IMF2017

In previous work, molecular dynamics simulations based on a first-principles-derived effective Hamiltonian for Pb_1-X(Sc_1/2Nb_1/2)O_3-X (PSN), with nearest-neighbor [Pb-O] divacancy pairs, was used to calculate X[Pb-O] vs. T, phase diagrams for PSN with: ideal rock-salt type chemical order; nanoscale chemical short-range order; and random chemical disorder. Here, we show that the phase diagrams should include additional regions in which a glassy relaxor-phase (or state) is predicted. Curves that locate the PNR-to-Relaxor transition are referred to as T*(X), where T* is a characteristic temperature at which a small but significant stiffining of polar nano-domains occurs; T* appears to be a weakly first-order transition, but these are numerical simulations, so order of transition, or crossover, can not be clearly established. With respect to phase diagram topology, these results strongly support the analogy between relaxors and magnetic spin-glass-systems. Similar calculations for PMN [Pb(Mg_1/3Nb_2/3)O₃] evaluated the effects of various cation configurations on T* follow the same trends as those for PSN: increased cation order implies reduced random fields, and therefore higher T*-values. As in PSN, T* most often appears to be a weakly first-order transition, with the same caveat that numerical simulations can't unambiguously predict the order of a transition.