4
$\begingroup$

Good evening everyone, With previous Transiesta versions, we would not have issues with finding the transport involving graphene. However, this was not the case with the latest version. We tried:

  • Decreasing energy cut off
  • Atomic force
  • Mixing pulay mixing factor
  • Changing BasisSize to less accurate ones DPZ to DZ

Despite getting perfect electrode cells and doing the analyze step. We relaxed our structure even and yet transiesta does not converge. I was wondering if we could get any advice on what further can we try.

Transport system via the attached POSCAR file Transport system from the scattering.fdf file

Here is the fdf file:

************************** Dump of input data file ****************************
SystemLabel scattering
SystemName scattering
SolutionMethod        Transiesta
%block ChemicalSpeciesLabel
 1 6 C
%endblock ChemicalSpeciesLabel
# Basis set variables
PAO.BasisType split
PAO.BasisSize       DZ
XC.functional   GGA
XC.Authors    PBE
# K-points
%block kgrid_Monkhorst_Pack
100 0   0   0.0
0   1   0   0.0
0   0   1   0.0
%endblock kgrid_Monkhorst_Pack
NumberOfSpecies        1
NumberOfAtoms         94
#Structure
LatticeConstant 1.0 Ang
%block LatticeVectors
 29.03931999 0.00000000 0.00000000
 -4.83546051 8.59115145 0.00000000
 -0.00002410 0.00001178 15.08028603
%endblock LatticeVectors
NumberOfAtoms 94
AtomicCoordinatesFormat Ang
%block AtomicCoordinatesAndAtomicSpecies
  0.12180800  1.37371700  7.54014700 1 #  1: C
 -1.08712900  3.52144600  7.54014400 1 #  2: C
 -2.29594100  5.66906700  7.54014100 1 #  3: C
 -3.50470500  7.81692300  7.54014400 1 #  4: C
  2.54169800  1.37367700  7.54014300 1 #  5: C
  1.33279500  3.52150200  7.54014200 1 #  6: C
  0.12401700  5.66921400  7.54013900 1 #  7: C
 -1.08484900  7.81693700  7.54013900 1 #  8: C
  4.96169300  1.37364700  7.54014400 1 #  9: C
  3.75284800  3.52153100  7.54014400 1 # 10: C
  2.54396900  5.66927600  7.54014300 1 # 11: C
  1.33514600  7.81696200  7.54014300 1 # 12: C
  7.38172100  1.37357900  7.54014700 1 # 13: C
  6.17284200  3.52134200  7.54014400 1 # 14: C
  4.96404200  5.66919200  7.54014500 1 # 15: C
  3.75522200  7.81698300  7.54014800 1 # 16: C
  1.32751800  0.64790300  7.54014500 1 # 17: C
  0.11859500  2.79581700  7.54014400 1 # 18: C
 -1.09017100  4.94343700  7.54014100 1 # 19: C
 -2.29901000  7.09112400  7.54014100 1 # 20: C
  3.74748600  0.64788400  7.54014200 1 # 21: C
  2.53860000  2.79579200  7.54014300 1 # 22: C
  1.32975200  4.94351300  7.54014100 1 # 23: C
  0.12092800  7.09123400  7.54013900 1 # 24: C
  6.16750000  0.64786200  7.54014500 1 # 25: C
  4.95859900  2.79571400  7.54014500 1 # 26: C
  3.74974100  4.94359000  7.54014500 1 # 27: C
  2.54095900  7.09134600  7.54014500 1 # 28: C
  8.58752100  0.64794400  7.54014900 1 # 29: C
  7.37857000  2.79562700  7.54014500 1 # 30: C
  6.16974300  4.94336400  7.54014300 1 # 31: C
  4.96094600  7.09124600  7.54014600 1 # 32: C
  9.80884929  1.28428982  7.54014751 1 #  1: C
  8.90792584  3.50100431  7.54014661 1 #  2: C
  7.56949711  5.62164311  7.54014930 1 #  3: C
  6.16488123  7.71072851  7.54014661 1 #  4: C
 12.04587741  1.35268747  7.54015470 1 #  5: C
 10.08577592  5.84389795  7.54015110 1 #  6: C
  8.51641225  7.86878636  7.54015380 1 #  7: C
 14.38130760  1.54842976  7.54015650 1 #  8: C
 12.88146544  3.61208026  7.54015380 1 #  9: C
 11.68798911  5.89703111  7.54015200 1 # 10: C
 10.83584711  8.08077261  7.54015110 1 # 11: C
 16.82711483  1.57341314  7.54015829 1 # 12: C
 15.48024078  3.67471390  7.54015650 1 # 13: C
 14.21442898  5.78377347  7.54015470 1 # 14: C
 13.19035357  7.98584050  7.54015290 1 # 15: C
 10.95285593  0.50411198  7.54014930 1 # 16: C
 10.10182400  2.68744344  7.54015200 1 # 17: C
  8.90802873  4.97178466  7.54015020 1 # 18: C
  7.40721639  7.03732497  7.54015380 1 # 19: C
 13.27309110  0.71560811  7.54015560 1 # 20: C
 11.70473809  2.74008629  7.54015380 1 # 21: C
  9.74336705  7.23108759  7.54014571 1 # 22: C
 15.62332254  0.87474637  7.54015739 1 # 23: C
 14.22086730  2.96370171  7.54015560 1 # 24: C
 12.88151893  5.08354039  7.54015380 1 # 25: C
 11.98042401  7.30045498  7.54015200 1 # 26: C
 18.02944081  0.82533015  7.54015919 1 # 27: C
 16.75488228  2.98438201  7.54015739 1 # 28: C
 15.47703940  5.08859455  7.54015560 1 # 29: C
 14.35553266  7.21529680  7.54015380 1 # 30: C
 19.27907400  1.47658400  7.54015000 1 #  1: C
 18.07010900  3.62423600  7.54016000 1 #  2: C
 16.86122900  5.77208000  7.54014500 1 #  3: C
 15.65245300  7.91986400  7.54012500 1 #  4: C
 21.69907900  1.47651500  7.54012600 1 #  5: C
 20.49015900  3.62427200  7.54017100 1 #  6: C
 19.28135500  5.77207800  7.54019200 1 #  7: C
 18.07239600  7.91986900  7.54013700 1 #  8: C
 24.11903300  1.47635700  7.54013100 1 #  9: C
 22.91007900  3.62410100  7.54013500 1 # 10: C
 21.70123800  5.77204200  7.54017100 1 # 11: C
 20.49240600  7.91980600  7.54017600 1 # 12: C
 26.53895600  1.47636400  7.54013800 1 # 13: C
 25.33004600  3.62412800  7.54013300 1 # 14: C
 24.12118800  5.77191100  7.54014100 1 # 15: C
 22.91228600  7.91973500  7.54015600 1 # 16: C
 20.48474800  0.75084200  7.54013400 1 # 17: C
 19.27585400  2.89867800  7.54016700 1 # 18: C
 18.06685600  5.04642600  7.54017200 1 # 19: C
 16.85806100  7.19413900  7.54012600 1 # 20: C
 22.90463300  0.75082000  7.54012300 1 # 21: C
 21.69564500  2.89854900  7.54014500 1 # 22: C
 20.48690900  5.04630000  7.54018600 1 # 23: C
 19.27807600  7.19423700  7.54017800 1 # 24: C
 25.32460700  0.75062400  7.54014800 1 # 25: C
 24.11574700  2.89843600  7.54012800 1 # 26: C
 22.90677300  5.04622800  7.54014600 1 # 27: C
 21.69792400  7.19407500  7.54017500 1 # 28: C
 27.74460500  0.75071600  7.54014900 1 # 29: C
 26.53568700  2.89842400  7.54013700 1 # 30: C
 25.32680600  5.04620300  7.54013900 1 # 31: C
 24.11798200  7.19407500  7.54014300 1 # 32: C
%endblock AtomicCoordinatesAndAtomicSpecies
NumberOfSpecies 1
%block ChemicalSpeciesLabel
 1 6 C
%endblock ChemicalSpeciesLabel
SpinPolarized    true
# General variables
MeshCutoff    200.0 Ry
MaxSCFIterations     100000
DM.MixingWeight       0.05          # New DM amount for next SCF cycle
DM.Tolerance          1.0E-4         # Tolerance in maximum difference
DM.UseSaveDM          true          # to use continuation files
DM.NumberPulay         6
PAO.SplitNorm    0.150000
PAO.EnergyShift    0.005000  Ry
# MD variables
MD.MaxForceTol 0.01 eV/Ang
MD.TypeOfRun        CG
MD.VariableCell     false
MD.NumCGSteps        0
MD.InitialTimeStep      1
MD.FinalTimeStep        1
MD.LengthTimeStep       1 fs
ElectronicTemperature 300 K
MD.MaxCGDispl 0.05 Bohr
SaveRho     true
SaveDeltaRho        true
UseSaveData     true
WriteCoorInital    true
WriteCoorStep    true
WriteForces     true
WriteKpoints        true
WriteEigenvalues        true
WriteKbands     true
WriteBands      true
WriteWaveFunctions      false
WriteMullikenPop        1
WriteDM     true
WriteCoorXmol       true
WriteCoorCerius     true
WriteMDXmol     true
WriteMDhistory      true
Diag.ParallelOverK  true
######TRANSIESTAPARAMETERS#####
TS.Voltage 0 eV
%block TS.ChemPots
  Left
  Right
%endblock TS.ChemPots
%block TS.ChemPot.Left
  mu V/2
  contour.eq
    begin
      c-Left
      t-Left
    end
%endblock TS.ChemPot.Left
%block TS.ChemPot.Right
  mu -V/2
  contour.eq
    begin
      c-Right
      t-Right
    end
%endblock TS.ChemPot.Right
TS.Contours.Eq.Pole 2.5 eV
%block TS.Contour.C-Left
  part circle
   from -40. eV + V/2 to -10 kT + V/2
     points 25
      method g-legendre
       opt right
%endblock TS.Contour.C-Left
%block TS.Contour.T-Left
  part tail
   from prev to inf
     points 10
      method g-fermi
%endblock TS.Contour.T-Left
%block TS.Contour.C-Right
  part circle
   from -40. eV -V/2 to -10 kT -V/2
     points 25
      method g-legendre
       opt right
%endblock TS.Contour.C-Right
%block TS.Contour.T-Right
  part tail
   from prev to inf
     points 10
      method g-fermi
%endblock TS.Contour.T-Right
%block TS.Contours.nEq
  neq-1
%endblock TS.Contours.nEq
%block TS.Contour.nEq.neq-1
  part line
   from -|V|/2 - 5 kT to |V|/2 + 5 kT
     delta 0.01 eV
      method mid-rule
%endblock TS.Contour.nEq.neq-1
%block TS.Elecs
  Left
  Right
%endblock TS.Elecs
%block TS.Elec.Left
  TSHS ./left.TSHS
  chem-pot Left
  semi-inf-dir -a1
  elec-pos begin 1
%endblock TS.Elec.Left
%block TS.Elec.Right
  TSHS ./right.TSHS
  chem-pot Right
  semi-inf-dir +a1
  elec-pos end -1
%endblock TS.Elec.Right
# TBtrans options
TBT.T.Eig 9
TBT.Elecs.Eta    0.0001000000 eV
%block TBT.Contours
  neq
%endblock TBT.Contours
%block TBT.Contour.neq
  part line
   from   -0.50000 eV to    0.50000 eV
    delta    0.00990 eV
     method mid-rule
%endblock TBT.Contour.neq
TBT.DOS.A T
TS.BTD.Pivot orb+none

I am happy to provide the files if needed but stack exchange does not seem to give this option.

Cheers, EL-Abed

#############################################################UPDATE

MAINLY WE HAVE CHANGED THE SYSTEM BY INCREASING THE NUMBER OF ELECTRODE LAYERS AND CHANGED THE MIXING FACTOR. THANK YOU ALL! #############################################################

$\endgroup$
3
  • 1
    $\begingroup$ I am pretty sure you are doing things wrong, there are tests in siesta with TranSiesta which converges quite fast. So one way or another something is off in your settings. $\endgroup$
    – nickpapior
    Sep 15, 2023 at 6:48
  • 1
    $\begingroup$ Could you please elaborate on the settings? Ideally showing your device and where the electrodes are in a picture. $\endgroup$
    – nickpapior
    Sep 20, 2023 at 6:30
  • 2
    $\begingroup$ @nickpapior sorry for all the mess. Hope everything has been fixed and we can get the physics going. If there is anything else I can help with please feel free to ask. $\endgroup$ Sep 20, 2023 at 13:32

1 Answer 1

3
$\begingroup$

The primary reason why convergence problems occur in TranSiesta is because of not properly testing the screening layers towards the electrodes. I also believe this is your main problem.
Again, problems are best understood by testing additional layers, then playing with SCF parameters, once a convergence can be reached, then one can go back to smaller systems and check if the same physics can be obtained with a smaller system, if needed.

Here would be my approach:

  1. Check whether additional screening layers can solve the issues.
  2. Reduce the mixing weight, a very low (it can be arbitrarily small) can cause very slow but certain convergence, a high value can result in non-convergence. The right mixing weight is system dependent and results in a fast, yet smooth convergence. There is no optimal mixing weight. I think this is different depending on the module in use. The force tolerance has nothing to do with the SCF convergence, it is not going to solve any problems.

Lastly, some comments. I think your k-points are wrong. You have no k-points along the b axis. This is clearly not ideal, and perhaps this can also be a problem for the convergence. Start with the electrodes, and find a suitable set of k-points.
The manual specifies that the k-points along the transport direction for the electrodes should be very high (to get a good accuracy). The transverse k-points requires the usual convergence approach. Just selecting 1, 1, 100 is a bad idea if one does not know the implications. Always, always ensure the k-point sampling converges the physical properties of the electrodes.

I can't see why reducing the mesh-cutoff would help, generally a larger mesh cutoff improves the precision. Same goes for the basis size.

$\endgroup$
7
  • $\begingroup$ Thank you very much Nick! I will test what was mentioned in your feedback and get back to you. But what I do not understand is the k points story. Why would increasing the k points help with convergence? Besides in the siesta manual, it was suggested to use 1 1 100 with the 100 in the transport direction which we did. Any further clarity would be helpful $\endgroup$ Sep 20, 2023 at 23:31
  • 1
    $\begingroup$ No that is not suggested in the manual. There are electrode k points, and device region k-points. The siesta manual does not suggest k-points in the transverse directions. Your foremost job is to get the electronic structure correct. And that requires you to test your electrodes. $\endgroup$
    – nickpapior
    Sep 21, 2023 at 4:37
  • $\begingroup$ I took it from page 134 of siesta 4.1.5: This implies that the proper bulk properties are obtained by a sufficiently high k-point sampling. If in doubt, use 100 k-points along the semi-infinite direction. Will try and change the k pts while changing other options and hope for the best $\endgroup$ Sep 22, 2023 at 0:55
  • $\begingroup$ Yes, and does that section refer to the electrodes, or the device? And does it mention anything about the transverse directions? :) $\endgroup$
    – nickpapior
    Sep 22, 2023 at 4:54
  • 1
    $\begingroup$ @nickpapior thank you very much! We added an extra layer and transiesta converged in much shorter time. Thank you! $\endgroup$ Nov 2, 2023 at 0:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .