Investigating the Reactivity of Single Atom Alloys Using Density Functional Theory
Hari Thirumalai and John R. Kitchin, “Investigating the Reactivity of Single Atom Alloys Using Density Functional Theory”, Topics in Catalysis 61(5-6), 462-474 (2018). https://doi.org/10.1007/s11244-018-0899-0
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
Single atom alloys, in which isolated atoms of a reactive metal sit in the surface of a relatively inert host, are known experimentally to be active and selective, especially for hydrogenation. The authors hypothesized that this behavior comes from anomalies in the electronic structure of the isolated atom, and asked two things: whether such alloys obey the d-band model and the usual correlations that hold for other transition metal alloys, and how a single Pd atom changes the energetics of simple probe reactions relative to the pure host (Thirumalai 2018).
Approach
All calculations used VASP with PAW potentials, the PBE functional, a 350 eV plane wave cutoff and 0.1 eV Gaussian smearing. Surfaces were 3x3, three-layer slabs of Cu, Ag and Au in which one surface atom was replaced by Pt, Pd, Ir, Rh or Ni, with the bottom two layers fixed. Atom-projected d-band densities of states (Wigner-Seitz radius 1.5 Å) were computed for the dopant atom with a hydrogen atom adsorbed in a hollow site on the fcc(111), (100) and (211) facets, and the d-band center and width (first moment and square root of the second moment) were extracted. Three Pd-containing alloys (AuPd, AgPd, CuPd) were then compared with the pure hosts and with Pd(111) for water dissociation, NO dissociation, and acetylene hydrogenation to ethylene, using potential energy diagrams built from electronic reaction energies (no zero-point or thermal corrections). Water dissociation barriers were found by constrained relaxations along the OH-H distance with ASE’s BFGS optimizer (0.05 eV/Å), followed by the improved dimer method. Gas-phase O2 energies were referenced to experimental formation enthalpies of water or NO to avoid the PBE O2 error. The authors explicitly did not treat alloy stability, segregation or poisoning, assuming such sites can exist under reaction conditions (Thirumalai 2018).
Key findings
The d-band density of states of the dopant atom in the Au and Ag hosts shows a sharp peak near the Fermi level, resembling the atomic orbitals of the isolated gas-phase metal atom. The authors attribute it to poor mixing with the host electronic states and to effective tensile strain, since most dopants are smaller than Au and Ag. In Cu hosts this sharp feature is absent, which they ascribe to compressive strain broadening the band offsetting the narrowing from electronic isolation (Thirumalai 2018).
Across pure metals and single atom alloys, the d-band width and d-band center remained roughly linearly correlated, so the isolated atom still behaves approximately according to the rectangular band model, with the alloys having narrower bands closer to the Fermi level and binding H more strongly than the pure metals. However, the dissociative adsorption energy of H2 showed large scatter against the d-band center, so the reactivity of these alloys is not captured by the d-band center alone. The authors suggest the unusual electronic structure, or effects such as Pauli repulsion from the noble metal host, matters, and that single atom alloys may escape the linear scaling relations that constrain transition metal catalysts. Adsorbed H sits offset toward the Pd atom rather than at the center of the host hollow site (Thirumalai 2018).
For water dissociation, water binds atop the Pd atom, and OH, H and O bind in hollow sites next to Pd. AuPd and AgPd bind the intermediates and products about 0.5 eV more strongly than pure Au and Ag at every step, and transition states are lower, consistent with Brønsted-Evans-Polanyi behavior. CuPd is less favorable than pure Cu. Final recombination and desorption are harder on the alloys because the atoms are bound more strongly. For NO dissociation, AuPd and AgPd again improve on their hosts, but the process remains endothermic on Au, AuPd, Ag and AgPd, while Cu and CuPd give favorable pathways with CuPd less active than Cu. For acetylene hydrogenation, acetylene prefers fcc sites on all surfaces (atop is less stable by 0.008 to 0.111 eV), while ethylene binds atop Pd in the alloys as a π complex rather than the di-σ mode on Pd(111). Vinyl intermediates bind strongly atop Pd, the alloy pathways after vinyl formation closely track that of Pd(111), and the ethylene desorption barrier on the alloys is negligible, which the authors interpret as favoring selectivity to ethylene (Thirumalai 2018).
Significance
The paper gives a systematic DFT picture of why dilute Pd in Au or Ag is more reactive than the host: a narrow, atom-like d state near the Fermi level that pulls adsorbates toward the dopant. It also shows that a common descriptor, the d-band center, fails to rank hydrogen binding on these surfaces, and that the host lattice matters, since compressive strain on Pd in Cu removes the sharp feature and the enhancement. The authors conclude that single atom alloys can behave quite differently from concentrated alloys and may lie outside the usual scaling trends (Thirumalai 2018).