Abstract
It is vital to understand the mechanical properties of flowing ice to model the dynamics of ice sheets and ice shelves and to predict their behaviour in the future. We can increase our understanding of ice physical properties by performing deformation experiments on ice in laboratories and examining its mechanical and microstructural responses. However, natural conditions in ice sheets and ice shelves extend to low temperatures (<< -10 degrees C), and high octahedral strains (> 0.08), and emulating these conditions in laboratory experiments can take an impractically long time. It is possible to accelerate an experiment by running it at a higher temperature in the early stages and then lowering the temperature to meet the target conditions once the tertiary creep stage is reached. This can reduce total experiment run-time by > 1000 h; however it is not known whether this could affect the final strain rate or microstructure of the ice and potentially introduce a bias into the data. We deformed polycrystalline ice samples in uniaxial compression at -2 degrees C before lowering the temperature to either -7 or -10 degrees C, and we compared the results to constant-temperature experiments. Tertiary strain rates adjusted to the change in temperature very quickly (within 3% of the total experiment run-time), with no significant deviation from strain rates measured in constant-temperature experiments. In experiments with a smaller temperature step (-2 to -7 degrees C) there is no observable difference in the final microstructure between changing-temperature and constant-temperature experiments which could introduce a bias into experimental results. For experiments with a larger temperature step (-2 to -10 degrees C), there are quantifiable differences in the microstructure. These differences are related to different recrystallisation mechanisms active at -10 degrees C, which are not as active when the first stages of the experiment are performed at -2 degrees C. For studies in which the main aim is obtaining tertiary strain rate data, we propose that a mid-experiment temperature change is a viable method for reducing the time taken to run low-stress and low-temperature experiments in the laboratory.