Enabling Real-Time, Low-Cost Measurement of Hazardous Air Pollutants: A Multidimensional Approach

TALHA JUBAIR, James Hurley, Anum Khan, Namrata Shanmukh Panji, Amanda Gao, Jesse Kroll, Gabriel Isaacman-VanWertz, Virginia Tech

     Abstract Number: 362
     Working Group: Instrumentation and Methods

Abstract
Hazardous air pollutants (HAPs) such as benzene, toluene, and acrolein pose significant health risks even at low concentrations. However, conventional measurement methods are often expensive, stationary, and lack the temporal and spatial resolution needed for effective air toxics surveillance in communities. To address this gap, we are developing a compact, low-cost system designed to measure HAPs in real time with compound-level resolution. This approach consists of short length of gas chromatography column used as an inlet for commercially available low-cost sensors. Compounds are coarsely separated by the column, the identified through the relative responses of low-cost detectors—including photoionization and metal oxide sensors—to capture the chemical signatures of the compounds as they elute from the columns. We pair this hardware with machine learning algorithms trained to interpret the sensor signals and estimate concentrations of specific HAPs. In this presentation we evaluate the system using known gas mixtures and compared against conventional reference instruments to quantify its accuracy, sensitivity, and reliability. Preliminary lab results show that the system can separate and detect key air toxics in under a minute, with clear potential for identifying and quantifying compounds in real-world conditions. In particular, we demonstrate detection limits for some detectors of down to tens of ppb and chromatographic peak widths of only a few seconds, enabling baseline resolution of compounds within a carbon number of each other (e.g., benzene and toluene). However, this sensitivity and resolution are achieved with different detectors, highlighting the engineering challenges of the low flows and high specificity needed for the goals of this project. In this work we will therefore examine solutions to these engineering challenges and present opportunities for validation through field deployment.