Nuclear Instruments and Methods in Physics Research B52 (1990) 461-468 North-Holland 461 U/Th and 14C ages of corals from Barbados and their use for calibrating the 14C time scale beyond 90~0 years zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJ B.P. * Edouard Bard ‘**)+ Bruno Ham&n ‘I* * Richard G. Fairbanks *), Alan Zindler I), Guy Mathieu 1, , and Maurice Arnold *) ‘I Iamont-Doherty Geological Observarov of Columhiu University. Palisades, New York 10964. USA ” Cenrre des Faib/es Rudioaciivit& CNRS- CEA, 91198 Gijf-sur- Yvette, France The mass specfrometry of ““Th and 234U has been developed on a thermal ionization mass spectrometer (VG Micromass 30). It is shown that the mass specfromefric defermi~tion is si~ificantly more precise and rcpr~ucible than the classical radioactive zyxwvu method (u decay). The age errors obtained by this new technique of U-Th measurement are now comparable to that achieved by the 14C method which will allow us to pursue the 14C calibration on carbonates. The comparison between the U/‘I’Tt and 14C ages obtained on the Holocene samples demonstrate that the U/Th ages are accurate because they are in agreement with the dendrochronological calibration. Beyond 9ooo yr B.P. it is shown that the 14C ages are systematically younger than the U/Th ages with a m~mum difference of about 3000 yr reached at about 20000 yr B.P. It is also shown that the two main problems are the thorium ionization efficiency and the mass separation of the magnet sector for the 234tJ/2’KU ratio measurement. Signilicant impr~vemenfs can thus be expected from mass spectrometry through the use of sputtering ion sources for (~pfirni7~ng the ioni~tion efficiency, additional magnetic and electrostatic sectors to improve the abundance sensitivity and ion counting detectors to increase the measurement statistics. The experimental method of U/Th dating by ther- mal ionization mass spectrometry was first developed at the California Institute of Technology and was demon- strated to be significantly more precise and accurate than the a-counting method [l-4]. We implements this technique at the Lamont-Doherty Geological Ob- servatory in order to compare directly U/Th ages and 14C ages obtained on the same coral samples by classi- cal P-counting and by AMS [S]. All isotopic measurements presented in this paper were performed on a VG Micromass 30, sin~e~oll~tor mass spectrometer, equipped with a Daly detector [6] for the analysis of small abundance isotopes (see fig. 1 for a schematic description of the Daly). The Daly has a gain of IO7 and the signal is measured in analog mode using an electrometer with a lo7 52 high-precision resis- tor (see fig. 1). This device provides an amplification factor of lo3 compared to the Micromass 30 Faraday cup. A new VG Isolab 54 mass spectrometer is being installed presently at L-DGO. The two main advantages of this new instrument for our me~urements are an * Invite4 paper. ** Present address: GCosciences de l’Environnement, Uni- versite Aix-Marseille III, 13397 Marseille, France. increased abundance sensitivity obtained by two elec- trostatic sectors (before and after the magnet), a Dafy detector that works in ion counting mode and a sputter- ing ion source. We are also working on the development of thorium and uranium isotopic analyses with this facility. 2. Chemical separation of uranium and thorium The corals were first rinsed with distilled water in an ultrasonic bath to remove surface contaminants. The chemical separation techniques of U and Th have been adapted from refs. [2,3] and are described briefly in fig. 2. The main difference with the chemistry procedure adopted by Edwards et al. [3] is that the thorium fraction is collected before the uranium fraction. This change was crucial for obtaining Th recovery better than 95%. During our chemical procedure some uranium is lost during the iron elution, but for most age dating this is of minor important since only a third of the actual uranium fraction is loaded on the filament (about 1 ug of U). The second step of purification of the U fraction is needed to obtain a clean deposit on the Re filament with optimal ionization characteristics. We tested the need of the HClO, procedure by comparing duplicate ages obtained on coral samples with and without the HClO, chemical treatment. No detectable difference (three duplicate couples) was ob- 0168-%3X/9O/SO3.50 8 1990 - Elsevier Science Publishers B.V. (North-Holland) IV(a). ARCHAEOLOGY