Hydrogen Dissociation and Spillover on Individual Isolated Palladium Atoms
Heather L. Tierney, Ashleigh E. Baber, John R. Kitchin, and E. Charles H. Sykes, “Hydrogen Dissociation and Spillover on Individual Isolated Palladium Atoms”, Physical Review Letters 103(24), 246102 (2009). https://doi.org/10.1103/physrevlett.103.246102
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Alloying a small amount of a reactive metal into a cheaper, more inert host is often described as simple dilution, with the minority element doing the chemistry. Hydrogen dissociation is a key step in CO2 hydrogenation, water gas shift and hydrogen separation, and it is slow and activated on Cu but facile on Pt-group metals. The paper asks whether single, isolated Pd atoms embedded in an inert host can dissociate H2 and let the resulting H atoms spill over onto the host, and whether the identity of the host (Cu versus Au) changes the outcome (Tierney et al. 2009).
Approach
The work combines low-temperature scanning tunneling microscopy (STM) experiments with density functional theory (DFT) calculations. Isolated Pd atoms were made by vapor-depositing 0.01 monolayer of Pd onto Cu(111) and Au(111) during controlled annealing, chosen to give roughly equal amounts of surface and subsurface Pd. Surfaces were then exposed to 10 L of H2 at 420 K and imaged at low temperature (down to 7 K). On the theory side, the heat of formation of 21 clean surface alloy configurations was computed for each host, and H adsorption was modeled in the fcc hollow on a 3x2 surface cell with four layers, using the PW91 functional, with PBE, revPBE and RPBE evaluated non-self-consistently to gauge functional sensitivity. Three sites were compared: S1 (an fcc site touching the Pd atom), S2 (the neighboring fcc site on host atoms only) and the pure host surface. Free energies were obtained with standard atomistic thermodynamics, and dissociation and nudged elastic band diffusion barriers were estimated (Tierney et al. 2009).
Key findings
STM showed that Pd forms isolated atoms in both hosts with no Pd-Pd dimers at low coverage. In Cu(111), alloying starts at ascending step edges, forming Pd-rich brims, while in Au(111) Pd place-exchanges at the elbows of the herringbone reconstruction before dispersing. Surface Pd images as a protrusion and subsurface Pd as a depression. DFT confirmed that Pd alloys exothermically into both surfaces and that isolated Pd surrounded by host atoms is most stable below a Pd fraction of about 0.4 (Tierney et al. 2009).
Bare Cu(111) showed no H uptake after the H2 dose, but the Pd/Cu(111) surface carried a large amount of hydrogen that clustered into islands on pure Cu regions, with H coverage decreasing away from the Pd atoms, consistent with dissociation at Pd followed by spillover. Pd/Au(111) with the same Pd loading showed no adsorbed hydrogen (Tierney et al. 2009).
With PW91, H adsorption relative to gas-phase H2 is favorable by about 0.24 eV/H at the Pd S1 site in Cu(111), and by about 0.14 eV/H at S2, slightly weaker than on Cu(111) itself. In Au(111), H binds favorably at Pd (about 0.15 eV/H, although RPBE gives a slightly unfavorable +0.01 eV/H), but S2 is endothermic at +0.07 eV/H, so recombination and desorption of H2 is preferred over spillover. The functionals agree on trends and differ by 0.09 to 0.16 eV/H per site. Including zero-point energy, pressure and temperature, dissociative adsorption at 420 K becomes highly unfavorable on the Au host but only slightly unfavorable on the Cu host, and favorable on Cu below 420 K. The H2 dissociation barrier over an isolated Pd atom in Cu(111) was estimated at only 0.02 eV (the authors note the method may underestimate it), compared with 0.5 eV calculated (0.3 to 0.4 eV measured) for Cu(111). Diffusion barriers for H moving from weaker to stronger binding sites were 0.12 to 0.15 eV. Two spillover pathways onto Cu were both judged feasible (Tierney et al. 2009).
Note: the extracted text loses some minus signs in the adsorption energies; the signs above follow the paper’s own description of each site as favorable or unfavorable.
Significance
The study shows that the host metal is not a passive diluent: an identical array of isolated Pd atoms is active for H2 dissociation and spillover in Cu but not in Au, because the host controls whether spilled-over H is thermodynamically stable. The authors argue that the same theoretical survey can be applied to other host and dopant combinations to design bifunctional alloy catalysts that dissociate otherwise unreactive molecules (they mention O2 on Au as an example) with tunable adsorption energies and spillover barriers (Tierney et al. 2009).