LTRIPLET: Difference between revisions

Revision as of 14:50, 11 January 2022

LTRIPLET = .TRUE. | .FALSE.

Default: LTRIPLET

= .FALSE.

Description: LTRIPLET selects a triplet ansatz for Bethe-Salpether-Equation (BSE) calculations. This is entirely synonymous to LHARTREE = .FALSE.

For a usual BSE calculation (LHARTREE=.TRUE.; LTRIPLET = .FALSE.), the excited state corresponds to a singlet if the ground state is not spin-polarized (ISPIN=1) or anti-ferromagnetic (ISPIN=2, total magnetic moment 0). That is because the ansatz for the BSE calculation involves a hole and an electron pair prepared within each spin channel. For instance, an electron with up spin is removed from the ground-state determinant and placed with the same spin into a previously unoccupied orbital. If there are no separate spin channels this ansatz results in a net spin zero, which corresponds to a singlet state.

If LTRIPLET=.TRUE., VASP assumes that an up electron is removed from the ground-state determinant and placed as a down electron into a previously unoccupied orbital (spin flip). This ansatz corresponds to a triplet state as it has a net spin of one. Without spin-orbit coupling (LSORBIT = .FALSE.), the transition probability from singlet ground state to the triplet solutions will be exactly zero. These excitations correspond to dark, potentially long-lived triplet excitons. However, VASP calculates the transition probabilities incorrectly, by assuming that a spin flip excitation has the same transition probabilities as for a usual singlet excitation. The reported transition probabilities ("opticaltransitions" in vasprun.xml) are hence incorrect, they should be all zero. Likewise, the contributions to the dielectric function are zero for triplet excitations, and incorrectly reported in the vasprun.xml file.

To obtain accurate meaningful transition probabilities for the singlet and triplet excitations, spin-orbit coupling should be included by setting LSORBIT = .TRUE. in the INCAR file throughout the calculation, i.e., for the ground-state and BSE calculation. In this case, the singlet and triplet excitations are calculated all simultaneously, and proper transition probabilities are assigned to each transition.

If a non-magnetic material is calculated using ISPIN=2 for both the groundstate and BSE calculation, and a usual BSE calculation is performed (LHARTREE=.TRUE.; LTRIPLET = .FALSE.), the singlet as well as one set of the triplet excitations (that with $m_S=0$ ) is calculated. If LTRIPLET = .TRUE. is set for the BSE calculation, the other two sets of triplet excitations (those with $m_S=1$ and $m_S=-1$ ) are determined. Note that again for LTRIPLET = .FALSE., incorrect non-zero transition propabilties are reported in the vasprun.xml file (they should be all zero).

For a BSE calculation (LHARTREE=.TRUE.; LTRIPLET = .FALSE.) based on a spin-polarized ground-state calculation (ISPIN=2), VASP yields one set of singlet solutions and one set of triplet solutions. With LTRIPLET=.TRUE., VASP will obtain the remaining two sets of triplet solutions.

Warning: LTRIPLET=.TRUE. has not been extensively tested in combination with spin-polarized ground states (ISPIN=2).

Warning: Not sure whether the BSEFATBAND approach can be combined with the LTRIPLET tag.

References

Contents

@@ Line 2: / Line 2: @@
 {{DEF|LTRIPLET|.FALSE.|}}
-Description: {{TAG|LTRIPLET}} selects a triplet ansatz for Bethe-Salpether-Equation (BSE) calculations.
+Description: {{TAG|LTRIPLET}} selects a triplet ansatz for Bethe-Salpether-Equation (BSE) calculations. This is entirely synonymous to  {{TAG|LHARTREE}} = .FALSE.
 ----
-For a usual BSE calculation ({{TAG|LHARTREE}}=.TRUE.; {{TAG|LTRIPLET}} = .FALSE.), the excited state corresponds to a singlet if the ground state is not spin-polarized ({{TAG|ISPIN}}=1). That is because the ansatz for the BSE calculation involves a hole and an electron pair prepared within each spin channel. For instance, an electron with up spin is removed from the ground-state determinant and placed with the same spin into a previously unoccupied orbital. If there are no seperate spin channels this ansatz results in a net spin zero, which corresponds to a singlet state.
+For a usual BSE calculation ({{TAG|LHARTREE}}=.TRUE.; {{TAG|LTRIPLET}} = .FALSE.), the excited state corresponds to a singlet if the ground state is not spin-polarized ({{TAG|ISPIN}}=1) or anti-ferromagnetic ({{TAG|ISPIN}}=2, total magnetic moment 0). That is because the ansatz for the BSE calculation involves a hole and an electron pair prepared within each spin channel. For instance, an electron with up spin is removed from the ground-state determinant and placed with the same spin into a previously unoccupied orbital. If there are no separate spin channels this ansatz results in a net spin zero, which corresponds to a singlet state.
+If {{TAG|LTRIPLET}}=.TRUE., VASP assumes that an ''up'' electron is removed from the ground-state determinant and placed as a ''down'' electron into a previously unoccupied orbital (spin flip). This ansatz corresponds to a triplet state as it has a net spin of one. Without spin-orbit coupling ({{TAG|LSORBIT}} = .FALSE.), the transition probability from singlet ground state to the triplet solutions will be exactly zero. These excitations correspond to dark, potentially long-lived triplet excitons. However, VASP calculates the transition probabilities incorrectly, by assuming that a spin flip excitation has the same transition probabilities as for a usual singlet excitation.
+The reported transition probabilities ("opticaltransitions" in {{TAG|vasprun.xml}}) are hence incorrect, they should be all zero. Likewise, the contributions to the dielectric function are zero for triplet excitations,
+and incorrectly reported in the {{TAG|vasprun.xml}} file.
+To obtain accurate meaningful transition probabilities for the singlet and triplet excitations, spin-orbit coupling should be included by setting {{TAG|LSORBIT}} = .TRUE. in the {{FILE|INCAR}} file throughout the calculation, i.e., for  the ground-state and BSE calculation.  In this case, the singlet and triplet excitations are calculated all simultaneously, and proper transition probabilities are assigned to each transition.
+If a non-magnetic material is calculated using {{TAG|ISPIN}}=2 for both the groundstate and BSE calculation, and a usual BSE calculation is performed ({{TAG|LHARTREE}}=.TRUE.; {{TAG|LTRIPLET}} = .FALSE.), the
+singlet as well as one set of the triplet excitations (that with <math> m_S=0</math> ) is calculated. If {{TAG|LTRIPLET}} = .TRUE. is set for the BSE calculation, the other two sets of triplet excitations (those with <math> m_S=1</math> and <math> m_S=-1</math>) are determined. Note that again for {{TAG|LTRIPLET}} = .FALSE., incorrect non-zero transition propabilties are reported in the {{TAG|vasprun.xml}} file (they should be all zero).
-If {{TAG|LTRIPLET}}=.TRUE., VASP assumes that an ''up'' electron is removed from the ground-state determinant and places a ''down'' electron in a previously unoccupied orbital. This ansatz corresponds to a triplet state as it has a net spin one. Without spin-orbit coupling ({{TAG|LSORBIT}} = .FALSE.), many (or maybe even all) of the triplet solutions will have zero transition probabilities. These correspond to dark, potentially long-lived triplet excitons. To obtain non-zero transition probabilities for the triplets, spin-orbit coupling should be included by setting {{TAG|LSORBIT}} = .TRUE. in the {{FILE|INCAR}} file throughout the calculation, i.e., also in the ground-state calculation.
 For a BSE calculation ({{TAG|LHARTREE}}=.TRUE.; {{TAG|LTRIPLET}} = .FALSE.) based on a spin-polarized ground-state calculation ({{TAG|ISPIN}}=2), VASP yields one set of singlet solutions and one set of triplet solutions. With {{TAG|LTRIPLET}}=.TRUE., VASP will obtain the remaining two sets of triplet solutions.

Revision as of 14:50, 11 January 2022

Related tags and articles

References