[SVC34-P02] Kinetic delay of crystal growth controls plagioclase-basaltic melt apparent disequilibrium
Keywords:plagioclase, crystallization, kinetics, basalt, geothermometry
We used the high-Al basalt lava from Waianae, Hawaii Oahu as the starting material for the cooling crystallization experiments. This experiment was conducted using the 1atm fO2-controlled furnace at University of Hawaii at Manoa. After 3 hours pre-heating at 1180℃, ~30K higher than the liquidus, the samples were cooled at cooling rates, Rc, of 0.1, 0.3, 1, 3, and 10K/min, and then quenched in water at four target temperatures, Tq, of 1150, 1120, 1090, and 1060℃. We used FE-EPMA (JEOL-JXA-8530FPlus) and EPMA (JEOL8800R) at Earthquake Research Institute, University of Tokyo, for textural and chemical analyses of the run samples.
As Rc and/or Tq increase, the abundance and size of plagioclase crystals decrease. As Rc increases, the shape of plagioclase changes from euhedral to dendritic. The Rc-dependent shape change is more obvious at lower Tq. Pyroxene crystallization was suppressed in the run samples coolied at higher Rc. In addition, melt boundary layers are observed around pyroxene in the samples cooled at 0.3-1K/min, quenched at 1090℃ and cooled at 0.3-3K/min, quenched at 1060℃, and around plagioclase in the samples cooled at >3K/min, quenched at <1120℃. Lower diffusivity of Al2O3 compared to those of FeO and MgO is responsible for lower threshold cooling rate of melt boundary layer formation around pyroxene.
The maximum An# [=100Ca/(Ca + Na)] of plagioclase was almost the same among all run samples. We applied both plagioclase liquidus and plagioclase-melt An-partitioning thermometers of Putirka (2008) to the pairs of plagioclase rim-boundary layer melt (BLM) and plagioclase rim-far field melt (FFM) to estimate temperatures recorded in their phase compositions. At Tq=1150℃, the estimated temperatures well represent Tq for both BLM and FFM. However, at lower Tq, both of the thermometers estimate temperatures higher than Tq; difference between estimated temperature and Tq, ΔT, increases as decreasing Tq. The two thermometers estimate similar temperatures for each sample, and the differences between the two estimated temperatures do not depend on Rc. These results suggest that the plagioclase-melt partition coefficient of An component does not depend on the Rc. Therefore, the increase of ΔT at lower Tq is not due to the Rc-dependence of the partition coefficient. It is attributed to the kinetic delay of the crystal growth at lower temperature. On the other hand, the coincidence between calculated temperatures and Tq at 1150℃ suggests that kinetic delay of plagioclase crystal growth is small at near liquidus temperature.
Present results show that plagioclase-melt thermometers can be applied regardless of the Rc at higher temperatures, while magmatic temperature tends to be overestimated due to kinetic delay of crystal growth at lower temperature.