structure relaxation problem

[Jamil Missaoui]

Dear all,

I tried to relax my system consisting of a layered structure with the presence of a molecule between the layers, the problem stills always the same with different positions.
In the first iteration there is a warning message "nstep= 800 was not enough SCF cycles to converge; although it stopped only in the 55th iteration, and in the second ionic optimization the deltaE(h) fluctuates without converging.
What I can understand from my output files is that the z-axis is pushed a lot from its initial position. I tried some solutions ( relaxation only along the z-axis and subdivide the interval of the difference between the two positions (before and after the first ionic optimisation) in order to find the minimum energy and from this value we do the relaxation again, but I find that the initial position is the ground state), but it still always the same problem.
My question is if there is someone how was confronted by the same problem or if there is some idea to beat this problem.

I attached the input of the
prtden3 1
kptopt3 1
ionmov3 2
optcell3 2
# strfact3 1
ntime3 300
tolmxf3 5.0d-7
# toldff3 5.0d-6
toldfe3 1.0d-10 # 1.0e-8 eV
#restartxf3 -1

PS: I cannot increase more the supercell.
Thanks in advance
Jamil

[ebousquet]

Dear Jamil,
I do not see your input file, right?
Do you relay the cell parameter with atomic positions? If so I would advise you to first relax the internal atomic coordinate at fixed cell parameter and then restart a cell relaxation with the new atomic positions.
Best wishes,
Eric