THE FE2-PYROXENE - A CRITICAL-ASSESSMENT OF ITS POTENTIAL AS A GEOSPEEDOMETER(, MG DISTRIBUTION IN ORTHO)

Extremely precise and accurate Fe2+,Mg distributions are required in o rder to derive meaningful cooling rates from the orthopyroxene geospee dometer. A discussion of X-ray data collection and refinement strategi es, using orthopyroxene from the Acapulcoite meteorite ALHA81261 (X-Fs = 0.11) as an example, discloses that the standard practice of one si ngle X-ray refinement can be inadequate. Therefore, a novel bivariate analysis is suggested that is based on both stepwise truncation of low order data and rejection of outliers (LOT analysis). Thereby, extinct ion and chemical bonding effects are reduced. Correlations between sit e occupancies and thermal vibration parameters are circumvented when a B(M2)/B(M1) ratio is imposed that provides invariance of the refined site occupancies with respect to the choice of data subsets. For calcu lating cooling rates, the Mueller (1967, 1969) rate equation is used, which requires that the Fe2+,Mg equilibrium distribution coefficient, K-D, and the rate constant, k(dis), for the (microscopic) disordering reaction are known as functions of both temperature, T, and compositio n, (X) over bar(Fs). The coefficients of the lnK(D) function were calc ulated from our own and published K-D values. Their analysis revealed internal inconsistencies related to the origin of the crystals (natura l or synthetic) and to the data collection and site refinement strateg ies. From the total set of lnK(D) values we have chosen three subsets, each of which appeared to be rather consistent internally. The regres sion analyses of these subsets resulted in quite different parameters. Rate constants k(dis) were (re)calculated from our own and published Fe2+,Mg site occupancies determined in isothermal kinetic experiments. Coefficients for the dependence of k(dis) on T and (X) over bar(Fs), assuming (1) a continuous, and (2) a stepwise behaviour of lnk(dis)((X ) over bar(Fs))(T) were derived. The results agree for 0 < (X) over ba r(Fs) < 0.55, but disagree for 0.55 < (X) over bar(Fs) < 1. Using a re gression line individually determined from lnK(D) values of Johnstown and Acapulcoite meteoritic bronzites, we have calculated the cooling r ate of an Acapulcoite crystal for which a LOT analysis had provided ex tremely precise sire occupancies (sigma = 0.0007). This calculation re sulted in an apparent equilibrium temperature T-ae = 388(17)degrees C and a cooling rate dT/dt = -0.2 degrees C/y at T-ae. The lnK(D) calibr ation of Ganguly & Domeneghetti (1996) differs significantly from ours . Using their calibration, Zema et al. (1996) found for an Acapulcoite crystal T-ae = 467 degrees C, which corresponds to a cooling rate tha t is faster than ours by a factor of 90. Although site occupancies der ived by the LOT method can be considered more precise and accurate tha n those obtained by standard practice, the inherent uncertainties in l nK(D)(T, (X) over bar(Fs)) and lnk(dis)(T, (X) over bar(Fs)) indicate that the orthopyroxene geospeedometer should be checked, whenever poss ible, against cooling rates determined by other methods, such as radio metric age determinations, before it is applied to unknown cooling his tories.