Palladium-on-gold nanoparticles (Pd/Au NPs) have proven very effective in the catalytic breakdown of trichloroethylene, a carcinogenic groundwater pollutant. In particular, studies have revealed that these bimetallic NPs are more active than the associated monometallic NPs [1]. Our group is modeling the temporal evolution of these nanoparticles at room temperature in order to determine what types of catalytically active surface structures may form over time. Modeling the NP structures and surface compositions is important to understanding why these Pd/Au NPs have higher catalytic activity than Au or Pd alone.
Pd/Au NPs are synthesized by a chemical reaction in which Pd is deposited on the surface of a Au nanoparticle. Thus, we assume a core/shell bimetallic structure by modeling these NPs as gold "magic clusters" of closed shells that are cuboctahedral in nature with an outer cuboctahedral Pd shell. The evolving cluster is simulated via application of the temperature-accelerated dynamics (TAD) [2] technique.
Figure 1. Snapshots of a 147-atom bimetallic cluster at three different times in a TAD simulation at 300K. Initially, the cluster features face-centered cubic cuboctahedral symmetry (top left). The Au and Pd atoms are colored blue and red respectively. Two distinct phenomena are observed: 1) cluster shape transformation (top right) and 2) interdiffusion of Au and Pd (bottom).
We have focused on 147- and 309-atom clusters at room temperature (see Figure 1). For the 147-atom cluster, the structure transformed very rapidly - on the order of 10 picoseconds - from the starting face-centered cubic cuboctahedral structure to a five-fold icosahedral symmetry. Subsequently, Au/Pd interdiffusion occurred on a timescale of microseconds with gold atoms diffusing toward the surface while palladium atoms moved to the bulk.
For the 309-atom cluster, no shape transition was seen. For this cluster, only interdiffusion of the gold and palladium atoms was observed on a time scale of microseconds. A nudged elastic band (NEB) [3] analysis of the minimum energy paths (MEPs) associated with the cuboctahedral to icosahedral transition outlined in the above paragraph explains why the transformation was not observed for the 309-atom cluster. As noted by A. L. Mackay [4], the rearrangement of an icosahedral packing of spheres occurs via transformation into a cuboctahedral packing.
Figure 2. Snapshots of a 309-atom bimetallic cluster at two different times in a TAD simulation at 300K. The Pd/Au NPs are idealized as cuboctahedral clusters (left). The Au and Pd atoms are colored blue and red respectively. Interdiffusion of both Au and Pd (right) is observed during the course of the simulation.
Figure 3. Minimum energy paths for the rearrangements of an icosahedral packing of atoms. The symbols i, Ei, and EO_h refer to the NEB state and its associated energy. It should be noted that we have assigned a value of 0.0 eV to the energy of the cuboctahedral cluster (Eo_h). We examine four separate cases for each cluster size: pure Au, pure Pd, Pd core with Au shell, and Au core with Pd shell. Top: Rearrangement of the 147-atom icosahedral cluster. Bottom: Rearrangement of the 309-atom icosahedral cluster.
Figure 3 shows the minimum-energy paths for rearrangements of 147- and 309-atom icosahedral clusters. A careful examination of the green curves of the two plots reveals an activation energy barrier of 0.12 eV and 2.38 eV for the 147- and 309-atom Pd/Au clusters respectively. The 0.12 eV activation barrier of the 147-atom cluster is easily traversed at room temperature. However, the 2.38 eV activation barrier of the 309-atom cluster is too large for a thermally activated shape transformation to occur. The dramatic difference in relative stability of the icosahedral and cuboctahedral shapes for the four different cases - pure Au, pure Pd, Pd core with Au shell, and Au core with Pd shell - results from the strain inherent in the icosahedral particle shape. This strain is partially relieved by placing large Au atoms in the outer shell and Pd atoms in the inner core.
[1] M.O. Nutt, K. N. heck, P. Alvarez, and M.S. Wong, Appl. Catal. B: Environ. 69, 115 (2006)
[2] M. R. Sorensen and A. F. Voter, J. Chem. Phys., 112, 9599 (2000).
[3] H. Jonsson, G. Mills, and K. W. Jacobsen, in Classical and Quantum Dynamics in Condensed Phase Simulations, edited by B. J. Berne, G. Ciccotti, and D. F. Coker (World Scientific, Singapore, 1998).
[4] A. L. Mackay, Acta. Cryst. 15, 916 (1962).
