[CPMD-list] geometry optimization of a carbon nanotube
Axel Kohlmeyer
akohlmey at vitae.cmm.upenn.edu
Sun Dec 4 03:45:34 CET 2005
On Sun, 4 Dec 2005, Hiroshi Amawashi wrote:
HA> Dear CPMD community,
dear hiroshi,
HA> I'm a beginner of CPMD and trying to do a geometry optimization of
HA> an armchair-type carbon nanotube with chiral vector of (5,5).
may i suggest that you should first gain a little practice in using
CPMD with a few smaller and less demanding problems.
HA> But the results were bizarre: 20-atom armchair ring in the unit cell
HA> had broken into two parts. When I set TESR = 8, the bond angles of
HA> armchir were `optimized' to around 100 degrees rather than the normal angle
HA> of 120 degrees. Obviously, these results are all wrong.
HA> Any useful suggestions would be greatly appreciated. I've included the
HA> input file that I'm using below, in case I'm making a stupid mistake......
thank you for the detailed description of your attempts.
i have a few remarks about your input. i won't say that you
are making a _stupid_ mistake, but it is more that you are not
considering two factors that are important for a system like yours,
and those are actually quite common mistakes for a beginner, hence
my suggestion above.
1)
your box in x-, and y- direction is too small. you have to keep in
mind, that your system cell must be large enough to hold all atoms
_and_ their electron density and that there should be enough distance
from the atom centers to the wall of the unit cell, that the periodic
images (of the density) does not touch or influence each other. so
changing the cell keyword to:
CELL ABSOLUTE
15.0 15.0 2.494153 0.0 0.0 0.0
should take care of that.
2)
the second problem is more grave. you are doing a gamma-point
only calculation with a very small cell dimension in z-direction
and no k-points. this is all the worse, since your nanotube should
be metallic and thus you would need a lot of them and/or do a
free energy functional type calculation (which requires quite a
bit of experience, since there are many parameters to fiddle
with in order to get a good convergence). to some degree, you
may be able to avoid the k-point issue, by replicating your
your box in z-direction a few times. but you should carefully
compare your results to what is published in the literature.
there should be a ton of those. you can find some examples of
using k-points (without the free energy functional) at:
http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/cpmd-tutor/part7.html#chap8
where you have different types and sizes of unit cells. unfortunately
i did not have the time to add an explanation of the results.
they should basically all give the same total energy per atom,
but you need either a many k-points or a lot of atoms to get
that, and doubling the unitcell (in all directions) should be
equivalent to doubling the number of k-points (in all directions).
the only good news i have is that you can probably 'get away'
with a smaller plane wave cutoff (e.g. 60ry), but that won't
by far compensate the additional effort from the other
required changes. also in that case TESR would not matter
and can be kept at the default (0).
BTW: a fast and convenient way of testing whether your
box is large enough, is doing a wavefunction optimization
with RHOOUT using a plane wave cutoff of 10ry or even 5ry.
you can then convert the resulting DENSITY file into cube
format with: cpmd2cube -center -dens DENSITY
and visualize the resulting density as an isosurface with
a very low isovalue (0.0001 or lower). the isosurface should
not touch the walls, or better there should be a small
safety margin. it took me longer to write this mail than
checking your input this way... ;-)
best wishes from cold philadelphia,
axel.
HA> Best wishes,
HA>
HA> Hiroshi Amawashi
HA>
HA> ------------------INPUT FILE
HA>
HA>
HA> &CPMD
HA> OPTIMIZE GEOMETRY XYZ
HA> ODIIS
HA> 4
HA> TSDC
HA> SPLINE RANGE
HA> 5.0
HA> BFGS
HA> STRUCTURE BONDS ANGLES
HA> MIRROR
HA> HESSIAN UNIT
HA> &END
HA>
HA> &SYSTEM
HA> SYMMETRY
HA> 6
HA> ANGSTROM
HA> CELL
HA> 10.0 1.0 0.2494153 0.0 0.0 0.0
HA> CUTOFF
HA> 80.
HA> TESR
HA> 8
HA> POINT GROUP
HA> AUTO
HA> &END
HA>
HA> &ATOMS
HA> Carbon
HA> C_MT_GIA_BLYP KLEIMANN-BYLANDER
HA> LMAX=P LOC=P
HA> 20
HA> 3.4377467708 0.0000000000 0.0000000000
HA> 3.1405379470 1.3982575813 0.0000000000
HA> 2.7811955599 2.0206568530 1.2470765814
HA> 1.7188733854 2.9771760353 1.2470765814
HA> 1.0623221745 3.2694914677 -0.0000000000
HA> -0.3593423871 3.4189144343 -0.0000000000
HA> -1.0623221745 3.2694914677 1.2470765814
HA> -2.3003015812 2.5547437240 1.2470765814
HA> -2.7811955599 2.0206568530 -0.0000000000
HA> -3.3626237558 0.7147477437 -0.0000000000
HA> -3.4377467708 0.0000000000 1.2470765814
HA> -3.1405379470 -1.3982575813 1.2470765814
HA> -2.7811955599 -2.0206568530 -0.0000000000
HA> -1.7188733854 -2.9771760353 -0.0000000000
HA> -1.0623221745 -3.2694914677 1.2470765814
HA> 0.3593423871 -3.4189144343 1.2470765814
HA> 1.0623221745 -3.2694914677 -0.0000000000
HA> 2.3003015812 -2.5547437240 -0.0000000000
HA> 2.7811955599 -2.0206568530 1.2470765814
HA> 3.3626237558 -0.7147477437 1.2470765814
HA> &END
HA>
HA> &DFT
HA> FUNCTIONAL BLYP
HA> &END
HA> ---------------------End of INPUT FILE
HA>
HA>
HA> ----------------------------------------------------------------
HA> Hiroshi Amawashi
HA> University of Gifu
HA> 1-1 , yanagido , Gifu City, Gifu, Japan
HA> E-mail: k3135005 at guedu.cc.gifu-u.ac.jp
HA> ----------------------------------------------------------------
HA>
HA> ----
HA> Hiroshi Amawashi k3135005 at guedu.cc.gifu-u.ac.jp
HA> _______________________________________________
HA> CPMD-list mailing list
HA> CPMD-list at cpmd.org
HA> http://cpmd.org/mailman/listinfo/cpmd-list
HA>
--
=======================================================================
Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu http://www.cmm.upenn.edu
Center for Molecular Modeling -- University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.
More information about the CPMD-list
mailing list