Matrix effect and partitioning of boron isotopes between immiscible Si-rich and B-rich liquids in the Si-Al-B-Ca-Na-O system: A SIMS study of glasses quenched from centrifuge experiments

The results of a study of isotopic fractionation of boron between silica-rich (LS) and borate-rich (LB) immiscible liquids in the system Si-B-Al-Ca-Na are reported. Partitioning experiments were performed on glasses produced by quenching immiscible coexisting Si-rich and B-rich borosilicate melt pairs that had been equilibrated at high temperature and separated with the aid of a high temperature centrifuge furnace [Veksler, I.V., Dorfman, A.M., Dingwell, D.B., Zotov, N., 2002. Element partitioning between immiscible borosilicate liquids: a high-temperature centrifuge study. Geochim. Cosmochim. Acta 66, 2603-2614]. Here B/Si and ¹¹B/¹⁰B ratios have been determined using secondary ion mass spectrometry (SIMS). The determinations encountered unexpectedly strong matrix effect (up to 12‰) due to contrasting chemical compositions of the immiscible melts. Employment of - 60 V high-voltage offset (not used during routine SIMS B isotope measurements) effectively eliminated the initially observed matrix effect. With this improved SIMS technique, we obtained the result that the coexisting quenched immiscible liquids have the same ¹¹B/ ¹⁰B ratio within ± 1.2‰ (2σ) analytical uncertainty. Thus little or no significant two-liquid fractionation of ¹⁰B and ¹¹B isotopes was observed in the temperature range between 950 and 1350 °C. This is despite the observation that significantly different proportions of trigonal vs. tetrahedrally coordinated B in these immiscible liquids have been determined by Raman spectroscopic analyses. The essential observations (i.e., a strong matrix effect and a lack of significant B isotope fractionation between the glasses) were both supported by independent thermal ionization mass spectrometry (TIMS) analyses of one selected centrifuged pair. Those glasses showed no difference in B isotopic composition within the quoted 2σ analytical error of ± 1.0‰.