My Community

Dear Users,

I am trying to calculate real and imaginary part of dielectric function using
LOPTICS tag. But in the output for Imaginary part and real part, I am getting 0 and 1 respectively. And at the end,
" The outermost node in the dielectric function lies at epsilon= -1.0 "
this is printing.
For your reference, I am attaching inputs and output (vasprun.xml and OUTCAR). Please help me out.

---
Chandan

Dear Chandan,

I found that some of the bands in your OUTCAR are partially occupied, but the PEAD method is not supported for metallic systems.
You can try to reduce the smearing width (SIGMA) to ensure that your system does have a bandgap.

Dear Alexey,

Thank you for your reply.
I reduced the SIGMA to 0.05 and, it worked!!
But, The plot is not smooth. How can I smooth the curve for publishable results.

---
Chandan

If you would like to apply larger smearing you can use the tetrahedron method (ISMEAR = -5). Also, you can try increasing the number of grid points in the dielectric function (NEDOS = 2001) to get a smoother spectrum.

Dear Alexey,

Thank you for your reply.
As I tried the tetrahedron method (ISMEAR = -5), but it printed the following error,
"Integration of the imaginary part of the dielectric function by means of the tetrahedron method is not supported for metallic systems."

Apart from this, I also tried to repeat the same calculation for hetrostructure and for that case same issue is there even I tried to reduce the SIGMA to 0.05.
Please comment on the same in details so that I can rectify the problem by myself.

---
Regards
Chandan

This error again is related to the fact that some of the orbitals are partially occupied.
Sorry, my previous answer might have been misleading. The broadening of the dielectric function is determined by the complex shift CSHIFT, so you can get a larger broadening of the dielectric function with ISMEAR=0 by increasing CSHIFT.

Dear Alexey,

Thank you for your reply.
it worked!!

---
Chandan