles: without filtration of the catalyst. Red circle: with removal of the catalyst by hot filtration after 80 min.

Full size image

Figure 3c shows the evaluation of the durability of Fe2P NC/TiO2 under a high substrate/Fe ratio (S/Fe = 307). 1.2 g of 1a was converted into 2a in 96% yield, with a turnover number (TON) of 295. This TON value is the highest reported for the iron catalysts (Supplementary Table 2). The durability of Fe2P NC/TiO2 was also highlighted in the recycling experiments (Fig. 3d). In contrast to the conventional air-unstable Fe catalysts with difficult handling, the spent Fe2P NC/TiO2 catalyst was easily recovered under ambient conditions through centrifugation and was reused without any additional treatment. Notably, Fe2P NC/TiO2 exhibited a consistently high activity without significant loss up to the fourth recycling experiment. 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. 3d), demonstrating the durability of this catalyst. The yield of 2a slightly decreased after the fifth recycle, which may be attributed to the reduced amount of catalyst during the recovery process (Supplementary Fig. 13). 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. 3e). 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. Elemental analysis of the filtrate using inductively coupled plasma-atomic emission spectrometry (ICP-AES) confirmed the absence of Fe and P species (detection limit: 0.004 ppm Fe, 0.001 ppm P). These results indicate 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. ICP-AES revealed that the quantities of Fe and P in the Fe2P NC/TiO2 before and after the reaction were comparable (Supplementary Table 3), while representative TEM images showed no significant changes in the morphology of the Fe2P NCs (Supplementary Fig. 14). In addition, the XPS spectra of Fe2P NC/TiO2 before and after the reaction were similar (Supplementary Fig. 15). Thus, overall, these observations confirm the excellent reusability of the Fe2P NC/TiO2 catalyst.

The substrate scope of the hydrogenation of various nitriles was explored using the optimized Fe2P NC/TiO2 catalyst and reaction conditions (Fig. 4a). 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, aliphatic and alicyclic nitriles, which are less reactive, were also hydrogenated to afford their corresponding primary amines in high yields (2dd–2qq). 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, succinonitrile, suberonitrile, sebaconitrile, and 3,3’-iminodipropionitrile were facilely converted into their corresponding diamines (2rr–2ww). Additionally, the feasibility of the practical application of Fe2P NC/TiO2 was investigated under gram-scaled conditions using four selected nitriles, providing the corresponding primary amines in 82–98% isolated yields (Fig. 4b). These findings highlight the broad scope and practical utility of the Fe2P NC/TiO2 catalyst for nitrile hydrogenation.

Fig. 4: Applicability of Fe2P NC/TiO2 in the nitrile hydrogenation.
figure 4

a Substrate scope of nitriles. Reaction conditions: Fe2P NC/TiO2 (0.1 g), substrate (0.5 mmol), 2-propanol (3 mL), MgO (0.1 mmol), H2 (3.8 MPa), NH3 (0.2 MPa), 3 h. Yield was determined by GC using the internal standard technique. a6 h. b12 h. cFe2P NC/TiO2 (0.2 g), substrate (0.25 mmol), H2 (4.4 MPa), NH3 (0.6 MPa), 24 h. b Multi-gram scale nitrile hydrogenations. Reaction conditions: Fe2P NC/TiO2 (5 g, Fe: 1.9 mol%), substrate (5 g), 2-propanol (15 mL), H2 (4.5 MPa), NH3 (0.7 MPa), MgO (2.5 mmol), 453 K, 12 h. dFe: 4.3 mol%, substrate (2 g), 24 h. eFe: 7.2 mol%, substrate (1.5 g), 24 h. fFe: 4.7 mol%, substrate (2.5 g).

Full size image

Discussion of the origin of catalysis by Fe2P NC/TiO2

As outlined in Fig. 5a, 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. Remarkably, this value is comparable to the activation energy of 60.3 kJ·mol−1 reported for the Raney nickel catalyst51, indicating the promising potential of Fe2P NC/TiO2 as a cost-effective alternative in industrial applications. Investigation of the dependency of the reaction rate on the substrate concentration and the H2 pressure shows that the initial rates were positively correlated to both parameters (Fig. 5b, c), 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 (Fig. 5d). 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. 3a. 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 H252, 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.

Fig. 5: Kinetic study of the hydrogenation of 1a and H2–D2 exchange experiments.
figure 5

a Arrhenius plot of the hydrogenation of 1a. Reaction conditions: Fe2P NC/TiO2 (0.1 g, Fe: 1.9 mol%), 1a (2 mmol), 2-propanol (3 mL), H2 (4 MPa), 432–453 K, 2 h. b Double logarithm plots of the concentration of 1a and the initial reaction rate. c Double logarithm plots of the partial pressure of H2 and the initial reaction rate. Reaction conditions: Fe2P NC/TiO2 (0.1 g, Fe: 1.9–15.2 mol%), 1a (0.25–2 mmol), 2-propanol (3 mL), H2 (1–4 MPa), 453 K, 2 h. The plots denote the data mean values, and the error bars show the range. d H2–D2 exchange experiments. Red, blue, green, and yellow show the results obtained by Fe2P NCs, Fe2P NC/TiO2, Fe2P NC/SiO2, and Fe2P NC/C, respectively.

Full size image

To gain additional insight into the origin of the hydrogenation catalysis of Fe2P NC/TiO2, 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 6a 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 (red line) and Fe2P NC/TiO2 (blue line) were considerably lower than that of FeO (orange line), and very close to that of the Fe foil (purple 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. 2. 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. 6b). 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 (Fig. 6c–f) showed that 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 regions 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 arising from an interfacial interaction within Fe2P NC/TiO2. This is also consistent with the XPS results shown in Fig. 2. The Fe–Fe species were further examined by EXAFS curve-fitting analysis (Supplementary Fig. 16 and Supplementary Table 4). 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 catalyst53.

Fig. 6: XAFS measurements and stability evaluation of Fe2P NC/TiO2.
figure 6

a Fe K-edge XANES spectra of Fe foil, FeO, Fe2P NCs, and Fe2P NC/TiO2. b Fourier transformation of k3-weighted Fe K-edge EXAFS spectra of Fe foil, FeO, Fe2P NCs, and Fe2P NC/TiO2. Wavelet transformations for k3-weighted EXAFS signals of c Fe foil, d FeO, e Fe2P NCs, and f Fe2P NC/TiO2. g Fe K-edge XANES spectra and h Fourier transformation of k3-weighted Fe K-edge EXAFS spectra of Fe2P NCs with increasing temperature under air. i Hydrogenation of 1a using Fe2P NC/TiO2 after heat-treatment in air. The columns denote the data mean values, and the error bars show the range. Reaction conditions: Fe2P NC/TiO2 (0.1 g), 1a (0.5 mmol), 2-propanol (3 mL), H2 (3.8 MPa), NH3 (0.2 MPa), 453 K, 3 h. Yield was determined by GC using the internal standard technique.

Full size image

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. 6g, the Fe K-edge XANES spectra have 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. 6h). 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. 6i). These results demonstrate that both the structure and the catalytic activity of the Fe2P NCs are highly stable in air.

In conclusion, Fe2P NCs were synthesized and the optimal TiO2-supported Fe2P NCs exhibited high catalytic activity for the hydrogenation of various nitriles to provide the corresponding primary amines in high yields under liquid-phase conditions. These results contrast sharply with those of conventional iron-based heterogeneous catalysts (i.e., iron nanoparticles), which exhibited no activity in this reaction and were unstable in an air environment. Moreover, the optimal Fe2P NC catalyst demonstrated high durability, wherein the spent Fe2P NC catalyst was easily recovered under air, and was reusable without any significant loss of activity. Characterization using a range of spectroscopic methods revealed that coordinatively unsaturated Fe species with low oxidation states play a crucial role in the nitrile hydrogenation reaction. This study therefore demonstrates that highly active Fe2P NCs, which do not require high-temperature pretreatment by H2, are promising candidates to replace conventional Fe NPs. Moreover, the stability of the prepared Fe2P NCs under air allows easy handling. In addition, further improvements to the catalytic performance can be achieved by the selection of the appropriate support material. Indeed, the use of a TiO2 support significantly increased the hydrogenation activity of the Fe2P NC catalyst through electron donation from TiO2 to Fe2P NCs. While this study mainly focused on the concerted effect between Fe2P NCs and TiO2, it is worth noting that combining Fe2P NCs with other support materials could lead to further improvements in a variety of reactions or the development of unexplored iron catalyses. Thus, the development of this catalytic system constitutes a breakthrough in terms of the stability and facile improvement of iron-based heterogeneous catalysts and significantly expands the applicability of iron catalysts beyond gas-phase hydrogenation reactions into liquid-phase organic transformations. Owing to the abundant, inexpensive, and low-toxic nature of iron, this study is expected to pave the way for establishing green, sustainable, and cost-effective methods for manufacturing valuable chemicals using iron catalysts.

Methods

Materials

All commercially available chemicals were used as received. Fe(CO)5 (>95%) was purchased from Kanto Chemical Co., Ltd (Tokyo, Japan). Hexadecylamine (>95%) and triphenylphosphite (>97%) were acquired from Tokyo Chemical Industry Co., Ltd (Tokyo, Japan). Acetone (>99.0%), chloroform (>99.0%), ethanol (>99.5%), 2-propanol (>99.7%), Fe(NO3)3·9H2O (>99.0%), triethylamine (>99.0%), 1,8-diazabicyclo[5.4.0]undec-7-ene (>97.0%), N,N-diisopropylethylamine (>97.0%), carbon (charcoal, activated, powder), and Fe2O3 were obtained from Fujifilm Wako Pure Chemical Corporation (Osaka, Japan). 2-propanol-d8 (99.5%) was purchased from Sigma-Aldrich (St. Louis, the United States). TiO2 (JRC-TIO-2), ZrO2 (JRC-ZRO-7), and CeO2 (JRC-CEO-1) were provided by the Catalysis Society of Japan (Tokyo, Japan) as reference catalysts. SiO2 (CARiACT, Q-6) was supplied by Fuji Silysia Chemical Ltd (Aichi, Japan). Bulk Fe2P (>99.5%, particle size: 1–5 µm) was purchased from Mitsuwa Chemicals (Osaka, Japan). MgO was obtained from Tomita Pharmaceutical Co., Ltd (Tokushima, Japan). The following nitriles were obtained commercially: Tokyo Chemical Industry Co., Ltd (Tokyo, Japan)— benzonitrile (>99%), o-tolunitrile (>98%), m-tolunitrile (>98%), p-tolunitrile (>98%), p-tert-butylbenzonitrile (>98%), o-chlorobenzonitrile (>98%), m-chlorobenzonitrile (>98%), p-chlorobenzonitrile (>98%), p-bromobenzonitrile (>98%), p-fluorobenzonitrile (>98%), 2,4-difluorobenzonitrile (>98%), 3,4-dichlorobenzonitrile (>98%), p-(trifluoromethyl)benzonitrile (>98%), p-aminobenzonitrile (>98%), p-(dimethylamino)benzonitrile (>98%), p-phenoxybenzonitrile (>98%), piperonylonitrile (>98%), o-methoxybenzonitrile (>98%), m-methoxybenzonitrile (>98%), p-methoxybenzonitrile (>98%), phenylacetonitrile (>98%), 4-cyanobiphenyl (>98%), 1-naphthonitrile (>98%), 5-cyanoindole (>98%), thiophene-2-carbonitrile (>98%), valeronitrile (>98%), octanenitrile (>97%), decanenitrile (>98%), lauronitrile (>98%), 3-ethoxypropionitrile (>99%), pivalonitrile (>98%), cyclopropanecarbonitrile (>98%), ethylene cyanohydrin (>97%), 3-methoxypropionitrile (>99%), isophthalonitrile (>98%), succinonitrile (>99%), suberonitrile (>98%), sebaconitrile (>98%), and 3,3’-iminodipropionitrile (>98%); Fujifilm Wako Pure Chemical (Osaka, Japan)—4-(methylthio)benzonitrile (>98%), 4-cyanopyridine (>98%), 3-cyanopyridine (>98%), 2-furancarbonitrile (>98%), 1-adamantanecarbonitrile (>97%), terephthalonitrile (>95%), and N-benzylidenebenzylamine (>96%); and Sigma-Aldrich (St. Louis, the United States)—cyclohexanecarbonitrile (98%). Dimethylaminopropionitrile54, 3-[(2-hydroxyethyl)amino]propiononitrile55, 3-isopropoxypropanenitrile56, and benzylideneamine57 were prepared according to the literature procedures.

The compounds employed in catalyst preparation, including iron pentacarbonyl, triphenylphosphite, and hexadecylamine were confirmed using EDX elemental analysis to have purities of >99.9%.

General considerations

XRD analysis was carried out with Philips X’Pert-MPD (PANalytical B. V., Almelo, Netherlands) using Cu Kα radiation (45 kV, 40 mA). Elemental analysis was performed using ICP-AES (Optima 8300, Perkin Elmer, Waltham, United States) or EDX (EDX-7200, Shimadzu Corporation, Kyoto, Japan). TEM analysis was carried out with JEM-ARM200F at 200 kV (JEOL Ltd., Tokyo, Japan). STEM coupled with Super-X EDX detection with elemental mapping and EELS was carried out at 300 kV with FEI Titan Cubed G2 60-300 (FEI Co. Japan Ltd., Tokyo, Japan). Elemental EDX mapping analysis was carried out using an Esprit detector. The Fe K-edge X-ray absorption spectra were recorded at the BL01B1 and BL14B2 lines, using a Si(111) monochromator at the SPring-8 facility of the Japan Synchrotron Radiation Research Institute (Harima, Japan). The acquired EXAFS data were normalized using xTunes software58. The k3-weighted χ(k) data of the Fe K-edge in the range of 3 ≤ k ≤ 13 were Fourier transformed to the R space. WT analysis was performed in the range of 1 ≤ r ≤ 4 using Morlet software to obtain information on the coordination environment of the Fe species59. XPS (Kratos Ultra2, Shimadzu Corporation, Kyoto, Japan) was performed using an Al Kα radiation source. The analysis area was 0.7 × 0.3 mm, and the C 1 s peak at 285.0 eV was used as the internal reference.

Product quantification

GC–flame ionization detection (GC-2014, Shimadzu Corporation, Kyoto, Japan) was performed using an InertCap for amines (GL Sciences, Tokyo, Japan, 30 m × 0.32 mm i.d.). The oven temperature was programmed as follows: the initial temperature was 473 K and maintained for 3 min. The temperature was increased to 533 K at a rate of 20 K·min−1, then kept constant at 533 K for 20 min. The other conditions were as follows: column flow rate (N2 carrier), 2 mL·min−1; split ratio, 17.5; vaporization chamber temperature, 523 K; and detector temperature, 533 K. The 1H and 13C NMR (JEOL JNM-ESC400, JEOL Ltd, Tokyo, Japan) spectra were acquired at 400 and 100 MHz, respectively. 1H NMR chemical shifts were reported in parts per million (ppm) using the following standard chemical shifts: tetramethylsilane (0.00 ppm), the residual proton signal in D2O (4.70 ppm at 303 K), or CD3OD (3.30 ppm). 13C NMR chemical shifts were reported in ppm using the following standard chemical shifts: dimethyl sulfoxide-d6 (DMSO-d6) (39.50 ppm), CD3OD (49.00 ppm), or 1,4-dioxane-d8 (67.19 ppm). NMR multiplicities were reported using the following abbreviations: s: singlet, d: doublet, dd: double doublet, t: triplet, q: quartet, sep: septet, m: multiplet, br: broad, J: coupling constants in hertz.

Catalyst preparation

Synthesis of the Fe2P NCs

Triphenylphosphite (10 mmol) and hexadecylamine (10 mmol) were added to a Schlenk flask and stirred at 393 K for 30 min under vacuum. After increasing the temperature to 473 K under an argon atmosphere, Fe(CO)5 (1 mmol) was injected. Subsequently, the temperature was further increased to 593 K at a rate of 50 K·min−1 and then held constant for 4 h to provide a black colloidal solution. After cooling the mixture to 298 K, the product was isolated by precipitation in acetone. Finally, the precipitate was washed with chloroform–acetone (1:1) mixture to afford the desired Fe2P NCs as a black powder.

Synthesis of the Fe2P NC/support

Fe2P NCs (40 mg) were dispersed in chloroform (100 mL) and stirred with the desired support (i.e., TiO2, SiO2, or C, 1.0 g) at 298 K for 6 h to afford the corresponding Fe2P NC/support (i.e., Fe2P NC/TiO2, Fe2P NC/SiO2, or Fe2P NC/C).

Synthesis of Fe NP/TiO2

TiO2 (1.0 g) was stirred in a 2 mM ethanolic solution of Fe(NO3)3 (50 mL) for 48 h at 298 K, followed by evaporation at 348 K. The resulting catalyst precursor was then reduced in a H2 flow with heating from 298 K to 1173 K at a rate of 5 K·min−1, followed by holding at 1173 K for 1 h to yield the desired Fe NP/TiO2.

Typical reaction procedure

The typical reaction procedure for the hydrogenation of nitriles using Fe2P NC/TiO2 was as follows. Fe2P NC/TiO2 (0.1 g) was placed in a 50 mL stainless-steel autoclave with a Teflon inner cylinder, followed by the addition of nitrile (0.5 mmol) and 2-propanol (3 mL). The reaction mixture was stirred at 453 K under 3.8 MPa of H2 and 0.2 MPa of NH3. After 2 h of reaction, the obtained solution was analyzed by GC to determine the conversion and the yield using diethylene glycol dimethyl ether as an internal standard. In addition, to obtain the hydrochloride salt, the crude reaction mixture was filtered to remove the catalyst, and ammonia was removed under vacuum conditions. The mixture was then added to a hydrogen chloride solution (1.25 M in 1,4-dioxane), and the solvent was removed to give the pure hydrochloride salt for NMR analysis.

The yields of primary amine and imine are calculated as follows (Eqs.1 and 2):

$${{{{{\rm{Yield}}}}}}\,\left(\%\right)\,{{{{{\rm{of}}}}}}\,{{{{{\rm{primary}}}}}}\,{{{{{\rm{amine}}}}}}=\frac{{{{{{\rm{mol}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{obtained}}}}}}\,{{{{{\rm{primary}}}}}}\,{{{{{\rm{amine}}}}}}\,{{{{{\rm{product}}}}}}}{{{{{{\rm{initial}}}}}}\,{{{{{\rm{mol}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{substrate}}}}}}}\times 100\%$$
(1)
$${{{{{\rm{Yield}}}}}}\,\left(\%\right)\,{{{{{\rm{of}}}}}}\,{{{{{\rm{imine}}}}}}=\frac{{{{{{\rm{mol}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{obtained}}}}}}\,{{{{{\rm{imine}}}}}}\,{{{{{\rm{product}}}}}}}{{{{{{\rm{initial}}}}}}\,{{{{{\rm{mol}}}}}}\,{{{{{\rm{of}}}}}}\,{{{{{\rm{substrate}}}}}}}\times 2\times 100\%$$
(2)

Recycling experiment

After the reaction, Fe2P NC/TiO2 was removed by centrifugation, and the primary amine yield was determined by GC. The spent catalyst was washed with 2-propanol for the reuse experiments. No other catalyst pretreatment was required.

Gram-scale experiment

The gram-scale reaction of benzonitrile was performed in a 100 mL stainless-steel autoclave with a Teflon inner cylinder at 453 K according to the above procedure. After the reaction, the crude reaction mixture was filtered to remove the catalyst, and the remaining ammonia was removed under vacuum conditions. Subsequently, the mixture was added to a hydrogen chloride solution, and the solvent was removed to give the pure hydrochloride salt. The TON was calculated based on Eq.3.

$${{{{{\rm{Turnover}}}}}}\,{{{{{\rm{number}}}}}}\,({{{{{\rm{TON}}}}}})=\frac{{{\mbox{mol}}}\,{{\mbox{of}}}\,{{\mbox{obtained}}}\,{{\mbox{primary}}}\,{{\mbox{amine}}}\,{{\mbox{product}}}}{{{\mbox{mol}}}\,{{\mbox{of}}}\,{{\mbox{Fe}}}\,{{\mbox{used}}}\,{{\mbox{in}}}\,{{\mbox{the}}}\,{{\mbox{reaction}}}}$$
(3)

Kinetic experiment

The kinetic analysis was evaluated by the conversion of benzonitrile. The initial reaction rates were determined at low conversions ( < 30%), and were calculated based on Eq.4.

$${{{{{\rm{Reaction}}}}}}\,{{{{{\rm{rate}}}}}}=\frac{{{\mbox{conversion}}}\,{{\mbox{of}}}\,{{\mbox{benzonitrile}}}\times {{\mbox{initial}}}\,{{\mbox{mol}}}\,{{\mbox{of}}}\,{{\mbox{substrate}}}}{{{\mbox{mol}}}\,{{\mbox{of}}}\,{{\mbox{Fe}}}\,{{\mbox{used}}}\,{{\mbox{in}}}\,{{\mbox{the}}}\,{{\mbox{reaction}}}\times {{\mbox{reaction}}}\,{{\mbox{time}}}}$$
(4)

H2−D2 exchange reaction

The H2−D2 exchange reactions were performed in a closed gas-circulation system equipped with an online quadrupole mass spectrometer (BELMass-S, BEL Japan, Inc., Osaka, Japan). The prepared Fe2P NC catalyst (Fe: 8 mg) was placed in a reactor and the air was then evacuated from the reactor under vacuum conditions. The sample was then heat-treated at 453 K under vacuum for 1 h. Subsequently, an equimolar mixture of H2 and D2 gases was introduced to the reaction system at 453 K and the total pressure was adjusted with Ar to 23 kPa. The gas phase was analyzed by monitoring the signals corresponding to m/z values of 2, 3, and 4.