Iron-manganese system for preparation of radiocarbon AMS targets; characterization of procedural chemical-isotopic blanks and fractionation.

Abstract

We report a practical system to mass-produce accelerator mass spectrometry (AMS) targets with 10-100 mu g carbon samples. Carbon dioxide is reduced quantitatively to graphite on iron fibers via manganese metal, and the Fe-C fibers are melted into a bead suitable for AMS. Pretreatment, reduction and melting processes occur in sealed quartz tubes, allowing parallel processing for otherwise time-intensive procedures. Chemical and isotopic ( (super 13) C, (super 14) C) blanks, target yields and isotopic fractionation were investigated with respect to levels of sample size, amounts of Fe and Mn, pretreatment and reduction time, and hydrogen pressure. With 7-day pretreatments, carbon blanks exhibited a lognormal mass distribution of 1.44 mu g (central mean) with a dispersion of 0.50 mu g (standard deviation). Reductions of 10 mu g carbon onto targets were complete in 3-6 h with all targets, after correction for the blank, reflecting the (super 13) C signature of the starting material. The 100 mu g carbon samples required at least 15 h for reduction; shorter durations resulted in isotopic fractionation as a function of chemical yield. The trend in the (super 13) C data suggested the presence of kinetic isotope effects during the reduction. The observed CO (sub 2) -graphite (super 13) C fractionation factor was 3-4% smaller than the equilibrium value in the simple Rayleigh model. The presence of hydrogen promoted methane formation in yields up to 25%. Fe-C beaded targets were made from NIST Standard Reference Materials and compared with graphitic standards. Although the (super 12) C ion currents from the beads were one to two orders of magnitude lower than currents from the graphite, measurements of the beaded standards were reproducible and internally consistent. Measurement reproducibility was limited mainly by Poisson counting statistics and blank variability, translating to (super 14) C uncertainties of 5-1% for 10-100 mu g carbon samples, respectively. A bias of 5-7% (relative) was observed between the beaded and graphitic targets, possibly due to variations in sputtering fractionation dependent on sample size, chemical form and beam geometry.