Charge Density Waves with Abinit. Possible?

[spamrefuse]

Hi,

I'm asked to calculate Charge Density Waves (CDW) with Abinit.

I kind of understand the very basics of this phenomenon: in conductors
a small rearrangement of the ions will lower the total energy by creating
a small bandgap at the fermi energy.

Can somebody help me find the important parameters to do this with Abinit?

I think one only needs a supercell with dimensions approximately ~ 1/kf
(with kf = wavevector k at fermi level) and let the ions relax. The lowest
energy should then exist when the ions arrange themselves such that
the new arrangement has a lattice constant which creates a gap at the
fermi level.

Is it then possible to find the CDW simply by doing an ionic relaxation in a
large enough supercell?

Or is my thinking too simplistic?

Thank you!

Rob.

[mverstra]

The simplest way to see charge density waves is to do a phonon calculation with the appropriate phonon q, and finding an unstable mode. If you want to do a frozen CDW, it's possible too, and you'll have to get a supercell corresponding to q (as you suggest), with appropriate displaced atoms, and see what comes out. If your q is incommensurate, this may be impossible!

Matthieu

[spamrefuse]

Thanks for the explanation.
Whether it is commensurate or not will then also depend on how large size of a
supercell I can choose, doesn't it? There is always a supercell (however large enough)
which has the periodicity of both the original structure and the new CDW structure.....right?

In case I can do a 'frozen CDW', should I then find this by letting the ions relax?
Will the mimimum energy procedure easily rearrange the ions such that for a metal
a gap is created right at the fermi energy?
If yes, I may need to switch off Abinit's use of symmetry, since that may prevent
Abinit to find the CDW arrangement?

I just wonder if somebody has experience with this. I kind of understand the basics
how a CDW comes to existence and what (basic) effects it has.....but don't see
clearly how a DFT can be used to deal with this.

Rob.

[mverstra]

Thanks for the explanation.
Whether it is commensurate or not will then also depend on how large size of a
supercell I can choose, doesn't it? There is always a supercell (however large enough)
which has the periodicity of both the original structure and the new CDW structure.....right?

if q is not commensurate (q*a/2 pi is not a rational fraction) there is no supercell large enough.

In case I can do a 'frozen CDW', should I then find this by letting the ions relax?
Will the mimimum energy procedure easily rearrange the ions such that for a metal
a gap is created right at the fermi energy?
If yes, I may need to switch off Abinit's use of symmetry, since that may prevent
Abinit to find the CDW arrangement?

yes. The easiest is to displace a few atoms, but of course this biases the direction of the relaxation. If you have the phonons, as I suggested in the last post, you know which displacements will go soft and you can try each of them to see which gives you the lowest energy result.

I just wonder if somebody has experience with this. I kind of understand the basics
how a CDW comes to existence and what (basic) effects it has.....but don't see
clearly how a DFT can be used to deal with this.

Rob.

never done it this way - some things must exist in the literature, though

Matthieu