[CPMD-list] equilibrated system?

[Axel Kohlmeyer]

On Mon, 25 Jul 2005, Nico Zobel wrote:

dear nico,

[...]

NZ> Since I could not find PBE-USPPs for neither Mg nor Li in any of the  
NZ> common
NZ> pseudopotential libraries I generated them myself with Vanderbilt's  
NZ> code.

they are not there, because they are quite difficult to get right.
also, since the vanderbilt code has so many options, it is rather
easy to get things wrong. that the pseudopotential program does
not complain about anything, does not mean you have a pseudopotential
that works well. i've been trying for quite a while now to produce
a good (and CPMD compatible!) uspp for mg, but could not succeed.
i currently have something that semi-works (i.e. could verify,
that - unlike with a 'minimal' pseudopotential - you can do a stable
CP dynamic (EKINC ~0.002a.u.), but is not good enough for serious
calculations and learning the black art of generating good 
pseudopotentials for uncooperative elements takes quite some time...

NZ> Does "semi-core potentials" mean that the number of electrons that  
NZ> are to
NZ> be treated as valence electrons should be larger?

yes. for metallic systems you may get away with only the minimal
number of valence electrons, but even then you have to include 
non-linear core correction or your results are most likely wrong.
for the semi-core potentials, you have to especially watch out
for 'ghost'-states, i.e. unphysical energy eigenvalues.
you need to do a series of tests on new pseudopotentials.
the simplest are just comparing the eigenvalue spectra for
different occupation number for the valence electrons
with the corresponding all-electron calculations. and then
i'd first check the lattice constant and bulk modulus of the 
pure metals but also for the oxides. since CPMD does not
support k-points with USPPs, you'd have to do that with a
really large system, or use a different code.

Alternatively, you could use the matching set of pseudopotentials
from the Goedecker library. there is all you need (H,Li,O,Mg,C) 
for PBE, and you could use k-points.  the only drawback is that
you'd need a much higher plane wave cutoff. i also should have
a matching set of Troullier-Martins pseudopotentials, that are
not so well tested (i'd appreciate it, if you were willing to
test them, though), but may work with a little bit smaller plane
wave cutoff.

NZ> > one more stupid question: is it intentional, that your methane
NZ> > is closer to the bottom of your slab than your top?
NZ> >
NZ> 
NZ> I'm interested in the hydrogen abstraction from methane. The  
NZ> configuration
NZ> you have seen is the educt configuration. The product configuration
NZ> will be a methyl radical and a hydrogen atom adsorbed at the surface.  
NZ> In this
NZ> configuration the methyl radical is situated approximately halfway  
NZ> between the
NZ> adsorbed hydrogen atom and the bottom of the MgO-slab.

as stated before, i see this as a minimal test case. you may
gain some information from it, but since the layer below the lithium
defect is kept fixed and there is still significant interaction
of the methane with the periodic image of the bottom of the slab,
i would be very careful on interpreting the results.

best regards,
	axel.

NZ> 
NZ> Best regards,
NZ> 
NZ> Nico.
NZ> 
NZ> 

-- 

=======================================================================
Dr. Axel Kohlmeyer   e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.