Browsing by Author "Mofokeng, Sipho Wiseman"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Open Access
    A comparison of the Nickel and the conventional geothermometers with respect to the Jagersfontein and the Matsoku kimberlite peridotite xenolits
    (1998) Mofokeng, Sipho Wiseman; Le Roux, A P; Tredoux, M
    The accuracy of the experimental (Canil, 1994; T-Canil) and the empirical (Ryan et al., 1996; T-Ryan) calibrations of the Ni geothermometer has been evaluated on two suites of geochemically and geothermobarometrically well characterised mantle xenoliths from Matsoku and Jagersfontein by comparison to the more commonly used conventional geothermometers. The two published calibrations of the Ni geothermometer are in agreement to within ±500C in the temperature range of ~900°C to 1200°C. Outside this temperature range, the two calibrations differ by between 75 and 150°C. The importance of the Ni geothermometer in diamond exploration and the studies of the mantle makes the resolution of this discrepancy very important. In addition to issues of calibration, errors in the determination of trace levels of Ni abundances in garnets may affect the accuracy of the Ni geothermometer. A 'reliable' Ni in garnet dataset was, therefore, required to minimise errors associated with Ni compositions used in temperature determination by Ni geothermometry. Thus, Ni compositions of garnets determined by PIXE, LA-ICP-MS and SIMS were compared to select the most 'reliable' dataset. Four matrix-matched secondary garnet standards were developed for the cross-checking and testing of the accuracy of data. The standards were developed by multi-method analyses, which included PIXE (using both the GeoPIXE and GUPIX software programmes for data reduction), LA-ICP-MS and solution ICP-MS. PIXE and LA-ICP-MS data were found to agree to within their 2σ errors of, respectively, 2 to 10% and 4 to 12% for a concentration range of ~15 to 112 ppm Ni. However, PIXE analyses were found to be superior in terms of smaller beam width, allowing several repeat analyses, and analysis of small and altered garnets thereby producing a relatively larger dataset.