I am doing some calculations on a MnO2 system. In particular I am interested in calculating the Frequency dependant dielectric constant.
I have a doubt about it. And I hope you can find me to help the solution.
My Brillouin zone is sampled with 91 k-points.
I did 2 kinds of calculation.
1) The first using
ISMEAR=0
ISPIN=2
MAGMOM=4*3.0 8*0.
NSW = 0
ICHARG=11
LORBIT = 12
NBANDS=210.
NEDOS=2000
LOPTICS=.TRUE.
NPAR = 1
with a Cartesian set of kpoints
K-point for Band Structure
91
Cartesian
0.0000 0.0000 0.0000 1.0000 0.0000 ! G
0.0022 0.0013 0.0000 1.0000 0.0161
0.0044 0.0026 0.0000 1.0000 0.0322
0.0066 0.0038 0.0000 1.0000 0.0482
.....
.....
From this calculation I get in the OUTCAR
frequency dependent IMAGINARY DIELECTRIC FUNCTION (RPA, no local field effects)
E(ev) X Y Z XY YZ ZX
--------------------------------------------------------------------------------------
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
0.547643 0.050135 0.059462 0.014276 -0.008105 0.000863 -0.000499
1.095285 0.134056 0.161762 0.037418 -0.024076 0.000638 -0.000369
1.642928 0.417343 0.534233 0.117466 -0.101575 -0.023645 0.013675
2.190571 4.703351 5.844855 0.956823 -0.991947 -0.119657 0.069202
2.738214 6.863573 8.069441 1.785169 -1.047892 1.062545 -0.614505
....
....
....
And this is a reasonable result for a system whose bandgap on Gamma is about 1 eV, that is the strongest peak is at energy (2.73 eV)above the bandgap value.
2) The second is exactly as the previous one but using
ISMEAR=-5
I create the IBZKPT file with same numbers of k-points (91) for the tetrahedra method.
Then I copy IBZKPT in the KPOINTS file and launch the calculation with exactly the same
tags as the previous case (only difference is ISMEAR=-5)
What I found here is that in OUTCAR I get
frequency dependent IMAGINARY DIELECTRIC FUNCTION (RPA, no local field effects)
E(ev) X Y Z XY YZ ZX
--------------------------------------------------------------------------------------
0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
0.054559 68.267761 68.268570 0.273189 -0.000893 3.469354 -2.006440
0.109119 129.268666 129.109779 0.526207 0.137705 6.557587 -3.792464
0.163678 149.383779 148.854669 0.641054 0.459384 7.553307 -4.368321
0.218237 127.217257 126.092461 0.609472 0.977080 6.387441 -3.694064
0.272797 98.280428 96.620450 0.559899 1.442194 4.886132 -2.825808
0.327356 74.903259 73.396012 0.535605 1.309609 3.709142 -2.145116
0.381916 57.684651 56.665510 0.551249 0.885468 2.860069 -1.654070
0.436475 48.473535 47.859413 0.623937 0.533536 2.410018 -1.393792
0.491034 44.708449 44.416469 0.747526 0.253610 2.228833 -1.289005
0.545594 43.258343 43.174422 0.917235 0.072799 2.152265 -1.244724
....
....
which is far different from the previous case ( I suppose that the two cases should give rise to the same results). The stronget peak is at energy (0.16 eV) far below the bandgap of the system.
In other words, using the same number of K-points (91), but different ISMEAR I get 2 far different results for what regards the absorption energy (2 orders of magnitude different).
Most of all, in this latter case (ISMEAR=-5) the absorption is well below the bandgap of the system that as I mentioned before is about 1 eV, making me quite skeptical about the correctness of the latter results.
Can someone help me in understanding this?
Am I missing some basic physical concept?
Thanks in advance for the cooperation,
Giacomo
