Pietro Bonfa - 2013-01-25

Dear Elk users,

I'm trying to reproduce the results obtained by Iimura and coworkers on LaFeAsOH with Wien2k .
First of all, here's my input:

tasks
  0
  43
spinpol
 .false.
mixtype
  2
nempty
  25
xctype
  21
trimvg
 .true.
sppath
  './sp/'
lradstp
 2
gmaxvr
 20.0 
rgkmax
 5
lmaxapw
 8
lmaxvr
 8
stype
 2
swidth
 0.01
scale
 10.784324

scale1
 1.0

scale2
 1.0

scale3
 1.0

avec
   1.000000000       0.000000000       0.000000000    
   0.000000000       1.000000000       0.000000000    
   0.000000000       0.000000000       1.531660000
atoms
   5                                    : nspecies
'La.in'                                 : spfname
   4                                    : natoms; atposl, bfcmt below
    0.00000000    0.25000000    0.14154000    0.00000000  0.00000000  0.00000000
    0.50000000    0.75000000    0.14154000    0.00000000  0.00000000  0.00000000
    0.00000000    0.75000000    0.85846000    0.00000000  0.00000000  0.00000000
    0.50000000    0.25000000    0.85846000    0.00000000  0.00000000  0.00000000
'H.in'                                  : spfname
   1                                    : natoms; atposl, bfcmt below
    0.25000000    0.00000000    0.00000000    0.00000000  0.00000000  0.00000000
'O.in'                                  : spfname
   3                                    : natoms; atposl, bfcmt below
    0.75000000    0.50000000    0.00000000    0.00000000  0.00000000  0.00000000
    0.75000000    0.00000000    0.00000000    0.00000000  0.00000000  0.00000000
    0.25000000    0.50000000    0.00000000    0.00000000  0.00000000  0.00000000
'Fe.in'                                 : spfname
   4                                    : natoms; atposl, bfcmt below
    0.25000000    0.00000000    0.50000000    0.00000000  0.00000000  0.1000000
    0.25000000    0.50000000    0.50000000    0.00000000  0.00000000  0.1000000
    0.75000000    0.50000000    0.50000000    0.00000000  0.00000000  -0.1000000
    0.75000000    0.00000000    0.50000000    0.00000000  0.00000000  -0.1000000
'As.in'                                 : spfname
   4                                    : natoms; atposl, bfcmt below
    0.00000000    0.25000000    0.65120000    0.00000000  0.00000000  0.00000000
    0.50000000    0.75000000    0.65120000    0.00000000  0.00000000  0.00000000
    0.00000000    0.75000000    0.34880000    0.00000000  0.00000000  0.00000000
    0.50000000    0.25000000    0.34880000    0.00000000  0.00000000  0.00000000
ngridk
  15  15  10

This input works perfectly when Oxygen is not substituted by Hydrogen.
Iimura and coworkers only report about the IBZ sampling (15x15x9) and rgkmax (= 9) in w2k.

With Elk, the above input file won't converge. After 3 or 4 self consistent loops the "linearization energy not found" warning comes up for Fe orbitals. This is probably probably related to the strange behavior of the charge in the mt of one of the iron atoms which is progressively reduced during the first self consistent steps.

For example:

+--------------------+
| Loop number :    1 |
+--------------------+

Energies :
 Fermi                       :     0.259168804121    
 sum of eigenvalues          :     -29262.1768218    
 electron kinetic            :      50981.8419783    
 core electron kinetic       :      49454.8332644    
 Coulomb                     :     -97976.1926183    
 Coulomb potential           :     -78414.7497705    
 nuclear-nuclear             :     -4579.08799822    
 electron-nuclear            :     -108379.459470    
 Hartree                     :      14982.3548496    
 Madelung                    :     -58768.8177331    
 xc potential                :     -1829.26902968    
 exchange                    :     -1380.72315929    
 correlation                 :     -24.7946060821    
 total energy                :     -48399.8684054

Density of states at Fermi energy :    261.0420605    
 (states/Hartree/unit cell)
Estimated indirect band gap :   0.6583245076E-04
 from k-point    240 to k-point    198

Charges :
 core                        :    302.0000000    
 valence                     :    187.0000000    
 interstitial                :    40.88251201    
 muffin-tins (core leakage)
  species :    1 (La)
   atom    1                 :    52.70027949     (  0.4973489587E-02)
   atom    2                 :    52.70027949     (  0.4973489587E-02)
   atom    3                 :    52.70027949     (  0.4973489587E-02)
   atom    4                 :    52.70027949     (  0.4973489587E-02)
  species :    2 (H)
   atom    1                 :   0.3713604113     (   0.000000000    )
  species :    3 (O)
   atom    1                 :    7.673914630     (  0.1006875885E-06)
   atom    2                 :    7.675033773     (  0.1002653187E-06)
   atom    3                 :    7.675033773     (  0.1002653187E-06)
  species :    4 (Fe)
   atom    1                 :    23.43715687     (  0.3033308715E-08)
   atom    2                 :    23.42658796     (  0.2907420082E-08)
   atom    3                 :    23.38218869     (  0.2996753068E-08)
   atom    4                 :    23.42658796     (  0.2907420082E-08)
  species :    5 (As)
   atom    1                 :    30.06212649     (  0.1541038813E-02)
   atom    2                 :    30.06212649     (  0.1541038813E-02)
   atom    3                 :    30.06212649     (  0.1541038813E-02)
   atom    4                 :    30.06212649     (  0.1541038813E-02)
 total in muffin-tins        :    448.1174880    
 total charge                :    489.0000000    
[...]
+--------------------+
| Loop number :    3 |
+--------------------+

Charges :
 core                        :    302.0000000    
 valence                     :    187.0000000    
 interstitial                :    41.61175904    
 muffin-tins (core leakage)
  species :    1 (La)
   atom    1                 :    52.81213091     (  0.5185657478E-02)
   atom    2                 :    52.81213091     (  0.5185657478E-02)
   atom    3                 :    52.81213091     (  0.5185657478E-02)
   atom    4                 :    52.81213091     (  0.5185657478E-02)
  species :    2 (H)
   atom    1                 :   0.3671138789     (   0.000000000    )
  species :    3 (O)
   atom    1                 :    7.078525890     (  0.1054435881E-06)
   atom    2                 :    7.078822604     (  0.1059136894E-06)
   atom    3                 :    7.078822604     (  0.1059136894E-06)
  species :    4 (Fe)
   atom    1                 :    24.92312462     (  0.2994950066E-08)
   atom    2                 :    24.01033422     (  0.3519415870E-08)
   atom    3                 :    20.50521783     (  0.3936833082E-08)
   atom    4                 :    24.01033422     (  0.3519415870E-08)
  species :    5 (As)
   atom    1                 :    30.27185536     (  0.1566028164E-02)
   atom    2                 :    30.27185536     (  0.1566028164E-02)
   atom    3                 :    30.27185536     (  0.1566028164E-02)
   atom    4                 :    30.27185536     (  0.1566028164E-02)
 total in muffin-tins        :    447.3882410    
 total charge                :    489.0000000

RMS change in effective potential (target) :    65.03838093     (  0.1000000000E-05)
Absolute change in total energy (target)   :    7.848225701     (  0.1000000000E-03)

With respect to the above input I tried:
* Reducing the rmt of the hydrogen species.
* Slightly changing the Fe species file as discussed in https://sourceforge.net/projects/elk/forums/forum/897820/topic/3441899
* increasing rgkmax up to 8
* changing mixtype

But none of the above trials solve the problem.

Can someone please give me a hint?

Thanks in advances for your time,
Pietro Bonfa'