Chemical potential is a crucial quantity in systems such as liquid solutions. In few words it indicates the free-energy needed to introduce new molecules in the system.

Being a free-energy difference, it is a non-trivial quantity to compute. One basically needs to know the cost of adding a new molecule in each and every point of the system, considering all possible equilibrium states of the system.

computing mu
Improved sampling of insertion energy (with Metadynamics, in red) results in a more accurate estimate of the chemical potential.

A widely used technique to achieve this is the Widom insertion method. It turns out that Widom method (and other widespread techniques), work pretty well in low-density fluids, but they have troubles in dense systems. That’s because one needs the simulation to produce very unlikely states, that exist as possible equilibrium states, but happen very rarely in ordinary molecular simulations.

To solve this problem, I have developed a technique that employs enhanced sampling (e.g. Metadynamics), to force the production of such unlikely states, boosting the calculation of chemical potential. This technique has shown promising results with simple model systems, but we now need to apply it to more challenging liquids…


The project is in collaboration with Federico Giberti, Michele Parrinello, Giovanni Piccini and Omar Valsson. Details can be found in these two articles: