XAS for ZnO

[asiyeh_shokri2]

Hello,
I am using the INCAR file explained for XAS calculations for Diamond to plot the k-edge and L-edge XAS for Zn in wurtzite ZnO. but I get a compleet different diagram comparing the references! I change CLL and CLN for P and S orbitals and also the L and K edge.
I am new in VASP and very appresiate any help.

Best Regards,
Asiyeh

[martin.schlipf]

Please provide a complete set of input and output files so that we can investigate whether there is anything suspicious in your setup. Could you also elaborate on what differences you see compared to the literature? You shifted the onset of the peak to the correct position because the absolute position of the peak will be incorrect in DFT, didn't you?

[asiyeh_shokri2]

Hello,
Sorry, I wanted to be sure and I also used GW pseudo potential but my result is different. I read the tutorial about shifting the start position but it is not the only difference. I am sending the input and output file for the O k-edge of wurtzite ZnO. I compared it with different articles. I am sending the result of the workshop of wien2k (http://www.wien2k.at/events/ws2015/Core ... -short.pdf) and another article (DOI: 10.1063/1.3275738 · Source: IEEE Xplore) for instance.

Best Regards,
Asiyeh

[martin.schlipf]

The first obvious issue is the width of the spectrum. To address this you need to be aware that you expect to get a total width that is about the difference from the Fermi energy to the highest empty state. In your case that is ~10eV which is much smaller than the 30~50eV used in the other calculations. Please, increase NBANDS until your highest state is a little bit more than the desired width of your spectrum away from the Fermi energy.

The next thing you need to consider is the postprocessing. Be aware that the dielectric function has multiple directional components and you currently look at the isotropic spectrum at the moment. To get the directional resolution like in the first figure you need to change the script to select different components. Note that the broadening of the spectra needs to be carefully tweaked to compare to experiments.

Finally, consider the convergence of your calculation. You will probably need to increase the k-point mesh for a production calculation. Also check if a tighter EDIFF impacts the spectrum.

[asiyeh_shokri2]

I apresiate your help. I followed your advice and everything is ok now.

Best Regards,
Asiye

[asiyeh_shokri2]

Hello,
Thank you again for your kind reply. I have now another question. I saw that someone else has also had such question and there was not any answer for it. So, I want to again bring the question back:

Then how to use the calculated dielectric tensor to obtain the XAS of a specific electric polarization direction. (like the theta, phi direction in spherical coordinate system).
Also In the bash file in Diamond example there are different summations that I can not understand:

if (\$mode==0) {\$sum=\$help[2]+\$help[3]+\$help[4]+\$help[5]+\$help[6]+\$help[7];}
if (\$mode==1) {\$sum=\$help[4];}
if (\$mode==2) {\$sum=\$help[2]+\$help[3];}
if (\$mode==3) {\$sum=\$help[2]+\$help[3]+\$help[4];}
if (\$mode==4) {\$sum=(\$help[1]*\$help[1]*(\$help[2]+\$help[3]+\$help[4]+\$help[5]+\$help[6]+\$help[7]))**0.5;}
if (\$mode==5) {\$sum=(\$help[1]*\$help[1]*(\$help[4]))**0.5;}
if (\$mode==6) {\$sum=(\$help[1]*\$help[1]*(\$help[2]+\$help[3]))**0.5;}
if (\$mode==7) {\$sum=(\$help[1]*\$help[1]*(\$help[2]+\$help[3]+\$help[4]))**0.5;}
print \$help[1]," ",\$sum,"\n";

For instance, why should we do the summation between the imaginary dielectric parameters in mode "4" and then multiply it by energy?
and I also did not understand the -tauc parameter. I know if I run the bash file and using -normal and -tauc it will plot the mode 4. But I dont understand what the -tauc is and why the relation between energy and dielectric parameters are like the one in bash file.
Best regards
Asiyeh

[martin.schlipf]

For rotating the dielectric matrix into a specific coordinate system, you need to apply the corresponding linear transformation to it. Unfortunately, there is currently not a supported way to do that with the tools that we provide, so you need to come up with your own solution. It can be a good idea to test this for a simple system by rotating the coordinate system (i.e. changing the POSCAR file). After you apply the transformation, you should end up with the same dielectric tensor as for the original POSCAR file. When you tested this for a few systems, you can be quite sure that you implementation is correct and use it for the production calculation.

Regarding the different modes: Note that you typically want to reproduce experimental results. Because of the experimental procedure, these have a finite resolution leading to a broadening of the spectrum compared to the bare ab-initio calculation. It requires a lot of experience to choose an appropriate broadening for XAS. One consideration is that the experimental broadening can be energy dependent, so that higher peaks are broadened more than the lower energy ones.