Page last updated August, 2009 by BDS
We use high pressure experimentation together with analytical and theoretical tools to understand melting, mass transfer, and differentiation in planetary interiors. This is an exciting time to be investigating the interior of the Earth and the terrestrial planets, as there are many new ideas regarding processes at the same time that new devices and instrumentation allow measurements that were previously intractable. A prime example of the former is increased recognition of the fundamental influence of deep planetary volatile cycles on both interior dynamics and surface geology. A good illustration of the latter is application of SIMS for accurate measurement of low concentrations of H2O in nominally anhydrous mantle minerals. Both are described in greater detail below.
In our research we employ solid media high pressure devices, including piston cylinder and multianvil apparatuses, for exploration of phase equilibria and transport properties of Earth materials up to ~15 GPa. Experiments are analyzed by a range of microbeam and spectroscopic techniques, including electron microprobe, high resolution FEG-SEM, SIMS, LA-ICP-MS, FTIR, and Raman spectroscopy. In another important research focus, we seek to improve and apply the MELTS algorithm for thermodynamic calculations of melting and mass transfer in the mantle.
Cycling of H2O between the mantle and the exosphere and between different reservoirs in the mantle is one of the critical processes governing Earth's geodynamical and geochemical evolution. For example, the distribution of H2O in Earth's interior influences transport properties including rheology, the style of convection and plate tectonics, observable geophysical properties, and the locus of phase transitions and of melting. The partitioning of H2O between Earth's internal and external reservoirs controls the mass of the oceans, and consequently also influences surface geology. However, the distribution and inventory of H2O in the mantle remains uncertain and the mechanisms of transport between principal mantle reservoirs are not fully known.
We are engaged in a program of high pressure experiments to determine the H2O storage capacities of mantle minerals and to determine the partitioning of H2O between minerals and between minerals and melts. Storage capacities are needed to constrain plausible masses of H2O reservoirs in different parts of the mantle and mass transfer processes that may occur at discontinuities. For example, we must document storage capacities of upper mantle phases to determine whether widespread hydrous melting may occur at 410 km depth associated with the conversion from wadsleyite to olivine (Hirschmann et al., 2005). Mineral/melt partitioning of H2O during melting is particularly important because it can be related directly to the influence of H2O on the solidus and hence constrains the locus of melting beneath ridges (Aubaud et al., 2004; 2008; Tenner et al., submitted).
To facilitate experimental investigations of H2O in nominally anhydrous minerals, we have invested considerable effort in developing and calibrating new SIMS methods for determining H in nominally anhydrous minerals. This is necessary because the small crystals produced in many experiments hinder more conventional (i.e., FTIR) analyses. The methodology for this technique was developed in collaboration with Erik Hauri at Carnegie Institution, but more recently we have performed the analyses at ASU in a collaborative effort with Laurie Leshin and Rick Hervig. We have developed low blank (<10 ppm H2O by weight) techniques that enable us to use SIMS to analyze small amounts of H in mantle silicates from high pressure experiments. The initial calibration study was published in Koga et al. (2003) and revised and improved versions have been published or are submitted (Aubaud et al. 2007; Tenner et al., submitted). We are using these techniques to determine mineral/melt partition coefficients for H (Aubaud et al., 2004; 2008; Tenner et al., submitted) and to determine the storage capacities of mantle minerals in the upper mantle and transition zone.
An important new experimental effort, lead by Ph.D. candidate Travis Tenner, is to make direct experimental determinations of the influence of H2O on the extent of melting of garnet peridotite. These H2O addition studies have direct applicability to dehydration partial melting deep beneath mid-ocean ridges and oceanic island environments and also may be related to the flux-melting that occurs above subduction zones.
Our work investigating the possible role of H2O in the deep parts of the upper mantle near the 410 km discontinuity is an interdisciplinary collaboration with several other faculty at Minnesota, including Justin Revenaugh (seismology), David Kohlstedt (rock and mineral physics) and Renata Wentzcovitch (molecular dynamics).
The flux of subducted carbonate from the surface into the mantle is known to be huge. This carbonate originates chiefly as hydrothermal calcite in altered sea-floor basalts, and metamorphic petrology establishes that much of it survives devolatilization during subduction. What is the fate of this carbon and what is the residence time of carbon in the deep mantle? What is the influence of carbon on melt formation, mass transfer, and the physical properties of the mantle? To address these questions Raj Dasgupta (Ph.D., 2006) investigated partial melting of carbonated eclogite and peridotite. Results on carbonated eclogite (Dasgupta et al., 2004; 2005) show that subduction delivers carbon to the deep mantle without melting, but materials returning to the upper mantle along oceanic geotherms melts at great depth (~400 km). Beneath oceanic ridges carbonated peridotite likely begins to melt at depths near 300 km (Dasgupta and Hirschmann, 2006). Efficient removal of carbon by melting deep beneath ridges may indicate a residence time for carbon in the mantle on the order of 1 billion years, which is about a quarter of previous estimates. This deep melting may also explain some of the deepest seismic features found beneath the East Pacific Rise and may have subtle but critical influences on geochemical mass transfer in the upper mantle, as the melts liberated at these depths may be highly concentrated in incompatible trace elements. Additionally, interactions between carbonate melts released at depth and shallower melting of silicates may have key influence on the compositions of melts formed in oceanic basalt source regions (Dasgupta et al. 2007a) and may have a critical role in the "drying out" and strengthening of lithosphere during melting beneath ridges (Dasgupta et al. 2007b).
Recently, we have extended our studies of volatiles in planetary interiors to Mars. The first study addresses the role of H2O in the Martian interior. The concentrations of H2O in Martian magmas is a source of considerable uncertainty and controversy. Lead by Tony Withers, we are determining the partitioning of H2O between basaltic melts and olivine and pyroxenes for compositions and conditions appropriate to the Martian mantle. These will establish the effect of dehydration melting on the lowering of the Martian mantle solidus, provide new constraints on the concentration of H2O in the Martian mantle, and bound the concentrations of H2O available to crust-forming melts.
The second Martian-related study, lead by graduate student Ben Stanley, will determine the concentration of CO2 liberated during partial melting of reduced, graphite-bearing, Martian mantle. Thermodynamic calculations (Hirschmann and Withers, 2008) calibrated from experiments on terrestrial compositions show that the liberation of CO2 from graphite-bearing mantle is retarded under the reducing conditions that are likely prevailing in the Martian mantle. This may present problems for hypotheses suggesting that volcanic ventilation of CO2 from the Tharsis region provided a thick CO2 greenhouse in the late Noachian. However, CO2 solubility in silicate melts is strongly compositionally dependent, and therefore experiments directly relevant to Martian compositions and conditions are required.
Geochemical diversity of oceanic basalts represents a fundamental constraint on the long-term dynamical history of Earth's mantle. There is increasing recognition from geochemists, geophysicists, and petrologists that the mantle is lithologically heterogeneous on a range of scales. Because the depth and extent of melting of the mantle varies depending on lithology, this heterogeneity has profound importance to dynamical and geochemical processes in basalt source regions. We have undertaken extensive studies of the partial melting behavior of a range of pyroxenite and eclogite lithologies at high pressure (Hirschmann et al., 2003; Kogiso and Hirschmann; 2001; 2006; Kogiso et al., 2003; 2004; Pertermann and Hirschmann, 2002; 2003a; 2003b; Pertermann et al., 2004; Dasgupta et al., 2006). These studies show that pyroxenite melts at greater depths and more productively than peridotite and that partial melts of certain pyroxenites bear strong resemblances to the silica-poor oceanic basalt common in many oceanic islands. This is particularly interesting because it was previously thought that partial melting of eclogites and pyroxenites always yields highly siliceous partial melts. Although this work continues, we have established much of the essential petrologic framework, and my focus is moving more towards the roles of volatiles in mantle melting (see above) and the role of small-degree melts in planetary evolution (see next paragraph). However, related studies will continue for the foreseeable future. Future directions will likely include studies of interactions between partially molten bodies and their surrounding peridotitic matrix, studies of sub-solidus hybridization between vein and matrix lithologies, and studies investigating the relationship between the length scale of heterogeneities and their impact on melt composition and production.
A new related direction is experimental study of very small degrees (0-5%) of melting of mantle lithologies at high pressures. Small degree partial melts may be of extraordinary importance in differentiation of planetary interiors, both in conventional basalt source regions as well as in deeper regions, where volatile-assisted melting may occur both above and below the transition zone. Additionally, small degree melting of garnet peridotite is probably one of the chief sources of oceanic island basalt and may be of critical importance to formation of highly enriched geochemical reservoirs in the convecting mantle and in the tectosphere. However, to date, direct experiments at low melt fractions have previously been intractable and there are no reliable experiments documenting the compositions of partial melts of any garnet peridotite below a melt fraction of 10%. We have recently developed a new approach employing iterative experiments to allow direct and accurate determinations of high pressure melting relations at the solidus of garnet peridotite and other complex lithologies (Hirschmann and Dasgupta, 2007). A proof-of-concept study has allowed determination of composition of near-solidus partial melt of carbonated peridotite at 6.6 GPa (Dasgupta and Hirschmann, 2007). More recently graduate student Fred Davis has been applying these methods to determination of small-degree partial melting of garnet peridotite near 3 GPa.
I have a long-standing interest in calibrating and applying thermodynamic models of magmatic phase equilibria by extending and applying the MELTS algorithm pioneered by Mark Ghiorso. This thermodynamic approach allows investigation of dynamic processes that are not easily approached experimentally. A prime example of this is adiabatic melting of the mantle. In collaboration with Ghiorso, Paul Asimow, and Ed Stolper, I performed thermodynamic calculations of adiabatic melting in the mantle that have had a significant influence on understanding on the production, energetics, and geochemical consequences of melting beneath mid-oceanic ridges. However, the utility of such calculations is limited by their accuracy and the conditions under which they are applicable. In collaboration with Ghiorso and Tim Grove, we have recently embarked on a new NSF-funded project to create the next generation of MELTS, termed xMELTS. xMELTS employs a novel equation of state for silicate liquids and consequently may be applicable to much higher pressures (30 GPa?) than previous generations (MELTS, pMELTS). It also involves a major revision of our thermodynamic model for silicate melts.
An offshoot of the MELTS development project is compilation of a comprehensive database of experimental mineral-melt phase equilibria. This database is known as LEPR (Library of Experimental Phase Relations) and is available to the public on line. Though database compilation is not the most intellectually creative activity, it is an essential step for advancement of our field. Why should understanding of magmatic phase equilibria be garnered through anecdotal and haphazard encounters with the published literature? As is the case for geochemical data presently being corralled into large online databases (Georoc, PetDB, etc.), magmatic phase equilibria data should be available to the community for calibration of thermodynamic and empirical models, for parameterizations incorporated into geodynamical codes, for planning new experimental studies, and for comparison to data from natural rocks and new experiments. A critical component of this effort is to incorporate enough additional information (metadata) to allow users to evaluate the quality and applicability of individual experimental studies.
- Marc M. Hirschmann
Page last updated August, 2009 by BDS