problem about berryphase term in calculation of ferroelectric polarization
Posted: Tue Dec 13, 2011 4:31 pm
When I used the berry phase method combined with non-collinear or spin-orbit coupling to calculate the ferroelectric polarization of multiferroics (i.e.,the polarization is induced by special magnetic orders which destroy the inversion symmetry), I found the berryphase terms "e<r>_bp" are nonzero along the three directions of cartesian coordinate.
While, actually, the polarization only presents in one direction.
For example, the real polarization is along the c axis in a material with R-3 symmetry, while the calculated berryphase terms along G1 and G2 (in ab plane) are
G1:
e<r>_ev=( 0.00000 0.00000 0.00031 ) e*Angst
e<r>_bp=( 0.06706 -0.03871 0.00000 ) e*Angst
G2:
e<r>_ev=( 0.00000 0.00000 0.00031 ) e*Angst
e<r>_bp=( 0.00000 -0.18442 0.00000 ) e*Angst
How should I understand the berryphase terms present in the ab plane?
In my opinion, the electron terms "e<r>_ev" can rightly reflect the polar direction except the minor values.
While, sometimes the electron term e<r>_ev equals zero and e<r>_ev has nonzero value. Does this means the polarization present?
Another question: the spin-orbit coupling should be switched on in the previous scf calculations or the succedent nscf process ( switch on berryphase ) ?
Thanks,
Junting Zhang
<span class='smallblacktext'>[ Edited ]</span>
While, actually, the polarization only presents in one direction.
For example, the real polarization is along the c axis in a material with R-3 symmetry, while the calculated berryphase terms along G1 and G2 (in ab plane) are
G1:
e<r>_ev=( 0.00000 0.00000 0.00031 ) e*Angst
e<r>_bp=( 0.06706 -0.03871 0.00000 ) e*Angst
G2:
e<r>_ev=( 0.00000 0.00000 0.00031 ) e*Angst
e<r>_bp=( 0.00000 -0.18442 0.00000 ) e*Angst
How should I understand the berryphase terms present in the ab plane?
In my opinion, the electron terms "e<r>_ev" can rightly reflect the polar direction except the minor values.
While, sometimes the electron term e<r>_ev equals zero and e<r>_ev has nonzero value. Does this means the polarization present?
Another question: the spin-orbit coupling should be switched on in the previous scf calculations or the succedent nscf process ( switch on berryphase ) ?
Thanks,
Junting Zhang
<span class='smallblacktext'>[ Edited ]</span>