Advancing Climate Models by Microfluidic Nucleation Experiments

Mineral dust particles act as ice-nucleating particles and play a crucial role in weather and climate processes. Accurate understanding of these mechanics is required to improve and validate climate models, as the composition of atmospheric dust directly influences both precipitation and radiation behavior. Suspensions containing pure or binary mixtures of microcline, montmorillonite, or quartz have been investigated by Nadia Shardt and co-workers using the previously developed Microfluidic Ice Nuclei Counter Zürich (MINCZ) platform, which enables the precise control of droplet composition and temperature, allowing high-throughput and repeatable freezing experiments. Detection of recorded droplets was automated by a deep neural network (MINCZ-Net), classifying freezing events with high confidence automatically, with uncertain cases being reviewed by experts.

Such a systematic approach was used to follow and understand the frozen fractions inside droplets as a function of temperature and composition. Significantly, it was found that freezing in binary mixtures follows the most ice-active mineral. Predictions of binary mixtures based on additivity of nucleation sites from pure minerals aligned well with experimental validation and application to the Arizona Test Dust. These findings support composition-aware parametrizations, offering a pathway to more accurate atmospheric ice nucleation modelling in climate simulations.

Written by Leon Krinn

Read the published article here.