The ABINIT Users Mailing List ( CLOSED )

[Matthieu Verstraete <mjv500@york.ac.uk>]

Hello,

> I was reading the archives in the forum. From your email, it sounds like it
> is not possible in abinit to calculate the e-ph matrix element for
> transitions of electrons states which are not on Fermi surface.
By default the elphon matrix elements are integrated over the Fermi 
surface, as defined by the input eigenvalues and fermi energy. For an 
insulator you will get 0 everywhere.

> For example, the electrons could be excited by laser and lead to a wide 
> distribution of occupations around Fermi level. I have limited 
> experiences using abinit. Maybe this has been implemented?
2 things are possible:

1) you can increase the smearing temperature in the anaddb run. This will 
mimick the population being smeared out. If T is large enough you will 
occupy the conduction band and the matrix elements will be non zero

2) The other possibility is to fix the fermi energy (for the elphon part) 
by hand with the elph_fermie variable to anaddb 
(http://www.abinit.org/Infos_v5.5/users/anaddb_help.html#elph_fermie)

This means you are calculating the coupling between the 
phonons and the electrons which are at that energy.

The feature has never been tested, even though the implementation is 
simple. Use at your own risk!

The energy reference for elph_fermie is the same as the band energies in 
the GS part, so if you set elph_fermie = the fermi energy from the ground 
state run (to all digits) you will get the same result as if you do not 
set elph_fermie.

Please report any success or problems to the forum: I am very interested!

Matthieu