Experimento em um sistema pequeno no ensemble microcanônico onde o calor específico se torna negativo: Em resumo “Upon melting, a large system converts added energy completely into potential energy, reducing continuously the fraction of its solid phase. The kinetic energy and thus the temperature remain constant. A small system, on the other hand, tries to avoid partly molten states and prefers to convert some of its kinetic energy into potential energy instead. Therefore the cluster can become colder, while its total energy increases. Negative heat capacities have now been found for melting clusters, fragmenting nuclei, and astronomical objects. What do these widely different systems have in common? The answer is that in these systems energy is not an extensive quantity; i.e., if such a system is divided into arbitrary subsystems the total energy is not simply the sum over the subsystems. The interaction between the subsystems has to be taken into account [2,14,15]. For example, in stars it is impossible to neglect the gravity between parts of the system [1,2]. Similarly in clusters and nuclei the interaction between subsystems is not negligible due to their small size.”
http://journals.aps.org/prl/abstract/10.1103/PhysRevLett.86.1191