SIMS Analyses for Lithium Concentrations and Isotope Ratios
Secondary ion mass spectrometry (SIMS) is a sensitive technique for the analysis of lithium. Positive ions of this element are easily formed during sputtering with a primary beam of oxygen (lithium has a low ionization potential) and this results in measured useful yields (Li ions detected/Li atoms sputtered) as high as 2 to 4% on the Cameca ims 3f and 6f SIMS at ASU (Hervig et al., 2006). To illustrate the high sensitivity, consider a grain of olivine (Fo90) containing 1 µg/g Li (~3 ppm atomic) Li. This concentration can be represented as ~3 x 1017 atoms Li/cm3 of olivine. If we sputter a cylindrical crater 20 µm in diameter but just 1 µm deep, we will consume ~300 µm3 or 3 x 10-10 cm3 of olivine. Using the atomic density of Li given above, this represents a volume containing ~100 million Li atoms. With a useful yield of 2-4%, we could collect 2-4 million counts of Li during this analysis. This simple calculation is too liberal because the ASU SIMS instruments are single collector mass spectrometers, and would miss counting some 6Li+ while counting 7Li+ (and any other species being monitored). In addition, some signal would be lost because we would partially close the entrance slit to the mass spectrometer to ensure obtaining flat-topped peaks. However, the fact remains that signals for Li will be very high while consuming minimal amounts of a natural sample or experimental run product.
An example of a calibration for lithium is shown in Figure 1. Bulk analyzed rhyolitic and basaltic glasses and two clinopyroxenes are shown. The 7Li+/30Si+ ion ratio was measured using ions with 75±20 eV excess kinetic energy (conventional energy filtering). A least squares regression is shown (forced through the origin). As discussed by Decitre et al. (2002), it does not appear that there is a strong effect of changing sample chemistry on the Li ion yield.
Possible pitfalls with Li analysis by SIMS
Earlier work has documented that Li isotope ratios can also be determined by SIMS on olivine, pyroxene, amphibole, and basaltic glass, and that any matrix effects for these analyses were not detectable (Decitre et al., 2002). However, we are aware of some issues that affect the quality of Li analyses and we describe them below.
The first item we worry about when attacking an analytical problem by SIMS is if there are other species at the same nominal mass/charge ratio as the isotopes of Li. Potential interfering species include 24Mg4+ and 12C2+ on 6Li+, and 28Si4+ and 6LiH+ on 7Li+. A high-resolution mass spectrum obtained on a water-saturated rhyolitic glass quenched from 500MPa (from Hervig et al., 2002, AGU abstract) is shown on Figure 2 below. We observe no sign of either 28Si4+ (simply because it is so much lighter in mass than 7Li) or 6LiH+ near 7Li+. Similarly, no interfering peaks were observed near 6Li+. We have not observed the 6LiH+ species in any matrix we have studied (as of the end of 2006).
Primary Beam Current
While studying the rhyolitic glass shown in Fig. 2, we noted a dramatic effect of primary beam current (16O- primary beam accelerated to 12.5kV with an impact energy at the sample of 17 keV) on the Li+ ion signal. This is shown on Fig. 3, where the beam current was increased from 0.1 to nearly 2 nA on the hydrous glass and the natural (~0.2 wt.% H2O) starting material. Note that the Li+/Si+ ratio decreased by a factor of ~5 as the primary current increased. The starting material is not strongly affected by the changing beam current, indicating that the high water content of this sample (>10 wt.%) influences the stability of the Li ion signal. At the same range of conditions, the Li isotope ratios were measured on this sample. Increasing the primary current changed the measured 7Li+/6Li+ ratio by ~ 65 ‰ while, again, the nearly dry, glassy starting material is not strongly affected by changes in primary beam current.
We interpret the measurements described above to indicate that Li in extremely H2O-rich rhyolitic glass is mobilized by the primary beam, possibly via a charge-driven diffusion process. This mobilization affects the light isotope more than the heavy isotope. This effect is not observed in the starting material containing low water, and the effect is reduced when the primary beam current is kept at ≤0.3 nA. Other hydrous samples we have explored do not show such a strong effect (but have much lower water than the above rhyolite). Lithium-poor, natural glasses (e.g., Hawaiian basalts) and minerals (e.g., mantle olivine) require high primary currents to obtain signals intense enough to obtain isotope ratio measurements. Preliminary tests of olivine do not show any relation between primary current and measured isotope ratio, but we have observed an effect on glasses (Fig. 4). As the figure indicates, reproducible analyses of glasses can be obtained using primary currents greater than a threshold value (about 10 nA in this example). How this threshold varies with current density and sample composition remains to be explored.
Analyses of some experimental run products indicate a significant gain in Li compared to the starting material. For example, we have observed a factor of 10 to 100 gain in the Li+/Si+ ion ratio in basalts held in Pt capsules from internally heated pressure vessels (IHPV) compared to the starting material. We suspect either that the water added to the capsule was Li-bearing or that the extrusion process contaminated the tubing at the manufacturing site with a material that is not removed during acid cleaning of the capsule. Other glassy run products from experiments using “nano-pure” water and capsules treated with a degreasing step show low lithium contents similar to their initial concentrations. Vacuum grease (sometimes used during the preparation of epoxy mounts containing multiple grains of standards and unknowns) can contain enormous amounts of Li (added as a "stiffener"), and can result in serious (and even untreatable) contamination of the sample with lithium. This material should be avoided in the sample preparation process if the researcher wishes to measure Li in their sample.
Figure 1. Calibration of the SIMS for lithium. The measured 7Li+/28Si+ ion ratio is corrected for silica content and compared with bulk analyses of lithium. The calibration curve intersects clinopyroxenes and basaltic and rhyolitic glasses, indicating that matrix effects for Li content are small (see also Decitre et al. (2002).
Figure 2. High resolution mass spectrum taken on a hydrated rhyolitic glass. Despite the high water content (>10 wt.% H2O expected in this sample) there is no sign of a 6LiH+ molecular ion at the mass/charge predicted (0.007 amu heavier than 7Li+). A signal from 28Si4+ would be observed 0.02 amu lighter than 7Li+ (if present). The starting glass contained ~3400 ppmw Li, giving (at these conditions) about 10 counts/s per ppm Li.
Figure 3. Effect of increasing primary beam current on 7Li+/30Si+ ion signal measured on macusani rhyolite quenched from 500 MPa (same sample as in Figure 2). The silicon ion signal increased linearly with the primary current, indicating that the Li+ signal decreased with increasing primary current.
Figure 4. Effect of increasing primary beam current on instrumental mass fractionation of lithium isotopes measured on "dry" (<0.2 wt.% H2O) basaltic glass. Instrumental mass fractionation is the deviation (in ‰) of the measured 7Li/6Li ion ratio from the L-SVEC standard ratio (7Li/6Li = 12.039). Error bars represent 2 standard errors of the mean ratio measured in each analysis (analyses represent data from two sessions- one on the 3f and one on the 6f SIMS). The actual Li isotopic composition of this basalt is ~+4‰.
Decitre, S., Deloule, E., Reisberg, L., James, R.H., Agrinier, P., and Mevel,
C. (2002) Behavior of Li and its isotopes during serpentinization of oceanic
peridotites. Geochemistry, Geophysics, Geosystems, 3(1), 10.1029/2001GC000178.
Hervig, R.L., Mazdab, F.K., Williams, P., Guan, Y., Huss, G.R., and Leshin, L.A. (2006) Useful ion yields for Cameca IMS 3f and 6f SIMS: Limits on quantitative analysis. Chemical Geology, 227, 83-99.