Plagioclase crystal size distributions, growth and nucleation rates in an anhydrous arc basaltic andesite

Abstract

We experimentally investigated plagioclase nucleation and growth in anhydrous arc basaltic andesite at 1 atm and Ni-NiO equilibrium. After equilibration at 1190 °C (15 °C above the liquidus) for 24 h, experiments were cooled at 1, 3, or 9 °C/h and quenched at 1175–1000 °C. New plagioclase grains nucleated near the liquidus, followed by minor amounts of Fe–Ti oxides and pyroxene below 1120 and 1050 °C, respectively. Plagioclase shapes varied from 2D tabular/elongated (1 and 3 °C/h) to hopper and swallowtail textures (9 °C/h), suggesting a transition from interface- to diffusion-controlled growth. Crystal shapes and sizes were correlated, with the smallest and largest having equant/elongated and tabular/bladed 3D shapes, respectively. To identify the most suitable method for inferring storage timescales in natural magmas, we calculated nucleation (J) and growth rates (G) with different methods: G_max from the average size of the 10 biggest crystals, G_mean from the entire crystal population, J_batch and G_batch from the number and proportion of plagioclase estimated by point counting, and J_CSD and G_CSD from the crystal size distribution (CSD). J and G were greatest near the liquidus and decreased during cooling; the decrease was minimal at slow cooling rates, making G nearly constant. G decreased with decreasing cooling rates (from 10⁻⁷ to 10⁻⁹ cm/s at 9 and 1 °C/h, respectively), stabilizing after ~ 20 h of cooling. These variations of G principally resulted from differences in experimental conditions, more than the calculation method considered. Given the uncertainties of CSD theory in closed systems and the size and crystallographic axis-dependence of growth rates, combining Gₘₑₐₙ and Gₘₐₓ appears to be the most effective method for experimentally determining growth rates. However, the batch method (J_Batch) still provides a good estimate of J.