Phase transitions often are everyday processes whose conceptual basis is easily explained by simple examples. But how the phase transition proceeds at the molecular level is a far more delicate question. Indeed, predictions of theoretical models often fail miserably when compared with experimental results.

Our research is focused on the details of nucleation using molecular simulations and on trying to improve nucleation theory.

Nucleation has received much attention for more than a century but there still is no theoretical model that correctly describes and predicts nucleation rates–not even for the simple case of condensation. The dilemma lies in having either a model that is too crude to capture all aspects or a model that requires an intimate knowledge of the involved interaction potentials, which rarely are accurately known. For the latter reason, the simple classical nucleation theory (CNT) is still the predominant theory.1-4

But CNT predictions often fail to describe experimental results by many orders of magnitude, most notably more than 26 orders of magnitude in the case of argon condensation.5  It is quite hard to imagine another field in present day science that shows equally large deviations between experiment and theory. So why are scientist still sticking to a theoretical model that fails so badly?

Partly because one actually can predict something for a wide variety of situations and substances. Also, CNT does does not perform badly in all respects. Even though the prediction of nucleation rates typically is wrong, CNT still captures the qualitative essence of nucleation quite well. It also often yields surprisingly accurate predictions of the critical cluster size (the cluster size that marks the top of the nucleation barrier).

Still, it is obvious that we need a more powerful theory. Ideally, this theory should be as easy-of-use and as applicable as CNT. One successful step in this direction has been introduced with the “Extended Modified Liquid Drop Model – Dynamical Nucleation Theory”, or EMLD-DNT.6,7 We are continuing our efforts on further improving the predictive power of this model and on simplifying its application.

References

1 M. Volmer and A. Weber, Z. Phys. Chem. (Leipzig) 119, 227 (1926).
2 L. Farkas, Z. Phys. Chem. (Leipzig) 125, 236 (1927).
3 R. Becker and W. Döring, Ann. Phys. 24, 719 (1935).
4 For a review see e.g. P. G. Debenedetti, Metastable Liquids: Concepts and Principles. (Princeton University Press, Princeton, 1996).
5 K. Iland, J. Wölk, R. Strey, and D. Kashchiev, J. Chem. Phys. 127, 154506 (2007).
6 D. Reguera and H. Reiss, Phys. Rev. Lett. 93, 165701 (2004).
7 D. Reguera and H. Reiss, J. Phys. Chem. B 108, 19831 (2004).