Gas loss and isotopic fractionation induced by pumping during ice core gas extractions

Abstract. Ice core trapped gases are crucial paleoclimate archives, yet various gas loss processes introduce fractionation that obscures climatic signals. Among these, fractionation induced by pumping during sample evacuation has remained poorly constrained. Here we quantify pump-induced fractionation through controlled pumping experiments on a horizontal ice core from coastal East Antarctica. We investigated the responses of gas ratios and isotopic compositions (δ¹⁸O_atm, δO₂/N₂ and δAr/N₂) to varied rotary or turbo pumping durations (0.5 – 90 min). We found that the extent of gas loss is pump-dependent (turbo > rotary) but independent of pumping duration. The δAr/N₂ – δO₂/N₂ fractionation slope from pair differences between samples subjected to different pumping durations was 0.86, significantly higher than the ≤ 0.5 slopes typical of bubble close-off and post-coring gas loss, and approaching the slope of ~1 characteristic of size-dependent gas diffusion in ice lattice. Isotopic enrichment in δ¹⁸O_atm correlated strongly with gas loss magnitude, yielding fractionation slopes of -0.0132 ‰ ‰^-1 against δO₂/N₂ and -0.0124 ‰ ‰^-1 against δAr/N₂, substantially steeper than those associated with natural gas loss processes. These observations imply that pumping preferentially evacuates gases from ice cracks, which are mainly influenced by mass-dependent fractionation, leaving bubble-resident gases fractionated by lattice diffusion. The coupled loss of O₂ and Ar insights corrections for δ ¹⁸ O_atm and δO₂/N₂ based on the covariations among δ¹⁸O_atm, δO₂/N₂and δAr/N₂. Application to the Dome Fuji ice core demonstrates that scatter in δO₂/N₂ records from the bubble-clathrate transition zone can be effectively reduced by correcting for gas losses using δAr/N₂ as a proxy. Our findings provide the first quantitative constraints on pump-induced fractionation and offer a feasible correction method for reducing data uncertainties, thereby enhancing the fidelity of ice core paleoclimate reconstructions.