# Dresden 2009 – wissenschaftliches Programm

## Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

# O: Fachverband Oberflächenphysik

## O 21: Methods: Electronic structure theory I

### O 21.2: Vortrag

### Dienstag, 24. März 2009, 10:45–11:00, SCH A316

**Scalar relativistic schemes for all-electron DFT with atom-centered basis functions** — Paula Havu, •Volker Blum, and Matthias Scheffler — Fritz-Haber-Institut der Max-Planck-Gesellschaft, D-14195 Berlin, Germany

Numeric atom-centered orbitals are an efficient, accurate basis choice for
all-electron electronic structure theory [1]. For seamless
efficiency and accuracy, a one-component (two with spin) Schrödinger-like
equation is computationally most convenient, but for most elements
(*Z*30), relativistic effects arising near the nucleus cannot be
ignored. Dirac’s equation can simply be rewritten in a
“scalar-relativistic” (one-component) form, but with a
separate Hamiltonian for each eigenstate. For some paradigm test systems
[e.g., the Au dimer; CO adsorption on
Pt(111)], we here benchmark the accuracy of a hierarchy of
scalar-relativistic schemes that circumvent the state dependence: (i)
the unsatisfactory “zero-order regular approximation” (ZORA), which simply
neglects the state dependence; (ii) a
restriction of ZORA to only the atomic center of each basis function
(“atomic ZORA”) and (iii) a perturbative rescaling of all ZORA eigenvalues
(“scaled ZORA” [2]), which both recover geometries and binding energies
within a few 10 meV of benchmark full-potential linearized augmented plane wave [FP-(L)APW] calculations; and (iv) a
separate, exact treatment of all non-overlapping core states, which then
necessitates only small further (scaled) ZORA-like approximations to the
extended semicore and valence states.
[1] V. Blum *et al.*, Comp. Phys. Comm., accepted (2008).
[2] E. van Lenthe *et al.*, J. Chem. Phys. **101**, 9783 (1994).