Assessing the influence of morphological variability and classifier arrangement on tandem particle classification analysis

Aerosol science relies on multiple classification techniques that separate particles based on distinct physical properties such as mass, mobility, and aerodynamic diameter. Instruments like the differential mobility analyser (DMA), aerosol aerodynamic classifier (AAC), and centrifugal particle mass analyser (CPMA) enable these separations. By combining two of these measurement methods in tandem, it becomes possible to infer additional particle characteristics, such as effective density, which are crucial for understanding aerosol morphology. In this work, we investigate how morphological diversity within a particle population and classification-induced asymmetries influence the retrieval of average aerosol properties in tandem measurements. Numerical simulations reveal that instruments such as the AAC and, to a lesser extent, the CPMA select particles asymmetrically about the mean of a mass–mobility distribution, leading to systematic shifts in the inferred effective density. Experimental measurements of soot aerosols confirm these predictions, showing that the order of classifiers in tandem setups alters the retrieved mass–mobility parameters, in some cases producing physically unrealistic exponents. These findings highlight that classification-induced biases, if unaccounted for, can lead to misinterpretation of ensemble-averaged morphology, particularly for morphologically diverse aerosols. We emphasise the need for careful selection of classifier pairings and correction strategies when comparing mass–mobility relationships across different instruments, studies, or laboratories. • Numerical simulations reveal that tandem AAC and CPMA classifiers select particles asymmetrically around the mass–mobility mean, systematically biasing effective density estimates. • Experimental soot measurements confirm that the sequence of mobility, mass, and aerodynamic classifiers alters retrieved mass–mobility exponents and can yield non-physical results. • A DMA-first tandem configuration greatly reduces bias, whereas upstream AAC or CPMA configurations introduce significant errors. • Practical recommendations for classifier pairing, sequencing, and bias-correction strategies are provided to ensure accurate and comparable mass–mobility analyses across studies.