MC-ICPMS isotope ratio measurements using an ultra-low flow sample introduction system

Abstract

This study characterises for the first time isotope ratio measurements by multi-collector ICPMS when performed at liquid flow rates as low as 10–15 µL min−1. An evolution of the torch integrated sample introduction system (TISIS) was employed, which combined an OpalMist nebulizer and a heated single pass spray chamber, allowing the transport of almost 100% of the solvent. The different factors potentially influencing the isotope ratio measured values and the associated sources of uncertainty (sensitivity and interferences, mass discrimination effects, repeatability of isotope ratio measurement, and rinsing time and memory effects) were investigated systematically under varying experimental conditions. Results showed that the liquid flow rate has a large impact on mass discrimination effects, making the control of this variable critical. From 5 to 30 µL min−1, when using the exponential model, the mass discrimination per mass unit changed from −1.6 to −2.1 and from −1.1 to −1.9, for 88Sr+/86Sr+ and 208Pb+/206Pb+ ratios respectively. Moreover, extrapolations from these results lead to the conclusion that a syringe pump may be required instead of a free aspiration regime to control the liquid flow rate and eliminate the possibility of undesired variations of isotope ratio results (typically, 0.05‰ error for 1–2% fluctuations at 10 µL min−1). The validity of the exponential model also depends on the experimental conditions selected. When working at 15 µL min−1 and heating the chamber walls at 60–80 °C, the performance was as good as it could be with a MicroMist/water cooled cinnabar combination operated at 200 µL min−1 (thus, the efficiency was 12 times better for the TISIS). Both systems were compared for the measurement of the Sr isotopic signature in honey samples. Since the TISIS allowed for a preconcentration of samples by a factor 5, the combined uncertainty on results could be improved by 1.6 to 4.4. The main drawback was, however, the longer rinsing time required to reach a stable background signal (10–12 minutes rather than 4).