[Pw_forum] Definition of Fermi level in PWscf

loc duong ding mambom1902 at yahoo.com
Mon Apr 13 02:32:06 CEST 2009



> http://www.democritos.it/pipermail/pw_forum/2008-August/009764.html
>
> http://www.democritos.it/pipermail/pw_forum/2008-August/009771.html
>
>>By definition: the Fermi energy is an energy which separates the valence (occupied) electrons from the >>conductivity (unoccupied) electrons.
>>In our DFT calculations there is no "real" Fermi energy (E_F), only the one for which \int_{E_min}^{E_F)N(E)dE = >>N is fulfilled (N(E) is the DOS, N is total number of valence electrons).
Because the DOS in the bandgap is 0, we can put the Fermi level everywhere in the bandgap. Thus, following this definition ( \int_{E_min}^{E_F)N(E)dE = N), we can have many (or infinite number of solution for Fermi level). But when using the code, there is only one value. I check DOS results. I see the Fermi level laying near the conduction band and It can be layed near the valance band. I wonder what is the critical of choosing Fermi level in Pwscf algorithm?

By the way, Fermi level is  equal to the chemical potential at 0K. So, It is a very important physical meaning and it must be a specific value at 0K in metal, semiconductor and insulator. Does choosing any value of Fermi level in the bandgap lose its physical meaning?

I post the input file of my calculation. It shows Fermi level moving around in the bandgap.

Model 1: C8O
&CONTROL
calculation = 'nscf',
prefix='C8O_LDA',
restart_mode = 'restart',
pseudo_dir ='./',
outdir='./'
tstress = .true. ,
tprnfor = .true. ,
nstep =  100  ,
etot_conv_thr = 1.0E-4 ,
forc_conv_thr = 1.0D-3 ,
dt = 20 ,
/
&SYSTEM
ibrav= 4,  celldm(1) = 9.4518, celldm(3)=2, nat = 9, ntyp = 2, nspin=2,
ecutwfc =20, ecutrho = 210, occupations='tetrahedra',
starting_magnetization(1) = 0.0,
starting_magnetization(2) = 0.3,
starting_magnetization(3) = 0.5
/
&ELECTRONS
startingwfc = 'atomic'
mixing_mode = 'plain'
mixing_beta = 0.6
conv_thr = 1.0e-6
electron_maxstep= 150
/
&IONS
upscale = 15
/
&CELL
cell_dynamics = 'bfgs' ,
press = 0.00 ,
wmass =  0.00150000  ,
/
ATOMIC_SPECIES
C  12.011  C.pz-rrkjus.UPF
O  15.9994 O.pz-rrkjus.UPF
ATOMIC_POSITIONS {angstrom}
C        4.132585975   4.832995059   0.302530806
C        5.456020552   5.597038625   0.302477533
O        4.794489489   5.215547072   1.572012872
C        1.698357183   4.846620591   0.084098745
C        2.888975237   5.584067069   0.084077187
C        2.927154758   2.718125473   0.084310286
C        4.161227149   3.380362792   0.084418771
C        0.411042607   2.684363968   0.116522861
C        1.676147051   3.414879352   0.116550940
K_POINTS {automatic}
9 9 1 0 0 0

Model 2: C8O2

&CONTROL
calculation = 'nscf',
prefix='C8O2_LDA',
restart_mode = 'restart',
pseudo_dir ='./',
outdir='./'
tstress = .true. ,
tprnfor = .true. ,
nstep =  100  ,
etot_conv_thr = 1.0E-4 ,
forc_conv_thr = 1.0D-3 ,
dt = 20 ,
/
&SYSTEM
ibrav= 4,  celldm(1) = 9.4518, celldm(3)=2, nat = 10, ntyp = 2, nspin=2,
ecutwfc =35, ecutrho = 210, occupations='tetrahedra',
starting_magnetization(1) = 0.0,
starting_magnetization(2) = 0.3,
starting_magnetization(3) = 0.5
/
&ELECTRONS
startingwfc = 'atomic'
mixing_mode = 'plain'
mixing_beta = 0.6
conv_thr = 1.0e-6
electron_maxstep= 150
/
&IONS
upscale = 15
/
&CELL
cell_dynamics = 'bfgs' ,
press = 0.00 ,
wmass =  0.00150000  ,
/
ATOMIC_SPECIES
C  12.011  C.pz-rrkjus.UPF
O  15.9994 O.pz-rrkjus.UPF
ATOMIC_POSITIONS {angstrom}
C        0.896514978   5.847447161   0.369684249
C        2.166323242   5.100308876   0.369426937
C        3.396879855   5.847835514   0.029522644
C        4.666783051   5.100388577   0.029864650
C        2.169717969   3.638682147   0.193058463
C        3.393684393   2.977581062   0.193531103
C        4.670322202   3.639102512   0.208350961
C        5.894369614   2.977777255   0.207910567
O        1.532383787   5.473020020   1.633770223
O        4.031020909   5.472856877  -1.234119799
K_POINTS {automatic}
9 9 1 0 0 0
-----------------------------------------------
Loc Duong Dinh
Ms-Ph.D Student