Random-field effects in the diluted two-dimensional Ising antiferromagnetRb2Co0.7

Abstract

We report a comprehensive neutron scattering study of the spin correlations in the diluted twodimensional (2D) Ising antiferromagnet Rb2Co07Mgp 3F4 in an applied magnetic field. As predicted by Fishman and Aharony, an applied field in this system produces a random staggered magnetic field. Random fields are expected to have drastic effects on the cooperative behavior of magnets although the detailed behavior remains controversial. It is found that the applied magnetic field, and by inference the concomitant random staggered fields, destroy the 2D long-range order for all temperatures and fields. The structure factor W (Q)= g-, e' o'( -' SS'-)r is well described as the sum of a Lorentzian plus a Lorentzian squared with the Lorentzian-squared term dominating at low temperatures.The integrated intensity of the Lorentzian-squared term exhibits the same temperature dependence as the Bragg intensity at zero field. The correlation length K and structure-factor peak intensity 5 (Qv) exhibit power-law dependences on the applied field H, tt-H', and P' (Qv)-H ' at low temperatures; the measured exponents agree reasonably with the values 2 and -4, respectively, deduced from theories with the lower marginal dimensionality dI---3. The effective exponents initially increase slightly with increasing temperature and then decrease dramatically as T~T~, taking on values of about 0.7 and -1.2, respectively, at T~--42. 5 K. For small but nonzero applied fields the inverse correlation length may be factorized into a random-field part and -3.4J /k~T a thermal part, K=KRF+Ky, with K&-e where J is the Ising exchange constant. It is also found that for T &20 K reproducible results are only obtained when the sample is cooled in the presence of the applied field; this history-dependent behavior is analogous to that found in spinglasses.

Structure factor of a relaxor ferroelectric

“…(19) are, however, in sharp contrast with those of short-range interactions with quenched random fields in three dimensions where the correlation functions decay exponentially. 40 We now discuss the near-neighbor correlations with compositional disorder. For temperatures slightly above T c , the near-neighbor longitudinal correlations with and without disorder show similar decay with an overall strength that decreases with increasing compositional disorder ( Fig.…”

Understanding complex magnetic order in disordered cobalt hydroxides through analysis of the local structure

“…3,4,19 In those random dilute systems, the formation of magnetic microdomains and the disruption of magnetic long-range order is driven by local random-field energy. 19 While a different experiment here, the magnetic reflections from neutron diffraction have significantly increased Lorenztian widths and low diffracted intensities, which also indicate disrupted long-range order of the magnetic structure. Here, we hypothesize that uncompensated moments from the local clustering of polyhedra disrupts the long-range magnetic order into microdomains, and thus decreases the diffracted intensity and broadens the reflections.…”

Correlation functions, free energies, and magnetizations in the two-dimensional random-field Ising model

“…(10)), and also by the Migdal-Kadanoff scheme in d = 2, then µ = 1/2 and Eq. (18) predicts the lineshape to be Lorentzian plus Lorentzian-squared, the mean-field form found when the disconnected contribution is taken into account [4,7,[12][13][14]. On the other hand, if ones goes by the saturation behaviour predicted in Ref.…”

“…The later realization that it corresponds, give or take a few (hopefully irrelevant) details, to the experimentally realizable dilute Ising antiferromagnet in a uniform applied field [2] brought new insights, and new questions as well; among the latter, is the interpretation of experimental data in a suitable theoretical framework. This has proved to be rather an intricate subject, even down to basic aspects such as whether the lower critical dimensionality for the problem was d = 2 or 3 [3][4][5]. Though by now this particular issue has been settled in favour of d = 2 [6], several important aspects (such as the scaling behaviour near the destroyed phase transition in d = 2, which will be of interest here) still require further elucidation [7].…”