How to handle imaginary phonon modes - Revision history

Csheldon: /* Step 2: Classify the modes */

2026-05-13T08:12:53Z

Step 2: Classify the modes

← Older revision		Revision as of 08:12, 13 May 2026
Line 49:		Line 49:
	\|}		\|}

	To identify the softest genuine optical mode with [[py4vasp]]:		To identify the softest genuine optical mode with {{py4vasp}}:

	<syntaxhighlight lang="python">		<syntaxhighlight lang="python">

Csheldon: /* Step 5: Extract the eigenvector and set up the IRC starting structure */

2026-05-13T08:12:35Z

Step 5: Extract the eigenvector and set up the IRC starting structure

← Older revision		Revision as of 08:12, 13 May 2026
Line 81:		Line 81:
	The intrinsic-reaction-coordinate (IRC) method requires the soft-mode eigenvector as a ''dimer-axis block'' appended to the {{FILE\|POSCAR}}. Use the high-symmetry saddle-point structure directly — no pre-displacement is needed because the forces are zero there by symmetry.		The intrinsic-reaction-coordinate (IRC) method requires the soft-mode eigenvector as a ''dimer-axis block'' appended to the {{FILE\|POSCAR}}. Use the high-symmetry saddle-point structure directly — no pre-displacement is needed because the forces are zero there by symmetry.

	Extract the eigenvector from [[py4vasp]] as shown in Step 2, or read it from {{FILE\|OUTCAR}}: locate the block "Eigenvectors and eigenvalues of the dynamical matrix", find the entry labelled <code>f/i=</code> for the imaginary mode, and copy the three Cartesian components listed for each atom.		Extract the eigenvector from {{py4vasp}} as shown in Step 2, or read it from {{FILE\|OUTCAR}}: locate the block "Eigenvectors and eigenvalues of the dynamical matrix", find the entry labelled <code>f/i=</code> for the imaginary mode, and copy the three Cartesian components listed for each atom.

	Append the eigenvector to the {{FILE\|POSCAR}} after a blank line following the last atomic position:		Append the eigenvector to the {{FILE\|POSCAR}} after a blank line following the last atomic position:

Schlipf at 06:33, 11 May 2026

2026-05-11T06:33:09Z

← Older revision		Revision as of 06:33, 11 May 2026
Line 66:		Line 66:
	See [[Computing_the_phonon_dispersion_and_DOS]] for the two-step procedure: a force-constants run followed by a dispersion run with {{TAG\|LPHON_DISPERSION}}{{=}}.TRUE. and a {{FILE\|QPOINTS}} file in line-mode format.		See [[Computing_the_phonon_dispersion_and_DOS]] for the two-step procedure: a force-constants run followed by a dispersion run with {{TAG\|LPHON_DISPERSION}}{{=}}.TRUE. and a {{FILE\|QPOINTS}} file in line-mode format.
	{{NB\|tip\|For cubic perovskite ABO₃ structures the key high-symmetry points are: Γ (0,0,0) — ferroelectric uniform polar displacement; R (½,½,½) — antiferrodistortive octahedron tilts; X (½,0,0) — cell-doubling along one axis; M (½,½,0) — cell-doubling along two axes. The labels and fractional coordinates depend on the space group of the specific structure.}}		{{NB\|tip\|For cubic perovskite ABO₃ structures the key high-symmetry points are: Γ (0,0,0) — ferroelectric uniform polar displacement; R (½,½,½) — antiferrodistortive octahedron tilts; X (½,0,0) — cell-doubling along one axis; M (½,½,0) — cell-doubling along two axes. The labels and fractional coordinates depend on the space group of the specific structure.}}

	{{NB\|mind\|This calculation does not include the macroscopic electric field responsible for LO-TO splitting near Γ. Ferroelectric branches are therefore degenerate at Γ and absolute optical frequencies near Γ may differ from experiment. To capture LO-TO splitting, obtain Born effective charges and the high-frequency dielectric tensor via {{TAG\|LEPSILON}}{{=}}.TRUE. or {{TAG\|LCALCEPS}}{{=}}.TRUE. and supply them via {{TAG\|PHON_BORN_CHARGES}} and {{TAG\|PHON_DIELECTRIC}}. This does not affect soft-mode identification.}}		{{NB\|mind\|This calculation does not include the macroscopic electric field responsible for LO-TO splitting near Γ. Ferroelectric branches are therefore degenerate at Γ and absolute optical frequencies near Γ may differ from experiment. To capture LO-TO splitting, obtain Born effective charges and the high-frequency dielectric tensor via {{TAG\|LEPSILON}}{{=}}.TRUE. or {{TAG\|LCALCEPS}}{{=}}.TRUE. and supply them via {{TAG\|PHON_BORN_CHARGES}} and {{TAG\|PHON_DIELECTRIC}}. This does not affect soft-mode identification.}}

Line 131:		Line 130:
	IRC (A): 0.05023 E(eV): -135.7812		IRC (A): 0.05023 E(eV): -135.7812
	</pre>		</pre>

	{{NB\|mind\|If no <code>IRC (A):</code> lines appear in {{FILE\|OUTCAR}}, the dimer-axis block in the {{FILE\|POSCAR}} was not read correctly. Check that a blank line separates the last atomic coordinate from the first eigenvector row, and that there is exactly one row per atom.}}		{{NB\|mind\|If no <code>IRC (A):</code> lines appear in {{FILE\|OUTCAR}}, the dimer-axis block in the {{FILE\|POSCAR}} was not read correctly. Check that a blank line separates the last atomic coordinate from the first eigenvector row, and that there is exactly one row per atom.}}

Line 146:		Line 144:

	Choose {{TAG\|NSW}} and {{TAG\|EDIFFG}} according to the required accuracy. For production calculations {{TAG\|EDIFFG}}{{=}}−0.01 is a common target; tighten to {{TAG\|EDIFFG}}{{=}}−0.001 if the relaxed structure will be used as input for a follow-up phonon calculation (Step 8).		Choose {{TAG\|NSW}} and {{TAG\|EDIFFG}} according to the required accuracy. For production calculations {{TAG\|EDIFFG}}{{=}}−0.01 is a common target; tighten to {{TAG\|EDIFFG}}{{=}}−0.001 if the relaxed structure will be used as input for a follow-up phonon calculation (Step 8).

	{{NB\|mind\|The frozen-phonon scan constrains displacements to a single mode eigenvector. Even at the energy minimum along λ, forces transverse to the scan direction may be non-zero when the mode is degenerate. The full {{TAG\|ISIF}}{{=}}3 relaxation in this step removes all residual forces and optimises the cell shape simultaneously.}}		{{NB\|mind\|The frozen-phonon scan constrains displacements to a single mode eigenvector. Even at the energy minimum along λ, forces transverse to the scan direction may be non-zero when the mode is degenerate. The full {{TAG\|ISIF}}{{=}}3 relaxation in this step removes all residual forces and optimises the cell shape simultaneously.}}

Schlipf: /* Step 1: Compute phonon frequencies */

2026-05-11T06:32:35Z

Step 1: Compute phonon frequencies

← Older revision		Revision as of 06:32, 11 May 2026
Line 27:		Line 27:
	The full displacement eigenvector for each mode is printed in the block "Eigenvectors and eigenvalues of the dynamical matrix" that follows the frequency list.		The full displacement eigenvector for each mode is printed in the block "Eigenvectors and eigenvalues of the dynamical matrix" that follows the frequency list.

	Alternatively, read frequencies and eigenvectors directly with [[py4vasp]]:		Alternatively, read frequencies and eigenvectors directly with {{py4vasp}}:

	<syntaxhighlight lang="python">		<syntaxhighlight lang="python">

Schlipf at 06:32, 11 May 2026

2026-05-11T06:32:15Z

← Older revision		Revision as of 06:32, 11 May 2026
Line 1:		Line 1:
	An imaginary (soft) phonon mode signals that the current structure is a saddle point on the potential-energy surface: displacing atoms along the mode eigenvector lowers the total energy. This page describes how to detect imaginary modes, locate the instability in the Brillouin zone, choose an appropriate working supercell, and follow the system to the true energy minimum using the intrinsic-reaction-coordinate (IRC) method or a frozen-phonon scan.		An imaginary (soft) phonon mode signals that the current structure is a saddle point on the potential-energy surface: displacing atoms along the mode eigenvector lowers the total energy. This page describes how to detect imaginary modes, locate the instability in the Brillouin zone, choose an appropriate working supercell, and follow the system to the true energy minimum using the intrinsic-reaction-coordinate (IRC) method or a frozen-phonon scan.

	{{NB\|mind\|An unrelaxed or poorly converged cell can produce spurious imaginary modes from [[Pulay stress]] or residual forces. If imaginary modes appear unexpectedly, first tighten the ionic relaxation ({{TAG\|ISIF}}{{=}}3, tight {{TAG\|EDIFFG}}) and recompute the phonons before drawing conclusions.}}		{{NB\|mind\|An unrelaxed or poorly converged cell can produce spurious imaginary modes from [[Pulay stress]] or residual forces. If imaginary modes appear unexpectedly, first tighten the ionic relaxation ({{TAG\|ISIF}}{{=}}3, tight {{TAG\|EDIFFG}}) and recompute the phonons before drawing conclusions.}}

Line 66:		Line 65:

	See [[Computing_the_phonon_dispersion_and_DOS]] for the two-step procedure: a force-constants run followed by a dispersion run with {{TAG\|LPHON_DISPERSION}}{{=}}.TRUE. and a {{FILE\|QPOINTS}} file in line-mode format.		See [[Computing_the_phonon_dispersion_and_DOS]] for the two-step procedure: a force-constants run followed by a dispersion run with {{TAG\|LPHON_DISPERSION}}{{=}}.TRUE. and a {{FILE\|QPOINTS}} file in line-mode format.

	{{NB\|tip\|For cubic perovskite ABO₃ structures the key high-symmetry points are: Γ (0,0,0) — ferroelectric uniform polar displacement; R (½,½,½) — antiferrodistortive octahedron tilts; X (½,0,0) — cell-doubling along one axis; M (½,½,0) — cell-doubling along two axes. The labels and fractional coordinates depend on the space group of the specific structure.}}		{{NB\|tip\|For cubic perovskite ABO₃ structures the key high-symmetry points are: Γ (0,0,0) — ferroelectric uniform polar displacement; R (½,½,½) — antiferrodistortive octahedron tilts; X (½,0,0) — cell-doubling along one axis; M (½,½,0) — cell-doubling along two axes. The labels and fractional coordinates depend on the space group of the specific structure.}}

Schlipf: Fix TAG template formatting in code blocks ({{TAG|X}} = Y, not {{TAG|X|Y}}); escape = in NB box; add backlinks (via VASP Copilot skill sync)

2026-05-08T15:08:59Z

Fix TAG template formatting in code blocks ({{TAG|X}} = Y, not {{TAG|X|Y}}); escape = in NB box; add backlinks (via VASP Copilot skill sync)

Show changes

Schlipf at 12:02, 8 May 2026

2026-05-08T12:02:22Z

← Older revision		Revision as of 12:02, 8 May 2026
Line 12:		Line 12:
	{{TAG\|NFREE\|2}} ! central differences (±POTIM)		{{TAG\|NFREE\|2}} ! central differences (±POTIM)
	{{TAG\|POTIM\|0.015}} ! displacement amplitude (Å)		{{TAG\|POTIM\|0.015}} ! displacement amplitude (Å)
	{{TAG\|PREC}} ~~= Accurate~~		{{TAG\|PREC\|Accurate}}
	{{TAG\|EDIFF\|1E-8}} ! tight convergence for reliable forces		{{TAG\|EDIFF\|1E-8}} ! tight convergence for reliable forces
	{{TAG\|NSW\|1}}		{{TAG\|NSW\|1}}

Schlipf: Add How-To: identifying and resolving imaginary (soft) phonon modes — covers mode detection, q-point mapping, supercell selection, IRC path, frozen-phonon scan, and relaxation (via VASP Copilot skill sync)

2026-05-08T12:00:41Z

Add How-To: identifying and resolving imaginary (soft) phonon modes — covers mode detection, q-point mapping, supercell selection, IRC path, frozen-phonon scan, and relaxation (via VASP Copilot skill sync)

Show changes

Schlipf: Created page with "_"

2026-05-08T11:58:45Z

Created page with "_"

New page

← Older revision		Revision as of 08:12, 13 May 2026
Line 81:		Line 81:
	The intrinsic-reaction-coordinate (IRC) method requires the soft-mode eigenvector as a ''dimer-axis block'' appended to the {{FILE\|POSCAR}}. Use the high-symmetry saddle-point structure directly — no pre-displacement is needed because the forces are zero there by symmetry.		The intrinsic-reaction-coordinate (IRC) method requires the soft-mode eigenvector as a ''dimer-axis block'' appended to the {{FILE\|POSCAR}}. Use the high-symmetry saddle-point structure directly — no pre-displacement is needed because the forces are zero there by symmetry.

	Extract the eigenvector from [[py4vasp]] as shown in Step 2, or read it from {{FILE\|OUTCAR}}: locate the block "Eigenvectors and eigenvalues of the dynamical matrix", find the entry labelled <code>f/i=</code> for the imaginary mode, and copy the three Cartesian components listed for each atom.		Extract the eigenvector from {{py4vasp}} as shown in Step 2, or read it from {{FILE\|OUTCAR}}: locate the block "Eigenvectors and eigenvalues of the dynamical matrix", find the entry labelled <code>f/i=</code> for the imaginary mode, and copy the three Cartesian components listed for each atom.

	Append the eigenvector to the {{FILE\|POSCAR}} after a blank line following the last atomic position:		Append the eigenvector to the {{FILE\|POSCAR}} after a blank line following the last atomic position:

← Older revision		Revision as of 12:02, 8 May 2026
Line 12:		Line 12:
	{{TAG\|NFREE\|2}} ! central differences (±POTIM)		{{TAG\|NFREE\|2}} ! central differences (±POTIM)
	{{TAG\|POTIM\|0.015}} ! displacement amplitude (Å)		{{TAG\|POTIM\|0.015}} ! displacement amplitude (Å)
	{{TAG\|PREC}} ~~= Accurate~~		{{TAG\|PREC\|Accurate}}
	{{TAG\|EDIFF\|1E-8}} ! tight convergence for reliable forces		{{TAG\|EDIFF\|1E-8}} ! tight convergence for reliable forces
	{{TAG\|NSW\|1}}		{{TAG\|NSW\|1}}