Abstract
New magnetic and mineralogical findings on self-reversing hemoilmenite (Fe2-y Tiy O3) grains from Pinatubo lavas (1991 eruption) provide important clues regarding the acquisition process of reverse thermoremanent magnetization (rTRM) in this solid solution series. Magnetic Force Microscopy indicates the presence of multidomain magnetic structures in coexisting strongly and weakly magnetic crystallographic regions having compositions of y0.54 and y0.53, respectively. Yet, continuous thermal demagnetization of natural and laboratory TRM carried out on both whole rock samples and single hemoilmenite crystals shows that the magnitude of a normal TRM (nTRM) component, observed at temperatures above the Curie point of the strongly magnetic self-reversing regions, is much too large to be carried by a phase that is entirely cation-disordered. Consistent with this observation are findings using Transmission Electron Microscopy (TEM) which, in contrast to that what is commonly assumed, reveals the weakly magnetic regions to be magnetically heterogeneous. Specifically, these regions are found to contain tiny (20-40nm) domains that are cation-ordered and evidently ferrimagnetic dispersed within the cation-disordered, presumably spin-canted antiferromagnetic matrix. Given these findings, we argue that the so-called nTRM-carrying x-phase is itself partially cation-ordered, and thus ferrimagnetic, as postulated first by Ishikawa and Syono (1962). We propose a “nanophase” self-reversal model for the ilmenite-hematite solid solution series in which the rTRM and nTRM components are carried by the cores and margins, respectively, of the tiny, partially cation-ordered nano-sized domains observed by TEM. Due to the partial cation order, both the core and the margin of each domain are expected to behave in a ferrimagnetic fashion at temperatures below their respective Curie points. However, given the kinetics of the ordering process, their cation distributions need be antiphase, which causes their magnetic moments to be oppositely aligned. Since it is most reasonable to consider each margin to be slightly more Fe-rich than the inside core, upon cooling the margins acquire a magnetic remanence first (a nTRM). Then, upon further cooling, given that the intralayer and interlayer nearest-neighbor superexchange interactions are ferromagnetic and antiferromagnetic, respectively, the net magnetic moment of the core material need be oppositely aligned (producing a rTRM). The nano-sized regions would indeed behave in a superparamagnetic (SP) fashion if magnetically uncoupled to adjacent material; however, the spins in the margins (the x-phase) must be locked through superexchange to those of the surrounding disordered matrix, which we also claim to be locally enriched in iron. If so, then the magnetization of the x-phase can be both highly-coercive and thermally stable, as observed experimentally. Upon stepwise thermal demagnetization, the self-reversed remanence measured at room temperature is not destroyed until the unblocking temperature of the disordered Fe-enriched aureole (approximately 410°C) is reached. Mineralogical considerations and magnetic evidence from previous works suggest that this model is generally valid for self-reversed dacitic pumice, in particular the Mt. Haruna dacite and the 1985 Nevado del Ruiz dacitic andesite.