University of Minnesota
University of Minnesota
David Kohlstedt

Department of
Earth Sciences

Rock and Mineral Physics Lab


Research Themes

Role of water in diffusion, deformation, and solid state reactions

A small amount of water present as hydrogen or hydroxyl ions in nominally anhydrous minerals dramatically affects many physical and chemical properties of rocks and minerals including creep rate, dislocation recovery kinetics, electrical conductivity, and ionic diffusion. In most cases, however, these observations are only qualitative. Does the presence of water enhance the rate of creep by enchancing diffusivity? We are investigating the dependence of Fe-Mg interdiffusion in olivine as a function of hydrogen concentration as a route to understanding the mechanism by which water influences physical properties. We are also exploring the diffusion and solubility behavior of hydrogen in olivine as constraints on the water content of Earth's upper mantle.To utilize observations determined from laboratory experiments to develop models of the geochemical evolution and geodynamical behavior of the mantle, it is critical to quantify the dependence of kinetic properties on water content. An important goal of our experiments is to determine quantitative relationships between cation (Fe-Mg) diffusivity and hydrogen/hydroxyl concentration for mantle minerals under well-defined thermochemical conditions. We analyze our results in the context of point defect thermodynamics to gain a fundamental understanding of the mechanism(s) by which water influences transport properties such as ionic diffusion and electrical conductivity. The resulting theoretical framework is essential for reliably extrapolating laboratory results to mantle environments and for utilizing experimental observations to develop models of the dynamic behavior of the Earth’s interior.

Rheology of lower crust/upper mantle materials

We are investigating the high-temperature, high-pressure rheological properties of clinopyroxene and olivine, important minerals in the lower crust and/or the upper mantle. Because even a small amount of water dramatically weakens nominally anhydrous minerals, emphasis is given to the contribution of water to the viscosity of single- and polyphase aggregates of clinopyroxene and olivine. In most studies, the water-weakening effect has been treated as an ‘on-off’ process – if water is present, rocks are weak; if water is absent, rocks are strong. Recently, however, we demonstrated that the viscosity of olivine-dominated aggregates is inversely proportional to OH concentration. To extrapolate from laboratory to geologic conditions, a major goal of our research is to determine detailed flow laws describing strain rate (i.e., viscosity) as a function of water fugacity (i.e., water concentration) as well as differential stress, grain size and temperature in both the diffusion creep and the dislocation creep regime.
Recent analyses of the depth of earthquakes in continents and of the correlation between crustal thickness and effective elastic thickness (Te) have led to the conclusion that the lower crust, at least in some places, is stronger than the uppermost mantle. This point of view is contrary to models of continental rheology such as the jelly-sandwich construction in which a weak lower crust lies between a strong upper crust and a strong upper mantle. Many models take the strength of the continental lithosphere to lie in the mantle beneath the Moho. Certainly, the relative strength of lower crust and upper mantle will depend on the presence or absence of water. Little is known about the strength of clinopyroxene relative to that of olivine, even when the two phases are present in the same rock as in the mantle. Our recent experiments suggest that these two minerals have similar strengths under anhydrous conditions but that clinopyroxene is weaker when water is present. A careful investigation of the dependence of viscosity on water concentration for aggregates of clinopyroxene and clinopyroxene + olivine is critical in order to address these issues and to model the dynamical behavior of the lower crust and upper mantle.

Transport properties of partially molten rocks

Geochemical contraints on the rate of melt extraction from beneath a mid-ocean ridge indicate that melt is transported from the MORB source region to the surface much more quickly than is possible by porous flow. Hence, we have investigated the role of deformation on the permeability structure of partially molten mantle rocks. In sheared samples of olivine + basalt, a distinct melt preferred orientation develops with melt-rich planes oriented sub-parallel to the shear plane and antithetic to the shear direction. We have calculated the permeability of these samples using lattice-Boltzmann simulations on 3-D reconstructions of the melt distribution. In sheared samples of olivine + chromite + basalt and of anorthite + basalt, not only does the melt develop a pronounced preferred orientation, but it also undergoes a profound segregation. As a result, melt-rich bands spontaneously form, separated by regions of relatively low melt fraction. These low-viscosity melt-rich bands are paths of high permeability and may correspond to the replacive dunite shear zone channels observed in the mantle section of ophiolites.

Rheology of partially molten rocks

The geometry of mantle flow beneath the lithosphere at mid-ocean ridges, hot spots, and subduction zones and the processes that accommodate melt extraction from these tectonic settings are controlled by the rheological properties of partially molten peridotite. An essential parameter in models of the relationship between melt extraction and mantle flow beneath mid-ocean ridges is the viscosity of the partially molten aggregate. To develop the constitutive equations needed to model the geodynamic behavior of partially molten regions of the upper mantle, we are carrying out high-pressure, high-temperature experiments on samples of lherzolite and of olivine + basalt. Our research explores the influence of melt on viscosity of mantle rocks. The results provide fundamental constraints for physical models for melt extraction and mantle flow and aid in the interpretation of the processes that result in the focusing of melt flow into channels near the lithosphere-asthenosphere boundary or within an upwelling mantle plume.

Melt migration and planetary core formation

Most recent models of formation of the cores of terrestrial planets involve gravitational sinking of molten iron with some nickel or iron-nickel sulfide through a partially or fully molten silicate mantle, often referred to as a “magma ocean.” Alternative models invoke percolation of metal melts along an interconnected network (i.e., porous flow) through a solid silicate matrix. Previous experimental studies performed at high pressures showed that, under hydrostatic conditions, these melts do not form an interconnected network, leading to the widespread assumption that formation of metallic cores requires a magma ocean. In contrast, our experiments demonstrate that shear deformation to large strains interconnects a significant fraction of initially isolated pockets of metallic melts in a solid matrix of polycrystalline olivine. Therefore, we argue that, in a dynamic environment, percolation may be an important mechanism for the segregation and migration of core-forming melts in a silicate mantle.

Rheological properties and planetary evolution

Venus is tectonically quite distinct from Earth. While both planets show abundant evidence of tectonism and similar ranges of topography, the plate tectonic style observed on Earth is almost totally absent on Venus. In addition, the topography on Venus is only partially gravitationally compensated. One explanation for the high topography with limited isostatic compensation is the presence of a stiff lithosphere and the absence of a weak asthenosphere in the venusian upper mantle. Our experimentally determined flow laws for dry diabase rocks are particularly appropriate for crustal deformation models of Venus because high surface temperatures and extensive volcanic degassing likely resulted in a dehydrated crust and gradual dehydration of the venusian interior. Our measurements predict a far stronger crust than previously expected and indicate no significant strength contrast across the boundary between the crust and mantle. The entire crust is stiff and likely strongly coupled to the upper mantle. On Earth, the presence of a weak region (asthenosphere) in the upper mantle allows the lithosphere to relax toward a state near isostatic equilibrium. The low viscosity of this region is due to the presence of water in Earth’s mantle. Given the depletion of water in the Venusian interior, an asthenosphere on Venus seems unlikely, so that rates of isostatic compensation are expected to be slower than on Earth. Hence, differences in the tectonic styles and rates of isostatic equilibration on Earth and on Venus are a direct result of the anhydrous nature of the venusian crust and the progressive dehydration of the venusian interior

Structure and properties of grain boundaries

Grain boundaries directly influece both the physical and the chemical properties of rocks, yet relatively little is known about their structure. Grain boundaries are paths of rapid diffusion and sites for storage of incompatible elements. We are using high-resoloution transmission and analytical electron microscopy techniques investigating both the structure and the local chemistry of grain boundaries in olivine-rich rocks. Our observations demonstrate that most grain boundaries do not contain a second (e.g., an amorphous) phase, yet, elements such as Ca, Ti and Al are concentrated at grain boundaries. The presence of these ions will influence the point defect structure of the grain boundaries and, hence, the kinetics of diffusion along grain boundaries.

Rheology of ice

The dynamical properties of glaciers, ice sheets and icy planetary interiors are controlled in large part by the grain-scale deformation of ice. To better understand the rheological behavior of ice and to develop a constitutive equation that can be used to model the flow behavior of natural ice bodies, we carried out laboratory deformation experiments on ice I. One of the exciting discoveries in our research is that ice exhibits an extensive creep regime in which flow is rate-limited by GBS, a regime often described as superplastic deformation. Our results provide the basis for state-of-the-art thermal/mechanical models of the modern Greenland ice-sheet and the ancient Laurentide ice-sheet to simultaneously explain the forms of these cryospheric structures, a situation not possible with the flow law previously used to describe the rheological behavior of ice.

  research themes
  Prof. David L. Kohlstedt
  Dept. of Earth Sciences
  University of Minnesota
  310 Pillsbury Dr. SE
  Minneapolis, MN 55455
  Phone: (612) 626-1544
  Fax: (612) 625-3819
  dlkohl at

  last modified
  November 21, 2010