A monometallic Au/SiO₂ catalyst was prepared by the met­hod in our previous report ^[19]. On a commercial silica support (Qingdao Ocean Chemical Plant, S_BET = 467 m²/g) was grafted 3-aminopropyltriethoxysilane (APTES, 99%, Acros Organics) by refluxing in ethanol at 80 °C for 24 h, followed with washing by ethanol and drying in air. The resulting material was denoted as SiO₂-APTES. To prepare the Au-Ni bimetallic alloy catalysts, 1 g of SiO₂-APTES was added into 20 mL HAuCl₄ solution (0.01 mol/L) and stirred for 30 min. After filtration and washing with water, the recovered solid was dispersed in 10 mL water, and 10 mL NaBH₄ solution (0.2 mol/L) was added dropwise under vigorous stirring for the reduction of AuCl₄^-. After stirring for 15 min, the solid was recovered by filtration and thoroughly washed with water to obtain 4.6 wt% Au/SiO₂. Then the Au/SiO₂ was re-dispersed into 0.05 mol/L Ni(NO₃)₂ solution with a defined Au/Ni ratio, followed by the addition of 0.1 mol/L tert-butylamine borane solution to reduce Ni²⁺. After filtration, washing with water, and drying at 80 °C for 12 h, the solid was calcined at 500 °C in air for 6 h and reduced at 550 °C in H₂ for 1 h to obtain the AuNi_x/SiO₂-cal and AuNi_x/SiO₂- cal-red catalysts, respectively (x refers to the atomic ratio of Au to Ni). For comparison, a Ni/SiO₂ catalyst was prepared by incipient wetness impregnation of the SiO₂ support with a Ni(NO₃)₂ solution, followed by the same calcination and reduction treatment as that for the Au-Ni/SiO₂ catalysts.

The catalytic testing for the chemoselective hydrogenation of nitroarenes was conducted in an autoclave equipped with a pressure gauge, magnetic stirring system, and water bath. A mixture of nitroarene substrate, solvent, and an internal standard of o-xylene with a total volume of 5 mL was put into the autoclave. Then, the autoclave was flushed with 10³ kPa hydrogen 5 times. After sealing, the autoclave was charged with H₂ to 300 kPa, then heated to 50 °C in a water bath with stirring to initiate the reaction. After reaction, the product was analyzed by GC-MS.

The Au and Ni loadings were measured with an inductively coupled plasma spectrometer (ICP-AES) on an IRIS Intrepid II XSP instrument (Thermo Electron Corp.). XRD analysis was determined on a PANalytical X’ pert diffractometer equipped with a Cu K_α radiation source operated at 40 kV and 40 mA. Temperature-programmed desorption of H₂ (H₂-TPD) was performed on an Auto Chem II 2920 automatic catalyst charact­erization system. TEM was carried out using a Tecnai G2 Spirit (FEI) microscope operating at 200 kV. High resolution transm­ission electron microscopy (HRTEM) was conducted with a Tecnai G2 F30S-Twin (FEI) microscope operating at 300 kV. The chemical composition of individual particles were measured by energy dispersive X-ray spectroscopy (EDS) on a Tecnai G2 F30S-Twin (FEI) electron microscope. X-ray absorption near edge structure (XANES) and EXAFS spectra at the Au LIII edge and Ni K edge were recorded at beamline 14W of Shanghai Synchrotron Radiation Facility, Shanghai, China.

It is a challenging task to form a homogeneous Au-Ni alloy due to the large difference in the reduction potential and immiscibility of the two metals at low temperature ^{[31, 32, 33]}. In our earlier work, we developed a two-step method for the preparation of Au alloy nanoparticles on a silica support including Au-Ag ^{[19, 20, 21, 22]}, Au-Cu ^{[23, 24]} and Au-Pd ^[27], in which NaBH₄ was used as the reducing agent in both reduction steps. However, a similar procedure failed in the preparation of Au-Ni alloy nanoparticles because Ni nanoparticles were prefere­ntially formed in the solution rather than on the preformed gold particle surface. Motivated by the noble metal induced reduction (NMIR) method developed by Li and coworkers ^[34], we modified the two-step method by using a weak reducing agent, tert-butylamine borane, in the second step, which allowed the preferential deposition of Ni nanoparticles on the gold surface based on the NMIR mecha­nism. The preparation procedure is illustrated in Fig. 1. The following calcination and reduction treatments were to remove APTES and to promote the formation of a homogeneous Au-Ni alloy phase ^[19].

Fig. 1. Schematic of the procedure for the synthesis of SiO2-supported Au-Ni alloy nanoparticles.

Fig. 2(a) presents the XRD patterns of the catalyst samples with different Au/Ni ratios. The monometallic Au/SiO₂ sample presented four broad peaks at 2θ = 38.4°, 44.2°, 65.1°, and 77.8° which were assigned to the reflections of the (111), (200), (220), and (311) lattice planes of Au, respectively. The broad peaks indicated the very small size of the Au particles. Using the Scherrer equation with the Au(111) peak, the average particle size of gold was calculated to be 3.3 nm. For the other three Au-Ni_x/SiO₂-cal-red samples, the reflection peaks were broader than that of the monometallic Au, indicating that the presence of a second metal Ni suppressed the sintering of the particles during high temperature calcination and reduction. This trend was in agreement with the Au-Ag ^{[19, 20, 21, 22]} and Au-Cu ^{[23, 24]} systems. More interestingly, the reflection peaks of the Au-Ni bimetallic samples were shifted to larger degree with an increase in Ni content, as highlighted in the narrow range XRD patterns (Fig. 2(b)). This provided strong evidence for the formation of the Au-Ni alloy phase. It was also noted that there was no peak assignable to the metallic Ni phase, even for the Ni-rich AuNi₃ sample. However, the absence of a monometallic Ni phase in the XRD pattern does not mean that all the Ni species were in the Ni-Au alloy phase because they could exist as highly dispersed Ni that cannot be detected by XRD.

Fig. 2. XRD patterns in a wide range (a) and in 2θ = 30°-50° (b) of the AuNix/SiO2-cal-red catalysts.

The TEM images in Fig. 3 provided more details of the particle size distribution. The metal particles in all the samples were uniformly dispersed on the silica support. No very big particles were observed in the different regions of the samples. The histograms in Fig. 4 showed that the average particle size of Au, AuNi_0.33, AuNi₁, and AuNi₃ were 3.69, 3.44, 3.28 and 3.41 nm, respectively, which was basically in agreement with that estimated from the XRD data. EDS analysis of individual particles revealed that the chemical composition (Au/Ni ratio) differed from one particle to another although most particles contained both Ni and Au (Fig. 5). Heterogeneity in composition is ubiquitous for bimetallic particles ^{[35, 36, 37]}, which in turn affects the catalytic performance.

Fig. 3. TEM images of the AuNix/SiO2-cal-red catalysts.

Fig. 4. Particle size distributions of the AuNix/SiO2-cal-red catalysts.

Fig. 5. EDS analysis of individual particles of the AuNi3/SiO2-cal-red catalyst.

EXAFS is a powerful technique for identifying the formation of an alloy phase ^{[38, 39]}. Hence, we performed in situ EXAFS on the AuNi₃/SiO₂ sample. To follow the formation of the Au-Ni alloy phase, both the calcined and calcined-reduced samples were characterized. The results are shown in Table 1. For the calcined sample, there were Au-Au and Au-Ni contributions in the Au L_III edge, with the coordination numbers (CNs) of 9.2 and 0.4, respectively, indicating that only a very small fraction of the Ni bonded with Au, while most of the Ni were on the surface of Au particles or existed as a separate monometallic phase. The same conclusion was reached from the result of the Ni K edge. There were three neighboring shells to Ni, which were Ni-Au, Ni-Ni and Ni-O with CN of 0.5, 9.4, and 4.9, respectively, in the AuNi₃-cal sample. From these results, we concluded that most of the Ni species were NiO on the surface of the Auparticles as a consequence of the high temperature calcination. After reduction in H₂ at 550 °C, the Au@NiO hetero-structure changed greatly. At the Au L_III edge, the CN of the Au-Au shell decreased from 9.2 to 5.0, while the CN of the Au-Ni shell increased from 0.4 to 3.4, which was indicative of much more alloying. In agreement with this, the Ni-O coordination number showed a clear decrease after reduction. As a consequence, the CN of Ni-Au and Ni-Ni at a distance of 2.48 Å (characteristic of the metal bond of Ni) was much increased. The presence of the Ni-O contribution after reduction implied that a small fraction of Ni strongly interacted with the silica support to from nickel silicate which cannot be reduced. However, this does not contribute to the catalytic activity in the hydrogenation reactions. The same change was reported earlier in the Au-Cu bimetallic system ^[40].

Table 1 EXAFS results of the AuNi3/SiO2 catalysts after calcination and reduction. Edge &data fitting range Sample Shell CN R (Å) σ2 (Å2) R-factor Au LIII edge Au foil Au-Au 12.0 2.86 0.007 0.0009 AuNi3/SiO2-cal Au-Au 9.2 2.83 0.008 0.0026 Au-Ni 0.4 2.69 0.006 AuNi3/SiO2-cal-red Au-Au 5.0 2.78 0.006 0.0017 Au-Ni 3.4 2.62 0.011 Ni K edge Ni foil Ni-Ni 12.0 2.48 0.005 0.0004 NiO Ni-O 6.0 2.06 0.006 0.0016 Ni-Ni 12.0 2.95 0.006 AuNi3/SiO2-cal Ni-Au 0.5 2.69 0.001 0.0002 Ni-Ni 9.4 2.99 0.014 Ni-O 4.9 2.06 0.007 AuNi3/SiO2-cal-red Ni-Au 4.3 2.62 0.008 0.0017 Ni-Ni 2.3 2.48 0.004 Ni-O 2.7 2.06 0.007 Ni-Ni 5.8 3.03 0.013 CN, coordination number of the absorber-backscatterer pair. R, the average absorber−backscatterer distance. σ2, the Debye-Waller factor. The accuracy of the above parameters were estimated as CN, ±20%; R, ±1%; σ2, ±20%; ΔE0, ±20%. Δk and ΔR are the data range used for data fitting in k space and R space, respectively. For Au LIII edge: Δk: 3.4-12.2 Å-1, ΔR: 1.6-3.3 Å; For Ni K edge: Δk: 3.3-13.1 Å-1, ΔR: 1.3-3.0 Å.

Table 1 EXAFS results of the AuNi3/SiO2 catalysts after calcination and reduction.

To summarize, EXAFS results demonstrated that a significant fraction of Ni was alloyed with Au in the AuNi₃ sample. This alloy phase is expected to show a synergistic effect in hydrogenation reactions.

To evaluate the catalytic performance of the different Au-Ni catalysts in the chemoselective hydrogenation of nitroarenes, we first studied the chemoselective hydrogenation of 3-nitrostyrene as it is the most demanding reaction owing to the presence of the -NO₂ and -C=C groups in one molecule ^[2]. The reaction was conducted under mild conditions, reaction temperature of 50 °C and H₂ pressure of 300 kPa. Table 2 shows the results obtained with the Au-Ni_x/SiO₂ catalysts with various Au/Ni ratios. The monometallic Au/SiO₂ catalyst afforded a moderate activity and excellent selectivity to 3-vinylaniline. The conversion and selectivity reached 79.3% and 99.6% after 8 h reaction, respectively. In comparison to Au/SiO₂, the 4.11%Ni/SiO₂ catalyst was more active but less selective. The conversion and selectivity reached 93.3% and 78.7% after 108 min reaction, respectively. Similar to 4.11%Ni/SiO₂, commercial Raney Ni also gave poor selectivity. On the other hand, the Au-Ni_x/SiO₂ bimetallic catalysts gave both high activity and selectivity, demonstrating a strong synergistic effect between the two metals. In particular, with an increase of Ni content in the Au-Ni bimetallic catalysts, the activity increased remarkably while the selectivity remained above 90% at high conversions, implying that Ni contributed mainly to the improvement of the activity while Au preserved the high chemoselectivity. AuNi₃/SiO₂ showed the best performance. The conversion and selectivity were 90.8% and 93.0% after a reaction time of only 70 min. For a better comparison, we also tested a standard Au/TiO₂ catalyst under identical reaction conditions and with the same Au/substrate ratio. In agreement with the literature ^[9], the Au/TiO₂ catalyst afforded a much better performance than Au/SiO₂: it took 180 min to obtain 99.6% conversion and 99.1% selectivity. Obviously, the redox property of the TiO₂ support played an important role in this reaction. In comparison to Au/TiO₂, our AuNi₃/SiO₂ catalyst was more active but a little less selective. It can be envisioned that if it was supported on a redox support such as TiO₂ or Fe₂O₃, the Au-Ni bimetallic catalyst could give a better performance than its counterpart on the SiO₂ support.

Table 2 Chemoselective hydrogenation of 3-nitrostyrene over different catalysts. Catalyst Time (min) Conversion (%) Selectivity (%) TOF a (h-1) Au/SiO2 480 79.3 99.6 5.5 AuNi0.33/SiO2 310 90.5 96.4 19.9 AuNi1/SiO2 160 92.5 92.1 21.9 AuNi3/SiO2 70 90.8 93.0 17.9 4.11%Ni/SiO2 108 93.3 78.7 25.5 1.37%Ni/SiO2 480 17.3 97.8 2.8 0.46%Ni/SiO2 960 11.2 99.2 1.3 Au/TiO2 b 180 99.6 99.1 — Raney Ni c 155 76.8 62.5 — 4.11%Ni/SiO2+Au/SiO2 108 92.9 80.1 — Reaction conditions: T = 50 °C, P = 300 kPa, 0.1 g catalyst, 0.5 mmol substrate, toluene as a solvent, o-xylene as an internal standard. a Calculated based on the metal dispersions in Table 3. b Catalyst amount: 0.3 g for the same Au/substrate ratio as Au/SiO2. c Catalyst amount: 15 mg Raney Ni + 0.1 g SiO2 powder.

Table 2 Chemoselective hydrogenation of 3-nitrostyrene over different catalysts.

To further demonstrate the synergy between Au and Ni, we performed a control experiment with a mechanical mixture of Ni/SiO₂ and Au/SiO₂ with the same metal contents as AuNi₃/ SiO₂, which was denoted as 4.11%Ni/SiO₂+Au/SiO₂. The result showed that this physical mixture catalyst performed similarly to the monometallic 4.11%Ni/SiO₂ catalyst, giving a selectivity of 80.2% at a conversion of 92.9%. Clearly, the improvement of the catalytic performance was due to a synergistic effect betw­een the two metals since it was not a simple additive effect.

Another interesting phenomenon was that all the bimet­allic Au-Ni catalysts gave higher catalytic activity or selectivity than their monometallic Ni/SiO₂ counterparts at the same Ni content (e.g., AuNi_0.33/SiO₂ vs. 0.46%Ni/SiO₂, AuNi₁/SiO₂ vs. 1.37%Ni/SiO₂, AuNi₃/SiO₂ vs. 4.11%Ni/SiO₂). The characteriz­ation results of the bimetallic catalysts have revealed that the Au-Ni nanoalloy phase was formed in these bimetallic catalysts. Even for the AuNi₃/SiO₂ catalyst that has excess Ni, there was still a significant amount of Ni alloying with Au. The formation of the alloy provided intimate contact between Au and Ni atoms, thus both geometric and electronic effects between the two metals occurred easily. H₂ chemisorption is able to provide useful information on the synergistic effect between the two metals. Table 3 lists the H₂ uptake of the Au-Ni bimetallic catalysts and the monometallic counterpart. Monometallic Au/SiO₂ does not adsorb H₂, which is consistent with that Au is a poor catalyst for hydrogen activation ^[14]. In contrast, the bimetallic Au-Ni catalysts adsorb H₂ well. The H₂ uptakes were 15, 27, and 12 μmol/g_Cat for AuNi₃, AuNi₁, and AuNi_0.33, respectively. These H₂ uptake values are remarkably larger than that of their monometallic Ni counterparts. Therefore, the formation of the Au-Ni alloy phase allowed H₂ to be activated more easily, thus enhancing the catalytic activity for the chemoselective hydrog­enation of nitroarenes. On the other hand, previous studies on the surface science of the Au-Ni alloy showed that Au is enriched on the surface in a H₂ atmosphere due to its lower surface energy than Ni ^{[29, 41]}, which facilitates the prefere­ntial adsorption of the nitro group on the Au surface and then the chemoselective hydrogenation of nitroarenes. As Au does not adsorb H₂, the metal dispersion was estimated from the average particle size obtained from TEM, rather than from the hydrogen uptake. The results are listed in Table 3. Then, we calculated the TOF (turnover frequency) for the chemoselective hydrogenation of nitroarenes. The results are shown in Table 2. The TOF of the AuNi₃/SiO₂ catalyst was 3 fold higher than that of Au/SiO₂ while preserving the selectivity above 90%. Although the 4.11%Ni/SiO₂ catalyst afforded an even higher TOF than the AuNi₃/SiO₂ catalyst, its selectivity to 3-vinyla­niline was significantly lower. For the other two monometallic Ni catalysts, the TOF were one order of magnitude lower than the corresponding Au-Ni alloy counterparts. The promising activity and chemoselectivity of Au-Ni bimetallic catalysts can be attributed to the improved H₂ activation and preserving of the preferential adsorption of nitro groups on the gold surface due to synergy between Au and Ni.

Table 3 Physical and chemical properties of the AuNix/SiO2 and Ni/SiO2 catalysts. Catalyst Metal loading (wt%) H2-uptake (μmol/g) Dispersion a (%) Particle size b (nm) Au Ni Au/SiO2 4.54 — — 27.1 3.69 AuNi3/SiO2 4.12 3.91 15 29.3 3.41 AuNi1/SiO2 4.73 1.37 27 30.5 3.28 AuNi0.33/SiO2 4.70 0.37 12 29.1 3.44 4.11%Ni/SiO2 — 4.11 5 11.4 8.81 1.37%Ni/SiO2 — 1.37 3 16.3 6.13 0.46%Ni/SiO2 — 0.46 6 35.2 2.84 a Estimated from D = 1/d (D: dispersion; d: average diameter of particles); b Estimate based on more than 200 particles in TEM images.

Table 3 Physical and chemical properties of the AuNix/SiO2 and Ni/SiO2 catalysts.

Finally, the best performance catalyst AuNi₃/SiO₂ catalyst was evaluated for the chemoselective hydrogenation of a variety of nitroarene substrates. As shown in Table 4, nitroar­ens possessing a C=C bond, carbonyl, nitriles, and halogens were all transformed to the corresponding substituted anilines with high chemoselectivity at high conversions, demonstrating the great potential of Au-Ni bimetallic catalysts for this type of reaction.