Can KKR methoud give a better band gap for semiconductor compared with LDA?

Posted on : August 10, 2012 (Fri) 22:18:55

by tfl

Dear all,

I want to calculate the band structure of some semiconductor. Using traditional LDA method, it is found that the band gap are obviously underestimated. Would you please to tell me that KKR method can provide more accurate band gap values for semiconductor?

[Re:01] Can KKR methoud give a better band gap for semiconductor compared with LDA?

Posted on : December 09, 2012 (Sun) 10:13:46

by Koji Kobashi

Since Akai-KKR uses the LDA, the calculated bandgaps are lower than the actual values of semiconductors. The ASA improves it but only slightly.

[Re:02] Can KKR methoud give a better band gap for semiconductor compared with LDA?

Posted on : May 06, 2014 (Tue) 23:43:48

by ming-qiu Tan

I don't agree with that the ASA improves the band-gap calculations. Indeed ASA is the origin of inaccuracy in most ASA based DFT codes such as LMTO-ASA.

[Re:03] Can KKR methoud give a better band gap for semiconductor compared with LDA?

Posted on : May 06, 2014 (Tue) 23:48:54

by ming-qiu Tan

In order to improve the band-gap calculation, I suggest that you might use the DFT package incorporated with GW correction.

[Re:04] Can KKR methoud give a better band gap for semiconductor compared with LDA?

Posted on : August 14, 2014 (Thu) 19:14:04

by Koji Kobashi

Back to your first question, I have reconfirmed for diamond that the ASA gives the bandgap approximately 15% larger than the bandgap without the ASA using the Akai-KKR package. The use of the GW correction or any other methods to obtain bandgaps closer to the experimental value is a different issue, and one must go back to the known discussions held in the past on what the calculated bandgap is in the DFT.