Band Structure of ICE I

[luca]

Hi all,

I am a bit confused.
I do a ground state calculation on Ice I using PBE functional.
I get a huge band gap of nearly 8 eV over all k points which seems to be correct as Ice is an insulator.
I do the same calculation using Abinit getting NO band gap and hence a metallic phase.

Abinit uses pseudo potentials and plane waves.
Attached is my Abinit ice I input file.

Bests
Luca

 

[Anders Blom]

First of all I didn't understand which program you used when you got the band gap. Second, although you can run Abinit from within ATK, support for Abinit is a bit limited if you are not using it from our interface. You have a very advanced input script for Abinit with many options, and I have to admit I don't know them. You may have better luck asking this on the abinit mailing list.

[luca]

I first used ATK to get the band gap of ice from band structure plot. This is a huge averaged band gap of 7-8 eV over all k points.
After that i used the external Abinit program which gives me no band gap for ice and
hence a metallic phase using the same functional for both programs (atk and abinit).

Any way, let me ask you in another way, what are the major differences between ATK and Abinit concerning band structure calculations ?

I asked it on Abinit Forum too, but still no answer.

Bests

[Anders Blom]

The difference between ATK and Abinit is, fundamentally, that ATK uses localized basis sets and Abinit plane waves. How this affects accuracy, performance, etc, is a very complex question.

The reason I can't really comment on your Abinit calculation is that, as mentioned, it contains a huge amount of very specialized keywords. As far as I can judge this is because of the use of GW, so you are not really using the same approach in ATK and Abinit, hence the results can't be compared. Also, the "basic" part with k-point sampling is commented out, so if I'm not mistaken you are just doing a Gamma point calculation - was that the case in ATK too?