Synthesis and characterization of RuNi bimetallic catalysts. A monometallic Ni supported on amorphous Al2O3 substrate was prepared based on structural topotactic transformation of layered double hydroxides (LDHs) precursors. Afterwards, a set of RuNi/Al2O3 bimetallic samples with various Ru loading (0.1−2 wt.%) were precisely synthesized by a galvanic replacement method to deposit Ru atoms onto the surface of Ni nanoparticles (NPs) (Fig. 1a). Inductively coupled plasma atomic emission spectroscopy (ICP−AES) measurement explicitly gives the element composition of Ru and Ni in these samples (Supplementary Table 1), which is close to the theoretical loading of the feed. SEM images (Supplementary Fig. 1) and BET results (Supplementary Fig. 2–9, Supplementary Table 1) clearly reveal that the morphology structure and specific surface area of RuNi/Al2O3 samples do not show obvious change during galvanic replacement process. The XRD patterns (Fig. 1b) show series of characteristic reflections at 2θ 44.3°, 51.6°, and 76.1°, indexed to the (111), (200), and (220) of a typical Ni (JCPDS 048–1548) phase. It is noted that the RuNi samples do not display metallic or oxidic Ru reflection, implying a high dispersion of Ru species. Transmission electron microscopy (TEM) measurements of monometallic Ni and RuNi samples (Fig. 1c1−c8) show that Ni nanoparticles are well dispersed and anchored onto the amorphous Al2O3 support with a close particle size (7.7−8.3 nm). HRTEM images (Fig. 1d1−d8) display a uniform lattice spacing of ~ 0.203 nm in both Ni and RuNi samples, corresponding to Ni (111) plane; yet the lattice fringes of Ru species are absent. This indicates that the Ru atoms are highly dispersed on the surface of samples, without influence on the lattice constants of Ni.
Catalytic behaviour for chemoselective hydrogenation of 4-nitrostyrene. The catalytic performance of RuNi samples were explored toward chemoselective hydrogenation of 4-nitrostyrene (4-NS) to 4-aminostyrene (4-AS). As shown in Fig. 2a, the pristine Ni catalyst gives an extremely low catalytic activity with 7.1% 4-NS conversion after 3 h, whereas 100% conversion is achieved in the presence of monometallic Ru catalyst under the same conditions. In contrast to monometallic samples, all the RuNi catalysts show significantly enhanced hydrogenation activity (conversion: ~100%; 3 h), and the reaction rate is positively correlated with Ru content. This implies that the Ru species play a crucial role during the hydrogenation process. Interestingly, the product selectivity (Fig. 2b) is significantly diverse over various catalysts. For pristine Ni catalyst with a rather low activity, 4-nitroethylbenzene (4-NE) is the main product (selectivity: 83.6%) and the selectivity of 4-AS is 11.3%. In the case of monometallic Ru, the C = C bond in 4-NS is peculiarly prone to undergo hydrogenation than −NO2, resulting in the formation of 4-nitroethylbenzene (4-NE) or 4-aminoethylbenzene (4-AE); yet the selectivity of 4-AS is merely 15.7%. Surprisingly, RuNi catalysts with relatively low Ru loading (0.1−0.4 wt.%) exhibit an excellent selectivity (> 99%) toward 4-aminostyrene (4-AS), indicating it is exclusively chemoselective for −NO2 rather than ethenyl. When Ru content is larger than 0.6 wt.%, the selectivity to 4-AS declines sharply whilst that of 4-AE augments instead, along with the increase of Ru loading. For the 2 wt.% RuNi catalyst, both nitro and ethenyl are highly hydrogenated with the selectivity toward 4-AE up to 98%. In addition, the turnover frequency (TOF) of RuNi catalysts was calculated for 4-nitrostyrene hydrogenation based on Ru sites, to further reveal the intrinsic catalytic activity (Fig. 2c). With the increase of Ru loading from 0.1 wt.% to 0.4 wt.%, the RuNi catalysts exhibit an almost constant TOF value (4271−4293 h−1), indicating that the Ru species is located as atomic dispersion (might form RuNi SAA) and serves as the main active sites. However, the TOF value decreases significantly with a further increase of Ru loading (within 0.6−2 wt.%), which implies a change of intrinsic active site structure (from atomic dispersion to Ru nanoclusters). Moreover, the TOF of 0.4 wt.% RuNi (SAA) catalyst, to the best of our knowledge, stands at the highest level among metal catalysts reported previously under similar reaction conditions (Supplementary Table 2), which further demonstrates RuNi single atom alloy affords superior catalytic performance. Supplementary Fig. 10 shows catalytic performance of 0.4 wt.% RuNi (SAA) catalyst at various H2 pressure and temperature, from which 1 MPa of H2 pressure and 60°C of reaction temperature are adopted as the optimal reaction parameters. The reusability of 0.4 wt.% RuNi sample was further evaluated (Fig. 2d), and no significant shrink in both activity and yield was observed within five successive recycles. The XRD pattern and TEM images (Supplementary Fig. 11−12) of the used 0.4 wt.% RuNi catalyst after 5 cycles do not show obvious change in crystal structure, indicating a high stability in the selective hydrogenation reaction. In addition, the 0.4 wt.% RuNi (SAA) catalyst also achieved excellent amines yield toward the chemoselective hydrogenation of other nitroarenes contained halogen, aldehyde and hydroxyl group (Supplementary Table 3), which demonstrates its good applicability.
To further study the remarkable chemoselectivity difference, the hydrogenation of styrene and nitrobenzene mixture (1:1) was also performed in the presence of monometallic Ni, Ru, 0.4 wt.% RuNi and 2 wt.% RuNi catalysts, respectively (Supplementary Fig. 13). As shown in Fig. 2e and 2f, the curve of ln (C0/C) as a function of reaction time give a straight line starting with the origin point, indicating a pseudo first-order reaction with respect to both nitrobenzene and styrene. Monometallic Ni catalyst shows extremely low hydrogenation rate constant toward both nitrobenzene (0.03 h−1) and styrene (0.05 h−1). Notably, a preferential styrene hydrogenation activity (rate constant: 0.89 h−1) is achieved over monometallic Ru catalyst, which is greatly larger than nitrobenzene hydrogenation (rate constant: 0.18 h−1). In the case of 0.4 wt.% RuNi (SAA) catalyst, nitrobenzene hydrogenation is more dynamically favorable than styrene hydrogenation (rate constant: 1.90 h−1 vs. 0.04 h−1), indicating a preferred hydrogenation of −NO2 group rather than C = C bond. For 2 wt.% RuNi catalysts, the rate constant of nitrobenzene hydrogenation (1.65 h−1) drops compared with that of 0.4 wt.% RuNi (but remains higher than monometallic catalysts), while the hydrogenation rate of styrene increases sharply (rate constant: 0.68 h−1). This further implies a significantly promoted catalytic activity and chemoselectivity toward −NO2 group over RuNi SAA with the synergistic effect between Ni and Ru.
Investigations on structure-selectivity correlations. Aberration-correction high-angle annular dark-field scanning transmission electron microscopy (AC–HAADF–STEM) imaging technique and energy-dispersive spectroscopy (EDS) elemental mapping were conducted to intuitively ascertain the dispersion state of Ru and Ni species. EDS elemental mapping of 0.4 wt.% RuNi sample (Fig. 3a, b) illustrates a highly uniform dispersion of Ru on Ni nanoparticles rather than Al2O3 substrate; corresponding AC–HAADF–STEM image (Fig. 3c) shows that a number of atom-sized bright spots (highlighted by the blue arrows) attributed to individual Ru atoms are distributed on the surface of Ni NPs, without observation of Ru clusters or nanoparticles. Enlarged STEM image along with the intensity profile further verifies that Ru atoms are atomically dispersed on Ni NPs (Fig. 3d), demonstrating the formation of RuNi single atom alloy. For the 0.6 wt.% RuNi sample (Fig. 3e), both Ru single atoms and few Ru ensembles are observed on Ni NPs, suggesting a slight aggregation of Ru atoms due to the increase in loading. In the case of 2 wt.% RuNi sample, a number of large Ru clusters are detected on Ni NPs in HAADF–STEM image (Fig. 3f) and EDS elemental mapping (Supplementary Fig. 14), indicating a considerable aggregation of Ru.
In situ DRIFTS spectra of CO adsorption (Fig. 4a) was performed to further explore structural details of 0.4 wt.%, 0.6 wt.% and 2 wt.% RuNi samples, in comparison with Ni/Al2O3 and Ru/Al2O3 samples. CO adsorption on Ru/Al2O3 sample produces dominant peak at 2060 cm− 1 and another broad peak at 1849 cm− 1, which are assigned to linearly adsorbed CO at Ru and bridged-bonded CO at two contiguous Ru atoms, respectively39,40. For monometallic Ni sample, only a strong peak at 2057 cm− 1 is observed, which is attributed to the linear CO on Ni site41,42. In the case of RuNi samples, in addition to the main peak at 2056 cm− 1, an obvious shoulder peak centered at ~ 2030 cm− 1 is observed. By using Gaussian peak fitting method, the profiles of RuNi samples within 2000−2100 cm− 1 are reasonably deconvoluted and fitted to two peaks including CO linear-type adsorption on Ni site (2056 cm− 1) and Ru site (2031−2039 cm− 1) (Fig. 4b). Interestingly, a remarkable red shift of the linear-bonded CO peak on Ru site occurs from Ru/Al2O3 (2060 cm− 1) to RuNi samples (2031–2039 cm− 1), and the red shift gradually became pronounced with the decrease in Ru content. This indicates an enhanced electronegativity of Ru species in RuNi samples owing to the electron transfer from Ni to Ru. Furthermore, for the 0.4 wt.% RuNi samples, no bridge adsorption peak is observed, implying that Ru species exists as isolated atoms (SAA) by Ni. In the case of 0.6 wt.% RuNi and 2 wt.% RuNi samples, the appearance of bridge-bonded CO confirms the existence of Ru multimer or clusters, which accords well with the AC–HAADF–STEM results.
The normalized Ru K-edge in situ X-ray absorption near-edge structure (XANES) spectra were conducted to study the electronic structure of RuNi samples in contrast to Ru foil and RuO2 samples. As shown in Fig. 4c, the intensity of white line shrinks gradually from Ru/Al2O3 to RuNi samples along with the decrease of Ru loading. Simultaneously, the white line intensity of Ni K-edge XANES spectra shows a slight increase from pristine Ni to RuNi samples (Supplementary Fig. 15). This indicates the electron transfer from Ni atoms to Ru atoms which results in an enhanced electronegativity of Ru (Ruδ−) 43. The Bader charge analysis of RuNi SAA (111) surface reveals that isolated Ru atom carry negative charges from sub-surface Ni atoms (Supplementary Fig. 16), which is in accordance with the in situ DRIFTS and XANES spectra results. The Fourier transforms of the extended X-ray absorption fine spectrum (EXAFS) in the R space was performed to investigate the detailed coordination structure of Ru (Fig. 4d). The 0.4 wt.% RuNi sample exhibits a sharp peak located at ~ 2.1 Å, which is in the region between Ru–O shell (1.5 Å) and Ru–Ru shell (2.4 Å) and can be assigned to the Ru–Ni coordination44,45. The EXAFS data-fitting results (Supplementary Table 4, Supplementary Fig. 17–20) manifest that the coordination number (CN) of Ru–Ni path is 5.4, while the Ru–Ru and Ru–O coordination are absent in 0.4 wt.% RuNi sample. This verifies that predominant Ru atoms are atomically dispersed and surrounded by Ni to form single atom alloy. Notably, the peak intensity of Ru–Ru coordination (~ 2.4 Å) arises in 0.6 wt.% RuNi sample and enhances in 2 wt.% RuNi sample. Exactly, the EXAFS curve fittings reveal that the Ru–Ru coordination number distinctly increases from 0 (0.4 wt.% RuNi) to 2.2 (0.6 wt.% RuNi) and further to 6.7 (2 wt.% RuNi), respectively, indicating the Ru atoms aggregate gradually upon increasing Ru loading. The k2-weighted wavelet transform (WT) for the Ru K-edge XAFS signals were further employed to study coordination environment of Ru species. As shown in Fig. 4e, the lobe of Ru foil at 2.3 Å, 9.7 Å−1 is ascribed to Ru–Ru contribution. The absence of lobes at k = 9.7 Å−1 and 5.3 Å−1 in 0.4 wt.% RuNi sample (Fig. 4f) excludes the central Ru bonds to Ru atom and O atom (Fig. 4g); meanwhile, a new lobe attributed to Ru–Ni contribution is observed at 2.1 Å, 7.1 Å−1, demonstrating the formation of SAA. Moreover, Ni K-edge EXAFS spectrum of different samples does not show significant differences (Supplementary Fig. 21), indicating that the coordination structure of Ni is less affected by the surface Ru atoms. In brief, the results of AC–HAADF–STEM, in situ CO–DRIFTS as well as in situ XAFS experiments substantiate the successful preparation of RuNi SAA catalyst, as well as the evolution of Ru species on Ni NPs from single atom to Ru multimers by increasing Ru loading. In addition, the HAADF–STEM image (Supplementary Fig. 22) and EXAFS spectrum (Supplementary Fig. 23) of the used RuNi SAA catalyst display that the dispersion state and coordination structure of Ru atoms do not show obvious change after 5 cycles, demonstrating a high stability of RuNi SAA catalyst.
Mechanism insight of hydrogenation of 4-NS on RuNi SAA. H2-TPD measurement was performed to explore dissociated adsorption of hydrogen on various catalysts, and the results showed that all these catalysts give strong H2 dissociation ability with desorption peaks at ~ 100°C (Supplementary Fig. 24). DFT calculation was carried out to further study the adsorption and dissociation behavior of hydrogen on RuNi SAA in comparison with monometallic Ni (Supplementary Fig. 25). For RuNi SAA sample, H2 molecule preferentially undergoes chemical adsorption at the top site of Ru single atom with an adsorption energy of −0.76 eV. Subsequently, the hydrogen dissociates into two H active atoms at Ru–Ni hollow site of RuNi SAA overcoming an energy barrier of 0.02 eV. In addition to the Ru site, H2 molecule can also be chemically adsorbs on the top site of Ni atom adjacent to Ru (adsorption energy: −0.38 eV), and then dissociates into two H atoms at Ru–Ni and Ni–Ni hollow sites with a barrier of 0.06 eV. In contrast, the adsorption energy and dissociation energy barrier of H2 molecule on Ni (111) surface are −0.40 eV and 0.09 eV, respectively. The extremely low energy barriers and subtle difference indicate that H2 is prone to dissociate on both Ni and RuNi SAA surface (either Ni site or Ru site), which is not the key factor influencing their catalytic performance.
The activation adsorption of specific functional group is of vital significant to selective hydrogenation of substrates. Hence, we conducted DFT calculations to investigate the possible adsorption configurations and active sites of 4-NS on RuNi SAA (111) surface, and the optimized results were shown in Supplementary Fig. 26. Clearly, the parallel configuration (Fig. 5a, Supplementary Fig. 26e), in which N atom is located at the Ru–Ni hollow site with two O atoms attaching on the Ru–Ni interface, displays the lowest level of adsorption energy (−3.29 eV). This indicates a more thermodynamically favorable adsorption mode in comparison to vertical and other parallel configurations (Supplementary Fig. 26a–d). Moreover, after the adsorption of 4-NS on the RuNi SAA (111), the bond length of N–O1 (L(N−O1)) in nitro-group is extended to 1.333 Å (Fig. 5a), which is significantly longer than that of gaseous 4-NS molecule (1.244 Å) (Supplementary Fig. 27) and even exceeds the L(N−O1) on Ni (111) (1.319 Å). This suggests a prominently enhanced activation adsorption of N–O1 bond onto RuNi SAA surface, in comparison with the pristine Ni (111).
To further verify the adsorption behavior of 4-NS, in situ FT-IR measurements were performed over monometallic Ni, monometallic Ru, 0.4 wt.% RuNi (SAA) and 2 wt.% RuNi catalysts (Fig. 5b). The FT-IR spectra of gaseous 4-NS shows three characteristic peaks located at 1603, 1532 and 1356 cm− 1, attributing to ν(C = C), νas(NO2) and νs(NO2), respectively46–48. In the presence of monometallic Ni, a red-shift of all these three bands is observed: v(C = C) (1595 cm− 1), νas(NO2) (1520 cm− 1) and νs(NO2) (1351 cm− 1), indicating both C = C and −NO2 groups undergo chemisorption on the surface of Ni (most likely parallel adsorption configuration). For monometallic Ru sample, a more pronounced red-shift of these three bands is found (1591, 1514 and 1348 cm− 1, respectively) relative to monometallic Ni, indicating a slightly enhanced adsorption of nitro-group and C = C bond on Ru. In the case of 0.4 wt.% RuNi (SAA), the ν(C = C) band is centered at 1596 cm− 1, rather close to monometallic Ni (1595 cm− 1), indicating vinyl tends to adsorb at the Ni sites of RuNi SAA. In addition, in striking contrast to the monometallic catalysts, the relative intensity of νs(NO2) band (1347 cm− 1) is significantly weaker than that of νas(NO2) (1512 cm− 1) on 0.4 wt.% RuNi (SAA), which is attributed to the cleavage of N–O bond in −NO2 to generate nitroso intermediate according to previous studies49,50. A similar phenomenon was also observed in 2 wt.% RuNi sample. The results above confirm that the synergy of bimetallic sites in RuNi SAA promotes the polarization and dissociation of nitro group, which accords well with the optimal adsorption configuration obtained by DFT calculations.
In situ XAFS spectroscopy was performed to investigate the dynamic evolution in electronic structure and coordination state of RuNi SAA during both 4-NS adsorption and catalytic reaction process. As shown in the Ru K-edge XANES spectrum (Fig. 5c), the absorption edge shifts remarkably to higher energy after adsorption of 4-NS on 0.4 wt.% RuNi SAA, accompanied with an enhanced white line intensity, indicating the Ru species undergoes partial oxidation due to the electron transfer from Ru to 4-NS. Furthermore, the phase correction EXAFS Fourier–transform spectra (Fig. 5d) of 4-NS adsorbed RuNi SAA show distinctly increased signals at ~ 1.7 Å and ~ 3.2 Å, which are ascribed to the formation of Ru–O coordination. After the introduction of H2 for 30 min, both the XANES and EXAFS spectra of 0.4 wt.% RuNi SAA recover to their original state. These phenomena indicate that the nitro-group adsorbs on Ru site based on electronic interaction via Ru–O bond. As for the Ni-K edge in situ XAFS spectroscopy (Supplementary Fig. 28), no obvious change is observed, which is possibly due to the dilutive effect of Ni atoms in the bulk phase on the surface Ni species. The projected density of states (PDOS) of RuNi SAA (Fig. 5e) show that the unoccupied states of nitro group over Femi level widen in adsorption state and shift below the Femi level, which further manifests electron transfer from d-states of RuNi SAA to the unoccupied states in −NO2. Charge density difference (Supplementary Fig. 29) and Bader charge analysis (Supplementary Fig. 30) display that the integral electrons density of 4-NS is accumulated after its adsorption on RuNi SAA (111) surface. Moreover, compared with the vinyl group in 4-NS, the charge density of −NO2 enhances more significantly by virtue of the electron transfer from Ru–Ni interface, indicating in the specific activation of the N–O bond in nitro group.
In situ FT-IR was performed to monitor the catalytic reaction process of 4-NS hydrogenation on catalyst samples (Fig. 6). For pristine Ni catalyst (Fig. 6a), only a slight decrease in band density of nitro groups (1561 and 1332 cm− 1) and C = C (1595 cm− 1) was observed along with flowing H2 for 12 min, indicating a rather weak activation for both −NO2 and C = C. In the presence of monometallic Ru (Fig. 6b), the ν(C = C) band (at 1591 cm− 1) shrinks rapidly within 0−12 min whist the bands of νs(NO2) and νas(NO2) decline very slowly. This manifests the preferential activation hydrogenation of vinyl-group, resulting in the formation of 4-nitroethylbenzene (4-NE). In the case of 0.4 wt.% RuNi (SAA) (Fig. 6c), with the inflow of hydrogen, the band of νs(NO2) (1347 cm− 1) disappears rapidly accompanied with a gradual recession of ν(N = O); whilst a new band centered at 1629 cm− 1 ascribed to N–H bending vibration is observed. Moreover, the band of ν(C = C) (1596 cm− 1) merely displays a rather slight decline after 12 min. This dynamic variation verifies the polarization and hydrogenation of −NO2 to −NH3 over 0.4 wt.% RuNi (SAA), in accordance with the unique chemoselectivity toward 4-aminostyrene. For 2 wt.% RuNi sample (Fig. 6d), in addition to the appearance of a new band attributed to δ(N–H) at 1628 cm− 1, the band of ν(C = C) decreases and fades away preferentially compared with the bands of nitro group (1561 and 1332 cm− 1). This indicates both C = C and −NO2 are effectively activated on account of the existence of Ru−Ru and Ru−Ni interface sites, respectively, corresponding to the formation of 4-NE and 4-AE over 2 wt.% RuNi catalyst.
The potential conversion paths of 4-NS over Ni (111) and RuNi SAA (111), including the hydrogenation of C = C group and cleavage of N−O bond were investigated by DFT calculations, so as to further reveal the decisive role of Ru−Ni interfacial sites for the target production of 4-AS. For Ni (111) surface (Fig. 6e), the energy barriers of N−O scission and the first hydrogenation step in vinyl group are 0.78 and 0.69 eV, respectively, demonstrating that 4-NS molecule preferentially undergoes C = C hydrogenation over monometallic Ni surface. In contrast, the energy barrier of N−O dissociation is merely 0.47 eV over RuNi SAA (111), much lower than that of C = C bond hydrogenation (0.72 eV) (Fig. 6f). This unambiguously confirms that the Ru−Ni interface sites effectively reduce the energy barrier of N−O scission in nitro group, resulting in the thermodynamically preferential reduction of nitro than C = C group on the surface of RuNi SAA, which is in accord with the consequence of experiments.
The reaction mechanism and computational energy profile (Fig. 7) of 4-NS hydrogenation on RuNi SAA were studied based on DFT calculations, and the detailed adsorption configurations for elementary steps were displayed in Supplementary Fig. 31. To optimize computational procedure, the formation of water molecule was excluded from the slab model calculations9,17. As shown in Fig. 7, firstly, the 4-NS molecule experiences parallel absorption on RuNi SAA with the two O atoms in nitro group bond to Ru–Ni interface sites (S0; step I). Subsequently, the N–O bond connecting with Ru site undergoes breakage, accompanied with the generation of nitroso intermediate (C8H7NO*) on Ru–Ni interfacial sites and O* on Ni hollow site (S0 → S1 via TS1; energy barrier: 0.47 eV; step II). The hydrogenation energy barriers of C8H7NO* intermediate (Supplementary Fig. 32, 33) indicate that the active H atoms from Ru–Ni hollow sites preferentially attack O atom (energy barrier: 0.76 eV) rather than N atom (energy barrier: 1.03 eV), giving rise to C8H7NOH* (S2 → S4; step III). Afterwards, the N atom in C8H7NOH* is hydrogenated to produce C8H7NHOH* after overcoming a barrier of 1.05 eV (S4 → S6; step IV), which acts as the rate-determining step of the whole reaction. The N–OH bond in C8H7NHOH* further experiences scission on Ru–Ni interface sites (S6 → S7; energy barrier: 0.43 eV; step V), followed by the hydrogenation of N atom on Ru–Ni hollow sites to yield C8H7NH2* (4-AS) with an energy barrier of 0.68 eV (S8 → S10; step VI). Finally, desorption of 4-AS occurs from the RuNi SAA surface and the catalyst recovers to its original state (step VII). This unique interfacial structure between Ru single atom and Ni substrate, accompanied by host-dopant synergistic effect in RuNi SAA, leads to the outstanding activity and chemoselectivity for 4-NS hydrogenation.