MAGMA CRYSTALLIZATION DURING EMPLACEMENT OF COLUMBIA RIVER FLOOD BASALT DIKES

The Chief Joseph Dike Swarm (CJDS), located in northeastern Oregon, is the interpreted magmatic plumbing system for much of the Columbia River Flood Basalts, Earth’s youngest flood basalt province. Dikes are generally interpreted as having been emplaced as near liquidus, crystal-poor magmas, with internal textural variations resulting from local cooling. In such a scenario, the result should be a textural coarsening into the interior of the dike. This was not observed in samples that were collected across several CJDS dikes of varying width (PR-1: ~2m, PR-6: ~18m, PR-9: ~8m). Textural analyses using electron backscatter diffraction (EBSD) allowed for a non-subjective interpretation of crystal grain boundaries, as well as generating robust datasets of several thousands of grains. These measurements were input into traditional stereological conversion toolkits (ShapeCalc for crystal shape estimation, CSDcorrections for crystal size distributions). The resulting crystal size distributions are primarily concave up, and within individual dikes, the distributions are remarkably similar. Any meaningful variance in crystal size distributions within a dike are present only in the proportions of large crystals. This is highlighted in dike PR-6, where the interior portion of the dike contains numerous large crystals, however, the groundmass is within error, identical to that of the margins. These results indicate that textural variation across dikes is not the result of in situ cooling. We propose an alternative crystallization scenario where texture is developed via degassing during ascent. For low crystallinity magmas, ascent through the volatile saturation curve would account for this effect. However, as crystallinity increases in the magma, the rheological properties of the magma change, and eventually the magma becomes a flow-resistant mush. Under these conditions, shear results in the formation of underpressured dilatant voids, which further act to quench the melt via highly localized decompression. These results suggest that the assumptions of near liquidus magma emplacement, and all of the associated magma dynamics implicit with this assumption, may need reconsideration.