High-precision, mass dependent Si isotope measurements via the critical mixture double-spiking technique

We have developed a new method for measuring mass dependent Si isotope fractionation via critical mixture double-spiking. Samples need to be spiked before column chemistry to guarantee full equilibrium between the sample and double-spike (²⁹Si-³⁰Si spike). An iterative addition of the double-spike to the sample, usually 2-4 times, is needed to generate a solution very close to the critically spiked mixture. We use a double-pass cyclonic quartz spray chamber, as it gives the highest signal-to-noise ratio. In conjunction with 6 μg ml^-1 Si solution to yield intense Si isotope beams, this setup results in an ∼25 V (with 10¹¹ Ω resistor) signal on ²⁸Si⁺, while on-peak noise is less than 0.06 V. A typical sample analysis comprises 8 repeats (n = 8) of an individual sample measurement (for each repeat n = 1, 168 second analysis time) normalised to bracketing measurements of critically double-spiked NIST SRM 8546 (commonly known as NBS28). Each of these n = 8 analyses consumes about 13 μg of sample Si and yields a mean δ ^30/28Si with a precision of approximately ±0.03‰ (2 s.e., 2 × standard error of the mean). Over a 16 month period, the reproducibility of the 11 mean δ ^30/28Si values of such n = 8 analyses of the silicate reference material BHVO-2 is ±0.03‰ (2 s.d., 2 × standard deviation), which is 2 to 8 times better than the long-term reproducibility of traditional Si isotope measurement methods (∼±0.1‰, 2 s.d., δ ^30/28Si). This agreement between the long-term and short-term variability illustrates that the data sample the same population over the long and short terms, i.e., there is no scatter on the timescale of 16 months additional to what we observe over twenty hours (the typical timescale in one analytical session). Thus, for any set of n repeats, including n >8, their 2 s.e. should prove a useful metric of the reproducibility of their mean. Three international geological reference materials and a Si isotope reference material, diatomite, were characterised via the critical mixture double-spiking technique. Our results, expressed as δ ^30/28Si_NBS28, for BHVO-2 (-0.276 ± 0.011‰, 2 s.e., n = 94), BIR-1 (-0.321 ± 0.025‰, 2 s.e., n = 27), JP-1 (-0.273 ± 0.030‰, 2 s.e., n = 19) and diatomite (1.244 ± 0.025‰, 2 s.e., n = 20), are consistent with literature data, i.e., within the error range, but much more precise.