MR14A-01 INVITED
Modelling dehydration reactions and deformation: theory and tests
There is no doubt that dehydration reactions affect rock strength as well as generating porosity. Dehydration reactions produce porosity in which fluid may not be at the same pressure as the rock matrix, once permeable pathways become established. Thus the rock matrix will deform as reaction proceeds, and this will feed back on fluid pressure and reaction progress. An internally consistent model encompassing dehydration reaction, fluid flow and deformation is presented here for the first time. Deformation implies that the rock matrix is moving, so the velocity field must be included as an evolving quantity, alongside porosity and fluid pressure. Matrix velocity is fully incorporated in our model, in contrast to some previous work. Our model presents a framework into which a variety of constitutive relationships can be inserted. These are: (1) a porosity-permeability relationship; (2) a reaction rate as a function of fluid and confining pressures; (3) a rheology (compaction rate) as a function of effective pressure. Such models contain many unknowns and demand testing. We use experiments on the dehydration of gypsum under a variety of conditions as a testbed for the numerical model. In these experiments, if one end of the gypsum polycrystal is maintained at a particular fluid pressure, higher pressures are developed elsewhere as a result of dehydration and hence spatial gradients are present. Because reaction, flow and compaction are linked, all of these processes become heterogeneous and this evolution provides several new ways in which to test the numerical model. Although tested in one system, the model serves as a guide for understanding the generic behaviour of rocks undergoing dehydration reactions.
MR14A-02
Solution Creep as a Deformation and Softening Mechanism in the Mid- and Lower- Crust
Chemical zoning preserved in grains of micas, feldspars, and amphiboles especially in foliated and lineated rocks implicates chemical processes in the development of these fabrics. Detailed electron microprobe analyses reveals monotonic and bell- or U-shaped patterns with the strongest chemical gradients present along the length of grains parallel to foliation and to lineation where present. Phyllosilicates preserve zoning in greenschist and lower amphibolite facies rocks in grains as small as 50 µm. Zoning in plagioclase of greater than 30% An content is preserved up to upper amphibolite facies. Core to rim zoning in amphiboles can range from 6.0 to 7.8 Si atoms p.f.u. We interpret these zoning patterns as growth zoning because pairs of parallel muscovite and chlorite, or muscovite and biotite flakes, and pairs of elongate amphibole and plagioclase grains are sympathetically zoned. Moreover, temperatures calculated using the amphibole- plagioclase thermometer correlate positively with Ti concentration in amphibole to 700°C, confirming both prograde and retrograde growth. These relationships strongly suggest that an instantaneous equilibrium existed between these pairs of minerals that crystallized simultaneously, but in an environment in which intensive variables were evolving during growth. Grains that are most strongly zoned along the length of the grain occur in rocks that are most strongly foliated; minimal zoning is found across the short dimensions of the grains. We interpret this to indicate that crystallization of these grains was syntectonic, parallel to the extension direction. Evidence of dissolution on the grain boundaries facing the shortening direction is less common, but present in the form of truncated grains and truncated zoning patterns within grains. Amphibole-plagioclase thermometry in amphibolites from Vermont and North Carolina suggests that solution creep is important to at least 730°C in rocks where amphibole interfaces were load bearing. However, in orthogneisses from Connecticut where networks of feldspar grains were load bearing, plagioclase is pervasively zoned, while relic magmatic amphibole grains show no evidence of recrystallization. We interpret that the development of a preferred orientation of amphiboles enhances deformation by solution creep in the direction of the c-axes of grains. Micas in phyllites and schists from New England show similar patterns, where crenulations tend to concentrate phyllosilicates into quartz- and feldspar-free layers, and in the limit, into cm or longer folia. These observations converge to suggest that solution creep is the dominant deformation mechanism. Thus foliations in both amphibole- and phyllosilicate-rich rocks are produced by a syntectonic incongruent dissolution-precipitation process (a type of metamorphic reaction) that tends to produce monomineralic folia. These in turn weaken the rock, lowering its viscosity. These reactions localize strain, and contribute to defining the base of the seismic zone at lower amphibolites facies conditions.
MR14A-03
Deformation of Ordinary Chondrite Under Very Reducing Conditons: Implications for Liquid Metal Compositions, HSE Partitioning and Enstatite Chondrites
One important method in which to gain insight into metallic liquid compositions and their ability to control HSE (highly siderophile element) distribution is through experimentation. Deformation experiments can additionally provide information into mechanisms and chemical consequences of dynamic liquid metal segregation under a variety of conditions. We report results on metallic liquid HSE compositions and their distribution from a set of deformation experiments on a natural H6 ordinary chondrite, performed under very reducing conditions and a series of phase equilibria experiments focused on HSE partitioning between Si-rich and S-rich Fe molten alloys. The deformation experiments were conducted at temperatures between 925°C and 950°C, at 1.3 GPa confining pressure with a strain rate of 10-4/s. Major element analyses of both silicate and metal phases show that they are considerably reduced and the typically lithophile elements are behaving like siderophiles. Fe-Ni-Si compositions are found in the shear zones produced during the deformation experiment. Metallic compositions also include (Mg,Fe,Ca)S, Fe-Ni-Si, FeP, and Fe-Ni-S quench metal. Silicate phases include forsterite (Fo92-96) and enstatite (En98). Highly siderophile element (HSE) concentrations have been measured in the sulphide ((Fe,Mg,Ca)S) and metal (Fe- Ni-Si) phases by LA-ICPMS and compared with results from an earlier set of experiments on the same material but which were not performed under reducing conditions. The partitioning of the PGE is modified by the changing conditions with elements such as Ir and Os having higher DMetal/Sulphide values under reducing conditions. Partitioning experiments between molten FeS and Ni-, Si-bearing molten Fe were performed at 1.5-5.0 GPa and 1500-1750° to further investigate this observation. The starting material is synthetic, doped with a range of trace and HSE elements. The results confirm the preference of the HSE for the metallic phase with DMetal/Sulphide > 100 in most cases, in contrast to Cu and Ag, which have D values near or below 1, respectively. Our results also suggest the possibility of significant PGE fractionation since D values are larger for Ir and Os and smaller for Pd and Au, with Pt, Ru, Rh having intermediate values. It is not clear with the present data set whether T and P variations can affect significantly HSE partitioning. These results have been applied to the most naturally reduced material we know, the Enstatite chondrites. Several E chondrites have bulk HSE data available, but no HSE data available on sulphide and metallic phases themselves. We have now a set of HSE data for individual metallic phases in several enstatite chondrites, both EH and ELs. The bulk data show that for elements such as Os and Pd, the abundances are positively correlated and overall Pd is much higher in abundance. We find in the experiments that DPd ranges between 10-100, but do not fully explain the bulk trends. Additional phases, such as FeP have therefore been analyzed and we find that Pd is concentrated in FeP and the presence of schreibersite may help explain the high Pd ratios (e.g. Pd/Ir) observed in the Enstatite chondrites.
MR14A-04 INVITED
Mineral growth in metamorphic rocks: relationships between chemical patterns, mineral microstructure and reaction kinetics
Mineral growth in metamorphic rocks is governed by kinetic processes, which are strongly related to geometrical and physical properties of the chemical pathways involved. Information on the nature of chemical pathways and potential effects of deformation on kinetic processes is stored in chemical patterns and mineral microstruture. By using high-resolution analytical techniques (i.e. EBSD, FE-EMPA and TEM/FIB) and numerical simulations we explored natural and experimental samples in order to reveal the kinetics and the microstrutural evolution of both transport controlled mineral reactions and exchange reactions. We found reaction rims between reactant minerals, which are composed of several grains. These grains are subdivided into subgrains perpendicular to the reaction fronts. The subgrains often show a systematic lateral orienation change, which is possibly caused by grain-scale plastic deformation in the form of creep polygonization. Such an arrangment of grain boundaries allows for short-circuit diffusion and amplifies the necessary mass transfer across reaction rims. Information about the contribution of grain boundary diffusion to bulk material flow in Grt is stored during the formation of asymmetric growth zonings and during micron- scale compositional variations along grain and phase boundaries. Our evaluation of these compositional patterns by diffusion modelling allows for the derivation of Digb/Divol ratios and rim growth rates. Information about phase boundary material properties is also stored in the zoning of Grt porphyroblasts, which formed during the retrograde Fe-Mg exchange between Grt and biotite. If temperature falls below a limiting value during cooling, non-equilibrated rim compositions start to develop along Grt-Qtz phase boundaries due to insufficient chemical transport rates. As the extent of non-equilibrated rim compositions reflects the transport capacity of the present interganular medium, our simulations by diffusion modelling of observed compositional patterns formed during cooling at Grt-Bt-Qtz triple junctions allows for the extraction of boundary material properties.
MR14A-05
Deformation-related trace element modification in mantle zircon and implications for Ti-in- zircon thermometry
A single zircon grain in a metasomatised mantle garnet websterite xenolith from Udachnaya, Siberia, preserves microstructures associated with crystal-plastic deformation that cumulatively accommodate up to 11° of misorientation. A complex CL pattern that mimics the deformation microstructure has resulted from modification of REE, particularly MREE enrichment relative to HREE, along fast-diffusion pathways associated with dislocation creep microstructures, e.g., low angle boundaries. Deformation-related enrichment of U, and to a greater extent Th has caused increase in the Th/U from 1.61 to 3.52. However, little spread in the U-Pb ages across the grain indicates that the deformation occurred at 1799 ±40 Ma soon after zircon growth at 1851 ±65 Ma. These ages are synchronous with U-Pb ages elsewhere in the Siberian Craton, and possibly correspond to metasomatic events at mantle conditions. The zircon grain records significant variations in Ti from 2.6 to 30 ppm, with Ti depletion that is spatially associated with deformation microstructure interpreted as defect-related enhanced diffusion. These results challenge the current paradigm of Ti retention in zircon, and demonstrate disturbance in the Ti-in-zircon thermometry that results in spurious temperatures ranging over several hundred degrees Celsius. Consequently, the recognition of deformation features in zircon is fundamental to the correct interpretation of zircon geochemistry.
MR14A-06
Developing Discordant Monazite During Ultra-High Temperature Metamorphism
The integration of the textural and chemical characteristics of accessory and silicate mineral assemblages provides key information when investigating the metamorphic history of a terrane. In particular the information extracted from mineral assemblages can be used to great effect in making inferences about poorly understood processes such as those involved in the generation and preservation of ultrahigh temperature (UHT) metamorphic assemblages. Coupling these observations with calculated metamorphic phase diagrams for specific whole rock compositions (P-T pseudosections) allows a clearer picture of the whole P-T-t evolution to be reconstructed. This approach is especially important in terranes that have undergone UHT metamorphism due to the uncertainty surrounding the ability of geochronometers to record the timing of peak metamorphism and there response to high temperature crustal metamorphism. In this paper we present texturally constrained SHRIMP U-Pb data on monazite and zircon from rocks that have undergone UHT metamorphism during Gondwana amalgamation. SHRIMP analysis of monazite yields a large number of discordant analyses and thus a total Th-U-Pb chemical age would yield anomalously young mean age. Zircon data gives a younger age than the monazite age, which is consistent with the pseududsection analysis that suggests zircon growth is related to the post-peak exhumation of the rocks.
MR14A-07
Reaction-driven fracturing during replacement processes and metamorphism
Hydration reactions involving igneous or high-grade metamorphic rocks often cause a significant increase in local volume at the site of reaction. Because such volatilization reactions are triggered by infiltration of external fluids and occur relatively far from thermodynamic equilibrium, they are often fast enough to produce sufficient stress perturbations to drive local fracturing. Microstructural studies show numerous examples of such small scale reaction-driven fracturing both within the reacting mineral grains and in the surrounding rock matrix. Small scale fractures often link up to form fracture networks that promotes further fluid infiltration, more reaction, more stress build up, more fracturing etc. etc. We believe that such self-accelerating coupling between reactions, fracturing, and fluid migration provides a first-order control on the rate of hydration of the Earth's crust both during metamorphism and during lower-temperature processes such as weathering. We present experimental studies that constrain under what conditions replacement reactions may cause fracturing near the reacting interface, both in synthetic and natural materials. Examples of natural reaction- driven fracturing at a wide range of scales will be presented with focus on serpentinization. Finally, a simple mechanical model will be presented to illustrate the most pertinent features of the hierarchical fragmentation process that arise from reaction-driven fracturing and demonstrate how this process may cause an overall acceleration of the hydration process. Relevant references Iyer, K., Jamtveit, B., Mathiesen, J., Malthe- Sørenssen, A., and Feder, J., 2007. Reaction-assisted hierarchical fracturing during serpentinization. Earth and Planetary Science Letters, 267, 503-516. Jamtveit, B, Austrheim, H., and Malthe-Sørensen, A., 2000. Accelerated hydration of the Earth's deep crust induced by stress perturbations. Nature, 408, 75-79 Jamtveit, B., Malthe-Sørenssen, A., and Kostenko, O., 2008a. Reaction enhanced permeability during retrogressive metamorphism. Earth and Planetary Science Letters, 267, 620-627. Jamtveit, B., Putnis, C., Malthe- Sørenssen, A., 2008b. Reaction induced fracturing during replacement processes. Contributions to Mineralogy and Petrology, DOI 10.1007/s00410-008-0324-y Malthe-Sørenssen, A., Jamtveit, B., and Meakin, P., 2006. Fracture patterns generated by diffusion-controlled volume changing reactions. Phys. Rev. Letters, 96, art no. 245501
MR14A-08
Grain-Recycling Zoning of Plagioclase and Metamorphic Fractionation
Quartzo-feldspathic gneisses make up much of the continental crust often having enjoyed a complex thermal history. Determining peak metamorphic conditions using conventional equilibrium thermodynamics is difficult because there are too many degrees of freedom. Zoned minerals are problematic, because of uncertainties in the exact equilibrium assemblage at any particular time, but provide a time-dependent measure of changes in equilibrium conditions. Zoning can arise due to diffusion of atoms into a homogeneous lattice from grain boundarys or through mineral growth under changing pressure, temperature or bulk rock composition. Conventional growth zoning considers a porphyroblast (commonly garnet) growing in an effectively homogeneous matrix with the growth rate controlled by reactions that produce new porphyroblast material. However, simulations of zoning developed by grain growth in a monophase domains of more complex rocks show boundary migration rates control the zoning geometry as shrinking grains are cannibalised by growing grains. This new grain-recycling zoning develops because chemical reactions change the composition of the material that is swept by the grain boundary without changing the mineral. A model of this process has been created using the Gibbs free energy minimisation software Theriak-Domino controlled by our custom written Matlab control program. This program assumes an initially homogeneous composition of equigranular plagioclase then uses the experimentally determined normal growth law for plagioclase to calculate the grain-size increase for a given time-step thus giving the amount of material swept. Assuming this is all available for reaction at the same time, the rest of the plagioclase is removed from the bulk composition, the equilibrium plagioclase composition calculated, and added to the growing grain. This fractionation alters the range of plagioclase compositions available over a given temperature range, changing PT estimates obtained from assuming whole rock equilibrium. An example of this new type of zoning is presented from the Lewisian Gneiss, NW Scotland, and a new PT path calculated for post deformation grain growth. This model highlights the need for careful consideration of what is in equilibrium and how it is constrained by the rock microstructure when applying thermodynamics to bulk compositions.
MR14A-09 INVITED
Modelling dislocation cores in Forsterite
Olivine (Mg,Fe)2SiO4 is considered as the main constituent of the Earth's upper mantle (down to 410 km deep). The rheology of, and convection in, the upper mantle is therefore controlled by the deformation mechanisms of this mineral. Numerous experimental studies have been undertaken leading to a good description of the deformation mechanisms and rheological properties of this mineral at ambient pressure. However, recent studies have show that [001] glide is enhanced over [100] glide when pressure increases or when trace amounts of water are dissolved in the crystals. These observations have a lot of implications on our understanding of the rheology of the upper mantle and call for a more detailed description of the dislocation cores and dynamics. The Peierls-Nabarro (PN) model including generalized stacking fault energies is a privileged tool to calculate core structures at a remarkably low cost. Moreover, the PN model, which is usually restricted to the description of planar cores, is very adapted to look for the most mobile core configurations. However, dislocation cores may exhibit distinct, low-energy, configurations that are not described by the PN model. We present here new calculations based on full atomistic calculations (using the THB1 potential) and a method coupling Peierls–Nabarro and element-free Galerkin methods. These techniques expand the possibilities of previously reported calculations, in particular in permitting modeling 3D dislocation cores. We show that, [100] dislocations may exhibit non collinear dissociation in the (010) plane following the reaction [100] = 1/6[3 0 1] +1/6[3 0 -1]. We also discuss several possible core structures for [001] screw dislocations, including non-planar core spreadings.
MR14A-10
Theory of Lattice Strain for Materials Undergoing Plastic Deformation
Radial x-ray diffraction is used to probe physical properties of materials including elastic and plastic properties. The theory used behind such an practice is the one developed by Singh (1993) in which the relation between lattice strain and elastic constants and macroscopic stress is derived. In this theory, the variation of inferred stress with the crystallographic planes, (hkl), is due to the elastic anisotropy. However, recent experimental studies showed that in many cases, the variation of stress with (hkl) far exceeds the value expected from this theory. I have developed a modified theory to rectify this problem with Singh's theory. In Singh's theory, the stress distribution in a polycrystalline material is treated only either unrelaxed or relaxed state. The role of plastic deformation is included only to the extent that plastic flow influences this stress state. Such an assumption corresponds to a Voigt model behavior, which is not an appropriate model at high temperatures where continuing plastic flow occurs with concurrent microscopic equilibrium, elastic deformation. This is a Maxwell model type behavior, and my model provides a stress analysis in a Maxwell material with anisotropic and non-linear power-law rheology. In this theory, the lattice strain corresponding to an imposed macroscopic strain-rate is calculated by three steps: (i) conversion of macroscopic strain-rate to macroscopic stress, (ii) conversion of macroscopic stress to microscopic stress at individual grains, and (iii) calculation of microscopic strain due to microscopic stress. The first step involves anisotropy in macroscopic viscosity that depends on anisotropy in crystal plasticity and lattice-preferred orientation. The second step involves anisotropic crystal plasticity and finally the third step involves elastic crystal anisotropy. In most cases, the influence of LPO is weak and in such a case, the lattice strain depends on (hkl) due to the anisotropy in both elastic and plastic properties of constituent crystal. In many cases, anisotropy in plasticity is much larger than elastic anisotropy. The theory explains many experimental observations and provides a new means to determine plastic anisotropy from radial x-ray diffraction when elastic anisotropy is known. A. K. Singh, Journal of Applied Physics 73, 4278-4286 (1993).