High-precision stable isotope analysis of <5 μg CaCO3 samples by continuous-flow mass spectrometry

Rapid Commun Mass Spectrom. 2020 Oct 15;34(19):e8878. doi: 10.1002/rcm.8878.

Abstract

Rationale: Oxygen (δ18 O) and carbon (δ13 C) isotope analysis of foraminifera and other CaCO3 samples has been a key technique for paleoceanographical and paleoclimatological research for more than 60 years. There is ongoing demand for the analysis of ever smaller CaCO3 samples, driven, for example, by the desire to analyse single specimen planktic foraminifera, or small samples of tooth enamel.

Methods: We present a continuous-flow mass spectrometric technique that uses cryo-focusing of sample CO2 to analyse CaCO3 samples in a weight range between 10 and 3 μg. These are considerably lower sample weights than achievable on most currently available standard instrumentation. The technique is automated, so that sample throughput lies at >60 samples per day. The method involves an on-line vial-flushing routine designed to remove machine drift due to blank CO2 build-up in the sample vials.

Results: In a series of experiments the effect of blank CO2 build-up is quantified, and outgassing from the chlorobutyl septa identified as the source. An improved flushing routine together with the use of a cryo-focusing step in the analysis is demonstrated to provide the analytical stability and sensitivity to analyse CaCO3 samples in a weight range between 10 and 3 μg at ≤0.1‰ precision (1σ) for both δ18 O and δ13 C values. The technique yields similarly precise results for the analysis of the structural carbonate fraction of small tooth enamel samples.

Conclusions: This study demonstrates that high-precision oxygen and carbon isotope analysis is possible on CaCO3 samples smaller than 5 μg by use of a continuous-flow isotope technique. Of key importance are (1) the application of a cold trap that drastically reduces sample gas loss, and (2) a modified flushing regime that eliminates increasing background CO2 build-up in sample vials during longer automated sample runs.