Abstract:

Speleothem oxygen isotopes and growth rates are valuable proxies for reconstructing climate history. There is debate, however, about the conditions that allow speleothems to grow in oxygen isotope equilibrium, and about the correct equilibrium fractionation factors. We report results from a series of carbonate growth experiments in karst-analogue conditions in the laboratory. The setup closely mimics natural processes (e.g. precipitation driven by CO2-degassing, low ionic strength solution, thin solution film) but with a tight control on growth conditions (temperature, pCO2, drip rate, calcite saturation index and the composition of the initial solution). Calcite is dissolved in water in a 20,000 ppmV pCO2 environment. This solution is dripped onto glass plates (coated with seed-carbonate) in a lower pCO2 environment (and rapid depletion of the dissolved inorganic carbon reservoir (rapid DIC-depletion). The impact of evaporation can be large so caves with high relative humidity are also preferable for palaeoclimate reconstruction. Even allowing for the maximum offsets that may have been induced by evaporation and rapid DIC-depletion, d18O measured in some of our experiments remain higher than those predicted by Kim and O’Neil (1997). Our new results are well explained by equilibrium at a significantly higher acalcite–water, with a kinetic-isotope effect that favours 16O incorporation as growth rate increases. This scenario agrees with recent studies by Coplen (2007) and Dietzel et al. (2009). Overall, our results suggest that three separate processes cause d18O to deviate from true isotope equilibrium in the cave environment. Two of these drive d18O to higher values (evaporation and rapid DIC-depletion) while one drives d18O to lower values (preferential incorporation of 16O in the solid carbonate at faster growth rates). While evaporation and DIC-depletion can be avoided in some settings, the third may be inescapable in the cave environment and means that any temperature to d18O relationship is an approximation. The controlled conditions of the present experiments also display limitations in the use of the Hendy test to identifying equilibrium growth.