LBERRY: Difference between revisions

From VASP Wiki
No edit summary
No edit summary
Line 18: Line 18:
This tag specifies the origin with respect to which the ionic contribution to the dipole moment in the cell is calculated. When comparing changes in this contribution due to the displacement of an ion, this center should be chosen in such a way that the ions in the distorted and the undistorted structure remain on the same side of {{TAG|DIPOL}} (in terms of a minimum image convention).
This tag specifies the origin with respect to which the ionic contribution to the dipole moment in the cell is calculated. When comparing changes in this contribution due to the displacement of an ion, this center should be chosen in such a way that the ions in the distorted and the undistorted structure remain on the same side of {{TAG|DIPOL}} (in terms of a minimum image convention).


\noindent
{\bf An example: The fluorine displacement dipole (Born effective charge) in NaF}


\noindent
== An example: The fluorine displacement dipole (Born effective charge) in NaF ==
First one needs to determine the electronic polarization of the undistorted NaF.\\


\noindent\underline{Calculation 1}\\
First one needs to determine the electronic polarization of the undistorted NaF.
It is usually convenient to calculate the self-consistent Kohn-Sham potential of the
undistorted structure, using a symmetry reduced (6$\times$6$\times$6)
Monkhorst-Pack sampling of the Brillouin zone.
Using for instance the following INCAR file:


\begin{verbatim}
Caluclation1:
PREC  = Med
It is usually convenient to calculate the self-consistent Kohn-Sham potential of the undistorted structure, using a symmetry reduced <math>6$\times$6$\times$6</math> Monkhorst-Pack sampling of the Brillouin zone. Using for instance the following {{TAG|INCAR}} file:
ISMEAR = 0
PREC  = Med
EDIFF  = 1E-6
ISMEAR = 0
\end{verbatim}
EDIFF  = 1E-6
{{TAG|KPOINTS}} file:
6x6x6
  0
Gamma
  6 6 6
  0 0 0
{{TAG|POSCAR}} file:
NaF
  4.5102
  0.0 0.5 0.5
  0.5 0.0 0.5
  0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5000000000000000  0.5000000000000000  0.5000000000000000
and LDA Na_sv and F PAW datasets.


\noindent KPOINTS file:


\begin{verbatim}
Calculation 2:
6x6x6
To calculate the electronic contribution to the polarization, along the reciprocal lattice vector <math>\bold{G}_{1} (i.e. \bold{P} \cdot \bold{G}_{1}), add the following lines to the {{TAG|INCAR}} file:
  0
LBERRY = .TRUE.
Gamma
  IGPAR  = 1
  6 6 6
  NPPSTR = 8
  0 0 0  
  DIPOL = 0.25 0.25 0.25
\end{verbatim}


\noindent POSCAR file:
Setting {{TAG|LBERRY}}=''.TRUE.'' automatically sets {{TAG|ICHARG}}=11, i.e., the charge density of the previous calculation is read and kept fixed, and only the orbitals and one-electron eigenenergies are recalculated for the new k-point set. This is advantageous, since the number of k-points used to evaluate the Berry phase expression can be quite large, and precalculating the charge density ({{TAG|ICHARG}}=11) saves significant CPU time.


\begin{verbatim}
The {{TAG|OUTCAR}} will now contain the following lines:
NaF
                                e<r>_ev=(    0.00000    0.00000    0.00000 ) e*Angst
4.5102
                                e<r>_bp=(    0.00000    0.00000    0.00000 ) e*Angst
0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5000000000000000  0.5000000000000000  0.5000000000000000
\end{verbatim}


\noindent and LDA Na\_sv and F PAW datasets.
  Total electronic dipole moment: p[elc]=(    0.00000    0.00000    0.00000 ) e*Angst


\noindent\underline{Calculation 2}\\
            ionic dipole moment: p[ion]=(    2.25510    2.25510    2.25510 ) e*Angst
To calculate the electronic contribution to the polarization, along the
reciprocal lattice vector ${\bf G}_1$ (i.e.~${\bf P} \cdot {\bf G}_1$),
add the following lines to the INCAR file:
\begin{verbatim}
LBERRY = .TRUE.
IGPAR  = 1
NPPSTR = 8
DIPOL = 0.25 0.25 0.25
\end{verbatim}


\noindent Setting {\tt LBERRY}=.TRUE. automatically sets {\tt ICHARG}=11,
i.e., the charge density of the previous calculation is read and kept fixed,
and only the orbitals and one-electron eigenenergies are recalculated for the
new $k$-point set.
This is advantageous, since the number of $k$-points used to evaluate
the Berry phase expression can be quite large, and precalculating the
charge density ({\tt ICHARG}=11) saves significant CPU time.


\noindent The OUTCAR will now contain the following lines:
Calculations 3 and 4:
\begin{verbatim}
The procedure mentioned under Calculation 2 now has to be repeated with {{TAG|IGPAR}}=2 and {{TAG|IGPAR}}=3 (again using the charge density obtained from Calculation 1), to obtain the contributions to the electronic polarization along <math>\bold{G}_2</math> and <math>\bold{G}_{3}</math>, respectively.
                                e<r>_ev=(    0.00000    0.00000    0.00000 ) e*Angst
                                e<r>_bp=(    0.00000    0.00000    0.00000 ) e*Angst


Total electronic dipole moment: p[elc]=(    0.00000    0.00000    0.00000 ) e*Angst


            ionic dipole moment: p[ion]=(    2.25510     2.25510     2.25510 ) e*Angst
Calculations 5 to 8:
\end{verbatim}
To calculate the change in the electronic polarization of NaF due to the displacement of the fluorine sublattice, one should repeat Calculations 1 to 4, using the following {{TAG|POSCAR}} file:
NaF
  4.5102
  0.0 0.5 0.5
  0.5 0.0 0.5
  0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5100000000000000  0.5100000000000000  0.4900000000000000
This corresponds to a displacement of the F ion by <math>0.01\times 4.51$\AA</math> along the <math>\hat{z}</math> direction. The output of the Berry phase calculation using {{TAG|IGPAR}}=1 should now look similar to:
                                e<r>_ev=(    0.00000    0.00000    0.00004 ) e*Angst
                                e<r>_bp=(    0.00000    0.18028    0.18028 ) e*Angst
  Total electronic dipole moment: p[elc]=(    0.00000     0.18028     0.18031 ) e*Angst


\noindent\underline{Calculations 3 and 4}\\
            ionic dipole moment: p[ion]=(    2.25510    2.25510    1.93939 ) e*Angst
The procedure mentioned under Calculation 2 now has to be repeated with
{\tt IGPAR}=2 and {\tt IGPAR}=3 (again using the charge density obtained
from Calculation 1), to obtain the contributions to the electronic
polarization along ${\bf G}_2$ and ${\bf G}_3$, respectively.\\


\noindent\underline{Calculations 5--8}\\
To calculate the change in the electronic polarization of NaF due to the
displacement of the fluorine sublattice, one should repeat Calculations 1--4,
using the following POSCAR file:
\begin{verbatim}
NaF
4.5102
0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5100000000000000  0.5100000000000000  0.4900000000000000
\end{verbatim}


\noindent This corresponds to a displacement of the F ion by
Collecting the results:
$0.01\times 4.51$\AA along the $\hat{z}$ direction.
The change in the electronic contribution to the polarization due to the F-sublattice displacement should be calculated as follows:
The output of the Berry phase calculation using {\tt IGPAR}=1 should
*Take the average of the <math>e<\mathrm{r}>_\mathrm{ev}</math> terms obtained in calculations 2 to 4. Lets call this <math>e<\mathrm{r}>_{\mathrm{ev,undist}}</math>
now similar to:
\begin{verbatim}
                                e<r>_ev=(    0.00000    0.00000    0.00004 ) e*Angst
                                e<r>_bp=(    0.00000    0.18028    0.18028 ) e*Angst


Total electronic dipole moment: p[elc]=(    0.00000    0.18028    0.18031 ) e*Angst
*Add the <math>e<\mathrm{r}>_{\mathrm{bp}} </math> terms obtained in calculations 2 to 4. Lets call this <math>e<\mathrm{r}>_{\mathrm{bp,undist}}</math>


            ionic dipole moment: p[ion]=(    2.25510    2.25510    1.93939 ) e*Angst
*The electronic polarization of the undistorted structure is then given by:
\end{verbatim}


\noindent\underline{Collecting the results}\\
<math> <\mathrm{r}>_{\mathrm{el,undist}}=e<\mathrm{r}>_{\mathrm{ev,undist}}+e<\mathrm{r}>_{\mathrm{bp,undist}} </math>
The change in the electronic contribution to the polarization due to the
F-sublattice displacement should be calculated as follows:
\begin{itemize}
\item
Take the average of the $e<{\rm r}>_{\rm ev}$ terms obtained in Calculations 2--4.
Lets call this $e<{\rm r}>_{{\rm ev,undist}}$
\item
Add the $e<{\rm r}>_{\rm bp}$ terms obtained in Calculations 2--4.
Lets call this $e<{\rm r}>_{{\rm bp,undist}}$
\item
The electronic polarization of the undistorted structure is then given by:
\[
e<{\rm r}>_{{\rm el,undist}}=e<{\rm r}>_{{\rm ev,undist}}+e<{\rm r}>_{{\rm bp,undist}}
\]
\item
Repeat the above three steps for the results obtained using the
distorted structure (Calculations 6--8), to evaluate $e<{\rm r}>_{{\rm ev,dist}}$,
$e<{\rm r}>_{{\rm bp,dist}}$, and $e<{\rm r}>_{{\rm el,dist}}$
\item
The change in the electronic contribution to the polarization due
to the F-sublattice displacement, $e\Delta<{\rm r}>_{\rm el}$ is then
given by $e<{\rm r}>_{{\rm el,dist}}-e<{\rm r}>_{{\rm el,undist}}$
\end{itemize}


\noindent To calculate the total change in polarization, $e\Delta<{\rm r}>$,
*Repeat the above three steps for the results obtained using the distorted structure (Calculations 6 to 8), to evaluate <math>e<\mathrm{r}>_{\mathrm{ev,dist}}</math>,
one should account for the ionic contribution to this change.
<math>e<\mathrm{r}>_{\mathrm{bp,dist}}</math>, and <math>e<\mathrm{r}>_{\mathrm{el,dist}}</math>
This contribution can be calculated from p[ion] as given above from
 
*The change in the electronic contribution to the polarization due to the F-sublattice displacement, <math>e\Delta<\mathrm{r}>_\mathrm{el}</math> is then given by <math>e<\mathrm{r}>_{\mathrm{el,dist}}-e<\mathrm{r}>_{\mathrm{el,undist}}</math>
 
 
To calculate the total change in polarization, <math>e\Delta<\mathrm{r}> </math>, one should account for the ionic contribution to this change. This contribution can be calculated from p[ion] as given above from
Calculations 2 and 5:
Calculations 2 and 5:
$\Delta{\rm p[ion]}={\rm p[ion]}_{\rm dist}-{\rm p[ion]}_{\rm undist}$.
<math> \Delta\mathrm{p[ion]}=\mathrm{p[ion]}_{\mathrm{dist}}-\mathrm{p[ion]}_{\mathrm{undist}} </math>.


\noindent
<math> e\Delta<\mathrm{r}> </math> is then given by <math> \Delta \mathrm{p[ion]}+e\Delta<\mathrm{r}>_\mathrm{el} </math>. In this example we find <math> e\Delta<\mathrm{r}>=0.04489</math> electrons <math>\AA</math>. Considering that the moved F-sublattice was displaced by 0.045102 \AA, this calculation yields a Born effective charge for fluorine in NaF of <math>Z^{*}=-0.995 </math>.
$e\Delta<{\rm r}>$ is then given by
$\Delta {\rm p[ion]}+e\Delta<{\rm r}>_{\rm el}$.
In this example we find $e\Delta<{\rm r}>=0.04489$ electrons~\AA.
Considering that moved the F-sublattice was displaced by 0.045102 \AA,
this calculation yields a Born effective charge for fluorine in NaF of
$Z^{*}=-0.995$.\\


\noindent N.B.(I) In the case of spinpolarized calculations ({\tt ISPIN}=2),
N.B.(I) In the case of spinpolarized calculations ({{TAG|ISPIN}}=2),the Berry phase of the orbitals is evaluated separately for each spin direction. This means a ''grep'' on "<math><\mathrm<r}> </math>" will yield two sets of <math> <\mathrm{r}>_{\mathrm{ev}} </math> and <math> <\mathrm{r}>_{\mathrm{bp}}</math> terms, which have to be added to oneanother to obtain the total electronic polarization of the system.
the Berry phase of the orbitals is evaluated separately for each spin direction.
This means a grep on "$<{\rm r}>$" will yield two sets of $<{\rm r}>_{\rm ev}$
and $<{\rm r}>_{\rm bp}$ terms, which have to be added to oneanother to
obtain the total electronic polarization of the system.\\


\noindent N.B.(II) One should take care of the fact that the calculated
N.B.(II) One should take care of the fact that the calculated "Berry phase" term <math> <\mathrm{r}>_{\mathrm{bp}}</math> along <math>\mathbf{G}_{i}</math> is, in principle, obtained modulo a certain period, determined by the lattice vector <math> \mathbf{R}_{i} </math> (<math>\mathbf{R}_{i} \cdot \mathbf{G}_{i} = 2 \pi </math>), the spin multiplicity of the orbitals, the volume of the unit cell, the number of k-point in the "perpendicular" grid, and some aspects of the symmetry of the system. More information on this particular aspect of the Berry phase calculations can be found in references <ref name="berryformalism1"/><ref name="berryformalism2"/><ref name="berryformalism3"/><ref name="berryformalism4"/><ref name="berryformalism5"/><ref name="berryultrasoft"/><ref name="berrymmars"/>.
"Berry phase" term $<{\rm r}>_{\rm bp}$ along ${\bf G}_i$ is, in principle,
obtained modulo a certain period, determined by the lattice vector
${\bf R}_{i}$ (${\bf R}_{i}\cdot{\bf G}_{i}=2\pi$), the spin
multiplicity of the orbitals, the volume of the unit cell,
the number of $k$-point in the "perpendicular" grid, and some aspects of
the symmetry of the system.
More information on this particular aspect of the Berry phase calculations
can be found in Refs.~\cite{Berry:formalism} and~\cite{Berry:mmars}.





Revision as of 13:27, 16 January 2017

LBERRY = [logical]
Default: LBERRY = .FALSE. 

Description: This tag is used in the the evaluationof the Berry phase expression for the electronic polarization of an insulating system.


As of VASP.5.2, calculating the macroscopic polarization and Born effective charges along the lines of the following example (using LBERRY=.TRUE. etc) is unnecessary. The use of LCALCPOL or LCALCEPS is much more convenient.

Setting LBERRY=.TRUE. in the INCAR file switches on the evaluation of the Berry phase expression for the electronic polarization of an insulating system, as modified for the application of USPP's and PAW datasets [1][2][3][4][5][6][7]. In addition, the following keywords must be specified in order to generate the meshof k-points:

IGPAR = 1|2|3

IGPAR tag specifies the socalled parallel or direction in the integration over the reciprocal space unit cell.

NPPSTR = number of points on the strings in the IGPAR direction

This tag specifies the number of k-points on the strings (with ).

DIPOL = center of cell (fractional coordinates)

This tag specifies the origin with respect to which the ionic contribution to the dipole moment in the cell is calculated. When comparing changes in this contribution due to the displacement of an ion, this center should be chosen in such a way that the ions in the distorted and the undistorted structure remain on the same side of DIPOL (in terms of a minimum image convention).


An example: The fluorine displacement dipole (Born effective charge) in NaF

First one needs to determine the electronic polarization of the undistorted NaF.

Caluclation1: It is usually convenient to calculate the self-consistent Kohn-Sham potential of the undistorted structure, using a symmetry reduced Monkhorst-Pack sampling of the Brillouin zone. Using for instance the following INCAR file:

PREC   = Med
ISMEAR = 0
EDIFF  = 1E-6

KPOINTS file:

6x6x6
 0
Gamma
 6 6 6
 0 0 0 

POSCAR file:

NaF
 4.5102
 0.0 0.5 0.5
 0.5 0.0 0.5
 0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5000000000000000  0.5000000000000000  0.5000000000000000

and LDA Na_sv and F PAW datasets.


Calculation 2: To calculate the electronic contribution to the polarization, along the reciprocal lattice vector and , respectively.


Calculations 5 to 8: To calculate the change in the electronic polarization of NaF due to the displacement of the fluorine sublattice, one should repeat Calculations 1 to 4, using the following POSCAR file:

NaF
 4.5102
 0.0 0.5 0.5
 0.5 0.0 0.5
 0.5 0.5 0.0
1 1
Direct
  0.0000000000000000  0.0000000000000000  0.0000000000000000
  0.5100000000000000  0.5100000000000000  0.4900000000000000

This corresponds to a displacement of the F ion by along the direction. The output of the Berry phase calculation using IGPAR=1 should now look similar to:

                                e<r>_ev=(     0.00000     0.00000     0.00004 ) e*Angst
                                e<r>_bp=(     0.00000     0.18028     0.18028 ) e*Angst 

 Total electronic dipole moment: p[elc]=(     0.00000     0.18028     0.18031 ) e*Angst
            ionic dipole moment: p[ion]=(     2.25510     2.25510     1.93939 ) e*Angst


Collecting the results: The change in the electronic contribution to the polarization due to the F-sublattice displacement should be calculated as follows:

  • Take the average of the terms obtained in calculations 2 to 4. Lets call this
  • Add the terms obtained in calculations 2 to 4. Lets call this
  • The electronic polarization of the undistorted structure is then given by:

  • Repeat the above three steps for the results obtained using the distorted structure (Calculations 6 to 8), to evaluate ,

, and

  • The change in the electronic contribution to the polarization due to the F-sublattice displacement, is then given by


To calculate the total change in polarization, , one should account for the ionic contribution to this change. This contribution can be calculated from p[ion] as given above from Calculations 2 and 5: .

is then given by . In this example we find electrons . Considering that the moved F-sublattice was displaced by 0.045102 \AA, this calculation yields a Born effective charge for fluorine in NaF of .

N.B.(I) In the case of spinpolarized calculations (ISPIN=2),the Berry phase of the orbitals is evaluated separately for each spin direction. This means a grep on "Failed to parse (Conversion error. Server ("cli") reported: "[INVALID]"): {\displaystyle <\mathrm<r}> } " will yield two sets of and terms, which have to be added to oneanother to obtain the total electronic polarization of the system.

N.B.(II) One should take care of the fact that the calculated "Berry phase" term along is, in principle, obtained modulo a certain period, determined by the lattice vector (), the spin multiplicity of the orbitals, the volume of the unit cell, the number of k-point in the "perpendicular" grid, and some aspects of the symmetry of the system. More information on this particular aspect of the Berry phase calculations can be found in references [1][2][3][4][5][6][7].


References

  1. a b R. D. King-Smith and D. Vanderbilt, Phys. Rev. B 47, 1651 (1993).
  2. a b D. Vanderbilt and R. D. King-Smith, Phys. Rev. B 48, 4442 (1993).
  3. a b R. Resta, Ferroelectrtics 136, 51 (1992).
  4. a b R. Resta, Rev. Mod. Phys. 66, 899 (1994).
  5. a b [R. Resta, in Berry Phase in Electronic Wavefunctions, Troisième Cycle de la Physique en Suisse Romande, Année Academique 1995-96, (1996).]
  6. a b [D. Vanderbilt and R. D. King-Smith, in Electronic polarization in the ultrasoft pseudopotential formalism, Unpublished report, (1998).]
  7. a b [Available online at http://cms.mpi.univie.ac.at/vasp/Welcome.html.]

Contents