Notes on Some Crystals of San Carlos Olivine and EPMA Standards

Natural Fo-rich olivine, of mantle xenolith origin, is a standard in many electron microprobe labs. Many labs use the San Carlos Fo90.1 standard developed by Gene Jarosewich and co-workers (1980) at the Smithsonian: USNM 111312/444. I recently encountered non-USNM-distributed crystals of San Carlos olivine, being utilized in scientific research (not for EPMA standards). EPMA of these showed slight but consistent differences: one set was more Fe-rich (Fo89.2) and the other, more Mg-rich (F91.1). This led to some further investigation of USNM San Carlos and the range of compositions of grains potentially used by EPMA labs for standards. Larger crystals of non-USNM San Carlos olivine are available commercially. Three 1-2 mm crystals from one source were analyzed by EPMA for Si, Mg and Fe, and a consistent value of Fo89.2 (±0.2) was found. Six 1 cm-size crystals from another source were analyzed and a range of compositions from Fo90.5 to Fo91.4 was found (average Fo91.1 ±0.3). 30-40 points per crystal were measured in all these cases. What then is the range of variability possible in the USNM San Carlos olivine standard? Amelia Logan of the Smithsonian supplied 2 small vials of the USNM111312/44 material. Jarosewich et al (1980) calculated "Boyd homogeneity indices" for USNM standards, with 100 measurements on 10 grains of each standard, and values <3 were OK. I report 71 measurements on 23 small (200-300um) grains of USNM 111312/44 in 1 vial, which show a bit wider range of hetereogeneity in Si and Mg than originally found in the 1980 paper: Si 2.32 (vs 0.81), Mg 3.05 (vs 1.00), and Fe 1.2 (vs 0.9). These "Boyd" numbers are less easy to comprehend than a simple k-ratio criteria: take the average x-ray count of a large number of points (which will represent the wet chemical analysis), and then take the lowest and the highest counts, to represent the worst case scenario. Here the results were: Si ranged from 0.947 to 1.017 (5% max difference), Mg 0.969 to 1.043 (4% max) and Fe 0.946 to 1.055 (6% max). However, 82% fell within the Fo90.1 (±0.2) window. Conclusions: (1) Crystals of "San Carlos olivine" available from mineral dealers cannot be assumed to be of the same composition as USNM 111312/444. (2) There is a small but not unreasonable probability that EPMA users who assume that any ONE grain of USNM Carlos olivine is EXACTLY the published composition could be making an error of ~5% in the measured K-ratio for Mg, Si or Fe. EPMAers need to acquire "a reasonably large number of counts on a reasonably large number of grains" (Jarosewich et al, 1980). (3) It is beneficial operating procedure for a lab to run several standards for an element and then compare the results for consistency. (4) Periodic use of applications such as "Evaluate" (Probe for EPMA software) provides one way to cross-check all standards and determine whether some grains of well known standards may not be exactly the published values and should have their compositions modified. QC Proposal: That a probe mount of at least 25 grains of USNM 111312/444 San Carlos olivine be made available to any EPMA lab for a short period of time, to run as a primary standard, to compare one's own few grains with and decide whether or not those grains' composition are of the exact composition being used, or whether small modifications might be justified.