M. Liu et al., THE EFFECT OF STRAIN-ENERGY ON GROWTH-RATES DURING THE OLIVINE-SPINELTRANSFORMATION AND IMPLICATIONS FOR OLIVINE METASTABILITY IN SUBDUCTING SLABS, J GEO R-SOL, 103(B10), 1998, pp. 23897-23909
We have studied growth kinetics during the transformation of Mg1.8Fe0.
2SiO4 San Carlos olivine to its high-pressure polymorphs wadsleyite (b
eta phase) and ringwoodite (gamma phase) at 800 degrees-1200 degrees C
and nominal pressures of 16-20 GPa. In experiments in which a large (
500-600 mu m) olivine single crystal (contained in a matrix of either
fine-grained olivine or NaCl) was transformed, reaction rims of wadsle
yite/ringwoodite form on the margins of the single crystal by incohere
nt grain-boundary nucleation and interface-controlled growth. Contrary
to theoretical expectations, the growth rate of these reaction rims d
ecreases sharply as a function of time; for instance, at 1100 degrees
C and 18 GPa, growth ceases on an experimental timescale after the rim
width reaches 20-25 mu m. In order to explain this observation, we de
velop an elastic model based on the theory of a misfitting inclusion.
Comparing the results of this model with the experimental data suggest
s that elastic strain energy, which develops because of the large volu
me decrease associated with the transformation, is responsible for the
decreasing growth rates. On the other hand, experimental and theoreti
cal results suggest that elastic strain energy is relatively unimporta
nt when grains of the product phase are randomly dispersed and distinc
t reaction rims do not form; this is the case when the nucleation rate
is low, the growth rate is fast, and the reactant olivine is fine gra
ined. On a geological timescale in subducting lithosphere, where the g
rain size of the olivine is large, the growth rates of grain-boundary
nucleated reaction rims are likely to be controlled by viscoelastic re
laxation. Therefore current kinetic models of olivine metastability in
subducting slabs, which are based on simple extrapolations of experim
ental data and on the assumption that the growth rate is constant at f
ixed temperature and pressure, need to be reevaluated.