Step decoration of chiral metal surfaces
Jeong Woo Han, John R. Kitchin, and David S. Sholl, “Step decoration of chiral metal surfaces”, The Journal of Chemical Physics 130(12), 124710 (2009). https://doi.org/10.1063/1.3096964
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
High Miller index cuts of metal crystals can produce intrinsically chiral surfaces, whose kinked step edges lack mirror symmetry and can bind the two enantiomers of a chiral molecule differently. One proposed way to strengthen this enantiospecificity is to decorate the kinked steps with atoms of a second metal. That only works if the added atoms stay at the kink sites instead of moving onto terraces or dissolving into the bulk. The paper asks which impurity/host metal pairs make such step decoration thermodynamically stable, and whether this can be predicted for all 3d, 4d and 5d transition metal combinations without computing each one (Han et al. 2009).
Approach
The authors treat the problem through the surface segregation energy of an isolated impurity, defined as the energy to move the impurity from the bulk to a given surface site in the dilute limit, so that a negative value means the surface site is enthalpically preferred. They evaluate it from differences of DFT total energies of surface and bulk supercells with and without one impurity. Only enthalpic contributions and isolated impurities are considered: configurational and vibrational entropy and impurity aggregation are neglected, as in the earlier flat-surface database of Ruban and co-workers. Calculations used VASP with ultrasoft pseudopotentials and the PW91 GGA functional, slabs equivalent to six (111) layers with 14 Å of vacuum, and impurities placed symmetrically in the top and bottom layers to cancel dipoles. Sites were sampled on flat (110), (100) and (111) surfaces (coordination numbers 7, 8 and 9), the achiral stepped (322) surface, and the chiral stepped (643) surface, whose sites range from CN 6 (the kink) to 11, with subsurface sites standing in for CN 12 (Han et al. 2009).
These direct DFT results for five impurities in each of four fcc hosts (Cu, Rh, Pd, Ir) were then used to fit a correlation in terms of physically motivated descriptors: the loss of coordination relative to bulk, a host/impurity bond strength difference taken from liquid surface tensions, and the shift in surface d-band center when the impurity substitutes into the unrelaxed close-packed host surface. Candidate models were compared by leave-one-out cross-validation (Han et al. 2009).
Key findings
For 19 impurity/host combinations on Cu, Rh, Pd and Ir(111), the method reproduced the flat-surface segregation energies of Ruban et al. with an rms deviation of 0.12 eV/atom. Computed lattice constants (3.64, 3.85, 3.96 and 3.89 Å for Cu, Rh, Pd and Ir) were close to experiment. Segregation energies became more negative as CN decreased; for Ag in Cu the kink sites of Cu(643) were the most favored. Direct DFT predicts stable kink decoration (negative energy at CN 6) for Ag on Cu; Cu and Pt on Rh; Ag on Pd; and Pd, Pt, Cu and Rh on Ir. Rh, Ti and W on Cu, Ti, V and Re on Rh, Pt, Rh and V on Pd, and Ti on Ir are instead predicted to dissolve into the bulk (Han et al. 2009).
The best correlation was linear in the coordination deficit (12 minus CN) multiplied by a weighted sum of the bond strength difference and the d-band center shift, plus a small constant (about -0.06 eV). Adding atomic size did not improve it, which the authors attribute to the d-band center already depending on atomic size, and a quadratic CN term gave no statistical improvement. Because the expression is linear in CN, moving an impurity from a (111) terrace (CN 9) to a kink (CN 6) costs the same as moving it from bulk to terrace, so existing flat-surface data can be reinterpreted for step sites. Against Ruban’s data for fcc metals plus Fe, Co and Ru, the standard deviation was 0.26 eV/atom, and the sign is reliable above that magnitude (88 pairs analyzed). Across all 3d, 4d and 5d metals the scatter was larger, mainly from bcc(110) hosts, since only fcc DFT data were used in fitting, so a 0.54 eV/atom threshold was applied. In total a firm prediction was made for 347 pairs: 206 favor kink decoration and 141 do not. Decoration is favored for many impurities on Mo, Ru, W, Re, Os and Ir, and unfavorable for nearly all impurities on Ag and Au (Han et al. 2009).
Significance
The work gives the first evidence that thermodynamically stable decoration of kinks on chiral metal surfaces exists, and a cheap descriptor-based screen for choosing candidate pairs. The authors caution that deposition and diffusion kinetics and impurity aggregation were not modeled, and that whether decorated surfaces actually improve enantioselective adsorption remains to be tested. They also point to bimetallic nanoparticles, which have many undercoordinated sites, as another place where stable step decoration could alter catalytic reactivity (Han et al. 2009).