# Can I identify phase (solid, liquid, gas) just from a snapshot of molecule positions?

Suppose I have a simulation of a bunch of nitrogen molecules (N_2). Just by looking at the positions of a bunch of the molecules in a volume of space, how can I tell whether the picture is showing a solid, a liquid, or a gas?

How far apart do they have to be before I conclude that it's a gas and not a liquid? How close together do they need to be before I conclude that it's a solid?

Motivation: I'd like to make simulations of various complexity to derive quantities like freezing point and boiling point at various pressures. For sub-regions e.g. voxels I want to be able to ask Is this solid, liquid, or gas?

Are these qualitative distinctions or quantitative?

• If you can see a bunch of molecules of Nitrogen (your example) then you should be able to tell whether they are in solid, liquid, or gas state. If solid, there should be regular structure to the orientation of the molecules since they are in a crystal lattice. If liquid, such regular structure does not exist. If gas, same as liquid but molecules are definitely further apart from each other. – K7PEH Oct 11 '15 at 18:44
• How to turn these qualitative statements into quantitative (algorithmic) ones? I can think of a few ways to detect lattices, so let's say solid vs. liquid is solved. But if the difference between liquid and gas is just spacing, then how do I get a threshold spacing to threshold my decision of liquid vs. gas? Do I need to look instead at mean free path between "collisions"? – Jabavu Adams Oct 11 '15 at 19:03
• When I gave my comment above, I particularly mentioned your example of Nitrogen because I could look up a few facts before answering. But, in general, for some unspecified substance, it would be complicated to come up with some algorithmic approach because different things are very different in some of these measures and a crystal structure does not always exist. – K7PEH Oct 11 '15 at 19:32

In a gas the distance between molecules is clearly higher. But between an amorpheous solid and an liquid, order and distances are equivalent (worse: remember that ice is less dense than liquid water) so you won't be able to tell. It will be more easy to detect the order for a cristalline solid, but polar liquids also have some short distance order.

Of course very close to transition (or worse, pass the triple point), you will not be able to tell the phase just from the positions. Note also that the phase is not limited to solid/liquid/phase: beside extreme states, there are also various reorganisation of cristals depending of temperature and pressure.

Identifying the solid phase is the easiest. you will see a crystal, an organized structure with six neighbors for each particle. (the Mermin–Wagner theorem does no apply here, because the interaction has a long enough range. see this article for simulations and snapshots of the solid phase). here is what the solid phase looks like:

there will be a hexatic phase (liquid crystal, I can't find snapshots).

The hard part is distinguishing the liquid and the gas. you should notice that for a high enough temperature there is no distinction between gas and liquid (above the critical point) so if you just look at the distance between particles, you can't say what's liquid and what is gas.

If you look at a temperature lower than the critical temperature, than depending on the pressure you apply, you can see gas, liquid, or drops of liquid within a gas. when I say drops of liquid within a gas, I mean that you will have clusters of particles and some free particles moving between them - see the picture in the liquid gas region.

if you are above the critical temperature there is no distinction between gas and liquid, we call this fluid (see the one-phase region in the picture).

If you want to distinguish between gas \ liquid \ fluid - there is a way:

what you can do is look at the density distribution. divide a snapshot of your simulation into boxes (you will need to play with the sizes, depending on the number of particles and density), like this picture:

ask what is the density in each box, and put the results in a histogram. do this for all the steps in your simulation, and average the results. see this article for more details.

if you have liquid mixed with gas, you will see two densities:

as you increase the temperature, the two densities will get closer, until they become one. this means you crossed the critical temperature, and you are now above the critical temperature.

one more note: finding the freezing point and the vaporization point can require different simulation techniques. as you get closer to the critical point, you will need much more steps to get the simulation equilibrated, so you will probably want to use the Gibbs ansamble to avoid this.