How to generate a mixed pseudopotential

How to generate a mixed
pseudopotential
Objectives
Generate a mixed pseudopotential to be used in the
Virtual Crystal Approximation or in
simulations at constant electric displacement
Most important reference
IOP PUBLISHING
JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 22 (2010) 415401 (16pp)
doi:10.1088/0953-8984/22/41/415401
An efficient computational method for use
in structural studies of crystals with
substitutional disorder
˜
Roberta Poloni, Jorge I´niguez,
Alberto Garc´ıa and Enric Canadell
Institut de Ci`encia de Materials de Barcelona (CSIC), Campus UAB, 08193 Bellaterra, Spain
E-mail: [email protected]
Received 11 June 2010
Published 23 September 2010
Online at stacks.iop.org/JPhysCM/22/415401
Abstract
We present a computationally efficient semi-empirical method, based on standard
first-principles techniques and the so-called virtual crystal approximation, for determining the
average atomic structure of crystals with substitutional disorder. We show that, making use of a
minimal amount of experimental information, it is possible to define convenient figures of merit
that allow us to recast the determination of the average atomic ordering within the unit cell as a
minimization problem. We have tested our approach by applying it to a wide variety of
How to compile the code to run the
pseudopotential mixer
The code is included in the Siesta distribution, within the Util/VCA directory
Assuming we are in the directory where the
sources
are stored, simply
I. Siesta
ENERGY
FUNCTIONAL
FOR A DIEL
$ cd Util/VCA
$ make
It will use the same arch.make file as for the compilation of SIESTA
(no need to copy it again to the Util/VCA directory)
mixps
fractional
This will generate two executable files:
(program to mix pseudopotentials)
(program that multiplies the strength of a pseudopotential by a
given fraction)
How to mix two pseudopotentials
Create a new directory in the directory where you generate your pseudopotentials
ENERGY
FOR A DIELECTRIC
INSIDE AN
Copy there the two pseudosI.you
want toFUNCTIONAL
mix (in this example,
O and F)
Cite1,2 .
I.
ENERGY FUNCTIONAL FOR A DIELECTRI
$ cd ../atom/Tutorial/PS_Generation
$ mkdir VCA-O-F
Cite$ 1,2
cd. VCA-O-F
$ cp ../O/O.tm2.psf ./O.psf
cd ../F/F.tm2.psf ./F.psf
cd $../atom/Tutorial/PS_Generation
$
$ mkdir
VCA-O-F
XC.functional
GGA
$ cd XC.authors
VCA-O-F
PBE
.true.
$ cp
../O/O.tm2.psf
./O.psf
Copy there the
twoSpinPolarized
pseudos you want to
mix
(in this example, O and F)
$ cd nsppol
../F/F.tm2.psf
./F.psf
2
# Number of spin polarizations
$
spinat 0.0 0.0 1.0
ixc
11
~/siesta/Util/VCA/mixps
occopt
3
O F
# Spin for atoms
# Integer for exchange-correlatio
0.9
# Occupation
Option
Mixing parameter:
InNumber
this example
of spinmetal)
pola
Fermi-Dirac smearing #(finite-temperature
90% of the #first
atomfor atoms
Spin
Temperature of smearing
(in Ha)
nsppol 2
#
spinat
0.0
0.0 1.0#
Labels of the two
atoms
involved
tsmear
0.01
ixc
11
10% of the second atom
# Integer for exchang
Output of
$ mkdir VCA-O-F
$ cd VCA-O-F
$ cp ../O/O.tm2.psf ./O.psf
the
mixing
of
the
pseudopotential
$ cd ../F/F.tm2.psf ./F.psf
$ ~/siesta/Util/VCA/mixps O F 0.9
nsppol 2
spinat
0.0 1.0
Labels of the two
atoms 0.0
involved
ixc
11
occopt
OF-0.90000.psf
Mixing parameter:# InNumber
this example
of spin pola
90% of the #first
atomfor atoms
Spin
10% of the second atom
# Integer for exchang
New files:
# Occupation Option
3
# Fermi-Dirac smearing (finite-tempera
# Temperature of smearing (in Ha)
Pseudopotential file with the mixture of the two pseudos
tsmear
0.01
Name of the mixed file: Symbols of the two original pseudopotentials,
followed by the mixing parameter (up to five decimal places)
OF-0.90000.synth
Bloch “SyntheticAtoms” to be used in SIESTA
%block SyntheticAtoms
1
2 2 3 4
2.000000 4.100000 0.000000
%endblock SyntheticAtoms
MIXLABEL
0.000000
Final label used
In this particular example a virtual with a
charge of 6.1 electrons is generated
0.9 * 6 (electrons in O) + 0.1*7 (electrons in F)
= 6,1 electrons in the Virtual Atom
2.0 electrons in the s chanel and 4.1 electrons
in the p channel
Some notes on the mixed pseudopotentials
Once SIESTA reads the new mixed pseudopotential, it proceeds as
usual, and generates:
- the local part of the pseudopotential
- the Kleinman-Bylander projectors
- the basis set
Those quantities are not a true mix of the corresponding quantities of
the individual atoms.
The basis set is generated by SIESTA using the mixed pseudopotential
(no mixing of the basis set has been implemented).
To see how to generate a basis set for a mixed atom, see the Tutorial
“How to run with a finite constraint electric displacement”
Uses of the Virtual Crystal Approximation
To study substitutional disorder
IOP PUBLISHING
JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 22 (2010) 415401 (16pp)
doi:10.1088/0953-8984/22/41/415401
An efficient computational method for use
in structural studies of crystals with
substitutional disorder
˜
Roberta Poloni, Jorge I´niguez,
Alberto Garc´ıa and Enric Canadell
Institut de Ci`encia de Materials de Barcelona (CSIC), Campus UAB, 08193 Bellaterra, Spain
E-mail: [email protected]
11 June 2010
ToReceived
perform
calculations at constrained electric displacement
Published 23 September 2010
see
the
Tutorial “How to run with a finite constraint electric displacement”
Online
at stacks.iop.org/JPhysCM/22/415401
Abstract
We present a computationally efficient semi-empirical method, based on standard
first-principles techniques and the so-called virtual crystal approximation, for determining the
average atomic structure of crystals with substitutional disorder. We show that, making use of a
minimal amount of experimental information, it is possible to define convenient figures of merit
that allow us to recast the determination of the average atomic ordering within the unit cell as a
minimization problem. We have tested our approach by applying it to a wide variety of