The ABINIT Users Mailing List ( CLOSED )

[Riad Shaltaf <shaltaf@pcpm.ucl.ac.be>]

Dear Hongyi Zhao,

In general surfaces are more demanding than bulk calculations. this
means that one shall must use the "lowest demanding parameters" that
would represent correctly the physics of the problem. 

Thus the idea was that to chose the smallest surface cell, this can 
be achieved simply by rotating our coordinates systems 45 degrees, thus
in this case the smallest surface unit cell will be one atom, this type
of unit cell is called (1x1) surface unit cell. Doing this can be
achieved by either taking coordinates of the atoms already projected on
the rotated coordinates system, in such case the unit cell will have
"rprim" of simple cubic, as I think done in VASP example.. 

or you keep the coordinates of the atoms as they are but you enforce the
rotation of coordinate system through rprim as done in the ABINIT
example.

now since the study of surface energy, will involve subtraction of
energy values, we must make sure that these quantities are calculated
with similar/close parameters ( kpoint grid, FFT grids, same cut off
energy) in order to eliminate/minimize the error that would come out if
we were using different parameters. This explains why we need to use 
unit cells in both surface and bulk calculations which are "compatible"


Best Regards

Riad
  




On Thu, 2007-04-19 at 12:18 +0800, Hongyi Zhao wrote:
> On Thu, Apr 19, 2007 at 01:06:57AM +0800, LEE Chin Chai wrote:
> > Hi HongYi,
> > 
> > To get the rpim vector defined by abinit, you can do a 45 degrees of 
> > clockwise rotation(you can use a 2D rotation matrix)of the rpim defined 
> > in the vasp course.  Hope it helps. It is really arbitrary of how you 
> > define your rpim parameter as long as it reproduces your slab correctly.
> 
> Thouth I can get the *rprim* vector defined by abinit through the way 
> you mentioned above, but I still have the following questions:
> 
> 1. If I use the *rprim* in the form defined by abinit, I can represent 
> all the atoms in the (1 0 0) plane with integral linear combinations of 
> the *rprim*;  but in the case of the vasp course, I think in order to 
> represent all the atoms in the (1 0 0) plane, we must use real linear 
> combinations of the corresponding *rprim*.  Is this true?
> 
> 2. I still cann't understand the following line's meaning I excerpt from 
> the Abinit's tutorial:
> 
> ---------
> In order to study the Aluminum (100) surface, we will have to set up a 
> supercell representing a slab. This supercell should be chosen as to be 
> compatible with the primitive surface unit cell.
> 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. We will first 
> compute the total energy associated with this doubled cell. This is not 
> strictly needed, but it is a valuable intermediate step towards the 
> study of the surface.
> ---------
> 
> Let me tell you the problems puzzlling my brains in the above paragraph:
> 
> a). It says that *The corresponding directions are [-1 1 0] and [1 1 
> 0]*, what dose this mean?  I mean, these two directions don't belong to 
> (100) surface at all, in fact, they belong to (001).
> 
> b). What's the mean by saying that *This supercell should be chosen as 
> to be·compatible with the primitive surface unit cell.*, would you give 
> me more detailed descriptions about this thing?
> 
> Thanks in advance.
> ---
> Hongyi Zhao
> GnuPG DSA: 0xD108493B
> 
> 
-- 
Dr. Riad Shaltaf                   UCL/SE/FSA/MAPR/PCPM
Tel: +32 (0)10 47 28 50            Bâtiment Boltzmann, a+1
Fax: +32 (0)10 47 34 52            1 place Croix du Sud
Mel: shaltaf@pcpm.ucl.ac.be        1348 Louvain-la-Neuve (Belgique)