Real-Time Detection of Narcotics and Explosives Using Single-Particle Mass Spectrometry
JOHANNES PASSIG, Marco Schmidt, Haseeb Hakkim, Aleksandrs Kalamasnikovs, Petra Hehet, Ellen Iva Rosewig, Guanzhong Wang, Heinrich Ruser, Michael Pütz, Martin Seipenbusch, Simone Vinati, Karsten Wegner, Thorsten Streibel, Robert Irsig, Andreas Walte, Sven Ehlert, Ralf Zimmermann, Mass Spectrometry Centre;Rostock University/Helmholtz Munich
Abstract Number: 274
Working Group: Indoor Aerosols
Abstract
Beyond air pollution, dangerous dusts pose significant risks to human health and safety, e.g. from accidental releases of hazardous substances or during the illegal transport and handling of drugs. The current opioid crisis in the U.S. has highlighted the lack of technologies that can detect such dusts in real-time. We developed an approach to address this gap. It includes (I) pulsed particle resuspension, (II) aerosol concentration, (III) SPMS detection, and (IV) real-time pattern recognition using machine learning.
The core detection technology of our approach is Single-Particle Mass Spectrometry (SPMS), which is capable of chemically characterizing individual particles in a complex aerosol (Pratt and Prather, 2012). Following the introduction of resonant ionization methods for particle-bound metals (Passig et al. 2020) and polycyclic aro-matic hydrocarbons (Passig et al. 2022), we are ready to tailor the SPMS technology to dusts of various drugs and explosives. We present the resulting single-particle mass spectra of drugs and show how interferences with other particle types in a complex environment can be overcome.
We developed a sampling system for the modified SPMS including particle redispersion using gas pulses and aerosol enrichment for direct sampling from contaminated surfaces. The results include measurements conducted at the DHL hub in Leipzig, where packages containing drugs could be identified using this technology. Furthermore, experiments in a former illegal drug lab demonstrated real-time detection of drug and tablet residues on contaminated surfaces and floors.
To realize the online screening of hazardous particles and real-time risk assessment, we develop real-time soft-ware, including preprocessing, screening, clustering, and machine learning. We explored supervised learning for particle screening and classification and show that they can outperform traditional unsupervised algorithms like ART-2a clustering in terms of speed and accuracy, facilitating real-time data analysis and immediate emergency response (Wang et al., 2024).