The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

First sorry for my previous answer. P. Boulanger told me that it was
pretty much cryptic, and in deeds it appeaars that I have forgotten 20
% of the words making understanding a challenge.

For your current questions,  the problem is all again about periodicity :
1)a) first suppose you got a plain bulk crystal of Al. Now instead of
doing a simulation from a single unit cell, you reproduce it along its
basis vector. So you get something of a supercell. And periodicity
makes it equivalent (not in term of computation time obviously) to the
unit cell. So basically, nothing as change and cell parameters, atomic
position etc.. are exactly the SAME as in your single atom computation
(unless you want to make some change) but with some enlargement.
Suppose you have taken a supercell of 2 times, in each directions,
your unit cell : you have now 8 atoms, with xred = (a,b,c) where a, b
and c will be all possible combinations (except (1,1,1) of 1 and 0.
And you have the choice: either you choose acell = 2*unit_cell_acell
or rprim=2*unit_cell_rprim.
b) Know you want a surface. This is simply impossible ! And that's may
be your understanding problem. Because of periodicity we won't be able
to get a real single surface. Instead we can have two. That is a SLAB.
b1) how to define it :
We will elongate our cell (whatever cristalline cell you want) in 1
direction following the ideas previously exposed. Then we just omit to
put into this supercell a few atoms. And suddenly a VOID is appearing.
It breaks our bulk crystal into interacting slabs.
b2) how to choose the parameters :
You told it you want a surface. So there is TWO things that matters :
- the interacting slabs should not interact. They must be appart one
from the other as much as needed to get no interactions across the
vacuum region.
- each surface of the slab should behave as if it was the surface of a
bulk crystal. That is again no interaction between the surface but
this time across the bulk. So you must have enough bulk to avoid
interaction.
Since you (and most of us) don't know exactly how much vacuum and how
much bulk is important you will have to find those parameters
yourself. Just choose bigger and bigger metalic slabs, bigger and
bigger vacuum ; converge your calculations and observe when increasing
those parameters don't change the results any more.

2) I think you got the answer from 1b2 right ?

3) If you start from the single atom unit cell of cfc you won't have a
100 surface. So what you are going to do is just find a unit cell with
bXc // 100.  Any of those cells wil fit. And you can repilcate it
along its "a"  axis as much as needed to achieve step 1b2. Yet one of
the possible cell is much better than the others. This is because it
is much smaller than the others along the two axis (b and c)  of no
interest for this study. If you draw all possible fcc cell with bXc //
100,  you'll soon discover that this "smallest" cell is the one you
mention. This is the reason why it is selected in the tutorial.
The fact that this cell is not primitive do not bother us. We are
going to work with a spercell anyway. Right ?

PMA


On 4/17/07, Hongyi Zhao <hongyi.zhao@gmail.com> wrote:
> Hi all,
>
> Now, I read the abinit's tutorial 4.4, that is *Determination of the
> surface energy of aluminum (100): changing the orientation of the unit
> cell.*, the corresponding url is:
>
> http://www.abinit.org/Infos/Tutorial/lesson_4.html
>
> I am wondering about the follwing things:
>
> 1.  For the Aluminum (100) surface which is representing by a
> slab within a supercell, how can I know the correspongding acell, rprim
> and xred?  For this case, what's the corresponding part should looks
> like in the input file?  In addition, I feel it is so difficult for me to
> visualization it in my mind, would anyone give me some tricks on this
> thing?
>
> 2.  How many atoms should I use for that supercell?
>
> 3.  The tutorial said: The corresponding directions are [-1 1 0] and [1
> 1 0]. The direction perpendicular to the surface is [0 0 1]. There is no
> primitive cell of bulk aluminum based on these vectors, but a doubled
> cell.
>
> But I just cann't understand this thing, would anyone give me some
> hints in detail?
>
> Thanks in advance
>
> ---
> Hongyi Zhao
> GnuPG DSA: 0xD108493B


--
Pierre-Matthieu Anglade