[gmx-users] Liquid/Gas Systems

[Jussi Lehtola]

On Thu, 02 Jun 2011 17:20:35 -0400
"Justin A. Lemkul" <jalemkul at vt.edu> wrote:
> Fabian Casteblanco wrote:
> > Hi Justin,
> > 
> > You had helped me earlier on calculating the heat of vaporization of
> > methanol and it worked great.  I'm just trying hard to understand
> > conceptually what is the difference between simulating a liquid
> > phase and a gas phase in Gromacs.  I mean technically if we throw
> > in 1000 molecules of component A, would it only be a gas if we made
> > the box super huge?  If we run NPT on a system, is that technically
> > when we are compounding it together to find a liquid density?

This all depends on the temperature (and pressure). If the temperature
is low, the system might be in a liquid (or solid) state even if
the box is *huge*. The matter would then be in small droplets,
i.e. clusters.

Whether or not the clusters will merge at some point depends on the
system and the temperature. The interaction between clusters might be
so weak that compounding does not occur.

> > I'm just a bit confused on the difference.  I've been simulating
> > liquids up to this point so when I ran only NVT on a single gas
> > molecule, I was trying to understand why we only run NVT on it (0
> > pressure so no pressure coupling).

NPT for a single molecule would not make sense, since pressure is a
macroscopic quantity and so you need a lot more molecules than one to
define a pressure. This is also a numeric difficulty, since the
(instantaneous) pressure fluctuates a lot even in "reasonably sized"
systems.

> The basic premise is that, in the gas phase, the gaseous species
> (atoms or molecules) do not interact (assuming an ideal gas), so yes,
> you absolutely could build a huge box that has N molecules in it to
> match your condensed phase system.

But they *do* interact in the gaseous state. The point is that the
interaction is weak compared to the kinetic energy, i.e. <V> << <T>.
Only an ideal gas has no interactions between molecules.

> What tends to happen is that molecules will eventually find one
> another and form clusters, which really (I think) is just an artifact
> of using condensed-phase parameters to simulate a gas.

Clusters *are* physical entities and can be found in gases. The
question about the validity of extending condensed-phase parameters to
the gaseous state is, of course, very important, so at a given
temperature (and pressure) you might end up with a wrong amount of
clusters.

>  If the atoms/molecules find each other, they condense.  This type of
> simulation would have to be done under NVT conditions, since gases
> fill the volume of their container, right?  NPT would indeed just
> compress all of your particles together into a liquid.

As I stated above, I'd say you can still have the "flying ice cube"
effect in NVT.

NPT won't compress the system into a liquid if the temperature is high
enough (and the pressure low enough).
-- 
------------------------------------------------------
Mr. Jussi Lehtola, M. Sc., Doctoral Student
Department of Physics, University of Helsinki, Finland
jussi.lehtola at helsinki.fi
------------------------------------------------------
Jussi Lehtola, FM, Tohtorikoulutettava
Fysiikan laitos, Helsingin Yliopisto
jussi.lehtola at helsinki.fi, p. 191 50632
------------------------------------------------------