[QE-users] Converging SCF for adsorbed molecules with EXX

[Guido Fratesi]

Dear all,

I'm running a calculation for phthalocyanines on Al surface. On top of 
the standard PBE calculation (that exposed no difficulty), I would like 
to run a SCF with PBE0.

For the molecule in the gas phase, the PBE0-SCF also went smoothly and I 
recover literature results for the energy levels.

However, for the molecule adsorbed on 5 layers of Al(001), the SCF loop 
(phi-psi) is not converging. At least, "dexx" fluctuates around 0.002Ry 
for more than 10 iterations. In another run with 3 layers, no progress 
is made after 30 iterations, and sometimes (say 1/10) also the phi-scf 
loop is not converging after several iterations.

I switched to NCPP for the EXX calculation and kept the "m-p" smearing 
as used for the surface, with nqx=1 for EXX. Part of the input is below, 
full I/O can be retrieved at: 
https://drive.google.com/drive/folders/1yTYWGrfWFUTcSkl_pXYzbOVYT9dsrXim?usp=sharing

To be noticed that the molecule (here, AlPh) may be spin-polarized (one 
electron in a pi orbital, in the gas phase) or not depending on the 
interaction with the substrate... Here I'm looking first to a 
spin-compensated case.

Thank you for your help,

Guido


&CONTROL
   calculation = 'scf' ...
/
&SYSTEM
   ibrav = 6
   a=14.2751509299088549285
   c=22.0
   nat = 182
   ntyp = 4
   ecutwfc = 70
   ecutrho = 280
   occupations = "smearing",
   smearing    = "m-p",
   degauss     = 0.02D0,
   vdw_corr='grimme-d3'
   input_dft='PBE0'
   nqx1=1,   nqx2=1,   nqx3=1
/
&ELECTRONS
   conv_thr    = 1.0d-7
     mixing_mode = "local-TF"
     mixing_beta = 0.7
/
ATOMIC_SPECIES
C 12.0 C.0.0.UPF
H  1.0 H.nc2.UPF
N 14.0 N.nc2.0.0.UPF
Al 26  Al.vbc.UPF
ATOMIC_POSITIONS {angstrom}


-- 
Guido Fratesi

Dipartimento di Fisica
Universita` degli Studi di Milano
Via Celoria 16, 20133 Milano, Italy