Negative specific heat in a thermodynamic model of multifragmentation

We consider a soluble model of multifragmentation which is similar in spirit to many models which have been used to fit intermediate energy heavy ion collision data. In this model $c_v$ is always positive but for finite nuclei $c_p$ can be negative for some temperatures and pressures. Furthermore, negative values of $c_p$ can be obtained in canonical treatment. One does not need to use the microcanonical ensemble. Negative values for $c_p$ can persist for systems as large as 200 paticles but this depends upon parameters used in the model calculation. As expected, negative specific heats are absent in the thermodynamic limit.