Assessment of the uncertainties and limitations of quantitative elemental analysis of individual fluid inclusions using synchrotron X-ray fluorescence (SXRF)

Abstract

Synchrotron X-ray Fluorescence (SXRF) analysis is a nondestructive analytical technique that provides compositional information from single fluid inclusions. A protocol for conducting quantitative analyses of metal concentrations in individual fluid inclusions has been developed. This has led to an understanding of the accuracy, precision, and detection limits of this technique, as well as the optimal shapes, sizes, and geometries required for reliable fluid inclusion analysis. Aqueous fluid inclusions containing known concentrations of SrCl 2 were synthesized for the development and the standardization of this technique. Strontium chloride was selected because it is highly soluble, its freezing-point depression is well known (allowing us to confirm the inclusion composition using microthermometric analyses), and the energetic Sr X-rays are only mildly attenuated by quartz. To confirm the composition of the synthetic standards, solutions were measured before and after each hydrothermal run using Atomic Absorption Spectroscopy (AAS), and the freezing-point depression for each fluid inclusion was measured. SXRF analyses were performed on beam line X26A of the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory using an 8 × 12 μm white X-ray beam. The analytical volume was calculated based on known beam dimensions and fluid inclusion geometry determined using a modified spindle stage. Elemental concentrations were determined by ratioing the Sr counts from an inclusion to the counts obtained from capillaries of known diameter containing similar solutions. Numerous inclusions from five samples, each with a different Sr concentration, were analyzed. Within a single population the mean is very close to the known concentration, but the precision is poor, with standard deviations (lσ) from 10–39% of the mean. Errors in determining the inclusion geometry are the main contributor to the poor precision. The poor precision requires that numerous inclusions within one population be analyzed and averaged to accurately estimate the metal concentration for that population. Selection of flat-lying, equant, regularly-shaped inclusions for analysis minimizes errors resulting from inclusion geometry if quantitative results are sought. The detection limit for Sr in synthetic fluid inclusions (typically 4–15 μm thick, and 5–100 pm below the upper polished surface) is approximately 2,000 ppm Sr.