My Community

Hi all,
I am working on simulating XAS by DFT in VASP for a group of MoxSy-based chalcogels, including MoS2, MoS4 and Mo3S13. Based on experimental data I should see XAS peak at around just above 20000 eV for K-edge of Mo; however when the SCH calculation is done, I just get the zero values for IMAGINARY part of the dielectric function in the OUTCAR file. I believe the code works fine, because when I caluclate the L-edges for Mo, I can see the absorption values for those within the expected electron binding energy ranges, however VASP outputs the dielectric function in fortran format 12F6. I think what's happening here is that for K-edge the dielectric function is too small such that the leading decimals are all zero. Dielectric function of material will be vanishingly small as the energy increases which should be obvious as it would be harder to excite core electrons and the deeper the more challenging it is. It should scale as 1/energy^2.

So, first I want to know your thought on this that do you think there is another issue going on with my calculation ? being said that I have made various changes to my input files for this run including different values for : NBANDS, ENCUT, K-POINT density, different pseudopotentials, ...

And, if do you think the issue is formating as mentioned above, as there any way we can do to increase the precision of VASP output to 12E6 so that it's in exponential terms?
Thank you,

Here is my INCAR:
ALGO = Normal
CH_LSPEC = True
CH_NEDOS = 10000
CH_SIGMA = 0.05
CLL = 0
CLN = 1
CLNT = 1
CLZ = 1.0
EDIFF = 1e-05
EDIFFG = -0.01
ENCUT = 400
ICORELEVEL = 2
ISMEAR = 0
ISPIN = 2
LCHARG = False
LORBIT = 11
LREAL = Auto
LWAVE = False
NBANDS = 200
NCORE = 4
PREC = Accurate
SIGMA = 0.05
SYMPREC = 1e-05
NSIM = 16
KPAR = 1

Please try the GW POTCAR of Mo, because it has much more energy channels at higher energies.
This could possibly provide the missing intensities.
It's this one:
Mo_sv_GW
from the available POTCARs found here:
wiki/index.php/Available_PAW_potentials

[font=]Thank you for your reply. I have also tried GW pseudopotentials, still I do not see any non-zero values; I think the code is working fine just as I get the other edges absorption values but I believe for K-edge of Mo, dielectric function values are so small in this energy range ( which is about 20000eV) that VASP round them up to zero when creating the output files, so I am just not sure how can I modify the output to for example to exponential terms to produce results at higher precision.

Ok there is an undocumented flag that I used for development.
It's called

CH_AMPLIFICATION

. You can set a real value for this and the spectrum is multiplied by this during evaluation before printing.
So with this precision problems can be circumvented and you can check if there are really peaks beyond numerical noise.

so like CH_AMPLIFICATION = 12E6 ? also Could this be used in VASP 6.2.0?
Thank you

Yes it's available and use it like that.

it worked!! Thank you so much!