Double dating in the Middle Pleistocene: assessing the consistency and performance of the carbonate U–Th and U–Pb dating methods

Abstract. The U–Th and U–Pb dating methods are widely used for radiometric dating of Pleistocene carbonates, such as speleothems and corals. The U–Th dating method has been incrementally refined over recent decades, largely through advances in mass spectrometry, and is now capable of providing accurate and precise ages for carbonates as old as 640 ka in ideal circumstances. Likewise, the U–Pb method, which was previously used exclusively for dating pre-Quaternary materials, has been adapted in recent times for dating carbonates as young as a few hundred ka. As a result, there is now considerable overlap in the applicable range of these two dating methods but, as yet, little data available regarding their consistency or relative performance when dating samples within this age range. In this study we undertake a systematic comparison of the U–Th and U–Pb dating methods focusing on a large part of their potential overlapping age range (∼430–630 ka). We achieve this by 'double dating' speleothems (secondary cave mineral deposits) from Corchia Cave, central Italy, that are well-suited to both methods and adopt state-of-the-art isotope dilution MC-ICP-MS analytical protocols. This includes a U–Th measurement protocol that collects the low abundance ²³⁴U⁺ and ²³⁰Th⁺ ion beams in a Faraday cup fitted with a 10¹³ Ω resistor, a protocol that is not only ideally suited to dating samples within this age range, but also permits measurement of ²³⁸U/²³⁵U ratios at a level of accuracy and precision appropriate for assessing natural ²³⁸U/²³⁵U variability.

The results of this comparison demonstrate excellent agreement between the U–Th and U–Pb dating methods and show that both methods are capable of producing accurate and precise ages over this interval. We find that U–Pb age uncertainties are generally less predictable than U–Th age uncertainties, but on average do not increase significantly over the age interval consid- ered, whereas U–Th age uncertainties tend to increase in a more predictable (approximately exponential) manner. Furthermore, U–Pb age uncertainties are highly dependent on the availability of sub-samples with a substantial spread in Pb/U ratios and/or highly 'radiogenic' (i.e. very low inherited-Pb) material. We also find that average U–Pb isochron age precision surpasses that of U–Th age determinations at ca. 520 ka, although the exact age overlap point is expected to vary significantly across different sample types and study sites. Overall, these findings support the prospect of obtaining accurate and internally consistent U-series based chronologies spanning the Middle Pleistocene and beyond, and suggest that for some carbonate samples the U–Pb chronometer may provide superior age precision to the U–Th chronometer prior to the latter reaching its upper age limit.