[Pw_forum] Functionals, pseudopotentials, etc. for H2O ice

[Axel Kohlmeyer]

On Sat, 9 Jul 2005, Steven Kirk wrote:

SK> Hello,

steve,

SK> Does anyone on this list have some recommendations for good functionals, 
SK> pseudopotentials and other simulation configuration data (e.g. cutoffs) 
SK> for PWSCF simulations of low-pressure tetrahedrally-coordinated H2O ice? 
SK> I already have some crystal structure data to work with, but I need 
SK> recommendations for the other simulation setup information.

i assume, that you are aware of the fact, that the investigation
of the structure ice and even more so of liquid water from various
models (classical or ab initio or anthing in between) is and has
been a controversially discussed topic for decades. so _the_ best
set of parameters does not exist. every choice so far had some
advantages and some shortcomings, so people tend to use what works
well enough for what they want to know.

SK> One important requirement is that the topology of the charge density be 
SK> accurate, so that charge density distribution on both covalent (sigma) 
SK> and 'H-bond' bonding interactions is as accurately represented as 
SK> possible. This suggests the use of ultrasoft pseudopotentials, but I'd 
SK> be grateful for some advice on this.

your main problem here should be the representation of the dispersion
interaction contribution at all, which is not very good in plain DFT.
there have been serveral suggestions on how to overcome this limitation: 
using an additional and empirically fitted attractive lennard-jones 
potential, modifying the nonlocal component of pseudopotentials using 
a pertubation theory based optimization scheme, and most rigorously 
using hybrid functionals, that include hartree-fock or exact-exchange.

to the best of my knowledge this is currently being implemented in 
pwscf and in at least one other similar package i know about, but 
not yet available in a released version.
unfortunately, using hybrid functionals, however successful and
efficient they are with local (gaussian) basis sets, pose a 
severe performance impact on calculations with pseudopotentials
and a plane wave basis set. 

in the end, you are probably best off starting with some
simple benchmark calculations to see how well the available
choices (pseudopotentials, functionals, parameters) fit your 
requirements and how much performance you are willing to sacrifice 
for how much additional accuracy.


with best regards,
	axel.

SK> 
SK> Many thanks in advance,
SK> Steve Kirk
SK> 

-- 

=======================================================================
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.