Calculation of effective charges always fails with PAW approach. Example of Effective charge tensors calculated with DFPT approach with norm conserving pseudo potentials:
Code: Select all
atom displacement
1 1 -1.362653E+00 -7.005936E-17 2.734595E-15
1 2 -7.005936E-17 -1.362653E+00 2.734595E-15
1 3 0.000000E+00 0.000000E+00 -2.526618E+00
The other way to calculate effective charges is to compute forces with finite electric field, here is example for the same system with NCPP.
Code: Select all
berryopt11 0
berryopt21 4
efield11 0.00000000E+00 0.00000000E+00 0.00000000E+00
efield21 1.00000000E-04 0.00000000E+00 0.00000000E+00
fcart11 1.8740197435E-27 2.8110296153E-27 4.3509916121E-05
1.8740197435E-27 2.8110296153E-27 4.3509916121E-05
0.0000000000E+00 9.3700987177E-28 4.0128747685E-06
0.0000000000E+00 9.3700987177E-28 4.0128747685E-06
0.0000000000E+00 9.3700987177E-28 -4.0128747685E-06
0.0000000000E+00 9.3700987177E-28 -4.0128747685E-06
-1.8740197435E-27 -4.6850493588E-27 -4.3509916121E-05
-1.8740197435E-27 -4.6850493588E-27 -4.3509916121E-05
fcart21 -1.3656955412E-04 -3.3732355384E-26 4.3633783390E-05
-1.3656955412E-04 -3.3732355384E-26 4.3633783390E-05
1.3656955412E-04 1.1244118461E-26 3.9358044430E-06
1.3656955412E-04 1.1244118461E-26 3.9358044430E-06
1.3656955412E-04 -3.7480394871E-27 -3.9358044430E-06
1.3656955412E-04 -3.7480394871E-27 -3.9358044430E-06
-1.3656955412E-04 2.6236276410E-26 -4.3633783390E-05
-1.3656955412E-04 2.6236276410E-26 -4.3633783390E-05
Thus for atom #1 Z*_xx = -1.3656955412E-04 / 1.00000000E-04 = -1.3656955412 wich is in good agreement with DFPT (see top).
The PAW method totally fails. Results with DFPT:
Code: Select all
Effective charge tensors
atom displacement
1 1 -5.059849E+00 9.548787E-16 -1.088279E-15
1 2 1.365618E-15 -5.059849E+00 -6.283949E-15
1 3 2.053695E-16 -2.053695E-16 -6.386698E+00
It is too high with respect to NCPP
the finite field method show an appearance of extra forces in Berry phase approach
Code: Select all
berryopt11 0
berryopt21 4
efield11 0.00000000E+00 0.00000000E+00 0.00000000E+00
efield21 1.00000000E-04 0.00000000E+00 0.00000000E+00
fcart11 0.0000000000E+00 0.0000000000E+00 -8.7573174285E-07
0.0000000000E+00 0.0000000000E+00 -8.7453507066E-07
0.0000000000E+00 0.0000000000E+00 -1.0379378587E-05
0.0000000000E+00 0.0000000000E+00 -1.0379283122E-05
0.0000000000E+00 0.0000000000E+00 1.0379378587E-05
0.0000000000E+00 0.0000000000E+00 1.0379283122E-05
0.0000000000E+00 0.0000000000E+00 8.7573174285E-07
0.0000000000E+00 0.0000000000E+00 8.7453507066E-07
fcart21 -7.0943300118E-04 0.0000000000E+00 4.2677714443E-05
-7.0943058316E-04 0.0000000000E+00 4.2399631840E-05
7.0941888806E-04 0.0000000000E+00 -1.3672525219E-05
7.0944469628E-04 0.0000000000E+00 -1.3818698694E-05
7.0941888806E-04 0.0000000000E+00 1.3672525219E-05
7.0944469628E-04 0.0000000000E+00 1.3818698694E-05
-7.0943300118E-04 0.0000000000E+00 -4.2677714443E-05
-7.0943058316E-04 0.0000000000E+00 -4.2399631840E-05
The finite field approach represent the same error in calculation of effective charges. The forces is really bad in x direction and even in z direction (the direction with electric field OFF) forces are totally wrong (these should be the same as the ground state forces). Is there something specific one have to include in input file when calculationg effective charges with PAW method? There was a thread in the forum related to error in DFPT part of code related to phonon calculations and that was fixed in Version 7.10.5 of ABINIT, but the calculation of effective charge is still broken.