Crystal clustering in magmas: mechanisms and significance

During crystallisation of molten rock, crystals often form clusters. Crystal clusters affect how easily magma - molten rock + crystals - can flow, as well as how easy it is to separate molten rock from crystals, which affects the chemical composition of the magma. Finally, crystal clusters can record information about the conditions they formed under. Clusters formed deep in the earth can be carried to the surface and erupted, acting like a message in a bottle from beneath a volcano. In this project we will use experiments to study crystal cluster formation in molten rock and compare our results to natural igneous rocks, that is, rocks that formed by solidification of molten rock. We will study clusters formed between two minerals, clinopyroxene and magnetite. Such clusters are especially common in basalt, itself the commonest igneous rock type on Earth. In our experiments, small (3 x 3 x 6 mm) platinum capsules containing glass with a composition similar to natural basalt will be compressed to pressures corresponding to a depth of 12km below the earths surface, and heated above the melting point of the glass (1200C) to create a synthetic molten rock. We will then cool the samples to different temperatures below the melting point and hold them at these temperatures for different amounts of time, allowing crystals and crystal clusters to form. At the end of each experiment, rapid cooling will freeze a snapshot of the samples appearance at high temperature. We will study how different crystallisation temperatures, experiment durations and amounts of water dissolved in the molten rock affect the crystal clusters formed in our experiments, focussing on two main types of information. Firstly, we will study the shape and distribution of crystals in two and three dimensions. To get a 3D image of the inside of our samples we will use X-ray computed tomography (CT), similar to the CT scans routinely used in medicine, but using higher energy X-rays, generated at a particle accelerator. Secondly, we will use a special technique in the scanning electron microscope to study the alignment between the crystal lattices of touching clustered crystals. Often, crystals that grow on other crystals align their crystal structure with the crystal they grow on, and so studying if and how the structures align can tell us about how the cluster formed. These alignments are also expected to change systematically with changing experimental conditions. Our experiments will show us if clusters form through different processes under different conditions as well as help to develop new ways of determining the conditions under which natural crystal clusters were formed. In the last stage of the project, we will compare our experimental results with basalt collected from Mt. Etna, Sicily, where clinopyroxene-magnetite clusters are common. The goal is to use our experimental data and the natural clusters to better understand the storage and movement of magma beneath Mt. Etna.

This project has greatly improved our ability to determine how clusters of crystals found in igneous rocks and experiments formed by studying the clusters themselves. We show how three-dimensional (using X-rays) and two-dimensional (using the electron microscope) observations of the shape and distribution of crystals in clusters can be combined with information about how the chemical composition of minerals varies in space and how the crystallographic orientations of touching crystals align to convincingly identify whether one mineral grain formed by "heterogeneous nucleation" on another (formation of a new crystal at an existing crystal surface). Using this result, we confirmed that certain alignments of the crystal structures of touching grains of the minerals clinopyroxene and titanomagnetite are excellent indicators of heterogeneous nucleation, as these "crystallographic orientation relationships" ("CORs") did not form when crystal clusters formed in other ways. Using our new criteria for identifying cluster formation mechanisms, we could fully reconstruct the crystallization history of our experiments. A key finding was that the availability of "seeds" for heterogeneous nucleation was more important than the temperature of each experiment for controlling the distribution and shape of crystals observed. We also found that the particular COR formed between clinopyroxene and titanomagnetite varied according to the difference between the actual temperature of the experiment and the temperature at which crystals should first form, although more research is needed to apply this insight to study natural rocks. Furthermore, in some experiments, titanomagnetite crystals clustered with clinopyroxene belonged to three different populations, with different sizes, shapes, and CORs, because cluster formation occurred in different ways and at different times relative to clinopyroxene. This is surprising, because our experiments involved only a single step of cooling. Therefore, igneous rocks with many different populations of crystals in clusters could form during a single cooling event. Previously, finding multiple different populations might have been used to argue for a much more complex cooling history for a rock. Due to heterogeneous nucleation, all our experiments crystallized very fast, reaching over 50% crystal fraction in less than five minutes - if this rapid crystallization occurred in a volcanic system, it would quickly slow or stop the flow of magma, potentially preventing eruption, or increasing the danger of explosive eruption. Because of this, we also studied the effects of rapid crystallization. We found that the crystal lattice of rapidly growing clinopyroxene shows a characteristic bending, which is different for different crystal growth directions and crystallization conditions, and could be used to get new information about rapid crystallization processes from natural samples. Finally, throughout the project we repeatedly demonstrated that crystal orientation measurement in the electron microscope (via "electron backscatter diffraction") is a powerful tool for studying crystallization.