A Experimental Investigation Into Random-Field Effects in the Mixed Jahn-Teller System Dysprosium Arsenic

An experimental investigation has been performed into the effects of microscopically random fields on phase transitions, in particular in the structural Ising compound rm DyAs_{x}V_{1 -x}O_4, in which a random strain field results from a mismatch of DyVO_4 and DyAsO_4 unit cell sizes. Experiments have been performed using several techniques, and results have been interpreted with particular reference to the Random-Field Ising Model (RFIM). Elastic constant measurements, made using a high -precision ultrasonic interferometer, enabled determination of the susceptibility critical exponent gamma. For pure DyVO_4 a value gamma = 1.15 +/- 0.05 was obtained, which is between the values gamma = 1 for a mean-field system and gamma = 1.25 for a three-dimensional Ising system with nearest-neighbour interactions. For random -field samples, a significantly higher value, gamma = 1.79 +/- 0.07 was obtained, with good agreement between samples, and in agreement with the two-dimensional nearest-neighbour Ising value gamma = 7/4 and with results of simulations by other authors, but in disagreement with certain scaling theory predictions. No other reliable experimental value for gamma exists for the RFIM in three dimensions. Results of optical birefringence measurements on the ultrasonics samples, along with numerical simulations, show that the effect of concentration inhomogeneities on the measurements of gamma was not significant. A new technique for measuring absolute birefringence in solids in also presented. Sensitive capacitance measurements of the dielectric susceptibility as a function of temperature and magnetic field history were used to investigate metastability in the distorted phase. The time-dependence and hysteresis observed are interpreted as being due to domain evolution. Domain behaviour has been investigated more directly in RF samples using high-resolution neutron scattering measurements of structural Bragg peaks in the basal plane. A splitting of the Bragg peaks into three or four peaks, characteristic of the "twinning" associated with {110} domain walls, was observed on cooling, revealing the relative populations of the different domain "orientations". Convincing evidence of finite-size broadening of the peaks was not obtained, as the observed broadening was almost entirely transverse, indicative not of finite domain size, but rather of mosaic spread, i.e. of a relative tilting of domains in the basal plane. This tilting has been interpreted as being due to a misalignment of neighbouring domain walls caused by random-field induced roughening of the walls. The transverse broadening is dependent on temperature and ordering-field history. The observed hysteresis is discussed in some detail.