Partially Molten Plumes and Melt-Fingers: Two Modes of Magma-Transport through the Mantle in Terrestrial Bodies

To understand the dynamics of partially molten mantle in terrestrial bodies, we carried out a linear perturbation analysis and 2-D numerical simulations of magma-matrix flow in a horizontal layer, where decompression melting generates magma that percolates through the convecting matrix. Our study shows that there are two regimes for the upward migration of magma, depending on the melt-buoyancy parameter B_m, which is the ratio of the Stokes velocity of matrix to the percolation velocity of melt, both driven by the melt-buoyancy. At large B_m, the magmatism-mantle upwelling (MMUb) feedback dominates the convective flow in the layer: decompression melting during upwelling enhances magma buoyancy, which further strengthens the upwelling. When a solid layer is overlaid on the partially molten layer, the MMUb feedback induces partially molten plumes that ascend through the solid layer by their melt-buoyancy. At lower B_m, in contrast, a perturbation in the melt-content in the partially molten layer propagates upward as a porosity wave: the perturbation induces a spatial variation in the rate of expansion or contraction of matrix caused by magma migration, leading to an upward shift of the perturbation. When a solid layer is overlaid, the porosity wave develops also along the layer boundary to induce a finger-like magma structure, or melt-finger, that extends upward into the solid layer. The threshold value of B_m for MMUb feedback suggests that it can explain volcanism forming Large Igneous Provinces, but not hotspot volcanism on Earth. Since B_m increases with decreasing matrix viscosity, volcanism caused by the MMUb feedback is likely to have been more important in earlier terrestrial planets where the mantle was hotter and softer. Melt-fingers are, in contrast, expected to have developed in the lunar mantle if a partially molten layer has developed at its base in the history of the Moon.

💡 Research Summary

The authors investigate how magma migrates through a partially molten mantle layer by coupling analytical linear‑perturbation theory with two‑dimensional numerical simulations of magma‑matrix flow. The system consists of a horizontal mantle slab in which decompression melting continuously generates melt that percolates upward through a solid matrix. The key nondimensional control parameter is the melt‑buoyancy number, Bₘ, defined as the ratio of the Stokes rise velocity of the solid matrix (driven by buoyancy) to the Darcy‑type percolation velocity of the melt (also buoyancy‑driven). Because Bₘ depends on matrix viscosity, porosity, and the buoyancy force, it encapsulates the relative importance of matrix motion versus melt percolation.

Two distinct regimes emerge. When Bₘ is large (matrix motion dominates), a positive feedback loop—termed magma‑mantle upwelling (MMUb)—operates. Upward matrix flow causes decompression, which enhances melt production; the added melt increases buoyancy, further accelerating matrix upwelling. This feedback concentrates melt into buoyant, columnar or spherical “partially molten plumes.” If a solid lid overlies the partially molten layer, the plume’s intrinsic buoyancy enables it to pierce the lid and rise into the overlying solid. The authors show that this regime can generate the high melt fluxes required for Large Igneous Provinces (LIPs) and that the threshold Bₘ for MMUb is consistent with the conditions expected in early, hotter, low‑viscosity terrestrial planets.

Conversely, when Bₘ is small (percolation dominates), a perturbation in melt fraction propagates upward as a porosity wave. The wave arises because a localized increase in melt fraction induces differential matrix expansion or contraction; the resulting volumetric strain advects the perturbation upward. When the wave reaches a solid‑melt interface, it spreads laterally along the boundary, producing narrow, finger‑like melt intrusions—“melt‑fingers”—that extend upward into the solid layer. Melt‑fingers are efficient conduits for modest melt fluxes and are favored in high‑viscosity, low‑Bₘ environments.

Numerical experiments confirm the analytical predictions. By varying Bₘ across several orders of magnitude, the authors identify a transition around Bₘ ≈ 10⁻². Above this value, MMUb dominates, producing robust plumes that breach the lid; below it, porosity waves and melt‑fingers prevail. Because Bₘ scales inversely with matrix viscosity, the MMUb regime would have been more prevalent on early Earth, Venus, or Mars when mantle temperatures were higher and viscosities lower. In contrast, present‑day Earth’s cooler mantle yields lower Bₘ, making melt‑finger dynamics more likely for contemporary hotspot volcanism. The lunar mantle, having possibly hosted a thin basal partially molten layer early in its history, would have operated in the low‑Bₘ regime, favoring melt‑finger formation.

The paper concludes that the MMUb feedback can explain the massive, short‑lived magmatic output of LIPs but not the sustained, localized fluxes of modern hotspots, which are better matched by melt‑finger transport. The work highlights how a single nondimensional number, rooted in mantle rheology, can dictate fundamentally different magma‑transport styles and thereby shape the volcanic and tectonic evolution of terrestrial bodies.