Rapid Identification of Aerosol Particles: Electrochemical Characterization Infused with Machine Learning

The rapid and accurate identification of aerosolized chemical and biological threats, such as viruses, anthrax, fentanyl, and other nanoparticles (NPs), represents a significant and ongoing challenge for defense, environmental monitoring, occupational health, and public safety. Traditional aerosol-detection technologies, primarily based on optical methodologies, struggle with sensitivity and robustness for low aerosol concentrations in the sub-?g/m3 due to low quantum yields that can uniquely identify the chemical fingerprint of the aerosol, e.g., Raman/fluorescence spectroscopy. In response, this proposal introduces the Electro-Chemical Aerosol Detector (ECAD), a novel detection platform that will employ electrochemical characterization combined with advanced machine learning (ML) techniques to detect and identify NPs chemically by their electrocatalytic temporal sequence combined with their voltametric characteristics. Electrocatalytic techniques are highly sensitive, down to single NP resolution for metallic and TiO2 semiconducting NPs in laboratory conditions. However, this ultimate resolution has not been achieved for carbon-based materials, or soft materials. Therefore, the proposed project will demonstrate extension to carbon-based materials and optimize detection technology towards field deployment. Phase I efforts will demonstrate the feasibility and efficacy of the ECAD platform through laboratory experimentation and preliminary prototype development. Efforts will focus on validating electrochemical detection techniques to identify single nanoparticles, leveraging a known relationship between electrocatalytic frequency oscillations and particle concentration for calibration. Additionally, this research will explore operational parameters for robust detection under realistic environmental conditions. Electrocatalytic temporal sequences will be statistically analyzed for: (1) automated data processing, (2) distinguishing meaningful signals from background interference, and (3) feature extraction towards ML input. We will develop ML algorithms to classify aerosolized threats rapidly and accurately using state-of-the-art transformer-based classifiers embedded with contextual information from cyclic voltammetry (for robustness in changing environmental conditions). A prototype will be constructed to demonstrate practical sampling capability, serving as a critical foundation for future development and field deployment of the ECAD system. If successful, the ECAD system can be developed further (Phase II and beyond) such that it will benefit sectors requiring rapid, precise aerosol detection capabilities, including military and defense operations, air quality monitoring, public health responses to biothreats, pandemic preparedness, law enforcement investigations related to drug exposure, and customs agencies conducting border security screenings. Anticipated Benefits/Potential Commercial Applications of the Research