The effect of dopants on phase equilibria: Implications for tests of Henry's Law behavior

Our current understanding of trace element behavior is in great part dependent on the experimental database that defines the trace element constraints for geologically significant systems. The accuracy of that database is dependent on a number of factors, including analytical precision, kinetics, and the relevance of the experiments to the specific system of interest. One of the cross cutting issues in this field relates to the use of dopants (added components) in trace element experiments. In designing such experiments, we must balance the improved analytical precision gained by addition of elements of interest against the concern that those added components will affect the behavior of the system. One of the most important of those effects relates to the Henry's Law limit - a concentration level at which there is an inflection point in the activity-composition behavior. Many previous investigations have run experimental tests designed to define the Henry’s Law limit in different systems - with varying results (e.g. Drake and Weill, 1974; Drake and Holloway, 1981; Nielsen et al 1992; Bindeman and Davis, 2000, and others). In all cases, the primary metric (although often not the only) for defining the upper limit of concentration at which an element has “trace” behavior was an increase or decrease in the partition coefficient as a function of composition. In our current study we chose a different perspective, tracking the effect of specific dopants on the liquidus temperature, % crystallization and phase composition. We focused on two systems, one a primitive MORB (starting composition CT14 from the West Valley of the JdF), and H87-3 an ankaramite from Hawaii. Our results indicate that the magnitude of the effects of dopant addition are correlated to the partition coefficient of the dopant element. Addition of dopants with compatible behavior increase the liquidus temperature of the phase in which they are compatible (e.g. Sr and plagioclase), and suppress phases in which they are incompatible (Sr and olivine). This outcome is consistent with the predictions one would make on the basis of our general knowledge of thermodynamics. What is most notable is the magnitude of the effect, and the consequent effect on % crystallization and phase composition at a specific temperature. For example, in the MORB composition experiments, we focused on dopant concentrations less than 4000 ppm total (0.4 elemental wt %). At a fixed temperature, Mg# of the melt increased by 5%, coexisting An contents increased 4% and crystalinity dropped from 40 to less than 20% (olivine and plagioclase) between 100 and 4000ppm dopant. These specific changes in phase composition are consistent with a suppression of the partition coefficient for trace elements with network modifying behavior - but do not represent a Henry's Law limit. Our results provide an alternative explanation for the observation of a negative correlation of D with dopant concentration (e.g. Bindeman and Davis 2000), and must be taken into account in our future experimental designs.