Microcanonical Thermostatistics, the basis for a New Thermodynamics, "heat can flow from cold to hot", and nuclear multifragmentation. The correct treatment of Phase Separation after 150 years of statistical mechanics

Abstract

Equilibrium statistics of finite Hamiltonian systems is fundamentally described by the microcanonical ensemble (ME). Canonical, or grand-canonical partition functions are deduced from this by Laplace transform. Only in the thermodynamic limit are they equivalent to ME for homogeneous systems. Therefore ME is the only ensemble for non-extensive/inhomogeneous systems like nuclei or stars where the $\lim_{N\to \infty,\rho=N/V=const}$ does not exist. Conventional canonical thermo-statistic is inapplicable for non-extensive systems. This has far reaching fundamental and quite counter-intuitive consequences for thermo-statistics in general: Phase transitions of first order are signaled by convexities of $S(E,N,Z,...)$ \cite{gross174}. Here the heat capacity is {\em negative}. In these cases heat can flow from cold to hot! The original task of thermodynamics, the description of boiling water in heat engines can now be treated. Consequences of this basic peculiarity for nuclear statistics as well for the fundamental understanding of Statistical Mechanics in general are discussed. Experiments on hot nuclei show all these novel phenomena in a rich variety. The close similarity to inhomogeneous astro physical systems will be pointed out. \keyword{Microcanonical statistics, first order transitions, phase separation, steam engines, nuclear multifragmentation, negative heat capacity}