[Pw_forum] Silicon DOS issue - Pw_forum Digest, Vol 115, Issue 18

[nicvok]

Dear Xufeng, 

from a first view your results seem reliable. 
Indeed you are right that for an isolated, k-isotropic band one would expect a sqrt(E) behaviour of the DOS near the band bottom (for perfect sampling and integration :D). I think if you are not to harsh with yourself, you can see this in your calculations for the first 80meV. 
Have in mind that depending on your calculation (LDA/GGA, strained etc..) a band folds in around 100-150meV above the CBM and changes the sqrt(E)-behaviour. 
Second the „effective mass“ is strictly speaking k-dependent as 1/m \propto del^2 E_{k} / del k^2 \propto del/del k v_{k}, as you can see in fig 3

https://img2.picload.org/image/rowwrwir/1.png <https://img2.picload.org/image/rowwrwir/1.png>

where v_k is shown 100meV above the CBM. I used only a coarse mesh of ~4000k, but get similar results as you. Ignoring the oscillations, the dependence of DOS^2 is almost linear, too. 

So again, it depending on what you want to calculate afterwards, your results look pretty okay.

bests,
Nicki

DTU Denmark, CAMd

> Am 18.02.2017 um 12:00 schrieb pw_forum-request at pwscf.org:
> 
> Message: 3
> Date: Fri, 17 Feb 2017 21:05:59 -0500
> From: Xufeng Wang <wang159 at purdue.edu <mailto:wang159 at purdue.edu>>
> Subject: [Pw_forum] Silicon DOS issue
> To: pw_forum at pwscf.org <mailto:pw_forum at pwscf.org>
> Message-ID:
> 	<CAH5zo9+vg-7TfvrM2hC+6n_AjtZBU7dXpfHythuiM_BoBAP2hA at mail.gmail.com <mailto:CAH5zo9+vg-7TfvrM2hC+6n_AjtZBU7dXpfHythuiM_BoBAP2hA at mail.gmail.com>>
> Content-Type: text/plain; charset="utf-8"
> 
> Dear community,
> 
> I am unable to obtain the square root energy dependency of the bulk Silicon
> conduction band DOS, despite using a fine k grid (100x100x100).
> 
> I expect the DOS near conduction band edge (below the second conduction
> band of silicon) has a square root dependence with energy. I plotted out
> the (DOS)^2 vs energy I obtained here: http://imgur.com/a/Wu07z <http://imgur.com/a/Wu07z>
> 
> As you can see, (DOS)^2 vs. energy is not linear, meaning DOS is not having
> a square root dependence with energy. Is this an known problem or am I
> doing something wrong? Here is the nscf calculation input deck:
> 
> -------
> 
> &control
> 
>    prefix='silicon',
> 
>    pseudo_dir='../pseudo/'
> 
>    outdir = './results',
> 
>    tstress=.false.,
> 
>    tprnfor=.false.,
> 
>    calculation = 'nscf',
> 
>    verbosity='high',
> 
> /
> 
> &system
> 
>    ibrav=  2, celldm(1) =10.2623466921, nat=  2, ntyp= 1,
> 
>    ecutwfc = 30.0, ecutrho=300.0, input_dft='PBE', nbnd = 8,
> 
> /
> 
> &electrons
> 
>    diagonalization = 'david',
> 
>    conv_thr = 1d-08,
> 
>    mixing_mode = 'plain',
> 
>    mixing_beta = 0.7,
> 
> /
> 
> ATOMIC_SPECIES
> 
> Si  28.086  Si.pbe-n-rrkjus_psl.0.1.UPF
> 
> ATOMIC_POSITIONS
> 
> Si 0.00 0.00 0.00
> 
> Si 0.25 0.25 0.25
> 
> K_POINTS automatic
> 
> 100 100 100 0 0 0
> -------
> 
> Thanks!
> 
> Xufeng

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170218/da4e1f33/attachment.html>