EarthRef.org Reference Database (ERR) -- Kennedy et al. 1993

The concentrations of 32 elements (major, minor, REE, Ba, Hf, Sc, V, Co, Ni, Sr, Y, Zr, Nb, Ge, Th and U) were measured using an ion microprobe in olivine and orthopyroxene, and in coexisting glass in the run products produced in isothermal and dynamic crystallization experiments on chemical compositions that produce porphyritic olivine, radial pyroxene and barred olivine textures. Cooling rates in the dynamic crystallization experiments ranged from 1-degrees-C/h to 2191-degrees-C/h. The mineral/melt partition coefficients (D) calculated from the measured concentrations for both olivine and orthopyroxene showed very little change between equilibrium experiments and dynamic experiments with cooling rates of up to 100-degrees-C/h. The data appear to define reliable equilibrium D values. Olivine D's range from 2 x 10(-5) for U to approximately 10 for Ni, and orthopyroxene D's range from 2 x 10(-5) for U to approximately 8 for Mg. There are regular relationships between the ionic radius, the valence of the trace element and the partition coefficients in olivine and orthopyroxene in all experiments. While there are some differences in the D values between olivine and orthopyroxene, they follow very similar trends. For experiments at high cooling rates there is an increase in the measured concentrations for incompatible trace elements in the crystals which yields apparent partition coefficients greater by up to 3 orders of magnitude from the equilibrium values obtained for isothermal equilibrium and slow cooling experiments. In contrast, compatible trace element D values are only slightly sensitive to cooling rate, typically varying by less than a factor of 2. SEM studies showed that melt inclusions were common in orthopyroxenes. It is shown that the increase in apparent D values for pyroxene observed with high cooling rates can be quantitatively explained by trapping of melt inclusions in crystals with equilibrium D values. Incorporation of melt inclusions may also quantitatively explain the increases in apparent D values in olivine at high cooling rates. However, no inclusions of melt were identified in olivine, even at the fastest cooling rate. If the high apparent D values obtained for olivine at high cooling rates were due to incorporation of trace elements in the lattice and not due to entrapment of melt, it would require incorporation of elements independent of valence and ionic radius.