Introduction

Bimetallic catalysts, consisting of a platinum-group metal and a late-transition metal, featured geometrical variation and electronic redistribution via metal–metal interactions and enhanced the activity and/or selectivity, strongly depending on the crystal-phase and the particle size19 or small amount (0.02–2.0 mol.%) of Pd on Cu particles (2–40 nm)5,24. PdCu bulk alloys with a majority of Pd (50–90 mol.%), typically adopting the fcc phase, promoted the selectivity towards ethylene but at the expense of activity46,57. Pd atoms served as the active sites for dissociating H2 while Cu atoms simply diluted Pd ensembles but did not interact electronically with Pd. Here, the B2 particle is characterized by the densely-populated surface Pd–Cu bond that possessed isolated Pd site with a lower coordination number, and thereby promoted the activity pronouncedly at room temperature under a stoichiometric feed gas condition. To some extents, it circumvented the problems encountered by the newly emerged single-atom alloys and the traditional bulk alloys for the selective hydrogenation of multiple carbon–carbon bonds.

In summary, the comprehensive experimental data and theoretical calculations justified that tuning the crystal-phase of bimetallic catalysts at the single-nanoparticle scale changed the atomic structure of the active sites. The densely-populated Pd–Cu bond on the chemically ordered B2 particle, featured by the low coordination numbers and the high-lying metal d-band centers, greatly facilitated H2 dissociation on the Pd atom and efficiently accommodated the activated H atoms on the top/subsurfaces, yielding a much higher activity. This finding fundamentally offers a new route to precisely tailor the geometric and electronic structures of the active sites over bimetallic particles for quantifying the structure−reactivity relationship at atomic accuracy, and practically provides an accessible approach to essentially promote the efficiency of metal catalysts by tuning their intrinsic activity via crystal-phase control at single-nanoparticle level.

Methods

Crystal-phase mediation of PdCu particles

Monodisperse PdCu colloids were prepared by reducing palladium and copper cations with ethylene glycol and using oleylamine as the cap** agent. 66.2 mg Na2PdCl4, 46.9 mg CuCl2·2H2O and 1 ml oleylamine were dissolved into 100 ml ethylene glycol at room temperature under Ar. The mixture was heated to 393 K under stirring and maintained at that temperature for 20 min; further heated to 473 K and kept at that temperature for 2 h. The PdCu colloids were dispersed into cyclohexane (250 ml).

Each PdCu colloid was then precisely coated with a thin silica shell using a water-cyclohexane reverse microemulsion method. PdCu colloids (37.4 mg), dispersed in cyclohexane (250 ml), were mixed with Triton X-100 (polyethylene glycol tert-octylphenyl ether, 80 ml) and ultra-sonically treated for 0.5 h at 293 K. Aqueous ammonia solution (29.4 wt.%, 5 ml) and aqueous hydrazine hydrate solution (80 wt.%, 5 ml) were added to mediate the pH value of the mixture to 13. 9.35 g tetraethyl orthosilicate, mixed with 250 ml cyclohexane, was added and the suspension was stirred for 1 h. The solid product was collected by centrifugation, washed with ethanol, and dried at 323 K under vacuum for 12 h, yielding silica-coated PdCu colloid.

Crystal-phase tuning was done by treating the silica-coated PdCu colloid with reactive gases (H2/O2) at 673–773 K. The ordered body-centered cubic (B2) phase, donated as B2 particle, was prepared by treating the silica-coated PdCu colloid with H2 at 673 K for 2 h. The face-centered cubic (fcc) phase, named as fcc particle, was obtained by calcining the B2 particle at 673 K in air for 4 h, followed by H2 reduction at 773 K for 2 h.

N2 adsorption–desorption isotherms revealed that the specific surface area was 177 m2 g−1 for the B2 particle while 161 m2 g−1 for the fcc particle, mainly contributed by the porous silica shell generated during the high-temperature treatments under the reactive gases (H2/O2).

STEM/ETEM analysis

STEM images were acquired over a JEOL-ARM 300 F microscope at 300 kV. Energy dispersive X-ray spectroscopy elemental map**s over the particles were collected using a JED-2300 T spectrometer. Environmental TEM observations on the particles under H2 and/or H2/C2H2 were done using an aberration-corrected Titan Themis ETEM G3 microscope at 300 kV. The sample was dispersed into ethanol, and the suspension was deposited onto a thermal E-chip that is equipped with a thin silicon nitride membrane. The sample was pretreated with H2 at 673 K for 30 min. The images were acquired at 303 K and under 1 mbar H2 or H2/C2H2 (molar ratio of 1/1). Image simulation was done using the JEMS software package and the Multi-slice module.

XAS spectra

XAFS spectra of Cu and Pd K-edges were measured at the BL14W1 beamline at Shanghai Synchrotron Radiation Facility, China. The sample (100–250 mg) was pressed into a self-supported wafer and mounted into a reaction cell, where it was treated with a 5.0 vol.% H2/N2 mixture (50 ml min−1) at 673 K for 1 h. The spectra were then recorded at 343 K. EXAFS data were processed according to the standard procedure using Athena and Artemis modules of the IFEFFIT software packages.

IR spectra

IR experiments were done with a dedicated ultrahigh vacuum (UHV) apparatus, combing a FTIR spectrometer (Bruker Vertex 80 v) and a multi-chamber UHV system (Prevac). 200 mg sample was pressed into an inert metal mesh and mounted on a sample holder. The sample was treated with H2 at 673 K for 1 h, exposed to CO at 110 K and gradually heated to 460 K at a rate of 3 K min−1. IR data were accumulated by recording 1024 scans with a resolution of 4 cm−1. Peak fittings on the spectra of CO adsorbed on Pd and Cu sites were performed by the Gaussian function.

Theoretical analysis

DFT calculations were done using the Vienna ab initio Simulation Package (VASP) code. To model the B2 and fcc particles, a 4-layer (3 × 2) B2(110) and a 4-layer (1 × 2) fcc (111) slabs were used. The bottom 2 layers were fixed in their bulk position while other metal atoms and adsorbates allowed to relax. The lattice constants of B2 and fcc phases were calculated to be 3.016 and 3.806 Å, respectively. The transition states of H2 dissociation were located using the dimer method and the convergence criteria were set to 0.05 eV Å−1. Additional computational details are reported in the Supplementary Methods.

Catalytic tests

Selective hydrogenation of acetylene was conducted with a continuous-flow fixed-bed quartz tubular reactor (inner diameter, 6 mm) at atmospheric pressure. 50 mg catalyst (40–60 mesh) was pretreated with a 5.0 vol.% H2/N2 mixture (50 ml min−1) at 673 K for 1 h. After being cooled down to room temperature (298 K), the catalyst was exposed to the reaction gas (1.0 vol.% C2H2/0.5–2.0 vol.% H2/He, 50 ml min−1) that was introduced via a mass flow controller. The outlet from the reactor was analyzed online using a gas chromatograph equipped with a thermal conductivity detector and a flame ionization detector. The reaction rate was measured by controlling the conversion of acetylene below 20% via varying the flow rate of the reaction gas or the weight of the catalyst. The activation energy was measured in the temperature range 268–318 K at H2/C2H2 ratio of 1/1–2/1. The reaction orders with respect to hydrogen and acetylene were determined at 298 K by adjusting the concentrations of H2 (0.5–3.0 vol.%) and C2H2 (1.0–3.0 vol.%) in the feed gases.