Jul
20
2012

Freezing Computational Water

One of the harder things to do in computational chemistry is to model phase changes. Theoretically it is almost impossible as one needs very large scale (infinite) calculations to account for putative large-scale fluctuations. None the less it is a fun problem.

I took 1000 waters, using molecular dynamics constant volume and temperature in the AMMP program, and the tuna potential set (”tuna” refers to the name of the program used to adjust parameters to fit independent data, the potential is similar to the SPC water potential and reproduces radial distributions etc quite well). The goal is to look at pair correlation times as a function of temperature. The pair correlation time (how long two different molecules are close together) should be very long for ice and rather shorter for water.

As a start, though, I looked at the energy of the system. There were no surprises with the total energy which increased more or less linearly with temperature. However, the standard deviations were different, Since Cv, the constant volume heat capacity, is related to the standard deviation of the energy (Cv = c (sigma/T)^2 ) I rescaled the data by the absolute temperature squared and plotted that.

There is a dip around 250-275K! which is just the right place.
Estimated Cv

More detailed, with more time and more temperature values, simulations around the putative phase change are shown below.
Cv calcualted on a longer run and with more temperatures

Cv calcualted on a longer run and with more temperatures

So it looks like it might work.

Written by Rob in: Uncategorized |

No Comments »

RSS feed for comments on this post. TrackBack URL

Leave a comment

Powered by WordPress | Aeros Theme | TheBuckmaker.com WordPress Themes