A detailed magnetic and mineralogical study of self-reversed dacitic pumices from the 1991 Pinatubo eruption (Philippines)

39 dacitic pumice and lithic samples from the 1991 eruption of Mount Pinatu bo were investigated through both magnetic and mineralogical means. As in a previous study, natural remanent magnetization (NRM) is found to be revers ed for most of the samples, with respect to the direction of the actual geo magnetic field direction. A few samples, amongst them ancient lithics trans ported by pyroclastic flows, show scattered NRM directions. From thermal de magnetization of these particular samples it is concluded that their orient ation changed after emplacement, The emplacement temperature is estimated t o be more than 460 degrees C from thermal demagnetization of lithic samples . Two magnetic minerals with large grain sizes are observed under the optical microscope: titanomagnetite (TM) and haemo-ilmenite (hem-ilm). Microprobe analyses yield x approximate to 0.10 for TM and y approximate to 0.52 and y approximate to 0.54 for two hem-ilm phases, in agreement with the observed Curie temperatures (similar to 480 degrees C for TM and similar to 250 deg rees C for hem-ilm), The hem-ilm particles display chemical zonation, which seems to be correlated with a change of the domain structure: typically, a ferrimagnetic (FM) phase with slightly higher titanium content is observed in the central part whilst the crystal margin, which is weakly ferromagnet ic (WF, due to spin-canted antiferromagnetism) is slightly poorer in titani um. Two different mechanisms for the origin and formation of the two observ ed phases are discussed: (1) chemical zonation of hem-ilm crystals due to a change in conditions in the magma chamber shortly before eruption; (2) sim ilar to the microstructures observed from synthetic samples, this zonation in the large natural hem-ilm could be the result of migration of the WF pha se towards the grain boundary during residence below the order-disorder tra nsition temperature in the magmatic chamber, The room temperature hysteresi s loop, which seems to be dominated by TM, provides multidomain (MD)-like p arameters: J(rs)/J(s) = 0.01 and H-cr/H-c = 20. The large coercivity of rem anence (15-40 mT), which is attributed to hem-ilm, may be intrinsic or due to interactions between WF and FM phases. The field dependence of the magni tude of the thermoremanent magnetization (TRM) is not linear: it increases first, reaches a maximum (negative) value for an applied field H close to 0 .5 mT, then decreases steadily. By extrapolation, it is estimated that the TRM should be zero for a field of about 12 mT and become positive beyond. T his total TRM is in fact the sum of several components, AF demagnetization of TRMs acquired in different fields shows the presence of at least three c omponents: a self-reversed (SR) component that contains both hard and moder ately hard components and a soft normal component. Independently of the val ue of H, the median destructive field of the SR component is of the order o f 70 mT, which suggests that the FM phase is not entirely locked by exchang e interaction with the WF phase and can experience noticeable wall movement s. However, some 20 per cent of the SR TRM component remains after AF treat ment of 220 mT. Although we believe that exchange interaction plays a key r ole in the process of SR TRM acquisition, torque curves in the high field o f samples both with and without SR TRM do not show clear evidence for unidi rectional anisotropy. Similarly, we do not observe any significant shift of major hysteresis loops or non-vanishing hysteresis losses, two characteris tics generally considered as indicative of exchange anisotropy and usually observed for synthetic single-domain;pseudo-single-domain (SD/PSD) hem-ilm. indeed, in addition to the exchange interaction the magnetostatic coupling must play some role here because of the large size of the hem-ilm grains.