Dynamics of Melting and Melt Migration as Inferred from Incompatible Trace Element Abundance in Abyssal Peridotites

To better understand the melting processes beneath the mid-ocean ridge, we developed a simple model for trace element fractionation during concurrent melting and melt migration in an upwelling steady-state mantle column. Based on petrologic considerations, we divided the upwelling mantle into two regions: a double- lithology upper region where high permeability dunite channels are embedded in a lherzolite/harzburgite matrix, and a single-lithology lower region that consists of partially molten lherzolite. Melt generated in the single lithology region migrates upward through grain-scale diffuse porous flow, whereas melt in the lherzolite/harzburgite matrix in the double-lithology region is allowed to flow both vertically through the overlying matrix and horizontally into its neighboring dunite channels. There are three key dynamic parameters in our model: degree of melting experienced by the single lithology column (Fd), degree of melting experienced by the double lithology column (F), and a dimensionless melt suction rate (R) that measures the accumulated rate of melt extraction from the matrix to the channel relative to the accumulated rate of matrix melting. In terms of trace element fractionation, upwelling and melting in the single lithology column is equivalent to non-modal batch melting (R = 0), whereas melting and melt migration in the double lithology region is equivalent to a nonlinear combination of non-modal batch and fractional melting (0 < R < 1). Given the nonlinear nature of the melting model and uncertainties in trace element data for the abyssal peridotite, we showed, with the help of Monte Carlo simulations, that it is difficult to invert for all three dynamic parameters from a set of incompatible trace element data with confidence. However, given Fd, it is quite possible to constrain F and R from incompatible trace element abundances in residual peridotite. As an illustrative example, we used the simple melting model developed in this study and selected REE and Y abundance in diopside from abyssal peridotites to infer their melting and melt migration history. The abyssal peridotites used in this study are from Central Indian Ridge [1] and Vema Fracture Zone along the Mid-Atlantic Ridge [2]. As one of the end-member cases, we chose DMM as our starting mantle composition and assumed melting initiates in the spinel lherzolite field. To invert for F and R from a given set of trace element data, we considered a range of Fd values (0-5%). Overall, the degree of melting inferred from these two sets of data is not sensitive to the value of Fd used in our inversion and ranges from 9% to 25%. The relative rate of melt suction, R, however, depends slightly on the choice of Fd and ranges from 0.67 to 0.99 for Fd = 5% and 0.55 to 0.97 for Fd = 0. Hence there is a strong component of fractional melting beneath the mid-ocean ridge with an average of 80% melt being extracted through dunite channels. Further, our estimated R is inversely correlated with F, a robust feature independent of the choice of Fd. The upward decrease of R in the upwelling mantle column can be understood in terms of an upward increase in the volume fraction of high permeability dunite channels in the double-lithology region. And finally, given F and R, we found that the relative mass flux of the melt percolating in the lherzolite/harzburgite matrix also increases as a function of F (or height) in the melting column. This is a natural consequence of concurrent melting and melt migration in an upwelling steady-state mantle column. [1] Hellebrand et al. (2002) J. Petrol. 43, 2305-2338; [2] Brunelli et al. (2006) J. Petrol. 47, 745-771.