Dear Gao:

In my opinion, you should build a thick slab to advoid the surface-surface 
interaction, and include a dipole correction perpendicular to the slab if your 
model of adsorption is not symmetric.
   
Best,
taoohee 




From: Yi Gao 
To: siesta-l 
Topics:[SIESTA-L] Surface relaxation 
 
Dear siesta users,


Recently I have been working on the molecule adsorbed on anatase surface. I got 
stuck on the relaxation of the atomic structure.


The system is like this: the organic molecule adsorbs onto the anatase surface 
and hydrogen atoms dissociate from the hydroxyl groups of molecule to bond with 
2-fold oxygen atoms of anatase, while the remaining oxygen atoms of the 
molecule bond with 5-fold Ti of anatase.


I optimized crystal structure from bulk anatase, and then relax the supercell 
of clean anatase surface, say, (101)-1x4 supercell. Everything is fine and 
forces exerted on each atom is smaller than 0.02 eV/Ang. The vacuum is more 
than 15 Ang.


When I put dissociated molecule onto substrate, things go well except the 
forces on the dissociated hydrogen atoms. The Broyden method can never make the 
force on H atoms smaller than 0.08 eV/Ang, no matter how small the dynamics 
time length is.


It is strange because if the hydrogen atoms are being pushed, there should 
always be a point where it will not take large force, since the surface area is 
large enough. Also the problem is due to the substrate lattice parameter since 
all other atoms are taking forces smaller than 0.04 eV/Ang.


It seems a disaster to relax large surface system, especially with molecules 
adsorbed. Is there any more proper way to do the relaxation?


BTW, I used 500 Ry meshcutoff so it could not be the convergence problem of 
electronic part.

Responder a