[CP2K-user] Keep obtain large negative pressure deviations when modelling slabs with fixed z-axes

Ivan Clayson i.g.... at gmail.com
Mon Feb 3 12:27:40 UTC 2020


Hello, I'm currently performing a CELL_OPT minimisation of a slab to model 
an external surface where there is approximately 14 Å of vacuum between 
each slab in the z-axis. As I want to maintain the slab model, I'm 
constraining the z vector whilst allowing the a and b lattice vectors for 
my slab to relax along with my nuclei. However, I keep obtaining negative 
pressure values above 100 Bar and my system is unable to converge as the 
cell keeps wanting to collapse.

I've tried,
- Changing the minimiser from BFGS to CG
- Constraining and fixing the cell angels
- Increasing the EPS_SCF to obtain more accurate forces
- Varying the multi-grid cutoff energies
But sadly I've had no success with obtain reasonable pressure deviations. 
None of the forces are particularly large on the atoms (the largest are on 
the order of E-5 a.u.) so I'm unsure why my pressure is so negative. 
Identical slabs can successfully converge with slightly smaller and larger 
vacuum sizes so I'm unsure why certain vacuum separations are unsuccessful. 
I'm running CP2K version 6.1 and I've attached my input and the final step.

Does anyone have any suggestions for me to try?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200203/c0644a0a/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hyd_Cu_slab_14_cg_res1.restart
Type: application/octet-stream
Size: 28580 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200203/c0644a0a/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hyd_Cu_slab_14_cg_res1.out
Type: application/octet-stream
Size: 158072 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200203/c0644a0a/attachment-0003.obj>


More information about the CP2K-user mailing list