Domain pattern observations in rock magnetism: progress and problems

Abstract

Magnetic domain pattern observations that have been conducted in the field of rock magnetism are discussed in the light of several models for domain structure. Comparison of predictions with observations can reveal which of the several internal energies and physical mechanisms exert strongest control upon fine particle behavior. Owing to their great importance as paleomagnetic carriers in many rocks, members of the titanomagnetite series receive special attention here. Magnetite and titanomagnetites that are low in titanium display the expected patterns of 180°, 109°, and 71° walls typical for cubic materials in which K 1 is negative. By contrast, intermediate to high x-value titanomagnetites largely exhibit complex patterns with a distinctly non-cubic style, although simple patterns of broadly-spaced, roughly planar walls are sometimes observed. Undoubtedly, complex patterns in Ti-rich compositions result from stress. Many of these non-cubic patterns consist of wavy walls, often with reverse spikes. The wavy patterns are diagnostic of a uniaxial, vertical anisotropy and they are directly analogous to patterns found on the basal plane of many uniaxial industrial ferrites. Analyses of the wavy patterns with the model of Szymczak yield stresses on the order of several hundred bars. Thus the very different domain pattern styles observed throughout the titanomagnetite series indicate the increasing importance of stress with titanium content. Whether this stress is intrinsic or an artifact of the surface preparation methods still remain a topic of debate. Throughout the titanomagnetite series, particles that contain simple arrays of planar walls yield the familiar dependence of the number of domains ( n) upon L 1 2 ( L = grain thickness), in accordance with equilibrium theory. Plots of n vs. L derived from observations invariably have a ‘scattered’ appearance. It is shown here that much of this scatter is intrinsic and reflects physical processes that drive grains into accessible domain states. At any given size, however, titanomagnetite particles contain far fewer domains than are calculated with simple models in unstressed material. In magnetite and the low x-values, this discrepancy could be resolved by accounting for closure domains and LEM states. Only rarely are closure domains associated with simple patterns in the higher x-values, however. It is therefore proposed that simple patterns in the Ti-rich compositions reflect high levels of internal stress, rather than the magnetocrystalline anisotropy which is usually presumed. It so, then all patterns observed to date on Ti-rich titanomagnetites have a stress origin. Preliminary experiments to observe both ‘simple’ and wavy patterns on natural x = 0.6 titanomagnetite as a function of temperature are reported. Surprisingly, little change in either the domain widths or the degree of waviness is observed with heating. When these results are interpreted with models for the planar and wavy structures at elevated temperature, they lend further support to the stress origin of these two very different types of patterns. Furthermore, these results suggest that both the exchange constant and the underlying stress anisotropy vary approximately as M s 2( T). If so, the nucleation of domain walls in certain particles would be an energetically ‘difficult’ process at the elevated temperatures where TRM is acquired. This could explain why both saturation remanence and TRM in the PSD range of these compositions may be attributable to particles which fail to nucleate domain walls.