Dating old ice with the 36Cl/10Be ratio

The bottommost sections of ice cores are often difficult to date, due to the low temporal resolution and possible disturbances, such as folding and missing layers. One possible tool for dating this ice is the ³⁶Cl/¹⁰Be ratio, which decays with a combined half-life of 384 kyr years. Individual radionuclides are created by galactic cosmic rays in the atmosphere, but the ratio has been modelled to remove the varying production signal. The chronology of the recently drilled Skytrain ice core from West Antarctica ends with an age of 126 kyr BP 24 m above bedrock. Our aim was to obtain age estimates for samples in the undated section below, while improving our understanding of the ³⁶Cl/¹⁰Be ratio as a dating tool. Two datasets were measured: an annually resolved record of the last few decades and a series of older samples from the Holocene, the last interglacial and five samples from the undated section. The data from recent decades was used to test whether the Skytrain site is affected by ³⁶Cl loss, which occurs at low accumulation sites, such as EPICA Dome C and Little Dome C in East Antarctica, where ³⁶Cl is gassing out as HCl. By measuring anthropogenic ³⁶Cl from nuclear bomb tests in the 50s and 60s, we were able to confirm that the peak is found at the expected depth and that no ³⁶Cl loss occurs. In older samples, there was a marked difference between glacial and interglacial data, with higher individual ³⁶Cl and ¹⁰Be concentrations in glacial times. This is observed at other sites as well and can most likely be attributed to a dilution effect. However, the ³⁶Cl/¹⁰Be ratio was also found to be higher in the last glacial period and correlated with the d¹⁸O signal, which likely results from the different physical and chemical properties of ³⁶Cl and ¹⁰Be. While ³⁶Cl can be found in its gaseous form or attached to particles, ¹⁰Be is always attached to particles, which yields different sensitivities to changes in temperature or precipitation. Possible mechanisms include a washout en-route, which may affect one radionuclide more than the other or an increased scavenging efficiency for ³⁶Cl in mixed-phase clouds. While not fully understood, the correlation with d¹⁸O was used to detrend the data and estimate the age of five samples below the dated section, the oldest being 541 ⁺⁵⁵_-61 kyr old.