(424e) Assessing the Driving Force of Stochastic Ice Nucleation from Aqueous Solution

The freezing of aqueous solutions is of great importance to many fields, but the kinetics of its first step, stochastic ice nucleation, is not well understood. Relevant studies in the literature typically assess the freezing behavior of micrometer-sized droplets due to their relevance to cloud microphysics. It is unclear, however, whether the findings of these contributions can be generalized and extended to larger volumes, such as those often used to freeze biopharmaceuticals. This is because the volume affects both process conditions, such as attainable cooling rates and heat transfer in general, and the inherent nature of nucleation: in large volumes, heterogeneous nucleation promoted by the presence of surfaces dominates, while in micrometer-sized droplets homogeneous nucleation is commonly considered dominant.

To accurately capture how the stochastic nature of nucleation manifests itself in larger volumes, we monitored a total of about 10,000 freezing processes in vials with a fill volume of 1 mL from solutions of ten different compositions. We build on earlier work where we developed and validated the methodology to measure ice nucleation temperatures under tightly-controlled conditions at mid-throughput, and where we showed how to estimate the parameters in the kinetic rate expression and their uncertainty (cf. 10.1016/j.ces.2023.118531).

The statistical analysis revealed that the kinetic parameters in the rate expression of stochastic ice nucleation are independent of solute type and concentration: we estimated the parameter values for all solution compositions, and we found that a single set of parameters was able to quantitatively describe the nucleation from all solutions. This is true whether the driving force for nucleation is expressed in its rate expression as a function of water activity or of temperature. While the former expression is used more frequently in the literature, the latter is significantly less computationally demanding and hence is considered the method of choice for pharmaceutical applications.