We report the characterisation of anthropogenic magnetic particulate matter (MPM) collected on leaves from roadside Callistemon trees from Lahore, Pakistan, and on known sources of traffic-related particulates to assess the potential of first-order reversal curve (FORC) diagrams to discriminate between different sources of anthropogenic magnetic particles. Magnetic measurements on leaves indicate the presence of surface-oxidised magnetite spanning the superparamagnetic (< 30 nm) to single-domain (~30-70 nm) to vortex size range (~70-700 nm). Fe-bearing particles are present both as discrete particles on the surface of larger mineral dust or carbonaceous particles and embedded within them, such that their aerodynamic sizes may be decoupled from their magnetic grain sizes. FORC diagrams of brake-pad residue specimens show a distinct combination of narrow central ridge, extending from 0-200 mT, and a low-coercivity, vertically spread signal, attributed to vortex and multi-vortex behaviour of metallic Fe. This is in agreement with scanning electron microscopy results that show the presence of metallic as well as oxidised Fe. Exhaust-pipe residue samples display a more conventional ‘magnetite-like’ signal comprising a lower coercivity central ridge (0-80 mT) and a tri-lobate signal attributed to vortex state and/or magnetostatic interactions. The FORC signatures of leaf samples combine aspects of both exhaust residue and brake-pad endmembers, suggesting that FORC fingerprints have the potential to identify and quantify the relative contributions from exhaust and non-exhaust (brake-wear) emissions. Such measurements may provide a cost-effective way to monitor the changing balance of future particulate emissions as the vehicle fleet is electrified over the coming years.

Identification of unknown micro- and nano-sized mineral phases is commonly achieved by analyzing chemical maps generated from hyperspectral imaging datasets, particularly scanning electron microscope - energy dispersive X-ray spectroscopy (SEM-EDS). However, the accuracy and reliability of mineral identification are often limited by subjective human interpretation, non-ideal sample preparation, and the presence of mixed chemical signals generated within the electron-beam interaction volume. Machine learning has emerged as a powerful tool to overcome these problems. Here, we propose a machine-learning approach to identify unknown phases and unmix their overlapped chemical signals. This approach leverages the guidance of Gaussian mixture modeling clustering fitted on an informative latent space of pixel-wise elemental datapoints modeled using a neural network autoencoder, and unmixes the overlapped chemical signals of phases using non-negative matrix factorization. We evaluate the reliability and the accuracy of the new approach using two SEM-EDS datasets: a synthetic mixture sample and a real-world particulate matter sample. In the former, the proposed approach successfully identifies all major phases and extracts background-subtracted single-phase chemical signals. The unmixed chemical spectra show an average similarity of 83.0% with the ground truth spectra. In the second case, the approach demonstrates the ability to identify potentially magnetic Fe-bearing particles and their background-subtracted chemical signals. We demonstrate a robust approach that brings a significant improvement to mineralogical and chemical analysis in a fully automated manner. The proposed analysis process has been built into a user-friendly Python code with a graphical user interface for ease of use by general users.