International Journal of Mass Spectrometry 237 (2004) 107–118 Toward epsilon levels of measurement precision on 234 U/ 238 U by using MC-ICPMS M.B. Andersen ∗ , C.H. Stirling, E.-K. Potter, A.N. Halliday Institute for Isotope Geology and Mineral Resources, Department of Earth Sciences, ETH-Zentrum, NO C62.2, Sonneggstrasse 5, 8092 Zurich, Switzerland Received 3 June 2004; accepted 12 July 2004 Available online 20 August 2004 Abstract Variations in 234 U/ 238 U have wide-ranging applications as tracers for ground- and river-water fluxes and is an essential component in U-series dating. Analytical developments for measuring 234 U/ 238 U have progressed from direct alpha-counting, with precisions at the percent level, to thermal ionization and multiple-collector inductively coupled plasma mass spectrometry (TIMS and MC-ICPMS, respectively) isotopic measurement techniques. However, 234 U/ 238 U is difficult to measure with better than permil precision because of the small atomic ratios for most geological samples (∼10 −4 range). Using a Nu Instruments Nu Plasma MC-ICPMS, we have developed two analytical techniques for the precise measurement of 234 U/ 238 U: (1) a conventional standard-bracketing protocol using multiple Faraday cups and electron multipliers with ion counting capabilities (FM) and (2) a standard-bracketing Faraday cup protocol (FF). Both are capable of measuring 234 U/ 238 U with precisions at the epsilon level (1 epsilon = 1 part in 10 4 ): (1) The conventional standard-bracketing FM measurements are conducted as static measurements with the minor 234 U isotope measured in a conventional discrete dynode electron multiplier (SEM) equipped with ion counter and a retardation filter. The Faraday-multiplier gain is measured using bracketing measurements of the U metal standard CRM-145. The external reproducibility of 234 U/ 238 U (reformulated into δ-notation as δ 234 U), interspersed with frequent measurements of the gain, is at the ±0.6‰ level (2σ) for both uraninite and carbonate standards, takes ∼75 min and consumes ∼120 ng of U per measurement. (2) The static standard-bracketing FF protocol measures all three natural U isotopes in Faraday collectors. This is not usually possible using a standard multiple-Faraday array due to the large differences in the abundances of naturally occurring U isotopes. In our study, this is achieved by replacing the standard 10 11  resistor for the 238 U Faraday cup with a 10 9  resistor. The 10 9  resistor enables the measurement of ion beams that are ∼100 times larger than can be accommodated by the normal 10 11  resistor, so 238 U and 234 U are measured simultaneously in Faraday cups with intensities of ∼9 × 10 −9 and ∼5 × 10 −13 A, respectively. All measurements are normalized to bracketing CRM-145 standard measurements (measured with similar 238 U signal intensities) thereby correcting for significant tailing from the large 238 U ion beam below the smaller 234 U beam. Measurements are conducted over 2 min on-peak and 400–650 ng of 238 U is required per analysis. External reproducibility for samples with low matrix/U ratios (e.g., uraninites) is better than ±0.3‰ (2σ). Coral samples show a slightly poorer external reproducibility of ±0.4‰ (2σ) due to a higher matrix/U ratios of these samples. Repeat measurements of CRM-145 give respective δ 234 U values of −36.44 ± 0.10‰ (2σ m , n = 9) and −36.50 ± 0.14‰ (2σ m , n = 54) using the FF and FM analytical technique, assuming Harwell uraninite (HU-1) is in secular equilibrium with respect to 234 U/ 238 U. The improved analytical precision achieved in this study for 234 U/ 238 U measurement is superior to any other reported measurements and is of great importance for U-series dating errors, particularly for samples older than 300,000 years before present. © 2004 Elsevier B.V. All rights reserved. Keywords: MC-ICPMS; Uranium; Isotope; U-series ∗ Corresponding author. Tel.: +41 1 632 6983; fax: +41 1 632 1179. E-mail address: andersen@erdw.ethz.ch (M.B. Andersen). 1387-3806/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ijms.2004.07.004