Paleointensity on volcanic glass of varying hydration states

We have characterised the magnetic properties of variably hydrated volcanic glasses that were collected from rhyolitic deposits at Blلhnْkur, Torfajِkull, Iceland. The glasses span the range from fresh obsidians to highly fractured perlites that contain >2 wt% water. Lava hydration plays a key role in the formation of perlite and, hence, these rocks are ideal to study hydration effects on remanence carriers and reliability of the paleomagnetic record. The total volatile content of the different samples was determined as a proxy for the degree of perlitisation/hydration. It was found that coercivity of remanence, saturation magnetisation and saturation of remanence decrease with increasing hydration, i.e. that magnetic remanence carriers get lost and that magnetic stability is reduced. Additionally, thermal demagnetisation of a three component isothermal remanence revealed that mainly the high coercive material is destroyed within the more strongly hydrated samples while lower coercive material seems to be less affected. Grain sizes of all but one samples are in the pseudo-single domain range (the one exception shows multi-domain characteristics). It was impossible to unambiguously identify the remanence carriers, but titanomagnetites are most likely responsible for the lower coercivity component while hemoilmenites possibly represent the higher one. A modified Thellier method was used to determine paleointensity values. As most of the samples are hydrated it is not astonishing that the overall paleointensity data is not of very high quality. However, it is important to note that there are hydrated samples with well-defined Arai-diagrams. Although seemingly of high quality, these paleointensity values are incorrect as there is a trend towards lower paleointensity values observed with increasing perlitisation. We attempted to test for magnetic anisotropy and cooling rate dependency, but this was hampered by alteration during the experiments. However, we argue that both anisotropy and cooling rate dependency are unlikely to be responsible for the observed trend in paleointensity. Thus, even well-defined paleointensity values can be erroneous when obtained from hydrated glass. This emphasises the need for unaltered samples and additional attention during paleointensity determinations.