Computing the phonon dispersion and DOS: Difference between revisions

Revision as of 09:02, 20 February 2024

After computing the force constants using the finite differences or density-functional-perturbation theory (DFPT) approaches, it is possible to compute the phonon dispersion relation as well as the phonon density of states (DOS). This is accomplished by Fourier interpolating the interatomic force constants from a supercell calculation to the primitive cell.

Phonon dispersion: Step-by-step instructions

Step 1: Compute the force constants

There are two possible approaches for computing the force constants and then building the dynamical matrix:

Using finite differences with (IBRION = 5, 6).
Using DFPT with (IBRION = 7, 8).

These calculations must be performed in a supercell so that the force constants vanish at large distances.

Important: The phonon frequencies need to be converged with respect to the supercell size.

Step 2: Provide q-points along a high-symmetry path

Create a QPOINTS file containing a q-points path at which the phonon dispersion is computed. This is accomplished using the line mode of the KPOINTS-file format.

Step 3: Compute the phonon dispersion

To compute the phonon dispersion, set LPHON_DISPERSION = true in the INCAR file. The amount of information written to the OUTCAR file can be tuned using the PHON_NWRITE tag.

Reading of force constants

Steps 1-3 can be performed in one VASP calculation. However, generating the finite displacements in the supercell to compute force constants is time-consuming. It is possible to skip that step by providing force constants from a previous run. Rename the vaspout.h5 output file from the previous calculation to vaspin.h5, set

 LPHON_READ_FORCE_CONSTANTS = True 
 LPHON_DISPERSION = True

and provide a QPOINTS file.

Phonon DOS: Step-by-step instructions

Step 1: Compute the force constants

Same as above. This can be skipped by providing force constants in vaspin.h5 and setting LPHON_READ_FORCE_CONSTANTS = True.

Step 2: Specify a uniform q-point mesh

Create a QPOINTS file that specifies a sufficiently dense, uniform q-point mesh.

Step 3: Compute the DOS

Set PHON_DOS > 0 in the INCAR file. The DOS is computed between $[\omega_{\text{min}}-5\sigma,\omega_{\text{max}}+5\sigma]$ with $\omega_{\text{min}}$ and $\omega_{\text{max}}$ the lowest and highest phonon frequency and $\sigma$ the broadening (PHON_SIGMA).

The number of energy points in this energy range is specified by the PHON_NEDOS tag. To use a Gaussian-smearing method for the computation of the DOS set PHON_DOS = 1 or to use the tetrahedron method set PHON_DOS = 2.

Polar materials

If the material is polar, i.e., two or more atoms in the unit cell carry non-zero Born effective charge tensors, the long-range dipole-dipole interaction has to be treated by Ewald summation. This is achieved by setting LPHON_POLAR=.TRUE., supplying the static dielectric tensor (PHON_DIELECTRIC) and the Born-effective charges (PHON_BORN_CHARGES). The values for these have to be obtained from a separate VASP calculation in the unit cell setting LEPSILON or LCALCEPS. Optionally, specify a reciprocal space cutoff radius (PHON_G_CUTOFF) for the Ewald summation.

@@ Line 1: / Line 1: @@
-After computing the force constants using the [[Phonons_from_finite_differences|finite differences]] or [[Phonons_from_density-functional-perturbation_theory|density functional perturbation theory]] approaches, it is possible to compute the phonon dispersion relation as well as the phonon density of states.
+After computing the force constants using the [[Phonons_from_finite_differences|finite differences]] or [[Phonons_from_density-functional-perturbation_theory|density-functional-perturbation theory]] (DFPT) approaches, it is possible to compute the phonon dispersion relation as well as the phonon density of states (DOS).
 This is accomplished by Fourier interpolating the interatomic force constants from a supercell calculation to the primitive cell.
@@ Line 7: / Line 7: @@
 There are two possible approaches for computing the force constants and then building the dynamical matrix:
 # Using [[Phonons_from_finite_differences|finite differences]] with ({{TAGO|IBRION|5, 6}}).
-# Using [[Phonons_from_density-functional-perturbation_theory|density functional perturbation theory]] with ({{TAGO|IBRION|7, 8}}).
+# Using [[Phonons_from_density-functional-perturbation_theory|DFPT]] with ({{TAGO|IBRION|7, 8}}).
-These calculations have to be performed in a supercell so that the force constants vanish at large distances.
+These calculations must be performed in a supercell so that the force constants vanish at large distances.
-{{NB|important|If the supercell is not large enough, the phonon frequencies will likely not be properly converged. Always perform convergence studies with respect to the superecell size.}}
+{{NB|important|The phonon frequencies need to be converged with respect to the supercell size.}}
-In systems where the unit cell is already sufficiently large, one may directly use the unit-cell geometry.
 === Step 2: Provide '''q'''-points along a high-symmetry path ===
-Create a {{FILE|QPOINTS}} file containing a '''q'''-points path at which the phonon dispersion will be computed.
+Create a {{FILE|QPOINTS}} file containing a '''q'''-points path at which the phonon dispersion is computed.
-This is accomplished using the [[KPOINTS#Band-structure calculations|line mode]] of the KPOINTS file format.
+This is accomplished using the [[KPOINTS#Band-structure calculations|line mode]] of the {{FILE|KPOINTS}}-file format.
 === Step 3: Compute the phonon dispersion ===
-To plot the phonon dispersion, the tag {{TAGO|LPHON_DISPERSION|true}} must be set in the {{FILE|INCAR}} file.
+To compute the phonon dispersion, set {{TAGO|LPHON_DISPERSION|true}} in the {{FILE|INCAR}} file.
-The amount of information written to the {{FILE | OUTCAR}} file can be tuned using the ({{TAG|PHON_NWRITE}} tag).
+The amount of information written to the {{FILE|OUTCAR}} file can be tuned using the {{TAG|PHON_NWRITE}} tag.
 === Reading of force constants ===
 Steps 1-3 can be performed in one VASP calculation.
-However, generating the finite displacements in the supercell is a time-consuming process.
+However, generating the finite displacements in the supercell to compute force constants is time-consuming.
-In some situations, it can therefore be beneficial to skip this step and read the force constants from a previous run.
+It is possible to skip that step by providing force constants from a previous run.
-To do this, you set {{TAGO|LPHON_READ_FORCE_CONSTANTS|true}} and rename the <tt>vaspout.h5</tt> output file from the previous calculation to <tt>vaspin.h5</tt>.
+Rename the <tt>vaspout.h5</tt> output file from the previous calculation to <tt>vaspin.h5</tt>, set
-Together with {{TAGO|LPHON_DISPERSION|true}}, this shortcuts the VASP calculation to perform a phonon calculation without having to explicitly calculate the force constants again.
+  {{TAGBL|LPHON_READ_FORCE_CONSTANTS}} = True
+  {{TAGBL|LPHON_DISPERSION}} = True
+and provide a {{FILE|QPOINTS}} file.
-== Phonon density of states: Step-by-step instructions ==
+== Phonon DOS: Step-by-step instructions ==
 === Step 1: Compute the force constants ===
 Same as [[#Phonon dispersion: Step-by-step instructions#Step 1: Compute the force constants|above]].
-As explained earlier, this can also be skipped if done previously by using {{TAG|LPHON_READ_FORCE_CONSTANTS}}.
+This can be skipped by providing force constants in {{FILE|vaspin.h5}} and setting {{TAGO|LPHON_READ_FORCE_CONSTANTS|True}}.
 === Step 2: Specify a uniform '''q'''-point mesh ===
@@ Line 39: / Line 40: @@
 Create a {{FILE|QPOINTS}} file that specifies a sufficiently dense, uniform '''q'''-point mesh.
-=== Step 3: Compute the density of states ===
+=== Step 3: Compute the DOS ===
-Set {{TAGO|PHON_DOS|0|op=>}} in the {{FILE|INCAR}} file. The density of states is computed between
+Set {{TAGO|PHON_DOS|0|op=>}} in the {{FILE|INCAR}} file. The DOS is computed between
 <math>[\omega_{\text{min}}-5\sigma,\omega_{\text{max}}+5\sigma]</math> with
 <math>\omega_{\text{min}}</math> and
@@ Line 47: / Line 48: @@
 <math>\sigma</math> the broadening ({{TAG|PHON_SIGMA}}).
-The number of energy points in this energy range is specified by the {{TAG|PHON_NEDOS}} tag. To use a Gaussian-smearing method for the computation of the DOS set {{TAGO|PHON_DOS|1}} to use the tetrahedron method set {{TAGO|PHON_DOS|2}}.
+The number of energy points in this energy range is specified by the {{TAG|PHON_NEDOS}} tag. To use a Gaussian-smearing method for the computation of the DOS set {{TAGO|PHON_DOS|1}} or to use the tetrahedron method set {{TAGO|PHON_DOS|2}}.
 == Polar materials ==
 If the material is polar, i.e., two or more atoms in the unit cell carry non-zero Born effective charge tensors, the long-range dipole-dipole interaction has to be treated by [[Phonons:_Theory#Long-range_interatomic_force_constants_.28LO-TO_splitting.29|Ewald summation]].
 This is achieved by setting {{TAG|LPHON_POLAR}}=.TRUE., supplying the static dielectric tensor ({{TAG|PHON_DIELECTRIC}}) and the Born-effective charges ({{TAG|PHON_BORN_CHARGES}}).
-The values for these have to be obtained from a separate VASP calculation in the unit cell setting the {{TAG|LEPSILON}} or {{TAG|LCALCEPS}} INCAR tags.
+The values for these have to be obtained from a separate VASP calculation in the unit cell setting {{TAG|LEPSILON}} or {{TAG|LCALCEPS}}.
-Additionally, the user might specify a reciprocal space cutoff radius ({{TAG|PHON_G_CUTOFF}}) for the Ewald summation.
+Optionally, specify a reciprocal space cutoff radius ({{TAG|PHON_G_CUTOFF}}) for the Ewald summation.
 ==Related tags and articles==