Roy, Priyatanu2021-09-242021-09-242021-06https://hdl.handle.net/11299/224653University of Minnesota Ph.D. dissertation. June 2021. Major: Mechanical Engineering. Advisor: Cari Dutcher. 1 computer file (PDF); xxix, 269 pages.Atmospheric aerosols are suspensions of microscopic chemically complex solid or liquid particles in the atmosphere. The composition and phase of aerosols play important roles in determining radiative forcing, cloud formation, atmospheric chemistry, visibility and human health. Temperature and relative humidity (RH) dependent aerosol particle phase states and phase transitions control interactions with the surrounding gas phase as well as with other particles, and the way the particles evolve with age. Due to acceleration of global warming, there is an urgent need to develop more accurate particle-resolved climate models to improve climate prediction. Aerosols remain the largest source of uncertainty in climate predictions. The main goal of this dissertation is to develop microfluidic instrumentation to measure aerosol droplet phase transitions such as liquid-liquid phase separation and ice nucleation as a function of temperature and relative humidity. First, liquid-liquid phase separation (LLPS) similar to that observed in atmospheric aerosol droplets is investigated with aqueous droplets containing organic and inorganic solutes in a static trap based microfluidic device. LLPS in an aerosol particle directly affects aerosol water uptake and formation of cloud drops. Temperature and RH dependence of LLPS and crystallization for model aerosol droplets with varying composition is explored. It is observed that temperature has a significant effect on some systems while having no effect on others depending on the organic to inorganic ratio (OIR) as well as the identity of the organic and inorganic phases. Second, a high-throughput droplet freezing counter based on flow-through droplet microfluidics was developed to estimate ice nucleation (IN) in liquid samples relevant to atmospheric cloud droplets. Automated detection and classification of frozen droplets from liquid drops was implemented through machine learning with a deep neural network. A case study with an ideal biological ice nucleating particle (INP), Snomax, was performed. Heating and aging of the sample were also performed to identify the molecular nature of ice nucleation. The device benchmarked well against literature data and provided the highest throughput of any existing INP counters. Finally, a large array based static trap microfluidic device was implemented to study both RH dependent phase and temperature dependent INP concentration of the same sample in situ using bulk sea water and sea surface microlayer (SSML) from a simulated waveflume experiment (SeaSCAPE). This study has implications in identifying origins of INPs in sea spray which make up a significant portion of atmospheric aerosols. Correlation between ice nucleation temperature and residual dry particle morphology showed that the bulk sample had lower INPs than SSML and the residual particles were significantly different between the samples. In this dissertation, instrumentation development and case studies have been performed to show the suitability of microfluidics as versatile, adaptable and highly customizable devices, which are applicable to studying phases of aerosols and has broad implications in climate science.enAerosolsAtmospheric ScienceIce nucleating particleMicrofluidicsThesis or Dissertation