Abstract

Natural systems provide time-controlled experiments on scales unreachable in the laboratory, and with complexities unapproachable by computer experiments. Since temperature (undercooling) cannot be directly scaled with time, in situ crystallization experiments provide information unavailable by other means. The kinetics of crystallization in simple silicate systems have been studied extensively for glass ceramic systems (e.g. Uhlmann, 1982; James, 1982), but predictive models for nucleation behavior, in particular, remain inadequate. Part of the problem in creating models of nucleation lies in the difficulty of experimentally isolating the nucleation process; crystal growth measurements are more accessible experimentally, and for this reason crystal growth is better understood (e.g. Dowty, 1980; Kirkpatrick, 1981; James, 1982; Baronnet, 1984). However, predictive models of crystal growth (e.g. Lasaga, 1982) require accurate models for the temperature and compositional dependence of both the thermodynamic and kinetic driving forces for crystallization, information that is currently limited. The CSE parameterization of the position of Ymax as a function of T, TL and T suggested by Dearnley (1983) provides a potentially useful model that requires further testing. Similarly, application of the adiabatic nucleation model of Meyer (1986) to simple systems suggests that the position of Tmax relative to TL has an approximately constant value of 0.55-0.60 (exceptions are low DSf minerals such as albite and quartz). Additionally, while experimental crystallization rate data is a necessary link between processes in natural melts and numerical models, experimental growth rate data is confined to simple systems and conditions of moderately large undercooling, while growth rate estimates in complex natural systems suggest that under most conditions, undercoolings remain very small. Quantitative textural studies of dikes suggest a pronounced dependence of growth rate on total crystallization time (e.g. Ikeda, 1977), which is in turn related to effective undercooling, a function of both total dike width and distance from the dike margin. By analogy, minimum crystallization rates may be estimated from programmed cooling experiments if a temperature interval of crystallization is assumed. Growth rates determined in this way show the same relationship between growth rate and total crystallization time seen in the dike studies, reflecting a systematic relationship between growth rate and cooling rate (effective undercooling) for plagioclase in basaltic systems. While the data are not available for other mineral systems, there is no reason to believe that similar patterns would not exist for other silicate minerals (i.e., the similarity in dendritic olivine growth rates estimated by Fowler et al.,1989, using fractal analysis to the growth rates of plagioclase under conditions of rapid cooling estimated from programmed cooling rate experiments). Analysis of measured crystal size frequencies provides a means of extracting effective (bulk) kinetic information from geologic systems that can help to constrain the dynamic feedback between physical models of magmatic systems and rates of chemical change.

You do not currently have access to this article.