PII S0016-7037(01)00778-5 Analysis of individual fluid inclusions using Synchrotron X-Ray Fluorescence microprobe: progress toward calibration for trace elements B´ EN´ EDICTE M´ ENEZ, 1,2, *PASCAL PHILIPPOT, 2 MICHELLE BONNIN-MOSBAH, 1 ALEXANDRE SIMIONOVICI, 3 and FRAN¸ COIS GIBERT 4 1 Laboratoire Pierre Su ¨e, CEA-CNRS, Batiment 637, C. E. Saclay, 91191 Gif Sur Yvette cedex, France 2 Laboratoire de Ge ´osciences Marines, CNRS-FRE 2316, Institut de Physique du Globe, case 89, T26-00 E3, 4 place Jussieu, 75252 Paris cedex 05, France 3 European Synchrotron Radiation Facility, FID group, ID22, BP 220, 38043 Grenoble cedex, France 4 CNRS, UMR 6524, Universite ´ Blaise Pascal, 63038 Clermont Ferrand, France (Received February 15, 2000; accepted in revised form July 17, 2001) Abstract—A critical problem for conducting quantitative analysis of individual fluid inclusions using Synchrotron X-Ray Fluorescence (SXRF) technique relates to the standardization and the calibration of the X-ray spectra. In this study, different approaches have been tested for calibration purposes: (a) the use of chlorine when Cl content can be estimated either from melting point depressions of undersaturated fluid inclusions or from saturation limits for halite-bearing fluid inclusions, (b) the use of calcium from synthetic fluid inclusions of known CaCl 2 content as an external standard. SXRF analysis was performed on individual fluid inclusions from the Chivor and Guali emerald deposits, Columbia. These well-known samples contain a single fluid inclusion population for which detailed crush-leach analyses are available, thus providing a relevant compositional reference frame. Concentration estimates were also compared to Particle Induced X-ray Emission (PIXE) analysis carried out independently on the same fluid inclusions. Results of the calibration tests indicate that major (Cl, K, Ca, Fe, Mn) and trace element (Cu, Zn, As, Br, Rb, Sr, Ba, Pb) concentration estimates can be performed without precise knowledge of the analytical volume and the inclusion’s 3D geometry. Although the standard deviation of the SXRF results can be relatively high depending on the calibration mode used, mean concentration estimates for most elements are in good agreement with PIXE and crush-leach analysis. Elemental distributions within single fluid inclusions were also established. Associated correlation diagrams argue for the homogeneous distribution of most elements in the fluid inclusion. In contrast, Br shows a bimodal distribution interpreted to reflect a significant enrichment of the vapor portion of the inclusion fluid. Copyright © 2002 Elsevier Science Ltd 1. INTRODUCTION Fluid inclusions represent the only direct samples of ancient fluids in many crustal rocks. They are considered to be repre- sentative of the fluid present during either the growth of min- erals or the later healing of fluid-filled cracks. Accordingly, their composition provides crucial information for studies of fluid–rock interactions. Many natural minerals contain several generations of fluid inclusions, each one being representative of a distinct hydrothermal event. Therefore, there is general rec- ognition of the uncertainties inherent in bulk analytical ap- proaches such as crush-leach analysis, which result in homog- enizing several fluid populations. The potential of Synchrotron X-Ray Fluorescence (SXRF) microprobe for conducting quantitative analysis of single fluid inclusions has recently been recognized (Frantz et al., 1988; Rankin et al., 1992; Vanko et al., 1993, 2001; Bodnar et al., 1995; Bu ¨hn and Rankin, 1999; Cline and Vanko, 1995; Ma- vrogenes et al., 1995; Philippot et al., 1995, 1998, 2000, 2001; Me ´nez et al., 1998a,b; Me ´nez, 1999; Vanko and Mavrogenes, 1998). The advantages of this technique remain its non destruc- tive character, a high spatial resolution, and the in situ multi- element analytical capability. Moreover, the recent develop- ment of third-generation synchrotron radiation sources has opened up the prospect of a highly sensitive instrument with potential sub-ppm detection limits (e.g., Chevallier et al., 1996). Previous studies have demonstrated the efficiency of the SXRF technique for the detection of ionic species in synthetic and natural fluid inclusions (see Vanko and Mavrogenes, 1998 for a recent review). However, the computational procedures used to estimate the elemental concentrations present two ma- jor difficulties and, as a consequence, alternative approaches are needed to improve on these estimates. First, corrections for X-ray self-absorption by the inclusion fluid and the host min- eral are required. Previous research has relied on optical tech- niques for estimating the pathlengths traversed by the incident beam and the fluorescent radiation through host mineral and fluid. Philippot et al. (1998) showed that the K/K ratio of an element present in solution is directly related to the thickness of material traversed by the radiation, thus providing improved pathlength estimates commonly better than 1 m Philippot et al., 1998, 2001). A similar approach can be found in Ryan et al. (1991, 1993) for identical problems in Particle-Induced X-ray Emission (PIXE) analysis. The other limiting factor associated with SXRF analysis concerns the standardization of the measurements. Calibration of the collected element X-ray intensities remains a major problem. As a consequence, most attempts display qualitative results (Bodnar et al., 1995) or semi-quantitative analysis yield- ing elemental ratios (Rankin et al., 1992; Vanko et al., 1993; Philippot et al., 1995; Vanko and Mavrogenes, 1998). For * Author to whom correspondence should be addressed (menez@ipgp.jussieu.fr). Pergamon Geochimica et Cosmochimica Acta, Vol. 66, No. 4, pp. 561–576, 2002 Copyright © 2002 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/-1898 $22.00 + .00 561