The ABINIT Users Mailing List ( CLOSED )

[zhangting <zhangting@pku.edu.cn>]

forumDear Erik Ylvisaker:

   Thanks very much for your help! 

        
==============================================================================
My own understanding of LO-TO splitting is fairly rudimentary, so anyone feel 
free to correct me where I'm wrong.  But I'll explain what I understand of it.
 
Optical phonons at the gamma point involve shifting of one or more atomic 
sublattices with respect to another.  In insulating materials, this creates a 
long range electric field which effectively increases the spring constant (if 
you want to think of it that way) and frequency of the vibrational mode.  
This causes the LO-TO splitting.  It's not an issue in metals because the 
electric field is screened out.  In the DFPT formalism, this effect is caused 
by a term which has 1/q in it, which presents problems for a Gamma point 
calculation, but other q points are treated correctly.  The resolution of the 
problem is to take the limit at q goes to zero along a particular direction 
of that term.  This results in LO-TO splitting which depends on direction, 
which is relevant for, say, a hexagonal system, but for a cubic system like 
CsCl I'd imagine there's only one relevant splitting.  LO-TO splitting is 
important to have the phonon dispersion along a particular direction be 
continuous from q=0 to q != 0 
 
You can't switch on LO-TO splitting in abinit while doing the phonon 
calculation.  You need to use the post-processing code anaddb.  There's some 
good discussion of how to do this in the second half of the response 
functions 1 tutorial.
                        

　　　　　　　　zhangting
　　　　　　　　zhangting@pku.edu.cn
　　　　　　　　　　2006-05-11