Influence of Grain Boundaries and their Composition on the Deformation Strength of High-purity, Synthetic Forsterite

Abstract

Grain boundaries are an important feature of the mantle. With recent studies suggesting the majority of the upper mantle deforms by grain boundary sliding (Hirth and Kohlstedt, 2003; Hansen et al., 2013), understanding the role grain boundaries play is key. As grain boundary sliding always requires an accommodation mechanism, directly determining the contribution of grain boundary sliding to total strain on a sample is important for modeling deformation in the mantle. Altering grain boundary composition can change the structure and viscosity of the boundary. Understanding the effects of grain boundary composition is necessary for comparing data sets of different olivine as well as for accurately extrapolating experimental data to represent the mantle. In Chapter 2, uniaxial deformation experiments on high-purity synthetic forsterite at high temperature and ambient pressure are used to characterize the contribution of grain boundary sliding to strain in diffusion creep. Experiments were conducted in a one-atmosphere deformation rig, which allowed the polished surfaces of the samples to be analyzed with atomic force microscopy. The high temperature necessary for deformation enabled a great deal of thermal grooving, which can dramatically alter the topography of an initially polished surface. A methodology was developed to correct for the effect of thermal grooving and determine the amount of grain boundary sliding as a function of grain size and stress. A comparison is also made between two popular methods for determining grain size: the line intercept method and the equivalent area circle method. The line intercept method consistently produces larger grain sizes than the equivalent area circle method. In Chapter 3, triaxial compression experiments on forsterite are used to determine the effect of grain boundary chemistry on deformation strength. High-purity synthetic forsterite was doped with either Ca or Pr and then deformed at high temperature and a confining pressure of 300 MPa. Both impurities made the sample stronger, and the presence of Ca induced abnormal grain growth. This supports the theory that grain boundary composition can have a large effect on deformation strength. The hypothesis that the difference in strength between natural and high-purity synthetic olivines is due to the difference in grain boundary composition is not supported by these results. In Chapter 4, the results of experiments on forsterite with a small amount of melt are detailed. Two methods of adding melt were used. The first involved adding Pr to forsterite in concentrations greater than can dissolve in the grain boundary, which induced melting as well as enhanced grain growth. Even with a grain size over an order of magnitude greater than the melt-free sample, the melt bearing samples were weaker than the melt free samples. The second method involved synthesizing forsterite with a composition in equilibrium with a synthesized anorthitic melt. Samples were created with melt fractions < 0.01 and then deformed at a temperature of 1300°C and a confining pressure of 300 MPa. The drop in viscosity at very small melt fractions predicted by Takei and Holtzman (2009) was observed, although the drop occurred over a shorter change in melt fraction than predicted. This result suggests that, at the onset of melting, the mantle will become significantly weaker. In addition, the presence of as little as 0.1% melt in a high purity, synthetic olivine sample brings its deformation strength into agreement with natural samples. This suggests that deformation experiments on natural samples are never entirely melt free. The results of this study establish the role of grain boundary chemistry on polycrystalline deformation. The presence of large cations in olivine grain boundaries makes diffusion creep slower, which limits the regions of the mantle predicted to deform in diffusion creep and expands the regions predicted to deform in a dislocation accommodated grain boundary sliding or dislocation creep. At the onset of melting, this changes, as the melt would remove the impurities from the grain boundaries. Future studies on different types of impurities will allow the grain boundaries of natural olivines to be more accurately modeled.