We used a Monte Carlo error model to optimize a 48Ca–43Ca double-spike technique for measuring Ca isotope ratios (δ44/40Ca and δ44/42Ca) by Multi-Collector Thermal Ionization Mass Spectrometry (MC-TIMS). The model considers errors for counting statistics and Johnson noise, as well as changes in collector cup efficiency (drift). For a 20V 40Ca ion-beam implemented in a three-hop, dynamic multi- collection routine, the model predicts that a wide range of 48Ca/43Ca and spike/sample ratios should yield internal precisions (2σSEM) of 0.015 – 0.020‰ for δ44/40Ca and 0.025 – 0.030‰ for δ44/42Ca. Using a Thermo Fisher MC-TIMS (Triton), we tested 48Ca/43Ca= 1.5 [43Ca/(48Ca+43Ca)= 0.40 mol/mol] and spike/sample= 0.66 [Cadsp/(Cadsp+Casmp)= 0.40 mol/mol] by repeatedly analyzing OSIL Atlantic seawater, NIST SRM 915a, NIST SRM 915b, USGS BHVO-1, and CaF2 over 4 sessions spanning 1 month. While the measured internal precision generally agreed with model predictions, external reproducibility (2σSD) was much worse than expected. For the 81 measurements made, the average external reproducibility was 0.223‰ for δ44/40Ca and 0.126‰ for δ44/42Ca. After processing raw data through the double-spike equations, nearly all fractionation-corrected ratios showed remnant fractionation patterns. Such patterns reflect deviation from ideal exponential mass-fractionation due to mixing of multiple, independently fractionating reservoirs on the filament. Additional model simulations, as well as comparison against δ44/40Ca values determined with a 43Ca–42Ca double-spike, support the concept of an “average mass rule,” which states that inaccuracies in fractionation-corrected data are greater for isotope ratios having an average mass further away from the average mass of the normalizing ratio. Until advancements are made to eliminate filament reservoir effects, 43Ca–42Ca and 46Ca–43Ca double-spikes should yield the most precise δ44/40Ca and δ44/42Ca values, respectively, when using MC-TIMS. Within the limits of the 48Ca–43Ca double-spike technique, we observed no evidence for 40Ca enrichments among the standards analyzed. Finally, we found that sample matrix effects do not influence the quality of Ca isotope measurements by MC-TIMS, and we tentatively propose that the external reproducibility determined from the repeated analysis of standards can represent the uncertainty of a single sample analysis.
Lehn G. O. and Jacobson A. D. (2015): Optimization of a 48Ca–43Ca double-spike MC-TIMS method for measuring Ca isotope ratios (δ44/40Ca and δ44/42Ca): limitations from filament reservoir mixing. Journal of Analytical Atomic Spectrometry 30: 1571-1581. DOI: 10.1039/C4JA00412D