Magnetic properties and morphology of particles

Air volume–normalized low-field magnetic susceptibility (κ) and saturation isothermal remanent magnetization (SIRM) of Helsinki filters exhibited pronounced temporal variations from 1962 to 2005 (Fig. 1a, b). The annual averages of κ and SIRM of Helsinki samples range between 3.1 and 18 × 10-6 SI and, 0.3 and 1.4 × 10-6 A m-1, respectively. This contrasts with those measured for the Virolahti background site in 2001 (κ: 0.0095 × 10-6 SI and SIRM: 0.12 × 10-6 A m-1) indicating that the magnetic record in the Helsinki urban filters is dominated by anthropogenic sources. Contributions from natural background sources are comparatively minor or masked by urban sources as was the case for urban and industrial sites in Spain11. Peak κ and SIRM values of Helsinki filters occurred in the 1960s, followed by a decline reaching minima in 1986–1987 ( < 50% of 1960s levels). κ subsequently rebounded between 1987 and 1992 before stabilizing at low values from 2000 to 2005. Seasonal trends showed higher κ and SIRM during early winter to spring (November–April) and lower values in late summer (July–September) (Supplementary Fig. 1a,b). Background site at Virolahti shows slightly elevated SIRM values during December–February compared to those in summer months (Supplementary Fig. 2b). The contrast with the rural background site demonstrates that multi-decadal magnetic archives can isolate an urban anthropogenic signal above natural background, a prerequisite for source-resolved trend analysis.

Fig. 1: Long-term variation in magnetic properties of aerosol filters in Helsinki 1962–2005.Fig. 1: Long-term variation in magnetic properties of aerosol filters in Helsinki 1962–2005.

a Air volume-normalized low-field magnetic susceptibility (κ). b Air volume-normalized saturation isothermal remanent magnetization (SIRM). c S-ratio. In all panels, the bold line shows a locally weighted regression (LOESS) fit. Selected years for mineralogical analyses are indicated by colored circles: 1962 (red), 1969 (orange), 1986 (green), 1987 (dark blue), 1996 (purple), and 2005 (light blue).

S-ratio analysis of Helsinki filters indicated a decreasing contribution of hematite and/or goethite from 1962 to 1990, with mean values of 0.85–0.95. From 1995 onward, S-ratio stabilized at 0.95–0.98, consistent with magnetite/maghemite dominance (Fig. 1c). Seasonal S-ratio variations of Helsinki filters peaked in summer and were lowest in winter (Supplementary Fig. 1c). These values are in clear contrast with those observed at the rural Virolahti background site, where mean S-ratio for the year 2001 filters is 0.78 (Supplementary Fig. 2c). Seasonal trends were not evident at the background site, and the relatively high contribution of high-coercivity minerals at Virolahti reflects weathering of the Fennoscandian Shield bedrock, producing hematite- and goethite-rich mineral assemblages30.

MDFARM and MDFIRM values revealed modest temporal changes in magnetic grain coercivity (Fig. 2; Supplementary Fig. 3), reflecting corresponding variations in grain size and the relative contribution of magnetite versus hematite. Filters from 1962 and 1969 had the lowest MDFARM (17.4–19.0 mT), whereas 2005 filters exhibited the highest (25.9 mT), suggesting gradual shift toward smaller, more uniform magnetite grains. Similar trends were observed from the κARM/SIRM ratio (Fig. 2c, d; Supplementary Fig. 4), which indicated minor variability in particle size and grain uniformity. While the κARM/SIRM values were similar, the coercivity of ARM (MDFARM) was slightly lower than those reported for PM10 and leaf-deposited PM data from Lancaster, United Kingdom31, indicating a comparable abundance of single-domain (SD, grain size: 0.03  μm–0.1  μm) grains but slight differences in their grain-size distribution. Our data points plot to an area likely characterized by magnetic interactions between superparamagnetic (SP, <0.03  μm) particles, which may promote SD-like magnetic behavior32.

Fig. 2: Magnetic grain size indicators for selected aerosol samples.Fig. 2: Magnetic grain size indicators for selected aerosol samples.

a Box plot of median destructive field of anhysteretic remanent magnetization (MDFARM) values. b Box plot of median destructive field of isothermal remanent magnetization (MDFIRM) values. c Ratio of anhysteretic susceptibility (κARM) to saturation isothermal remanent magnetization (SIRM) κARM/SIRM versus MDFARM, (d)  κARM/SIRM versus MDFIRM. In (a) and (b) box plots, median values are indicated with a solid horizontal line, mean with open rectangles, outliers with black diamonds, whereas the purple boxes represent 25–75 % data range, and error bars are represented as 1.5 interquartile range (IQR). In (c) and (d), years are indicated by colored circles: 1962 (red), 1969 (orange), 1986 (green), 1987 (blue), 1996 (purple), and 2005 (black). For reference, grain size indicators32 (in μm; open gray rectangles and solid gray triangles), PM2.5 data from Southern West Africa21 (pink shaded area), and PM10 and leaf-deposited PM data from Lancaster, United Kingdom31 (green shaded area), are shown for comparison.

Hysteresis ratios plot to a limited area, between the SD/multi-domain (MD, >1 μm)–SD/SP mixing field33 of the Day plot34 (Fig. 3a). Despite minor temporal variations, this suggests that the magnetic assemblage is consistently dominated by grains in the SD–MD transition range (vortex-state), with smaller contributions from MD and SP grains. However, the contributions from other magnetic mineral phases and magnetic interactions may shift the hysteresis ratios giving apparent SD–MD magnetic behavior. The values are similar to those reported for the reference filter data from urban and industrial city areas of Spain11; they are also comparable to Rome PM10 data6, and with gasoline, diesel, and brake wear extractions12.

Fig. 3: Hysteresis properties of Helsinki aerosol filter samples for selected years.Fig. 3: Hysteresis properties of Helsinki aerosol filter samples for selected years.

a Hysteresis ratios plotted in the Day plot34 with adapted theoretical mixing curves33 (gray dashed lines). Reference data from urban and industrial city areas of Spain11 (gray rectangles), and PM10 samples from Rome6 (open diamonds), including gasoline (red area), diesel (yellow area), and brake wear (green area) sources12, are shown for comparison. b Low-temperature hysteresis curves (200 K, solid line; 20 K, dashed line) for samples collected in January 1969 (orange). c Low-temperature hysteresis curve for a sample collected in June 1996 (purple).

Hysteresis loops measured at low temperatures for samples from January 1969 and June 1996 revealed higher coercivities at 20 K compared to 200 K (Fig. 3b, c). The sample from 1969 exhibited an increase in coercivity by rising from 12.4 mT at 200 K to 22.6 mT at 20 K, while the sample from 1996 showed a more pronounced increase in coercivity, from 4.1 mT to 25.7 mT. The observed increase in coercivity of the samples is attributed to the contribution of SP particles35. These particles do not display remanence at room temperature; however, when cooled below their blocking temperature, they exhibit remanence and increased coercivity. This suggests increased SP contributions from 1969 to 1996.

Low temperature FC-ZFC and RT-SIRM analyses revealed minor variations between seasons and years (Fig. 4). Verwey transitions ( ~ 100–122 K) were observed in all RT-SIRM curves, although less pronounced in 1996 samples compared to 1969. In FC-ZFC curves, subtle Verwey transitions were observed in 1969 and 1987 but were again less pronounced in 1996 samples. This suggests progressive oxidation of magnetite from 1969 to 199636. The finer particle sizes inferred for 1996 would have a higher surface-area-to-volume ratio, which would facilitate surface oxidation and result in a suppressed Verwey transition.

Fig. 4: Low-temperature thermomagnetic measurements of seasonal aerosol samples.Fig. 4: Low-temperature thermomagnetic measurements of seasonal aerosol samples.

a, b Field-cooled (FC, solid lines) and zero-field-cooled (ZFC, dashed lines) curves for winter (a) and summer (b). c, d Room-temperature saturation isothermal remanent magnetization (RT-SIRM) curves for winter (c) and summer (d), with cooling shown as solid lines and warming as dashed lines. Data are shown for samples collected in 1969 (orange), 1987 (blue), and 1996 (purple).

SEM–EDS analysis of magnetic separates from Helsinki filters showed that most particles were anthropogenic iron oxide spherules ( < 1 μm; Fig. 5), often associated with minor silicate phases and trace metals including Co, Pb, and Ti (Supplementary Fig. 5). Clusters of spherical particles were frequent, with diameters commonly ranging from the resolution limits of 0.02 μm to 0.3 μm. Angular morphologies were also present, while euhedral iron oxide were less abundant compared to spherical particles. Sharp-edged, triangular iron oxide particles were also identified (Fig. 5f), along with various aggregates displaying irregular morphologies. Among them, framboid-like particles were observed (Fig. 5h–j). These particles, though variable in size, contributed to the overall diversity of particle forms observed in the sample. Anthropogenic iron oxide spherules and angular particles were also detected in the Virolahti filters (Supplementary Fig. 2d, e). Their presence likely reflects regional or long-range atmospheric transport of dust particles as documented in previous studies37.

Fig. 5: Scanning electron microscope images of iron-rich particles from aerosol filters.Fig. 5: Scanning electron microscope images of iron-rich particles from aerosol filters.

Extracted iron-rich particles from the 1967, 1976, 1986, 1995, 1996 and 2005 samples acquired in backscattered electron mode. a–l Fe-rich spherules, angular and euhedral particles were detected. d A bright Fe-rich spherule among Si-rich spherules (d); sharp-edged particle (f); euhedral Fe-rich crystals among Fe-rich spherules (g, l); framboid-like particles (h–j). l Amagnified view of the area in (k). Numbers in the upper right corner denotes collection month and year (mm/yyyy).

Magnetic component identification using coercivity unmixing and multivariate statistics

Coercivity unmixing and complementary statistical analyses were applied to the four-decade archive to resolve semi-quantitative source components and their temporal evolution. To link grain size and domain state to potential emission sources, coercivity spectra were assessed for representative summer and winter filters from 1969, 1987, and 1996.

Log-Gaussian fitting identified three magnetic components (Supplementary Fig. 6). A low coercivity component (B < 10 mT, log10 B < 1; 31–42% of the IRM signal) corresponds to coarse MD magnetite/maghemite from traffic-related abrasion, such as brake and tire wear6. An intermediate coercivity component (B ≈ 40–63 mT) constitutes the bulk of the IRM (50–67%) and represents SD to vortex-state magnetite/maghemite particles, characteristic of high-temperature anthropogenic emissions such as traffic and combustion-related sources11. However, this coercivity range may also include a contribution from fly-ash-derived hematite38,39, as well as magnetite from natural sources6,11,40. A high coercivity component (B ≳ 1000 mT, log10 B; 3 in Supplementary Fig. 6) representing 7.7–14.3% of the IRM signal is attributed to hematite. Such high coercivity component can be obtained from various sources, such as traffic and combustion emissions, brake pads and natural mineral dust41. In winter, the high-coercivity component (hematite) is apparent, with its contribution decreasing over time (absent by 1996).

Multivariate statistical analysis of magnetic and elemental variables further supports the presence of distinct particle populations. Varimax-rotated principal component analysis (PCA) resolved three components explaining more than 87% of the total variance (Fig. 6; Supplementary Table 1). The first component is dominated by remanence parameters (κARM, SIRM) together with Pb and Zn. The second component is characterized by Ca, K and magnetic susceptibility (κ). The third component reflects opposing loadings of S-ratio and HIRM, indicating variability in the relative contribution of high-coercivity phases.

Fig. 6: Principal component analysis of magnetic and elemental variables.Fig. 6: Principal component analysis of magnetic and elemental variables.

Varimax-rotated principal component loadings for magnetic and elemental variables from Helsinki aerosol filters (n = 59). PC1 (67% of variance) represents a remanence-bearing combustion component characterized by strong loadings of SIRM, κARM, Pb and Zn. PC2 (13%) reflects lithogenic dust/non-exhaust emission inputs, defined primarily by Ca, K and magnetic susceptibility (κ). S-ratio plots near the origin on the PC1–PC2 plane because it loads predominantly on PC3 (7%; not shown in the Figure), which is controlled by opposing loadings of S-ratio and HIRM and captures variability in high-coercivity hematite/goethite contributions. Together, the first three components explain >87% of the total variance.

Consistent with this, Pearson correlations show strong associations between magnetic concentration parameters and Fe, as well as moderate correlations with several trace elements, whereas S-ratio is strongly coupled to HIRM and largely decoupled from bulk elemental concentrations (Supplementary Tables 2 and 3). Together, the coercivity unmixing and statistical analyses indicate the presence of magnetic particle populations dominated by magnetite/maghemite-like remanence carriers, with additional contributions from lithogenic and high-coercivity components.

Characteristics and source of magnetic particles

Our analyses indicate that high-temperature anthropogenic particles dominate the magnetic PM in Helsinki with combustion-derived material representing a major component, as evidenced by the prevalence of magnetite and maghemite and their association with elements commonly enriched in combustion-related emissions, and their rather uniform grain size distribution. Room-temperature hysteresis properties, κARM/SIRM ratio, and median destructive fields (MDFARM, MDFIRM) reflect the magnetic behavior of predominantly uniform grain sizes in the stable SD to vortex state range, with additional contributions from MD ( > 1 μm) and SP ( < 0.03 μm) grains. These results are in agreement with previous studies of PM10 and total suspended particles, which reported their magnetic properties6,21,31,42.

Microscopic observations of the magnetic separates further confirm the predominance of anthropogenic iron oxide spherules, with lithogenic euhedral crystals being less abundant (Fig. 5). The predominance of iron oxide spherules aligns with prior observations of urban particles, commonly formed under high-temperature conditions, including both combustion processes5,43 and frictional sources such as brake wear and rail-wheel interactions4,44. Abrasion-related particles, including sharp-edged brake wear particles, were also detected, alongside framboid-like aggregates indicative of coal45,46. In addition to framboids, euhedral crystals commonly occur among fly ash, reflecting the presence of iron-bearing minerals in raw coal45. Framboid-forming pyrite is often the main iron-bearing mineral in coal, which oxidizes to iron oxide during combustion45 and further contributes to the magnetic fraction of fly ash. Obtained high-coercivity hematite is attributed to fly ash from coal combustion, where hematite is commonly present39. In addition, hematite can be also of lithogenic origin, derived from the road gritting material extracted from the Fennoscandian crystalline bedrock, containing hematite and goethite30. Collectively, these observations support the conclusion that the majority of the magnetic particles in the filters originate from anthropogenic sources.

The magnetic mineralogy inferred from coercivity unmixing further supports the presence of multiple particle sources. Three magnetic components were resolved from IRM spectra, representing low-coercivity, intermediate-coercivity, and high-coercivity populations. The intermediate-coercivity component dominates the magnetic signal and is consistent with magnetite/maghemite grains typical of high-temperature emissions. The low-coercivity fraction likely reflects coarser MD magnetite associated with mechanical abrasion processes such as brake and tire wear. A smaller high-coercivity component corresponds to hematite-like magnetic phases. Hematite in urban aerosols may originate from coal combustion fly ash39, vehicular exhaust41, or frictional sources although lithogenic sources may also contribute, particularly through road dust and winter road gritting materials derived from the Fennoscandian crystalline bedrock that contains hematite and goethite30. The presence of this component primarily in earlier winter samples and its subsequent decline over time is consistent with changes in fuel use and emission controls.

Multivariate statistical analyses reinforce the interpretation of mixed but anthropogenically dominated magnetic particle sources. PCA resolved three orthogonal components explaining >87% of the total variance (Fig. 6; Supplementary Table 1). Together, coercivity unmixing and statistical analyses indicate a dominant magnetite/maghemite remanence-bearing fraction (high-temperature emissions), superimposed on substantial lithogenic inputs (non-exhaust and/or natural) and a smaller, persistent high-coercivity (hematite) component (combustion/non-exhaust/natural). Bivariate Pearson correlations (Supplementary Tables 2 and 3) show strong associations among magnetic concentration parameters and Fe, as well as moderate correlations with combustion-related elements (Zn, S, V), while S-ratio remains primarily coupled to HIRM and largely decoupled from bulk elemental concentrations.

The first component ( ~ 67%), interpreted as an anthropogenic combustion component, is strongly loaded by magnetic parameters κARM and SIRM together with Pb and Zn, consistent with traffic- and fuel-combustion emissions3. This is in line with the bivariate correlation matrix, where SIRM and HIRM show moderate to strong correlations with the combustion and traffic associated elements (Supplementary Table 2). These relationships persist after detrending (Supplementary Table 3), suggesting that the anthropogenic magnetic particle association is not only driven by shared long-term temporal trends but also reflects short-term covariability in particulate inputs.

The second PCA component ( ~ 13%) reflects lithogenic dust (natural and/or non-exhaust emission inputs), characterized by Ca, K, and κ. This interpretation is supported by positive correlations between Fe and crustal elements both in raw and detrended data (Supplementary Table 1), indicating that substantial fraction of Fe-bearing particles is associated with lithogenic dust, either natural and/or non-exhaust derived.

The third PCA component ( ~ 7%) is controlled by opposing loadings of S-ratio and HIRM, capturing variability in high-coercivity hematite/goethite contributions. Consistent with this, S-ratio shows negligible correlations with elemental concentrations after detrending but remains strongly anticorrelated with HIRM (Supplementary Table 3).

Together, our results indicate that the Helsinki aerosol archive contains a mixture of magnetic particle populations dominated by magnetite/maghemite from anthropogenic high-temperature sources, with additional contributions from lithogenic dust (natural/non-exhaust) and a smaller fraction of high-coercivity iron oxides. Relationships among PCA components, coercivity parameters and magnetic grain-size proxies further illuminate key controls in observed variability in the magnetic properties. Annual average of MDFIRM does not correlate with any component, implying complex mixtures within the magnetite/maghemite assemblage. In contrast, MDFARM correlates negatively with PC2 (R² ≈ 0.67; Supplementary Fig. 7), indicating that increased lithogenic dust (natural/non-exhaust) input corresponds to lower coercivity of the magnetite/maghemite fraction. Correlations between PC3 and κARM/κ and κARM/SIRM (Supplementary Fig. 7) suggest coupled variability among magnetic mineralogy (magnetite/maghemite vs. hematite/goethite) and grain-size of the magnetite/maghemite. Overall, Helsinki aerosol record reflects combustion intensity, lithogenic dust (natural/non-exhaust) input, and magnetite/maghemite domain‑state evolution.

Notably, the anthropogenic component derived from PCA analysis exhibits temporal covariance with Pb/V ratios (Supplementary Fig. 8), a geochemical proxy previously used to distinguish coal-dominated from oil-dominated combustion47. Higher Pb/V ratios and Component 1 scores during the 1960s suggest stronger coal-related emissions, whereas lower Pb/V values during later decades is interpreted to indicate the combined effects of decrease in coal use, effects of air quality regulations, and increased traffic. This archive-based record demonstrates the multi-decadal persistence of anthropogenic iron-oxide spherules and associated changes in their domain-state characteristics.

Wind-rose analysis for the Kaisaniemi filters (1968–2005) demonstrated that directional variations in κ are largely controlled by source location rather than meteorology (Supplementary Figs. 9 and 10). In 1968–1979, κ peaked under northeasterly winds (45°), reflecting emissions from the newly commissioned Hanasaari A (1960) and Hanasaari B (1974) coal-fired power plants (Supplementary Fig. 9), Sörnäinen harbor, and associated industrial and traffic activities. During 1980–1989, peaks shifted to easterly winds, highlighting the growing influence of eastern industrial and harbor sources in Helsinki and potentially episodic long-range transport from Eastern Europe and the former Soviet Union. During 1990–1999, elevated κ occurred under southwesterly ( ~ 225°) and northwesterly ( ~ 315°) winds, indicating a complex mix of Salmisaari coal-fired power plant emissions and increasing traffic contributions, respectively. By 2000–2005, overall κ declined, though easterly winds ( ~ 90°) still showed a maximum, consistent with reduced point-source emissions and a rising contribution from diffuse urban and traffic sources.

Tracing urban air quality transitions using magnetic parameters and trace elements

Magnetic PM shows clear temporal trends over the study period. Magnetic susceptibility (κ) and SIRM indicate that magnetite/maghemite particle concentrations in Helsinki were highest in the 1960s, declining by more than 50% by 1986. While SIRM remained low until 2005, κ exhibited fluctuations reflecting contributions from single-domain and multi-domain magnetite/maghemite grains32. Low-temperature magnetic measurements confirm an increasing SP fraction over time, with localized events, such as streetcar championship races48, in June 1995 and 1996 producing transient spikes in κ. Seasonal patterns show elevated magnetic particle concentrations during early winter to spring (November–April) (Supplementary Fig. 1) consistent with higher emissions from energy production and reduced atmospheric mixing49. To our knowledge, capturing a > 50% multi-decadal decline in magnetite/maghemite particles, along with resolving associated shifts in domain state from the same urban archive, is unprecedented.

Long-term meteorological changes in Helsinki are modest. Average annual precipitation increased by ~5% between the 1961–1990 and 1991–2020 climate normals (from 621 to 653 mm yr⁻¹), while summer–autumn precipitation remains roughly 200% and winter precipitation ~150% of spring values50. The 10-year moving average temperature was stable from 1960 to the mid-1980s and rose by ~1 °C thereafter due to urbanization and anthropogenic climate change50. These changes are small relative to the >50% decline in magnetite/maghemite particle concentrations, indicating that meteorology does not drive the observed long-term decadal trends.

S-ratio values, reflecting the relative contribution of high-coercivity hematite, declined from 1960 to 1990. After 1990 values stabilized at 0.97–0.98, reflecting that magnetite and/or maghemite dominate the magnetic mineral assemblage. Seasonal increases in hematite (October–March) likely reflect increased district heating, with minor contributions from abrasion of road surfaces and wintertime gritting practices. However, magnetite/maghemite dominates the magnetic signal due to higher susceptibility and remanence22,51. Hematite is a common product of combustion, for example in coal fly ash38. Winter road maintenance provides an additional source of hematite-bearing particles via the combined mechanical wear between studded tires, road, and the applied crushed local bedrock for traction (grit). Here the local bedrock is granitic, where hematite is a common accessory mineral. During winter, these anthropogenically derived lithogenic inputs are enhanced due to accumulation and reduced removal by precipitation, explaining the observed relative increase in high-coercivity hematite in urban dust. Importantly, post-collection oxidation of magnetite to hematite is unlikely to explain the observed temporal patterns, as such transformations proceed extremely slowly under ambient environmental conditions52,53,54 and the variability in S-ratio is most consistent with changes in emission sources and seasonal dust dynamics rather than mineral aging.

The observed temporal trends in κ reflect the impact of Finland’s energy transition, increase in vehicle numbers, and progressively stricter emission regulations from the 1960s to 2005 (Fig. 7). κ peaked in the 1960s, reflecting high emissions from coal- and heavy-oil combustion and the rapid expansion of traffic in Helsinki, a period during which household-scale heating dominated. The commissioning of the Hanasaari A (1960) and B (1974) coal-fired power plants shifted heating toward centralized district systems, allowing the decommissioning of the old Suvilahti (1909–1974; Sörnäinen) coal-fired power plant. In addition, the coal-fired Salmisaari A (1953) was later expanded with Salmisaari B (1984). This led to a steady decline in susceptibility, reaching less than half of 1960s levels by 1986–1987. From 1987 to 1992, κ rebounded, coinciding with a surge in vehicle numbers (exceeding 2 million by the late 1990s) and preceding the introduction of European vehicle emission standards. New vehicle registrations continued rapidly increasing after the early 1990s economic recession55, reflecting accelerated turnover of the vehicle fleet and the integration of improved vehicle technologies. Following the mid-1990s, κ stabilized despite continued traffic growth, reflecting the implementation of Euro 1–3 regulations and the transition to cleaner fuels. The commissioning of the Vuosaari natural gas–fired CHP plant (1991) and decommissioning of Hanasaari A (2000) further contributed to reduced particulate emissions. Overall, the temporal patterns in κ mirror the combined effects of district heating expansion, vehicle proliferation, fuel improvements, and progressively stricter air quality regulations. Comparing these trends with emission inventories from the Finnish Environment Institute56 show a shift in dominant PM10 sources from energy and industry in 1993 to a more balanced contribution from traffic and energy/industrial sources by 2000 (Supplementary Fig. 11).

Fig. 7: Long-term trends in magnetic susceptibility and emission-related indicators in Helsinki.Fig. 7: Long-term trends in magnetic susceptibility and emission-related indicators in Helsinki.

Time series of magnetic susceptibility (κ, black dashed line), lead-to-vanadium ratios (Pb/V, red solid line) and first vehicle registrations in Finland55, (bold and lilac solid line) from 1962 to 2005. Horizontal bars indicate operational periods of major local power plants and district heating facilities (Suvilahti, Hanasaari A and B, Salmisaari A and B, and Vuosaari natural gas plant). Vertical dashed lines mark the introduction of Euro 1, Euro 2, and Euro 3 vehicle emission standards. Star symbols indicate years for which low-temperature hysteresis measurements were conducted (1969, orange; 1996, purple). The figure illustrates the relationship between magnetic particle abundance, fuel combustion tracers, and major changes in energy production and traffic emissions.

Improved combustion technologies, mitigation measures, and stricter emission standards may explain the apparent shift in particle size distributions towards smaller, SP-dominated grain sizes, resulting from e.g., more complete fuel combustion. This shift would be reflected in decreased mass emissions, which are dominated by the largest particles. Similar trends have been observed when comparing older and newer vehicle fleets, where the generation of nucleation-mode nanoparticles ( < 0.02 μm) may even increase in newer fleets57.

These data reveal not only long-term reductions in emissions due to cleaner fuels and stricter regulations but also seasonal and interannual variability linked to energy use, traffic patterns, and meteorological conditions. Together, these observations demonstrate that long-term magnetic measurements can reliably reconstruct historical PM trends, distinguish anthropogenic sources, capture seasonal and interannual variability, and assess the effectiveness of air quality interventions, particularly in high-latitude urban environments. The direct alignment of magnetic source components and trace-metal proxies with dated policy and energy transitions indicates that magnetic archives can resolve regulatory imprints on iron-bearing PM that are not evident from mass trends alone.

Health implications of magnetic particulate matter

The outcomes of this study highlight health implications associated with magnetic PM in urban air, particularly in relation to the identified anthropogenic sources2,9,15. The presence of magnetic particles, such as magnetite and maghemite, is concerning due to their potential to induce oxidative stress and inflammation in the respiratory system13,14. Exposure to PM, especially fine and ultrafine particles, has been linked to a range of adverse health effects, including respiratory and cardiovascular diseases, lung cancer, and neurodegenerative disorders2,3,4,10,15,58,59. The observed seasonal patterns of magnetic PM concentrations, with peaks during winter months, further emphasize their relevance for health awareness. Increased emissions combined with stagnant atmospheric mixing conditions can lead to elevated pollutant levels and associated health risks, particularly for vulnerable populations such as children and the elderly. As urban areas continue to expand, it becomes increasingly essential to understand the health implications arising from anthropogenic sources to guide air quality management that effectively reduce exposure and protect public health. Recent legislative developments in the EU, particularly Directive (EU) 2024/288160 on ambient air quality, underscore this imperative by introducing reinforced standards and obligations. These include stricter limit values, average exposure reduction targets, and requirements for monitoring, modelling, and reporting, all grounded in the latest scientific evidence and WHO guidelines61. Aligning health impact assessments of PM sources with these regulatory benchmarks enables the translation of mechanistic and source-apportionment findings into actionable strategies. In particular, the magnetic iron-rich particles enriched in urban PM sources exhibit high oxidative potential and are mostly nanoscale, making them especially relevant when assessing compliance with evolving EU air quality criteria and prioritizing mitigation approaches.