Three magmatic components in the 1973 eruption of Eldfell volcano, Iceland: Evidence from plagioclase crystal size distribution (CSD) and geochemistry (original) (raw)
Related papers
Lithos, 2010
A systematic study of textural and compositional zoning (An% and FeO variation) in plagioclase phenocrysts of historic (pre-1971) and recent (post-1971) lavas at Mount Etna was made through back-scattered electron (BSE) images and electron microprobe analyses (EMP). The textures considered include oscillatory zoning and several types of dissolution, resorption and growth textures at the phenocryst cores and/or rims. Two patterns of oscillation were recognized from the combined An–FeO variation: 1) Low Amplitude–High Frequency (LAHF) and 2) High Amplitude–Low Frequency (HALF). The first pattern is interpreted here as due to kinetic effects at the plagioclase/melt interface which developed during crystallization in closed reservoirs. The second, which sometimes involves thin dissolution surfaces marked by irregular edges, angular unconformities and complex dissolution–regrowth patterns, might imply crystallization in a more dynamic regime, probably driven by chemical and physical gradients of the system (e.g., convection in a steadily degassing open-conduit). Dissolution and resorption textures at the core vary from patchy (exclusive to plagioclases within pre-1971 lavas) to strongly sieved, and can be related to increasing rates of decompression under H2O-undersaturated conditions. Thick sieve-textured envelopes at the phenocryst rims, generally coupled with marked An–FeO increase, result from mixing with more primitive and volatile-rich magmas. In the same crystals from recent activity, An and, to a lesser extent, FeO increase, consistent with the mixing of H2O-rich magmas similar in their mafic character to the resident magma (cryptic mixing). Two types of growth textures were also recognized at the crystal rims: 1) stripes of regularly-shaped melt inclusions and 2) swallow-tailed, skeletal crystals. In the first instance, the concordant An–FeO decrease suggests crystallization caused by fast ascent-related decompression accompanied by volatile loss. In the second, An decrease at effectively constant FeO contents may indicate crystallization at a high level of undercooling from already degassed magma, followed by rapid quenching; such a feature might be acquired during syn- or post-eruptive conditions.Although textures found in historic lavas are rather similar to those in the recent ones, some differences occur, such as lack of crystals with patchy cores in recent products and lower An contents in crystals of historic ones. The available data allowed us to obtain information on the dynamics of the feeding system, highlighting their possible modifications over time. In particular, historically erupted magmas, generally acknowledged to be volatile-poor, may have ascended through the deep portions of the plumbing system under H2O-undersaturated conditions at lower rates than the recent ones, recognized as more volatile-rich. Eruption triggering mechanisms from closed reservoirs in the shallow portions of the feeding system are similar for both historic and recent events, and may be generally favoured by a recharging phase of more primitive, undegassed magma or by a few episodes of important fracture opening (e.g., in response to an earthquake swarm).
Journal of Volcanology and Geothermal Research
Plagioclase phenocrysts in dacites from the high-K calc-alkaline Capraia Volcano were investigated for major, trace element and Sr isotope variations in order to gain better insight into the proposed open-system behaviour of the volcano. Repeated dissolution zones in plagioclases from the early-erupted Monte Castello (MC) and later Monte Rucitello (MR) dacitic units are associated with a sharp increase in An content and variation in isotopic compositions (87Sr/86Sr = 0.70872–0.71004), which argue for repeated magma mixing. In particular, petrographic and mineralogical evidence suggest that recharge with hot, basaltic magma occurred repeatedly during the lifetime of the Capraia volcanic system. We attempted to correlate shared crystallization and resorption histories using chemical (mol% An, Fe) and Sr isotopic zoning in plagioclase. In this correlation framework, the observed mismatch between chemical and isotopic data in some phenocrysts is interpreted to reflect either growth of phenocrysts in a separate magma chamber, or compositional zoning within a single magma chamber. A minor role of crustal contamination is inferred. Instead, the intra-crystal variations in isotopic composition in plagioclase reflect the existence of distinct components that may have thoroughly mixed at the end of the MC plagioclase growth history. The dominant radiogenic component is similar to lamproites occurring in Tuscany. Recharge with this component occurred throughout the genesis of the MR dacites, but was restricted to the early growth history of the MC dacites. The MR plagioclases are inferred to have crystallized from separate, more differentiated batches of magma that were likely to have evolved at the top of a zoned magma reservoir. Decrease in the An content after most dissolution zones in all phenocrysts, and low temperature estimates from zircon saturation thermometry (< 800 °C) also emphasize the importance of crystal fractionation during the evolution of the volcano, in agreement with the occurrence of a significant amount of rhyolitic material. Thus, it is inferred that a rather large magma chamber may have existed during the first cycle of eruptive activity.
The textural variation and compositional zoning of plagioclase in prehistoric and historic basaltic andesite lava flows from Ngauruhoe volcano reveals extensive crystal recycling from a multi-level magma system. Most phe-nocrysts have a calcic (~An 80–90) resorbed core with diffuse or no zonation that is depleted in Fe and Mg. Some cores display patchy zonation from replacement by high An crystallization prior to resorption. The cores are man-tled by oscillatory-zoned rims of lower An content (b An 60), and are enriched in Fe and Mg. Rim zones vary in relative thickness and textural complexity, and include sieve-textured bands, and/or cyclic calcic growth following dissolution events. A subordinate crystal population display similar features, but lack a resorbed core. These latter crystals display overall rimward enrichment in An, Fe and Mg. The resorbed cores crystallized from magmas more mafic than those erupted at Ngauruhoe, and slow cooling and prolonged storage resulted in loss of An zoning patterns and depletion of Fe and Mg by diffusion. These crystals are likely to have originated from deep cumulates or intrusions, and were subsequently entrained in ascending magmas. Patchy-textured cores were produced during decompression in a water under-saturated magma and staged ascent. The diversity in crystal cores reflect different conduits and ascent histories. The crystal rims grew in a more differentiated magma reservoir, and are in equilibrium with the erupted melt. Most of the zoning patterns in the rim zone require water pressure and/or temperature changes. These changes could have been caused by con-vective self-mixing in a closed system and/or the intrusion of hydrous melts of similar bulk composition. Other crystals display rimward elemental enrichments consistent with mafic recharge. Previously reported rimward enrichment in 87 Sr– 86 Sr compositions can be explained by the recycled origin of the crystal cores and progressive crustal assimilation at shallower depths in the magma system where subsequent crystal growth occurred. A steady-state balance of intrusion and crystallization in the deep crust, and remobilization and entrainment of co-magmatic crystals in newly formed and ascending magmas has produced relatively monotonous batches of basaltic andesite throughout the volcano's life. Trace element zonation (Fe and Mg) in the plagioclase mostly reflects variable diffusive loss through the crystal history, and the patterns in different crystals are highly variable. This, along with uncertainties in estimating trace element partitioning coefficients, highlights the difficulty of reconstructing melt compositions and subterranean residence time in such magmas.