The ABINIT Users Mailing List ( CLOSED )

[Josef Zwanziger <jzwanzig@gmail.com>]

A few comments: first, h-ZnO undergoes a phase transition at about 6
GPa to a cubic structure, that may be why you found two local minima
so easily. Second, optcell 1 is not appropriate for this problem
because it will make all three cell edges change equally, which isn't
physical for such a case (isotropic pressure does not lead to
isotropic strains in a tetragonal cell). The best procedure would be
to do an optimization run with optcell 0 and ionmov 2, to get the
internal degree of freedom close to ideal, and then a second run with
the new xred values, using optcell 2 and ionmov 2. optcell 2 will
maintain the cell symmetry, so that a=b will be inforced; so a=b and c
will be optimized independently (2 degrees of freedom). I have a paper
in press at the moment where I studied ZnO with abinit as a function
of pressure and this procedure works well.

good luck,
Joe

On Fri, Mar 20, 2009 at 11:14 PM, ren <opticalcase@gmail.com> wrote:
> Dear All:
>
> I am doing geometry optimization for ZnO which has a hexagonal
> geometry. The lattice a and b is equal. Therefore, a and b should
> change by the same amount. Since it has a space group of P63 mc,
> therefore, only the oxygen atoms is able to vary in its z (or u i
> suppose) direction , internal coordinate.
>
> I read up the optimization notes  by L. D. Marks, which is meant for
> Wien2k, but i guess it should be applicable universally. What is done
> is that, having 5 different volumes having the same a:b:c ratio, the
> internal coordinates are optimized. Thereafter, for 5 c/a ratio, each
> with 5 different volume, the internal coordinate are optimized.
>
> At the moment, I tried the first step. But I tried exploring a larger
> range of volumes (26 different volumes). By plotting the volumes vs
> total energy, it appears that there is 2 local minimal (See the
> picture attached. I supposed one would choose the minimal with the
> lowest energy ? The volume obtained for the lowest energy is
> 313.992080 bohr^3. The other local minimal with a higher energy has a
> volume of 294.765030 bohr^3
>
> The problem occurs when I tried to use optcell=1 to see one could get
> similar result. It appears that the result obtained has a optimized
> volume similar to 294.765030 bohr^3 which has an energy similar to the
> local minimal that has a higher energy as obtained from my first
> study. I have also attached the email with both input files together.
>
> For optcell=1, here is the options i used
>
> optcell 1 # with optimization of volume only, do not modify rprim and
> allow an homogeneous dilation of the 3 component of acell
> ionmov 2 # Broyden Steps
> ntime 40
> dilatmx 1.05
> ecutsm 0.5
> tolmxf 5.0d-5
> toldff 5.0d-6
>
> I am using Troullier-Martins psp.
>
> I tried to increase ntime, tolmxf to 5.0d-7 and toldff to 5.0d-7 but
> the same results are obtained.
>
> May I know if anyone can advised me what i have missed out that caused
> this discrepancy?
> Should using optcell=1 be more accurate than manually optimized the
> internal coordinate with different volumes with fixed a:b :c ratio?
>
> Lastly, i read from the mailing list and it is always recommended to
> use optcell=2? Because of this spacegroup, does anyone have experience
> that optcell=2 ensure that lattice a=b always expand and contract by
> the same amount and that only the z internal coordinate is optimized?
>
> Hope like to hear from your advices!!
>
> Sorry for the lengthy email.
>
> Cheers,
> Yee Yan
> School of Materials Science and Engineering
> Nanyang Technological University (Singapore)
>



-- 
Josef W. Zwanziger
Professor of Chemistry
Canada Research Chair in NMR Studies of Materials
Director, Atlantic Region Magnetic Resonance Centre
Dalhousie University
Halifax, NS B3H 4J3 Canada
tel: +1 (902) 494-1960
fax: +1 (902) 494-1867
web: http://jwz.chem.dal.ca
jzwanzig@gmail.com, jzwanzig@dal.ca