C-14 AMS data quality assessment: A key practice at the Rafter Radiocarbon Laboratory

Albert Zondervan1, Jocelyn Turnbull1, Cathy Ginnane1, Margaret Norris1, Rafter Radiocarbon Team1

1GNS Science, , New Zealand

The Rafter Radiocarbon Laboratory, in operation since 1951, transitioned from decay counting to AMS in the second half of the 1980s. In the decades since, many improvements were made to sample preparation and carbon-isotope measurement techniques. The most significant of these were an upgrade of the Cs-sputter source (year = 1996) and the charging system of the tandem accelerator (2000), a shift to fast isotope-beam switching and new data acquisition system (2006), complete replacement of the AMS system (2010), introduction of elemental analysis for organic samples (2010), replacement of the graphitisation system (2012), CO₂ extraction from air for high-precision ∆¹⁴C analysis (2012), development of high throughput capacity for modern tree ring samples via accelerated solvent extraction (2017), selective combustion of organics through ramped pyrolysis (2020), and establishment of pretreatment protocols for tiny macrofossil samples (2020). Development projects currently underway are adding a dedicated graphitisation system to facilitate more reliable preparation for < 0.3 mg C samples and an overhaul of vacuum pumping on all sample preparation lines.

Following a review in 2015 of mission-vision-values, the laboratory’s main purpose is to support New Zealand geosciences research and international collaborations that benefit New Zealand society. In practice, this means that our focus is less on profitability (competition on price, fast turnaround analyses) and more on value with an emphasis on quality and promotion of laboratory involvement in research collaborations contributing to appropriate sample selection and laboratory methodology, suitable calibration, and interpretation of results.

Generally, our approach to data quality is (i) that the reported error on each C-14 analysis should capture all sources of uncertainty, not just the Poisson counting error, and (ii) that these must be quantified from repeated analyses in order to minimize the risk of bias, i.e. loss of accuracy. We tune XCAMS, our present AMS system, on a graphitisation blank and IAEA-C6 sucrose. Oxalic-acid-I remains our primary standard and every batch measurement on ≤ 40 cathodes contains at least 6 of those. We use a few key radiocarbon inter-comparison materials repeatedly, to ascertain deviations from and non-Poisson scatter around their consensus value. Their results (mean, variance) over a period of 6—12 months allow us to determine the residual, i.e. non-Poisson uncertainty, specific to each sample type and size.

IAEA-C1 carbonate is used to blank correct solid carbonate and dissolved inorganic carbon samples. Wood from a Kauri tree from a well-preserved fossil forest (MIS 7) is used for blank-correcting results on all organic samples according to preparation. No correction is applied for the machine blank as this is incorporated in the process blank corrections. Contamination of the surface of cathodes prepared from materials with ¹⁴C/C < 10^−14, presumably by absorption of ¹⁴C-modern CO₂, is easily recognisable during data reduction. Blanks are characterised separately for samples < 0.3 mg C.

We will present results for all control materials, to highlight dependencies on date of preparation, preparation method, and sample size.


Albert Zondervan is a physicist at GNS Science, New Zealand. As senior scientist he is responsible for all aspects regarding XCAMS, the AMS system for 14C/10Be/26Al measurements. His main accomplishments are in applying AMS in the geosciences, such as surface exposure-age dating and high-precision radiocarbon analysis of atmospheric CO2.

  • 00


  • 00


  • 00


  • 00



Nov 17 2021


1:30 pm - 1:50 pm