Solution LAB1 bands DOS diamond

From Wiki Max
Jump to navigation Jump to search

Band structure and DOS of Diamond

Given the calculations run previously to determine the converged set of k-points, ecutwfc, and lattice parameter, here we set the following:

 nk = 8
 ecutwfc = 60.0     # could be lowered further
 alat = 6.6694212

As a preliminary step:

  • make a new directory, e.g. called Diamond_BANDS
  • In a situation like this it is always a good idea to start over from scratch with a fresh scf calculation.
  • prepare an input file for a single scf run according to the parameters above.
 &CONTROL
   calculation="scf"
   [...]
 /
 &SYSTEM
   alat=...
   ecutwfc=...
 /
 [...]
 K_POINTS {automatic}
 8 8 8  0 0 0
  • run the calculation as:
 mpirun -np 2 pw.x < scf.in > scf.out
  • The use of mpirun -np 2 in the example above makes the calculation running on two MPI processes.

We start by addressing the calculation of the DOS.

We need now to write the input file for a non-self consistent calculation using a regular k-point grid. The input file looks like:

 >$ cat nscf_dos.in
 &CONTROL
   calculation="nscf"
   verbosity="high"     # not strictly needed but useful sometimes
   [...]
 /
 &SYSTEM
   alat=...
   ecutwfc=...
   nbnd=20              # let's include some empty states
 /
 [...]
 K_POINTS {automatic}
 20 20 20  0 0 0        # here we use a much denser kpt grid

Then we run it

 mpirun -np 2 pw.x < nscf_dos.in > nscf_dos.out

Given the number of k-points requested, the calculation may take some time.

At this point we can prepare an input file for the dos.x program,

 $> cat  dos.in
 &DOS
    prefix = 'diamond'
    outdir = './SCRATCH/'    
    degauss=0.025  !  [Ry]  broadening parameter
    fildos = 'dos_diamond_0.025Ry.dat'
 / 

The variable degauss controls the gaussian broadening used to approximate Dirac delta's in the calculation.

Try to change this value to get a feeling of the effect is has on the computed DOS.

An example is show in the figure below.

DOS of Diamond for different values of gaussian broadening.