Continuous Elastic Phase Transitions in Pure and Disordered Crystals

TL;DR

This paper reviews the theory of second-order ferroelastic phase transitions in crystals, highlighting the effects of anisotropy, fluctuations, and disorder on critical behavior and phonon dynamics.

Contribution

It provides a comprehensive overview of elastic phase transitions, including the impact of disorder and defects on critical phenomena and local order parameter condensation.

Findings

Critical behavior is classical for one-dimensional soft sectors.

Sound velocity vanishes as |T - Tc|^{1/2} near transition.

Disorder can induce phase transitions and inhomogeneous scattering features.

Abstract

We review the theory of second--order (ferro--)elastic phase transitions, where the order parameter consists of a certain linear combination of strain tensor components, and the accompanying soft mode is an acoustic phonon. In three--dimensional crystals, the softening can occur in one-- or two--dimensional soft sectors. The ensuing anisotropy reduces the effect of fluctuations, rendering the critical behaviour of these systems classical for a one--dimensional soft sector, and classical with logarithmic corrections in case of a two--dimensional soft sector. The dynamical critical exponent is $z = 2$, and as a consequence the sound velocity vanishes as $c_s \propto | T - T_c |^{1/2}$, while the phonon damping coefficient is essentially temperature--independent. Disorder may lead to a variety of precursor effects and modified critical behaviour. Defects that locally soften the crystal may induce the phenomenon of local order parameter condensation. When the correlation length of the pure system exceeds the average defect separation $n_{\rm D}^{-1/3}$, a disorder--induced phase transition to a state with non--zero average order parameter can occur at a temperature $T_c(n_{\rm D})$ well above the transition temperature $T_c^0$ of the pure crystal. Near $T_c^0$, the order--parameter curve, susceptibility, and specific heat appear rounded. For $T < T_c(n_{\rm D})$ the spatial inhomogeneity induces a static central peak with finite $q$ width in the scattering cross section, accompanied by a dynamical component that is confined to the very vicinity of the disorder--induced phase transition.