Oxygen in the trapped air: identifying primary atmospheric signals and secondary bubble close-off fractionation

Ice cores offer a unique opportunity to study past atmospheric composition because the trapped gases are ultimately derived from ancient atmosphere. This distinctive ability has motivated some pioneering efforts in the late 1980s to measure the elemental composition of the trapped air. The goal at that time was to reconstruct past atmospheric O₂ concentration variations (PO₂), the primary (atmospheric) signal. However, these measurements quickly revealed that oxygen, argon, and nitrogen concentrations in the ice (expressed as δO₂/N₂ and δAr/N₂) are altered by gas-trapping processes. This is the secondary (and confounding) signal formation process.

Such alterations have long thwarted the application of ice core δO₂/N₂ to reconstruct true PO₂. Empirically though, δO₂/N₂ in the trapped air has a high coherency with local summer insolation, but the cause is not well understood. Presumably, the intensity of sunlight on the surface of the ice sheet determines the extent and nature of snow metamorphism, which in turn modulates the magnitude of O₂ and Ar losses relative to N₂ at the “bubble close-off depth” (typically 70-120 m in polar regions).

Here, we discuss recent efforts to identify the secondary signals in δO₂/N₂ and extract the primary, atmospheric PO₂ signals. First, using δAr/N₂ as a proxy for insolation, we show that PO₂ remained stable prior to the Mid-Pleistocene Transition (MPT), the shift from 40- to 100-kyr glacial cycles. After the MPT, however, PO₂ began to decline, possibly linked to enhanced to weathering as a result of glacier expansion and, to a lesser degree, more exposed areas of the continental shelves.

Next, we proceed to exploit the secondary signal itself. That is, δO₂/N₂ and δAr/N₂ could be used as a direct proxy of local insolation. By examining the relationship between elemental ratios and temperature proxies, we explore the relationship between high-latitude southern hemisphere insolation and Antarctic temperature in three time slices in the Pleistocene. The implications for the nature of 40-kyr glacial cycles will be discussed.