A Comparison of 14C/12C Data Quality Measured via Gas and Graphite Targets from an “off the shelf” MICADAS-AMS

Jennifer Walker1, Brett Walker1, Simon Fahrni2, Xiaomei Xu3

1University of Ottawa, Ottawa, Canada, 2Ionplus AG, Dietikon, Switzerland, 3University of California Irvine, Irvine, United States

Comparisons of achievable graphite vs. gas ¹⁴C accuracy and precision were reported >10 years ago from a prototype Mini Carbon Dating System (MICADAS) system (Wacker et al. 2010; Christl et al. 2013). Today, >30 MICADAS instruments have been produced with many improvements. This study evaluates the performance and precision of an “off the shelf” MICADAS, specifically for sample sizes <100 µgC. Modern carbon (MC) and dead carbon (DC) standards were combusted in large quantity (~4 mgC), cryogenically purified on a vacuum line, and split in size-series for gas and graphite measurements. Analyses were completed at Ionplus AG on a recently assembled MICADAS equipped with a hybrid Cs sputter source, Gas Interface System (GIS) and Elemental Analyzer (EA). Graphite was produced at UC Irvine using the small sealed-tube Zn method (Walker and Xu, 2019). Performance of the EA gas peripheral was assessed using standards weighed into pre-baked (300°C/2h) tin capsules. Background corrections were made following the methods of Santos et al. (2007).

Gas samples gave on average ~0.3 µA ¹²C+/µgC (~7.5 µA maximum for >30 µgC). Each titanium cathode allowed 15 minutes of measurement time before exhaustion, with NIST Oxalic Acid I totaling 44,000 counts. The internal error of 30-50 µgC samples measured via EA or ampules was roughly equivalent (±6.3-7.3‰). However for 6-15 µgC gas samples, the ampule cracker outperformed EA analysis with an internal error of ±7.5-9.1‰ vs. ±10.2-19.9‰, respectively. Both methods had ~0.1 µgC of DC blank. However, the EA MC blank was >4x higher than the GIS cracker (0.4 µgC) due to the addition of tin carbon during sample encapsulation. The resulting propagated error on EA and ampule cracker ~5 µgC samples was ±32.3‰ and ±18.6‰, respectively.

Graphite sample currents averaged ~0.13 µA ¹²C+/µgC, but had much higher maximum current yields ( 50 µA ¹²C+ for samples >500 µgC). Due to longer analysis times and higher carbon ionization efficiency, total counts were much higher compared to gas. NIST Oxalic Acid I averaged ~2700 ¹⁴C counts/µgC and full size samples (>500 µgC) totaled >1 x 10⁶ counts within 45 minutes, without target exhaustion. The internal error for samples >150 µgC was ±1.5‰, while samples sizes of ~40, 10, and 5 µgC, had internal errors of ±3.1‰, ±5.1‰, and ±8.1‰ respectively, thus outperforming gas analyses. However, the estimated MC and DC blanks of 0.5 and 0.7 µgC, resulted in propagated errors of ±9.8‰, ±35.2‰ and ±80.1‰. Therefore, while graphite measurements allow for improved counting statistics, the multi-step graphitization procedure introduced more extraneous carbon, thereby decreasing achieved precision.

These tests show for sample sizes <50 µgC (assuming total extraneous carbon additions are ≤1 µgC) precision is controlled by the cleanliness of sample preparation, and not MICADAS measurement precision. Our presentation will discuss in more detail the balance between accuracy and precision, the need for size-matched standards and reducing uncertainties.

Wacker et al., (2010) Radiocarbon 52, 252-262.
Christl et al., (2013) NIM-B 294, 29-38.
Walker and Xu, (2019) NIM-B 438 58-65.
Santos et al., (2007) NIM-B 259, 293–302.


Biography to come

  • 00


  • 00


  • 00


  • 00



Nov 08 - 19 2021