mirror of https://gitlab.com/QEF/q-e.git
90 lines
4.4 KiB
Plaintext
90 lines
4.4 KiB
Plaintext
This example shows how to calculate final-state core-level-shift (CLS)
|
|
using the core-excited pseudo-potential technique. The procedure has
|
|
been used in several works and references for the underlying theoretical
|
|
concepts can be found in the articles listed below.
|
|
|
|
E. Pehlke and M. Scheffler - http://link.aps.org/doi/10.1103/PhysRevLett.71.2338
|
|
J. N. Andersen, D. Hennig, E. Lundgren, M. Methfessel, R. Nyholm, and
|
|
M. Scheffler - http://link.aps.org/doi/10.1103/PhysRevB.50.17525
|
|
|
|
The definition of a FS core-level-shift is based on a difference
|
|
of energies and it's then necessary to take a reference atom in the
|
|
configuration studied to obtain the relative core level shifts. In this
|
|
example a very simple calculation regarding the SCLS (surface core level
|
|
shift) in the rhodium 011 slab is presented.
|
|
|
|
First the slab has to be defined with the correct parameters (how many
|
|
layers are needed to find a bulk-like atom, the separation between the
|
|
periodic repetitions of the slab, DFT convergence parameters, ecc...) and
|
|
the atomic positions have to be relaxed. All these steps are described
|
|
in detail in other examples and are not treated here. The structure
|
|
used in this example should not be taken as reference for an accurate
|
|
calculation but the parameters are chosen in order to keep the example
|
|
fast and instructive.
|
|
|
|
In the slab geometry it's natural to take the atom in the central layer,
|
|
that mimics the bulk environment, as reference and calculate all CLS from
|
|
the difference w.rt. this one. In this example the other interesting
|
|
atoms are the atoms in the surface layer and the one in the first
|
|
subsurface layer. Once the desired atoms are identified the procedure
|
|
is straightforward and can be defined in few steps:
|
|
|
|
1) Make a regular SCF calculation of the slab where the core-excited
|
|
pseudo-potential is used for the reference atom.
|
|
|
|
2) Make several other SCF calculations, one for each selected atom in
|
|
which only this one is described by the core-excited pseudo-potential
|
|
|
|
3) Calculate the ground state (GS) energy difference between each
|
|
of these SCF calculations and the one for the reference atom.
|
|
These differences are the FS core-level shifts.
|
|
|
|
----------------
|
|
|
|
1) For this simulation, and all the following ones, it's necessary to
|
|
define a normal pseudo-potential and a core-excited one for Rhodium. The
|
|
two potentials have to be consistent with each other (functional,
|
|
parameters, ecc..), being the core-excited one a PP for the same atomic
|
|
type with a different, core-excited, electronic configuration. (The
|
|
instructions to generate of a core-excited PP can be found in the ld1.x
|
|
manual.)
|
|
|
|
Once the PP and the core-excited PP are defined the calculation is
|
|
a regular SCF run with the only difference that the bulk atom, the
|
|
reference, is defined by the core-excited PP. ONLY the reference atom
|
|
is defined in this way and ntyp variable in the &system namelist has to
|
|
be defined including the new core-excited type. All the other parameter
|
|
are defined following the normal guidelines for a SCF calculation.
|
|
|
|
*** Keep in mind that the core-excited atom is a new atomic type in
|
|
the configuration and all the precautions of possible interaction have
|
|
to be considered. In the example a slab 1x1 is used only to let the
|
|
example run on an average single CPU, again this is just a reference
|
|
structure. It's possible, and in fact true, that a bigger supercell is
|
|
needed, for example a 2x2 or a 3x3, to keep all the core-excited atoms
|
|
enough separated, avoid an interaction between them.
|
|
|
|
(input=rh011bulk.scf.in, output=rh011bulk.scf.out)
|
|
|
|
2) All the other simulations are identical to the bulk-reference one but
|
|
this time the atom defined with the excited PP is different. For every
|
|
simulation ONLY ONE atom has to be defined by the core-excited PP and
|
|
no relaxation as to be done, only one SCF calculation. (If one wants
|
|
to calculate the CLS of three different atoms three SCF calculations
|
|
are needed).
|
|
|
|
It's clear that to keep consistency all the other SCF parameters
|
|
(k-points, energy cut-offs, ecc..) of these new calculations have to
|
|
be identical to the reference SCF calculation.
|
|
|
|
(input=rh011surf.scf.in, output=rh011surf.scf.in)
|
|
|
|
3) Once obtained the energy for all the atoms identified the CLS are
|
|
defined as the difference between the GS energy of the particular SCF
|
|
calculation and the GS energy of the reference SCF one:
|
|
|
|
SCLS = energy_gs(surface core-excited) - energy_gs(bulk core-excited)
|
|
CLS = energy_gs(other atom core-excited) - energy_gs(bulk core-excited)
|
|
|
|
(see final-state.txt)
|