Programs in Physics & Physical Chemistry
|[Licence| Download | New Version Template] acpv_v1_0.gz(10517 Kbytes)|
|Manuscript Title: A self-consistent surface Green-function (SSGF) method for the calculation of isolated adsorbate atoms on a semi-infinite crystal.|
|Authors: J. Bormet, B. Wenzien, M. Scheffler|
|Program title: fhi93ssgf|
|Catalogue identifier: ACPV_v1_0|
Distribution format: gz
|Journal reference: Comput. Phys. Commun. 79(1994)124|
|Programming language: Fortran.|
|Computer: IBM RISC System/6000.|
|Operating system: AIX 3.2, UNICOS 7.0, OSF/1.|
|RAM: 9999K words|
|Word size: 64|
|Keywords: Solid state physics, Crystal field, Green function, Adsorbate, Layer kkr, Dft-lda, Total energy, Hellmann-feynman forces.|
Nature of problem:
The computer code allows to calculate the Green function of an adsorption problem with a single, isolated adsorbate atom (so-called "adsorbate system") on a semi-infinite metal surface. The following physical quantities are available as output: change in electron density for the adsorbate system, change in density of states, total energy of the adsorbate system, and the Hellmann-Feynman forces on the adsorbate atom. The program uses density-functional theory within the local- density approximation for the exchange-correlation functional and ab initio, norm-conserving pseudopotentials.
The Green function of the clean substrate (so-called "reference system") has to be calculated in advance. This reference Green function is needed as input for this code. It is obtainable with the layer KKR method [1,2,3] and has to be projected onto a localized basis of Gaussian orbitals [4,5]. The code described below then solves the Dyson equation self-consistently for the effective potential of the adsorbate atom with the projected Green function of the reference system.
At this time, only one single adsorbate atom can be handled by the code, although the input is made for a finite number of adsorbate atoms. For the evaluation of the exchange-correlation functional the electron- density change, Delta n**v(r), is evaluated on a mesh in real space. This mesh is restricted to be of cubic shape. The treatment of f- electron systems is not possible with the present code, although there are no limitations in principle.
One iteration on a CRAY Y-MP (single processor) takes 82 seconds, on an IBM RS/6000-350 it takes 493 seconds. About 40 iterations are necessary to converge a typical problem which has a linear dimension of 108 in the Gaussian basis and about 40**3 points in the real-space mesh for Delta n**v(r). There are three time consuming parts: * Solution of the Dyson equation. * Projection of the effective potential onto Gaussians. * Transformation of the density matrix (in the Guassian basis) to the real space mesh.
|||F. Maca and M. Scheffler, Comput. Phys. Commun. 38(1985)403, 47(1987)349.|
|||F. Maca and M. Scheffler, Comput. Phys. Commun. 51(1988)381.|
|||B. Wenzien, J. Bormet, and M. Scheffler, Green function for crystal surfaces I. Submitted to Comput. Phys. Commun.|
|||Ch. Droste, Ph. D. thesis, Fachbereich 4 (Physik) der Technischen Universitat Berlin (1990).|
|||B. Wenzien, J. Bormet, and M. Scheffler, Green function for crystal surfaces II: Projection onto Gaussian orbitals. To be published.|
|Disclaimer | ScienceDirect | CPC Journal | CPC | QUB|