[Pw_forum] Shift of Energy
ttduyle at gmail.com
Mon Apr 4 17:04:51 CEST 2011
On Mon, Apr 4, 2011 at 10:32 AM, pari shok <parishok at gmail.com> wrote:
> Dear All,
> Thank you very much for your help.
> Please let me ask my final question w.r.t DOS shift of energy of SiC
> The total energy of the 8-atom cell of SiC is: -76.276 Ry, while that of
> 72-atom SiC is -694 Ry. Neglecting almost -7 Ry, the total energy is scaling
If you are doing correctly, the results must be identical. So make
sure to check the structures, and the convergence of total energy vs.
cutoff energy and k-point sampling.
> However, I still see the shift of energy in DOS diagram of two slabs (the
> max and min of bandgap shifts, although the gap is almost 2.1 eV for both
> 4H-SiC slabs). I was wondering whether this shift of energy is inevitable or
> it is as the result of my input file.
Slab? I though you are performing bulk calculation. Anyway, the shift
of energies' levels is normal. You should set the reference as center
of the gap (set it to be zero) then compare the DOS or band
By the way, if your bigger slab experiences the reconstruction, the
two slabs (8 atoms and 72 atoms) won't be identical => you should not
get the same results in this case. But 7Ry seems to be huge.
> I kept k points (w.r.t. crystal), and cut off energy the same for both
> structures. Should I scale the cut off energy as well. If positive, is there
Okie, but you should scale the k-point sampling accordingly because
the BZ of 72 atoms cell is smaller. But this won't harm the results if
total energy is already converged vs. k-point sampling.
> any relation between scaling the cell and cut off energy.
No. Cut off energy depends on only pseudo potential.
> As you know, I need to find a reference for my further calculations.
> I really really appreciate your help.
> P Shok
Department of Physics
University of Central Florida.
"Men don't need hand to do things"
> Dear All,
> let me clarify a little point.
> The potential (hence the eigenvalues) of any calculation in
> periodic boundary conditions, not just a plane-wave one , is defined up
> to an additive constant... In a finite system a reeference can be fixed
> by requiring that the potential goes to zero at infinite distance but in
> a periodic system this cannot be defined as there is nothing like a
> point at infinite distance ...
> In most codes (and in pw.x) this arbitrary constant is fixed by
> setting the average electrostatic potential to zero. If the supercell is
> properly constructed, with the properly scaled dimensions, the same
> cutoff, equivalent k-points etc.. the two calculations should be
> equivalent and the eigenvalues should be exactly mapped, the average
> potential should be the same, the total energy should be an exact
> multiple of the one of the original cell.
> If this does not happen one is NOT doing the supercell calculation properly.
> So if when doing a supercell you do not get the scaled result check your
> system definition.
> On 04/04/2011 11:46 AM, Giuseppe Mattioli wrote:
>> Dear all
>> I suspect that a a little misunderstanding has been going on here...
>> I try to reformulate the question: in the case of a 8-atoms cell the
>> band maximum (VBM) has been found around 10 eV and the conduction band
>> minimum (CBM) around 12 eV with a 2 eV band gap. So there is a shift in
>> position of VBM and CBM when calculated in a 72-atoms supercell, but the
>> gap value is the same. Then, the answer to P Shok could be: plane wave
>> pseudopotential calculations do not ensure an universal alignment of
>> eigenvalues. You should use some "internal" reference, like the 1s
>> of an He atom, in your cell; or trust that the two VBM values are aligned
>> On Monday 04 April 2011 09:39:18 Gabriele Sclauzero wrote:
>>> I perfectly agree with Eyvaz! Moreover the experimental bandgap is
>>> underestimated within LDA or GGA, and for 4H-SiC you would get about
>>> 2.2-2.3 eV with LDA.
>>> How do you measure the bandgap? Beware that in the fundamental cell
>>> bandgap is not a direct one, but it might become such when you use a
>>> supercell (because of the refolding of the Brillouin zones).
>>> Il giorno 03/apr/2011, alle ore 10.59, Eyvaz Isaev ha scritto:
>>>>> The bandgap of 8-atoms is around 10-12 eV,
>>>>> but that of 72-atom is around 5-7 eV.
>>>> This is unbelievable, as SiC is an indirect band gap semiconductor and
>>>> the bandgap is around (2.5 - 3)eV depending on crystal modification
>>>> (3C-SiC, 2H-SiC; 4H-SiC; 6H-SiC, etc.).
>>>> See http://www.matprop.ru/SiC_bandstr
>>>> So, check carefully your input/output files.
>>>> Prof. Eyvaz Isaev,
>>>> Department of Physics, Chemistry, and Biology (IFM), Linkoping
>>>> University, Sweden Theoretical Physics Department, Moscow State
>>>> of Steel& Alloys, Russia, isaev at ifm.liu.se, eyvaz_isaev at yahoo.com
>>>> From: pari shok<parishok at gmail.com>
>>>> To: pw_forum at pwscf.org
>>>> Sent: Sun, April 3, 2011 1:34:51 AM
>>>> Subject: [Pw_forum] shift of energy
>>>> Dear Paolo,
>>>> DOS of 72-atom SiC shows a shift of energy with respect to 8-atom
>>>> bandgap of 8-atoms is around 10-12 eV, but that of 72-atom is around 5-7
>>>> eV. Would you please help me to understand this shift.
>>>> Thanks again.
>>>> P Shok
>>>> On Apr 1, 2011, at 21:36 , pari shok wrote:
>>>>> The DOS of 72-atom SiC (supercell) shows a shift of energy.
>>>> a shift with respect to what?
>>>> Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
>>>> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
>>>> Phone +39-0432-558216, fax +39-0432-558222
>>>> Pw_forum mailing list
>>>> Pw_forum at pwscf.org
>>> § Gabriele Sclauzero, EPFL SB ITP CSEA
>>> PH H2 462, Station 3, CH-1015 Lausanne
> Pw_forum mailing list
> Pw_forum at pwscf.org
More information about the users