Anisotropic constant pressure MD

Summary

This exercise studies a well-known phase transition in potassium chloride, see ref. [Parrinello1981], using constant pressure molecular dynamics. The objective is to develop the best practice in using such algorithms and to learn how phase transitions can be induced, detected and monitored in a simulation.

Background

Potassium chloride at ambient temperature and pressure adopts the cubic rocksalt structure, in which each ion is surrounded by six ions of opposite charge in an octahedral arrangement. Under high pressure this structure transforms to something more close packed - the so-called caesium chloride structure, where the nearest neighbour coordination rises to eight ions. (Using the model potential adopted here, this occurs at about 1.4 GPa.) In this exercise the student will have the opportunity to see this phase transition using the method of anisotropic constant pressure molecular dynamics. Commencing with the rocksalt crystal structure and applying a fixed external pressure it is possible to induce the phase transition in a simulation. Similarly it is possible to see the reverse transition back to rocksalt. However it is not necessarily trivial to make these transitions happen in any given simulation (though you may be lucky the first time!) Your task will be to find the conditions under which the phase transition occurs. This will not be entirely a matter of finding the right conditions of temperature and pressure, but will also involve setting up the control parameters for the simulation so as to encourage the phase transition to occur. (Even if the transformation is thermodynamically permitted, it does not follow that it will happen in the lifetime of a simulation.)

Task

First download the FIELD, CONTROL, CONFIG files. The last of these is a crystal of potassium chloride at ambient temperature and pressure (i.e. in the rocksalt structure). You should proceed as follows:

  1. Load the CONTROL file in your favourite text editor. Select the constant stress barostat – see ensemble, and set appropriate relaxation times for the thermostat and barostat. Choose an appropriate starting temperature and pressure and run a reference simulation of the system at ambient temperature and pressure (i.e. set DLPOLY.Z running - 2000 time steps is quite sufficient). Examine the resulting OUTPUT file and display the final REVCON file and simulation RDFs to see what structure you have.
  2. Repeat the simulation at a different state point, where you might expect a phase transition to occur. Examine the result graphically once again (using the REVCON file and a visualiser such as VMD) and try to deduce how the phase transition occurred. Look at the RDF plots (which can be generated from the RDFDAT output file) and try to determine what phase the structure is now in.
  3. If you do not see a phase transition, experiment with the control parameters (e.g. change the relaxation times, temperature or pressure, as you think fit) until you see one. Be as systematic as you can, using whatever insight you gain to rationalise what’s going on.
  4. If you believe that you have obtained the phase transition, a number of other options are open to you:
    • Look in the STATIS or OUTPUT files (details in the manual) and see if the variables catalogued there can provide an independent ‘signature’ of the phase transition.
    • Continue to experiment with the control parameters and see how the system responds, thereby strengthening your understanding about how to ‘drive’ the program. Try some of the other ensembles available.
    • Explore the phase diagram at little more (change T and P). Try and obtain the reverse transition.
[Parrinello1981]
  1. Parrinello and A. Rahman, Polymorphic transitions in alkali halides. A molecular dynamics study, Journal de Physique Colloques, 42 C6, p. C6, 1981, doi: 10.1051/jphyscol:19816149, URL https://hal.archives-ouvertes.fr/jpa-00221214.