[DFTB-Plus-User] MgO crystal GFN2-xtb does not converge

Evgeny Permyakov permeakra at yandex.ru
Tue Mar 5 19:22:37 CET 2024


Hello!

I'm working on a project involving calculations of doped MgO structures. 
At first step I'm trying to model solid MgO and decided to use /xtb/ 
because the required parameters are available for most elements.

However, when I tried to run calculation on MgO x2x2x2 supercell, I 
found that scc does not converge for GFN2. It does converge for GFN1. I 
tried to fiddle with various parameters but it didn't help.

As far as I can see, GFN2 uses d-orbitals for Mg while GFN1 doesn't. I 
don't know if there is any other difference that might prevent convergence.

I would prefer to use GFN2 if possible since published spin constants I 
need for calculations with 3d metals are, apparently, for GFN2.

So, is there any way to achieve convegence on GFN2 on MgO ? If not, how 
bad would the spin constants from [1] work with GFN1 ?


[1] https://doi.org/10.1002/jcc.27185
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20240305/8f4e666f/attachment.htm>
-------------- next part --------------

Geometry = GenFormat { 
# 8 atoms in fractional coordinates
    8 F
  Mg O
# index type x y z
  1 1 0.0  0.0  0.0
  2 2 0.5  0.0  0.0
  3 2 0.0  0.5  0.0
  4 1 0.5  0.5  0.0
  5 2 0.0  0.0  0.5
  6 1 0.5  0.0  0.5
  7 1 0.0  0.5  0.5
  8 2 0.45  0.45  0.45
# origin
  0.0 0.0 0.0
# lattice vectors
  4.212 0.0   0.0
  0.0   4.212 0.0
  0.0   0.0   4.212
}

Hamiltonian = xTB { 
  Method = "GFN2-xTB" 
  KPointsAndWeights = SupercellFolding {
    3   0   0
    0   3   0
    0   0   3
    0   0   0
  }
 SCCTolerance = 1e-10
}

Driver = GeometryOptimization {
  Optimizer = FIRE {}
  LatticeOpt = Yes
  MaxSteps = 300
  OutputPrefix = "MgO.pert.geom.opt.out"
  Convergence {GradElem = 1E-8}
  AppendGeometries = yes
}

Options { 
  WriteDetailedOut = Yes  
  WriteDetailedXML = Yes
  TimingVerbosity  = 1

}

Analysis {
  CalculateForces = Yes
  ProjectStates {
    Region {
      Atoms = Mg
      ShellResolved = Yes
      Label = "MgO.pert.geom.dos_Mg"
    }
    Region {
      Atoms = O
      ShellResolved = Yes
      Label = "MgO.pert.geom.dos_O"
     }
  }
}

Parallel {
 UseOmpThreads = Yes
 Groups = 4
}



More information about the DFTB-Plus-User mailing list