Quantification of Uncertainty in Experimentally Determined Heterogeneous Nucleation Rates

Abstract:

An experiment that recorded freezing temperatures of 1 µL water droplets on linearly cooled treated mica surfaces was conducted to try and determine if the surface treatments had any impact on the heterogeneous nucleation rate. Based on approximately 50 total freezing measurements for each surface, estimations of the heterogeneous nucleation rate at different temperature intervals were made. Since the nucleation rate in some temperature ranges were ultimately estimated from only a few (<10) freezing events, there is significant statistical uncertainty in these estimates.

To quantify this statistical uncertainty, simulations that mimic the structure of the experimental data were conducted. Using a stochastically generated ensemble with intrinsic nucleation rates spanning 5 orders of magnitude and sample sizes ranging from 3-50 freezing events per temperature interval, the simulations yield uncertainty estimates that are mostly consistent with previously presented theoretical expectations; this helps to justify claims that samples with different surface treatments have distinct nucleation rates, despite a modest number of observed freezing events.