We present a theory which shows that, in a closed system of fixed volume capable of undergoing a phase transition, the transition state can be thermodynamically stable against the bulk phases if a certain material parameters criterion is fulfilled. In a small system below the critical size the transition state turns into a globally stable phase that can be observed experimentally. This effect is analogous to stabilization of icosahedral structures in clusters of certain sizes and energies. Stabilization of the transition state in small systems of limited resources allows us to conjecture that, in the case of a melting/freezing transition in pure substances, this state corresponds to an amorphous phase. Although unstable in open systems, this phase may be observed experimentally due to slow kinetics of its decomposition at low temperatures. The material-parameters criterion should help experimenters select the materials for the experimental verification of the phenomenon. In the present paper we consider thin films where the phase separation is permitted only parallel to the plane of the film. The calculations, however, hold true for 3D small systems: e.g., nanoparticles.