UC Berkeley

Bimetallic nanoparticles can lead to catalysts with improved turnover rates and selectivities, but many synthetic protocols, such as impregnation or precipitation, typically form particles of non-uniform size and composition. Colloidal methods may be able to improve their uniformity, but often require reagents that poison catalytic surfaces (ex. S, B, P). Such compositional non-uniformity and ubiquitous impurities have prevented rigorous conclusions about the consequences of alloying on reactivity and selectivity. Herein, we describe a sequential galvanic displacement-reduction (GDR) colloidal synthesis method using precursors containing only C, H, O, and N, that leads to bimetallic AuPt and PtPd nanoparticles narrowly distributed in size and composition.

Au3+ or Pt4+ precursors were added to monometallic Pt or Pd clusters, respectively, whose surface atoms are thermodynamically driven to reduce and deposit the solvated cations onto cluster surfaces due to the lower reduction potentials (E0) of the seed metal relative to the precursor cations (E0Au > E0Pt > E0Pd); oxidized Pt and Pd surface atoms subsequently return to cluster surfaces upon reduction by the solvent, a reductant (ethanol or ethylene glycol, respectively). Such methods have been previously used to synthesize AuPd clusters from Pd seed clusters.

TEM micrographs confirm that initial seed cluster sizes increase monotonically with increasing Au3+ or Pt4+ content, with final bimetallic cluster dispersity values near unity indicating a narrow size distribution. UV visible spectroscopy of AuPt cluster suspensions show no plasmon resonance features characteristic of Au nano-sized surfaces, indicating the presence of Pt atoms at bimetallic surfaces, as expected for GDR processes. Elemental analysis by EDS confirmed the formation of strictly bimetallic particles with the mean composition of the synthesis mixture.

The GDR model requires that bimetallic growth be proportional to the initial seed surface area, with the number of precursor atoms deposited per surface metal atom of the seed constant and independent of seed metal size. Elemental analysis using EDS supports this hypothesis for thermodynamically favorable alloys such as PtPd and AuPd, but not for AuPt, an unfavorable alloy. These differences appear to reflect the segregation of metals within AuPt clusters during synthesis, placing the metal with the lower surface energy, Au, at cluster surfaces, and decreasing the availability of Pt0 surface atoms for GDR. Consequently, autocatalytic Au3+ reduction on Au0 sites becomes a competitive Au3+ reduction pathway during the synthesis of AuPt clusters.

Polymers such as polyvinylpyrrolidone (PVP)—which bind to metal surfaces during synthesis via charge-transfer interactions—were required in colloidal suspensions to prevent particle agglomeration in solution, but must be removed prior to catalysis. We show that after depositing clusters on SiO2, PVP can be removed from particle surfaces by post-synthetic treatments at mild temperatures (≤ 423 K) in reductants such as H2 and/or EtOH without significant particle agglomerations. Reductants compete with the polymer at the metal surface, thus breaking the polymer-metal bond. The absence of surface residues was confirmed by the similar cluster sizes derived from O2 chemisorption and TEM measurements. Larger cluster sizes and surfaces that chemisorb oxygen more weakly—such as Pt relative to Pd—were found to facilitate the removal of PVP from metal particles due to weaker metal-polymer bonds.

The model catalytic materials prepared in this study are of both fundamental and practical interest to probe the effects of alloying. Using AuPd and AuPt, we investigate the consequences of alloying with Au on the reactivity of catalyst surfaces saturated with either chemisorbed CO* (CO oxidation) or O* (H2 oxidation) that bind strongly to Pt and Pd surfaces and inhibit rates. Singleton Pt-CO* bond energies, reflected in vibrational CO* stretches, were decoupled from dipole-dipole coupling effects using isotopic dilution methods, and were shown to decrease with increasing catalyst Au content. Despite lower CO* binding energies, CO oxidation turnover rates (normalized per metal surface atom) on AuPt catalysts decreased with increasing Au content. These results show that CO oxidation rates depend weakly on CO* binding energy—consistent with the reported structed insensivity of this reaction—and that Au acts primarily as an inert diluent of the active Pt ensembles required for catalysis.

In contrast, H2 oxidation turnover rates (normalized per metal surface atom) on AuPt and AuPd catalysts increase with increasing Au content (up to 11 % at. Au content on AuPt and up to 67 % at. Au content on AuPd), indicating that the reactivity of O* saturated surfaces is more sensitive to changes in adsorbate binding energy than surfaces saturated in CO*, consisted with the reported structure sensitivity of reactions on O* saturated surfaces. Reconstruction of CO* adlayers is facile due to highly mobile CO* molecules, thus allowing CO* adlayers to access configurations that help mitigate strong CO* binding and introduce vacancies. O* adlayers, meanwhile, are more strongly bound to Pt and Pd metal surfaces and less mobile. H2 oxidation rates thus depend more strongly on adsorbate binding energy than CO oxidation rates.