Attempts to understand potential deficiencies in chemical procedures for accelerator mass spectrometry (AMS)

Since 2009, the DREAMS (DREsden Accelerator Mass Spectrometry) facility offers users to do their own sample preparation for producing AMS-targets. Several projects are aiming at analysing ¹⁰Be, ²⁶Al, and ³⁶Cl (as BeO, Al₂O₃, AgCl). In cooperation with other AMS-facilities, also actinides (coprecipitation as Fe₂O₃) and 60Fe (as Fe₂O₃) are investigated.
Hence, essential steps are hydroxide or AgCl precipitation. For the determination of in-situ or atmospheric ²⁶Al in marine and terrestrial sediments, we had sometimes unaccountable low chemical yields, which might be explained by redissolving Al(OH)₃ in the last washings. Thus, we investigated these potential losses by ICP-MS as a function of alteration (waiting) times. Indeed, up to 31% of the precipitated Al was redissolved by immediate triple washings. After 2 h of waiting, this could be reduced to 11%. Further waiting (over-night) resulted in losses of 6% of Al (equally for Be) only.
We also tested the behaviour of Fe(III), U(VI) (also as analogue for ~Pu(VI) and Np(VI)), and Er(III) (as analogue for Am(III), Cm(III), Pu(III)) when Fe(OH)₃ is washed. Even including the supernatant, total losses of 3-times washing of over-night altered hydroxides are as low as 2.6-3.5%. Thus, repeated washing cycles are very advisable to reduce ions such as NH₄ ⁺ and Cl^- before drying and ignition.
For a single project, we explored the possibility to measure ³⁶Cl and ^natCl by (isotope dilution) AMS in “dirty” permafrost ice wedge samples as heavy as 1.6 kg. The chemical yield of AgCl was only 20-35% and is a function of total ^natCl. Thus, we tested preconcentration steps like ion exchange (DOWEX 1x8, 5 ml), which look promising.
Very often DREAMS projects focus at analysing quartz for in-situ-produced ¹⁰Be and ²⁶Al. Dissolving quartz only will minimise other troublesome elements such as Al, Be, and Ti from coexisting mineral phases. Obviously, low stable Al leads to higher ²⁶Al/²⁷Al, i.e. better ²⁶Al-statistics. However, low Ti is also helpful for fewer problems in Be-chemistry, i.e. better ¹⁰Be-statistics. For correct calculation of exposure ages and erosion rates, “pure quartz”, i.e. similar target elements as the calibration site used for production rates and no natural ⁹Be, is needed, too. One of the earliest quartz cleaning methods is routinely used at DREAMS: H₂SiF₆/HCl on a shaker table at room temperature. It produces up to 1.8% residue of the original “quartz” mass with a mean maximum value of 0.6% (values from >100 samples from six different projects). SEM-EDX identified the most prominent minerals to be zircon Zr[SiO₄], white sillimanite Al₂[O|SiO₄], transparent to blue kyanite Al₂[O|SiO₄], black chromite Fe²⁺Cr₂O₄, and orange rutile TiO₂. For comparison, we treated 3.5 g of these residues by microwave (MW) digestion resulting in further dissolving 31% of the original residue. SEM-EDX analyses of the MW-residue showed mainly pristine kyanite and heavily-attacked sillimanite only. ICP-MS of the MW-solution validated the dissolution of Al, Ti, Cr, Fe, and Zr. For a typical 50 g-“quartz sample” the MW-method would add more than 3 mg of Ti, over 7 mg of Al and, and, worst of all, about 25 μg of Be to the sample.
Ackn.: Thanks to B. Bookhagen, A. Gärtner, T. Opel, P. Steier, F. Quinto & S. Weiss.