The ABINIT Users Mailing List ( CLOSED )

[shaltaf@pcpm.ucl.ac.be]

Hi

As paul has already said, you have to do it manually. It is not difficult, for
example, suppose we want to passivate one side of Si slab. Since each Si
atom will have two dangling bonds on the surface, this means that
we need to add two H atoms to saturate the dangling bonds of each
one, you may try to guess the initial positions of H atoms using
some knowledge of the Si-H angle and bonding length from SiH4
molecule. Then relax the H atoms, it recommended here that you
keep all the Si atoms in the slab at this stage fixed on their ideal
bulk positions, just allowing  H atoms to move.  
you may also passivate both sides, and try to
compare, the Si-H angles and bonding lengths on both sides, in
this case you can try to plot the energy band structure, to make
sure that no surface states do exist.

Once the H atoms are relaxed, you keep the last one or two Si MLs
that in H side frozen, and you may relax the the reset of slab.

Hint:

after the slab is relaxed, you may check, the difference of final
and initial positions for each atom in the slab, call it 'delta
r'. If your slab is large enough, you will find that 'delta r' has
its largest value for atoms on the surface, and decreases as we go
deep in the slab, until it is very small for atoms that exist in
deep MLs, especially those in contact with the frozen MLs.


Riad



Quoting Hongyi Zhao <hongyi.zhao@gmail.com>:

> Hi all,
> 
> I've find that castep has the capability to perform the surface 
> passivation by some kind of atom, say, Hydrogen, in the compution of 
> slab model.  Let me describe this thing as follows:
> 
> Because the research on the surface is more demanding than bulk 
> calculations and thus we will use a supercell for that thing.  So the 
> problem is obvious: we must have adequate layers in our model as well as
> 
> the thickness of the vacuum layer.  In order to minimize our computation
> 
> workload and optimize our computaional results, we can perform a 
> passivation on one surface by some kinds of atoms, say, Hydrogen, to 
> rivet the dangling bonds.  In this way, the vacuum's thickness on one 
> side of our model can be decreased to some extent with the same 
> computational precise.
> 
> But I'm wondering on how can I do this thing with abinit?  Would anyone
> 
> give me some hints?
> 
> Thanks in advance.
> ---
> Hongyi Zhao 
> GnuPG DSA: 0xD108493B
> 


-------------------------------------------------
This mail sent through IMP: atlas.pcpm.ucl.ac.be