Cumulate formation and melt extraction from mush-dominated magma reservoirs: the melt flush process exemplified at mid-ocean ridges

International audienceVolcanism is the surface expression of extensive magmatic systems, with their intrusive counterpart representing ~80% of the total magma budget. Our knowledge of igneous processes therefore largely relies on our understanding of deep plutonic processes. In continental or oceanic environments, most of the intrusive igneous rocks bear geochemical cumulate signatures (e.g., depletion in incompatible elements and enrichment in compatible ones) that are commonly explained by mineral-melt segregation during differentiation. Deformation-assisted compaction aided by melt buoyancy is usually referred to as the main process involved in melt extraction. However, buoyancy alone is not sufficient, and a number of cumulative rocks are lacking any compaction evidence, opening the potential for the involvement of other processes. In addition, our view of magmatic systems has shifted in the last decades from large melt-rich bodies to crystal-rich magma reservoirs. This paradigm shift challenges some of the long-established first-order igneous concepts like the idea that melt differentiation at depth is mainly governed by (fractional) crystallization; alternatively, the presence of mush potentially favors additional processes such as melt-mush reactions. We propose a novel igneous process for the formation of igneous cumulates, consistent with the mushy nature of oceanic igneous reservoirs, their continuous/cyclic replenishment by primitive melts, and the widespread occurrence of reactive porous flow (RPF) during magma differentiation identified in a growing number of magmatic systems. The “melt flush” process relies on melt-mush reactions between the primitive recharge melt(s) and crystal mush. Replacement of the more evolved interstitial melt by the primitive recharge melt leading to reactions (dissolution+crystallization), and concomitant extraction of the more evolved melt from the cumulate by buoyancy participate in the acquisition of the final cumulate signature. This process relying on oceanic igneous systems considers for the first time melt inputs and not only melt extraction, and matches the petrographic (e.g., mineral dissolution evidence) and geochemical constraints (trace element signatures) brought by natural oceanic samples. We tested various melt-mush reactions likely involved in the early stages of the melt flush process during RPF to investigate their thermodynamic feasibility with the Magma Chamber Simulator. First-order results show that one-step equilibration of primitive melts with primitive to moderately differentiated mush crystals triggers mineral assimilation. Together with the constraints established from the natural rock record, it strengthens the idea that RPF is a potential key process for magma differentiation in magma reservoirs at different evolution stages. The proposed melt flush process eventually adds to other processes involved in cumulate formation like magma compaction or crystal settling, and is likely to apply to any other magmatic system from various settings sharing similar reservoir characteristics