Compound-Specific Radiocarbon Analysis (CSRA) by Elemental Analyzer-Accelerator Mass Spectrometry (EA-AMS): Precision and Limitations.

We examine instrumental and methodological capabilities for microscale (10-50 µgC) radiocarbon analysis of individual compounds in the context of paleoclimate and paleoceanography applications for which relatively high precision measurements are required. An extensive suite of data for 14C-free and modern reference materials processed using different methods and acquired using an elemental analyzer-accelerator mass spectrometry (EA-AMS) instrumental setup at ETH-Zurich was compiled to assess the reproducibility of specific isolation procedures. In order to determine the precision, accuracy and reproducibility of measurements on processed compounds, we explore results of both reference materials and of three classes of compounds (fatty acids, alkenones and amino acids) extracted from sediment samples. We utilize a MATLAB code developed to systematically evaluate constant contamination model parameters, which in turn can be applied to measurements of unknown process samples. This approach is computationally reliable and can be used for any blank assessment of small-size radiocarbon samples. Our results show that a conservative lower estimate of the sample sizes required to produce relatively high-precision 14C data (i.e., with acceptable errors of <5% on final 14C ages) and high reproducibility in "old" samples (i.e., F14C ≈ 0.1) using current isolation methods are 50 µgC and 30 µgC for alkenones and fatty acids, respectively. Moreover, when the F14C is > 0.5, a precision of 2% can be achieved for alkenone and fatty acid samples containing ≥15 µgC and10 µgC, respectively.