ABINIT Discussion Forums

Dear ABINITers,

is there a way to directly calculate the effective mass with ABINIT? A naïve approach would be to fit a curve (locally: 2nd order polynomial) to the band structure and take the second derivative. Does this work? How can I get the band structure, e.g. near Gamma, with sufficient accuracy? Wannier interpolation?

Thanks & Best regards,

Martin Haeufel
TU Munich, WSI (T33)
http://www.wsi.tum.de

Martin wrote:Dear ABINITers,

is there a way to directly calculate the effective mass with ABINIT? A naïve approach would be to fit a curve (locally: 2nd order polynomial) to the band structure and take the second derivative. Does this work?

certainly - you just have to be careful with the units for the k-vectors: to get a correctly dimensioned m* you need to know the energy unit but also the delta k which you will be dividing or using to fit. The easiest probably is to rescale the branches of band structure which you are fitting, to express the k axis in bohr-1, then fit and you will have m* in clean atomic units. Just multiply the reduced k-point you input to abinit by the reciprocal lattice vectors (search for PRIM in the output file), and don't forget a 2*pi factor where needed: the reciprocal lattice in abinit is strict bohr-1, without the 2pi

How can I get the band structure, e.g. near Gamma, with sufficient accuracy? Wannier interpolation?

it would certainly be efficient, but the simplest would be a non-self consistent band structure calculation (as in the tutorials) with just a few k-points around gamma in the directions you are interested in.


matthieu

Thanks Matthieu!

matthieu wrote:the simplest would be a non-self consistent band structure calculation (as in the tutorials) with just a few k-points around gamma in the directions you are interested in.

Is there a way to combine this with GW or does one have to use interpolation techniques? Are there any tutorials/papers on this? Are there any systematic ways/ external tools to get a weighted set of k-points that contain the high-symmetry points for the given structural symmetry? I used to have access to Materials Studio at Qimonda, which I think had this as a built-in feature. Or would this just lead to a very inefficient k-point set and one should use interpolation anyway?

Martin