Mean-field cluster model for the critical behaviour of ferromagnets

Two separate theories are often used to characterize the paramagnetic properties of ferromagnetic materials. At temperatures T well above the Curie temperature, T-C (where the transition from paramagnetic to ferromagnetic behaviour occurs), classical mean-field theory(1) yields the Curie-Weiss law for the magnetic susceptibility: chi (T) proportional to 1/(T - Theta), where Theta is the Weiss constant. Close to T-C, however, the standard mean-field approach breaks down so that better agreement with experimental data is provided by critical scaling theory(2,3): chi (T) proportional to 1/(T -T-C)(gamma) , where gamma is a scaling exponent. But there is no known model capable of predicting the measured values of g nor its variation among different substances(4). Here I use a mean-field cluster model(5) based on finite-size thermostatistics(6,7) to extend the range of mean-field theory, thereby eliminating the need for a separate scaling regime. The mean-field approximation is justified by using a kinetic-energy term to maintain the microcanonical ensemble(8). The model reproduces the Curie-Weiss law at high temperatures, but the classical Weiss transition at T-C = Theta is suppressed by finite-size effects. Instead, the fraction of clusters with a specific amount of order diverges at T-C, yielding a transition that is mathematically similar to Bose-Einstein condensation. At all temperatures above T-C, the model matches the measured magnetic susceptibilities of crystalline EuO, Gd, Co and Ni, thus providing a unified picture for both the critical-scaling and Curie-Weiss regimes.