Contour plots for VB CB carrier densities

[ams_nanolab]

Dear Sir,

Is it possible to make contour plots of conduction band valleys as in

http://arxiv.org/pdf/1201.5284.pdf

http://www.sciencedirect.com/science/article/pii/S0038109812000889

with ATK?

[Anders Blom]

Sure, it's a fun little exercise in Python

Attached are 3 scripts which produce the plot also posted. There are endless possibilities for the contour plots, colors, etc, but the key to getting this to work is to set up a regular grid of k-points and compute the band structure for these, then extract the lowest conduction band.

I start with a standard graphene calculation. Then, I run "calc_bs.py" which does the regular sampling. Here you need to edit the number of points, depending on the accuracy of the plot you want.

Then run "cb_plot.py". In this you can play with the axis limited and the color/contour settings on lines 34-38, the rest of the code is more or less generic.

Have fun!

[ramkrishna]

how to specify the highest valence band in the "cb_plot.py"?

[Anders Blom]

Code: python

cb = [numpy.take(e,numpy.where(e<=0))[0][-1] for e in energies]

(You may want to change the variable name )

[ams_nanolab]

Could you modify the code to give normalized values?

[Anders Blom]

k or E? Normalized to what? Isn't it easier to see what's going on if you have absolute energies?

[ams_nanolab]

How should this script be modified for a simple orthorhombic unitcell.

[Anders Blom]

It's a lot simpler of course. You don't need the k-point mapping and griddata conversion, just do (from line 14)

Code: python

kx = bs.kpoints()[:,0]
ky = bs.kpoints()[:,1]

Then skip to

Code: python

P.contour(kx,ky,cb,20,colors=['w'])
P.contourf(kx,ky,cb,60,cmap=P.cm.hot)

And, in the other script (to generate the band structure) you can probably reduce the range of k-points to

Code: python

x = numpy.linspace(-1.0,1.0,Nk)

Of course, we are now assuming the kz doesn't matter, i.e. you are working with a 2D material.

[ams_nanolab]

I modified the script as you said (attached herewith) but I'm getting an error. What I'm doing wrong?

[Anders Blom]

Right, cb is just a long array, it needs to be reshaped to match the k-point grid. Try

cb = cb.reshape(len(ky),len(kx))
(possibly other order of kx and ky, if they are different)

after kx and ky are defined.

[ams_nanolab]

Still gives error:

Traceback (most recent call last):
  File "c:\users\ams\appdata\local\temp\2211826568161225.py", line 20, in <module>
    cb = cb.reshape(len(kx),len(ky))
ValueError: total size of new array must be unchanged

[Anders Blom]

If your NC file is not too large, can you send it? Should be easy but hard to solve without an example to work on.

[ams_nanolab]

sending via e-mail  for confidentiality

[Anders Blom]

See attached script, it's a bit different (simpler) since you can directly make the regular grid.

Run as

atkpython orthoplot.py myfile.nc

and if you want to save the plot automatically you can uncomment the last line.

[ams_nanolab]

It's working now, thanks