Date: Fri, 27 Sep 2002 13:53:24 -0700 (PDT)
From: Thomas Cheatham
Subject: Re: Is H-bonding really there?


> any velocity of water molecules till I cross the
> temprature of 273K because they should be in a rigid

What is the diffusion constant of water in ice?
What is the rotational correlation time of water in ice?

> 3d cage structure (ice). why is there this apparent paradox?

At any temperature above 0 K there is some thermal energy present. This
means that the water will move and it will have a finite velocity.
Whether it will move within the "ice" cage (i.e. water will exchange
positions) depends on whether there is enough thermal energy (or time) to
observe this possibility.

Ice is not "locked" into a lattice structure except on average in a
macroscale sense; on the microscale, there are motions of the water
including reorientation, exchange, proton-transfer, etc. MD with a
classical empirical (molecular mechanical) potential cannot see
proton-transfer (except perhaps within an empirical valence bond
formalism).

There is a recent paper in JACS on simulation of atomic positional
fluctuations in ice (using the TIP4P) model; JACS 124, 8085-8089 (2002)
by Tanaka & Monhanty. This paper suggests that these lattice exchanges do
not occur until after the glasslike transition temperature (~200 K).
However, even below 200K the water is moving!

The observation that big movement (exchange of water) doesn't occur until
after 200K could suggest that equilibration shouldn't start at low
temperatures, but maybe 200K+? However, it should be remembered that the
point of equilibration is not only to relax the initial water density and
move the waters to more favorably solvate the solute, but to disperse
hot-spots in the micro-simulation cell due to potential bad overlaps, etc.
These local hot-spots can have a rather high temperature despite the
simulation being overall cold. In other words, if you started at
200-300K, you could have a hot spot that has a much higher temperature and
this could lead to SHAKE failure or the MD integrator failing...
Minimization can remove some of these hots spots and conservatively, low
temperature MD ramping up to higher temperature should smooth out the
rest. Finally, the initial water positions that are likely observed in
any water box you create-- unless it was created from a known ice
structure-- are likely not those of ice. To form the ice would require
properly annealing the water structure, i.e. it will take some time in MD
simulation to see this.