Shearing Melt Out of the Earth: An Experimentalist's Perspective on the Influence of Deformation on Melt Extraction

Melt extraction from the Earth’s mantle requires some form of channelized flow to produce the observed disequilibrium between the crust-forming lavas and their mantle sources. In this review, we discuss the influence of deformation on melt-extraction processes, with an emphasis on the understanding gained from a wide range of laboratory experimental investigations. Rheological properties of partially molten rocks are very sensitive to melt distribution, and melt distribution is profoundly influenced by deviatoric stress in a viscously deforming partially molten rock, establishing a coupling between deformation and melt distribution. During deformation, this coupling can lead to organization of melt into melt-enriched shear zones at length scales longer than the grain size. We discuss the current state of understanding of this process, with some speculation on its role in the ensemble of processes that constitute melt extraction from the Earth. 561 A nn u. R ev . E ar th P la ne t. Sc i. 20 09 .3 7: 56 159 3. D ow nl oa de d fr om a rj ou rn al s. an nu al re vi ew s. or g by C ol um bi a U ni ve rs ity o n 04 /2 7/ 09 . F or p er so na l u se o nl y. ANRV374-EA37-23 ARI 23 March 2009 15:28 Melt migration: movement of melt down a pressure gradient over distances greater than the grain size Melt segregation: formation of meltenriched and meltdepleted regions at length scales longer than the grain size but shorter than the compaction length Melt extraction: ensemble of processes that lead from formation to emplacement of magmas via segregation and migration by porous flow, fracturing, and meltrock reactions INTRODUCTION Melting in Earth’s mantle occurs primarily at plate boundaries and in plumes, which are also regions of intense deformation. Because melt formation and rock deformation coexist in time and space on a large scale, the importance of melt on the viscosity of partially molten rocks has long been appreciated and studied in experiments. Initial laboratory investigations focused on the influence of a melt phase on deformation behavior of partially molten mantle rocks, with emphasis on the dependence of rock viscosity on the amount of melt present as reviewed by Kohlstedt & Zimmerman (1996). More recent experimental studies emphasized the intimately related problem of the effect of viscous deformation on redistributing melt, both by alignment at the grain scale and segregation at larger scales (Kohlstedt & Zimmerman 1996, Zimmerman et al. 1999, Holtzman et al. 2003a,b). This perspective reflected the observation that not only the presence but also the distribution of the melt phase influence large-scale convective flow within our planet (C. Goetze, personal communication). Importantly, the distribution of melt directly affects its rate of transport through the asthenosphere. Studies of the coupling between melt transport and rock deformation started in the 1970s with analysis of melt migration through the mantle in buoyancy-driven fractures (Weertman 1971) and by porous flow (Ahern & Turcotte 1979, McKenzie 1984). The recognition that chemical disequilibirium between erupted basalt and the rock through which it traveled motivated hypotheses for the development of high-permeability channels (Spiegelman & Kenyon 1992, Hart 1993). Subsequent investigations of the formation of high-permeability, melt-enriched channels followed two distinct but almost certainly coupled routes. First, a series of papers explored channel formation by a reactive infiltration instability mechanism (Daines & Kohlstedt 1994; Aharonov et al. 1995, 1997). As melt ascends adiabatically, it becomes undersaturated in pyroxene. An instability then develops as a result of positive feedback that amplifies local perturbations in melt distribution and thus in permeability (Ortoleva et al. 1987). Second, researchers identified the possibility of channel formation by stress-driven melt segregation (e.g., Stevenson 1989, Kelemen & Dick 1995, Holtzman et al. 2003a,b). In this case, an instability develops as a result of the dependence of rock viscosity on melt fraction. Melt is drawn to local regions slightly enriched in melt as they are weaker and thus at lower mean pressure than those with lower melt content (Stevenson 1989). Again a positive feedback is established that leads to melt segregation and channel growth. Both reactive melt infiltration and stress-driven melt segregation produce melt-enriched, high-permeability channels that isolate the melt phase from the solid residuum, allowing chemical disequilibrium to exist between the ascending magma and the host rock. Over the past 20 years, researchers have undertaken theoretical analyses, numerical models, field observations, and laboratory experiments on both analog and natural systems to investigate the dynamics and kinetics of spontaneous melt segregation. In the present paper, we review progress in the area of melt extraction from the mantle, starting with an historical perspective on physical/mechanical mechanisms for channelizing the flow of melt in the asthenosphere. We emphasize in particular experimental investigations of the influence of stress on melt distribution both at the grain scale and at the rock scale, carried out under well-controlled thermo-mechanical conditions with well-defined boundary conditions. Finally, we examine applications of stress-driven melt segregation to Earth both at plate boundaries and in plumes, discussing the process of extrapolating the results of laboratory experiments to the longer temporal and spatial scales of geological processes. HISTORICAL BACKGROUND Our understanding of the mechanisms and kinetics of melt extraction from Earth’s mantle has evolved as constraints from field observations, theoretical analyses, and experimental 562 Kohlstedt · Holtzman A nn u. R ev . E ar th P la ne t. Sc i. 20 09 .3 7: 56 159 3. D ow nl oa de d fr om a rj ou rn al s. an nu al re vi ew s. or g by C ol um bi a U ni ve rs ity o n 04 /2 7/ 09 . F or p er so na l u se o nl y. ANRV374-EA37-23 ARI 23 March 2009 15:28 Ophiolite: fragment of oceanic crust and uppermost mantle usually formed at a spreading center and subsequently obducted onto the continental crust measurements provide new insights into the physical paths, spatial extent, and temporal duration of melt transport. Evidence for melt movement both by porous flow through hot, viscously deforming partially molten rocks at depth and by injection of magma into fractures in cooler, more brittle rocks in the shallower lithosphere are observable in ophiolites (Nicolas 1989, Kelemen et al. 1997) and orogenic lherzolites (Le Roux et al. 2008, Kaczmarek & Müntener 2008). The mantle sections exposed in ophiolites reveal evidence for both brittle and viscous processes of magma transport. To address the question of which processes dominate melt transport in the mantle, the effects of physical and chemical conditions on the behavior of partially molten rocks must be investigated with observations from experiments and nature. As illustrated in the melt-migration mechanism map in Figure 1, a range of driving forces and of response mechanisms define thermo-mechanical regimes for melt transport through the mantle. The driving forces for movement of melt include buoyancy forces due to the differences in density between the melt and the residual rock and pressure gradients caused by shear deformation of a partially molten rock. Possible response mechanisms involve magma injection into dikes that form by brittle deformation and porous flow of melt through melt-enriched channels that develop during viscous deformation of partially molten rock. Here we discuss each quadrant in Figure 1 in an historical context; in addition, we introduce the role of melt-rock reaction in magma transport. Shear stress Buoyancy Driving force for melt migration a d