REASSESSMENT OF UNCERTAINTY EXPANSION BY LINEAR ADDITION OF LONG-TERM COMPONENTS FROM TOP-DOWN INFORMATION

G Salazar,S Szidat

doi:10.1017/rdc.2021.96

Abstract

ABSTRACTSince radiocarbon accelerator mass spectrometry (14C AMS) is considered a high-precision technique, reassessment of the measurement uncertainty has been a topic of interest. Scientists from analytical and metrological fields have developed the top-down and bottom-up measurement of uncertainty approaches. The 14C quoted error should approximate the uncertainty of long-term repetitions of the top-down approach in order to be realistic. The novelty of this paper is that the uncertainty of both approaches were approximated to each other. Furthermore, we apportioned the graphitization, instrumentation, and bias components in order to additively expand the quoted error. Our results are comparable to error multipliers and to long-term repeatability studies reported by other laboratories. Our laboratory was established in late 2012 with N2 as stripper gas and 7 years later, we changed to helium stripper. Thus, we were able to compare both gases, and demonstrate that helium is a better stripper gas. In absolute F14C units, the ranges of graphitization+bias combined uncertainties were (0.7 to 4.1) × 10–3 for N2 and (0.7–3.0) × 10–3 for He depending on the standard 14C content. The error multiplier for He defined as the expanded uncertainty over quoted error, in average, was 1.7; while without the bias, the multiplier was 1.3.

Highlights

In radiocarbon accelerator mass spectrometry (14C AMS), it has been observed that, most of the time, the quoted error for single measurements is an underestimation when comparing with replications of the same 14C sample (Boaretto et al 2002; Scott et al 2007)
Taking advantage of the linear trend of the plots of uRw and ubias versus F14C; the instrumentation, graphitization and bias components are added in quadrature to obtain an expanded uncertainty for the 14C range from blank to Oxa2
This expanded uncertainty is compared with long-term repeatability studies from other laboratories

Summary

INTRODUCTION

In radiocarbon accelerator mass spectrometry (14C AMS), it has been observed that, most of the time, the quoted error for single measurements is an underestimation when comparing with replications of the same 14C sample (Boaretto et al 2002; Scott et al 2007). A more accurate report should include random and systematic effects as recently proposed by a unified theory of measurement errors and uncertainties (Huang 2018). In this way, the bottom-up and the top-down approaches can be coherent. Some long-term components have been taken into account like long-term repeatability and bias for only modern 14C samples (Miller et al 2013; Turnbull et al 2015), blank long-term uncertainty, error multipliers and the error propagation for graphitization and chemical treatment (Scott et al 2007; Schuur et al 2016). Specific chemical treatment and field sampling levels were not considered

METHODS

Findings

CONCLUSIONS