Rapid Diffusion of Ca Along Migrating Grain Boundaries in Calcite

Grain boundary diffusion is an important process at moderate to high homologous temperatures and one that has the potential to redistribute isotopes and other chemical species over significantly greater length scales than volume diffusion alone. When the grain boundaries are moving, this redistribution can be greatly enhanced because the incorporation of tracers into the grain interiors is not limited by the short length scale of volume diffusion, as it is in the case of static boundaries. An experimental charge consisting of a layer of fine (< 5 μ m) calcite powder containing 25% ⁴⁴Ca-enriched calcite (98% ⁴⁴Ca compared with the natural abundance of 1.95%) was sandwiched between cylinders of single crystal calcite and Carrara marble and hot isostatically pressed at 900 ° C 200 MPa for 10 hours. Examination of the experimental run products using the SEM in orientation contrast mode shows that during the experiment the fine calcite layer underwent static recrystallization followed by normal grain growth. The resulting mean grain size was 60-80 μ m. The single crystal also statically recrystallized in a 5 mm wide layer immediately adjacent to the powder layer. The resulting grain size was 100 μ m to 1 mm. The driving force for this latter recrystallization was the small amount of strain energy introduced during the initial pressurization of the charge as the powder layer compacted. There were no noticeable changes in the Carrara marble layer, which retained its initial grain size of about 150 μ m. Isotopes ⁴⁴Ca and ⁴⁰Ca were imaged using the Cameca ims4f microprobe at Edinburgh University, and ion beam traverses were also made with an estimated effective beam diameter of 5 μ m. Combined images of ⁴⁴Ca/(⁴⁴Ca+⁴⁰Ca) reveal extensive but very heterogeneous zoning of ⁴⁴Ca in the recrystallized single crystal, and relict cores of ⁴⁴Ca-enriched calcite in recrystallized grains in the powder layer. On the whole the powder layer homogenised to a composition of 22-24% ⁴⁴Ca, and provided a fairly fixed composition source of ⁴⁴Ca to diffuse into the recrystallizing single crystal. Penetration into the single crystal along mobile grain boundaries was extensive, with zones of up to 8% ⁴⁴Ca present locally at > 300 μ m from the tracer layer. Complex isotopic zoning patterns are interpreted to result primarily from variations in boundary migration velocity. There is good evidence that parts of the original single crystal have been swept by multiple grain boundaries, and that some boundaries reversed migration direction. Mishin and Razumovski (1992) developed an analytical model for the concentration (C) of tracer in a grain behind a grain boundary that is moving parallel to an interface beyond which there is a constant composition source of tracer. They found that the gradient reaches a steady state given by d(lnC)/dx = (V/D_GBδ )^0.5, where V is the velocity of migration, D_GB the grain boundary diffusion coefficient and δ the grain boundary width. Although our grain geometries are more complex and our ion probe profiles are not ideally located, it is possible to estimate maximum values for d(lnC)/dx parallel to moving boundaries and minimum values for V based on the width of compositional zones and the duration of the experiment. We estimate minimum values for D_GBδ of 10^-18 m³/s, about 500 times greater than the value reported by Farver and Yund (1996) in static grain boundaries.