Trac-related activities driven dust-bound magnetic pollution in urban road-system: Case study of Xiamen island, China

Sources of magnetic phases in urban dust include trac-related activities, industrial products and detrital minerals. Exposure to magnetic phases in urban dust co-associated with trace metals is associated with respiratory ailments in sensitive organisms. Here, we collected samples of surface roadside dust and topsoil from Xiamen island, Fujian, China and then magnetic properties were determined using rock-magnetic methods. Magnetic particle concentrations in surface roadside dust was found to be signicantly high compared with magnetic content in topsoil using magnetic susceptibility ( 𝜒 LF), saturation isothermal remanent magnetization (SIRM) and susceptibility of anhysteretic remanent magnetization ( 𝜒 ARM). Roadside dusts had much lower percent frequency-dependent susceptibility (1.83 ± 0.12%) than topsoil (2.96 ± 0.14%). Given the 𝜒 ARM/SIRM vs 𝜒 fd% and rst-order reversal curves, most of the magnetic particles in roadside dust were large than in topsoil. Variable temperature dependence of magnetization, isothermal remanent magnetization decomposition spectra and rst-order reversal curves suggested the enhancement of dust 𝜒 LF, SIRM and 𝜒 ARM were related to the exogenous pseudo-single-domain magnetite-like materials. This study provides a better understanding of magnetic ageing of dust particles in coastal cities. These results will be useful in optimizing environmental policies aimed at regulating magnetic dust particles pollution in coastal cities. Molspin pulse magnetizer and an AGICO JR6 molspin magnetometer at 1000 mT and − 300 mT. Classical percent frequency-dependent susceptibility (χfd%) and S − 300 were estimated using Eqs. (1) and (2). respectively. Basic interpretation of magnetic variables in urban dust have been given in (Muñoz et al., 2017) and are not stated in this study.


Introduction
Iron-containing minerals are widespread in soil, sediment and atmospheric dust in industrial environments (Cao et al., 2015;Szuszkiewicz et al., 2015), urban road-systems (Wang et al., 2019b;Aguilera et al., 2020;O'Shea et al., 2020), and dust-loaded leaves Cao et al., 2015). Iron of iron-containing minerals was substituted by other elements, e.g. Cu, Ni, Cr Pb or Zn during the aging of iron oxides (Aguilar Reyes et al., 2013;Jordanova et al., 2014;Švédová et al., 2020) and these iron oxides may include maghemite, magnetite, hematite, goethite and other iron oxides or α-iron. Variation in the abundance of those iron oxides in soil, sediment and atmospheric dusts leads to variation in magnetic signatures.
The magnetic particle sizes of primary magnetic minerals were mainly in the ultra-ne nano-sized superparamagnetic range and single domain grains (Reyes et al., 2011). The next most abundant magnetic minerals, including the products of coal combustion, industrial emission, and chemical formation are characterized by a low value of SIRM/ARM and χLF/ARM Cao et al., 2015;Zhao et al., 2020). Magnetic particles enrichment of trace metals has been found in the magnetic phases of the multi-domain range (Bourliva et al., 2016;Liu et al., 2019). Elevated concentrations of magnetic particles are related to tra c activities and industrial contributions (Aguilera et al., 2020;Zhao et al., 2020).
Exposure to magnetic particles co-associated with trace metals could cause severe adverse effects in sensitive organisms.
The occurrence and characteristics of iron-containing minerals and their environmental implication in topsoil, surface dust, and particles suspended in the air have been studied in Chinese cities e.g. Shanghai (Wang et al., 2019a;Wang et al., 2019b), Nanjing (Li et al., 2014;Leng et al., 2018;Wang et al., 2019c), Wuhan (Zhu et al., 2013) and Hokkien triangle, Fujian (Yang et al., 2020). Changes in magnetic concentration parameters, magnetic dependent grain-size and iron-oxide phases depend on pedogenesis and anthropogenic magnetic inputs. Polluted atmospheric particulates have a higher proportion of magnetic phase in the total magnetic mineral assemblage with coarse magnetic size (Zhao et al., 2020).
However, to our knowledge, how the magnetic behaviors of surface dust in main roadsides and in topsoil in park lawns respond to variation in anthropogenic activities has not been established for coastal cities. The main goal of this study was to understand the aging of iron-containing minerals in the surface roadside dust and park lawn top-soil at different sample sites.

Sampling locations
Xiamen, the special economic zone of southern Fujian, China, is situated in the mid-latitudinal belt between 24°23′ to 24°54′ N and 117°52′ to 118°26′ E. The average annual precipitation is 1200 mm, which is mainly concentrated in autumn. The average annual temperature is 20.9°C and there is no winter season in this coastal area. Total mileage of public roads was 1051.22 km in 2020 and 2200 km in 2018. Xiamen Island, the urban center of Xiamen City, has attracted more than 35.23 million trips in 2011 and 100.00 million trips in 2020. Population density of Xiamen Island was 13,957 people km − 2 in 2013 and 16,700 people km − 2 in 2017. The main sources of particulate matter in Xiamen include vehicle emissions and combustion of biomass, waste and soil (Zhao et al., 2011).
In July 2019, a total of 36 dust samples were collected from the roadside in Xiamen Island (Fig. 1). At each sampling point, four subsamples were obtained using a clean polyethylene brush at a comparable area of 1 m 2 and then homogenized in a polyethylene bag. Speci cally, the sampling time was selected after the ground had been dry for over 7 days (Wang et al., 2019a). A total of 81 soil samples (0-20 cm) were collected at 21 public gardens in Xiamen Island ( Fig. 1). At each sample point, three subsamples were obtained using a stainless-steel trowel.

Magnetic measurement
Mass magnetic susceptibility (χLF, χHF) was measured at two frequencies (470 and 4700 Hz) (Dearing, 1999) using a MS2B magnetic susceptibility meter (Bartington Led., UK). Susceptibility of anhysteretic remanent magnetization (χARM) was processed using a DTECH 2000 AF demagnetizer (ASC scienti c, USA) at a peak alternating feld of 100 mT and a biasing feld of 0.04 mT (Tan et al., 2018). Saturation isothermal remanent magnetization (SIRM) and IRM − 300 were separately acquired using an ASC Scienti c Molspin pulse magnetizer and an AGICO JR6 molspin magnetometer at 1000 mT and − 300 mT. Classical percent frequency-dependent susceptibility (χfd%) and S − 300 were estimated using Eqs. (1) and (2). respectively. Basic interpretation of magnetic variables in urban dust have been given in (Muñoz et al., 2017) and are not stated in this study. 1 2 High-temperature thermomagnetic curves were processed using a 34 mT direct current eld in a variable eld translation balance (Petersen Instrument, Germany). Samples were heated in air to 700 •C at a rate of 10 •C min − 1 , and then cooled to room temperature at the same rate. Isothermal remanent magnetization (IRM) curves and rst-order reversal curves (FORC) were measured using a MicroMag 3900 vibrating sample magnetometer (Lake Shore Cryotronics, Inc. USA). IRM decomposition spectra were analyzed using the software package IRMUNMIX V2.2 (Heslop et al., 2002) and IRM_CLG1.0 (Kruiver et al., 2001). The rst-order reversal curves were performed with the FORCinel software (Harrison and Feinberg, 2008) and the results displayed on a FORC diagram.
For low-temperature measurements, samples were given an IRM in a eld of 2.5 T at 300 K and cooled to 10 K in a eld of 0 T in a MPMS-XL7 magnetic property measurement system (Quantum Design, USA) at the State Key Laboratory of Physical Chemistry of Solid Surfaces, Xiamen University. Magnetization was recorded at intervals of 5 K heated to 300 K in zero eld. The temperature dependence of alternating current magnetization was measured in a peak amplitude of 0.3 mT at two frequencies (1 and 100 Hz) within the 10-300 K temperature range at intervals of 5 K.

Spectral measurement
Diffuse re ectance spectra were measured in a Lambda 1050 + UV/Vis/NIR spectrophotometer (PerkinElmer, USA) at Tan Kah Kee innovation laboratory, Xiamen University. The samples were recorded within the 380-710 nm wavelength range at intervals of 2 nm.

Distribution of magnetic particles in topsoil and surface dust
Magnetic concentration parameters χLF, χARM, and SIRM were signi cantly higher in surface roadside dust than in topsoil. χfd was signi cantly lower (40.78%) in surface roadside dust than in topsoil (Table 1). S − 300 , the relative importance of soft coercivity components and hard magnetic fractionations in total magnetic mineral assemblage, did not show any signi cant difference between surface roadside dust and topsoil. There was no signi cant difference in magnetic dependent grain-size, χARM/χLF and χARM/SIRM.

Variability of magnetic characteristics
All temperature dependence of magnetization ( Fig. 2a1-d1) have a magnetization peak around 510 ℃ and then display a major drop around 580 ℃. The magnetization in the cooling curves were much higher than in the heating curves between room temperature and 510 ℃. The temperature dependence of magnetization of topsoil (e.g. P81, P83) was characterized by a negative magnetization after exposure to 610 ℃.
IRM acquisition curves of topsoils (e.g. P81 and P83) were not saturated at 1000 mT ( Fig. 2a2-b2) and two coercitivecomponents were identi ed using the software packages IRMUNMIX V2.2 (Heslop et al., 2002) and IRM_CLG1.0 (Kruiver et al., 2001) (Fig. 2a3-b3, a4-b4). About 47.9 ~ 93.0 % of the magnetization was carried by the low coercivity component with median coercive eld (B 1/2 ) of 47.9 to 67.6 mT. A mineralogical phase with high coercivity component is characterized by the B 1/2 of about 104.7 to 638.7 mT and dispersion parameter (DP) of 0.12 to 0.43. All dust samples reach over 95% of saturation isothermal remanent magnetization by 0.3 T (Fig. 2c2-d2) and all coercivity components were likely subjected to the low coercivity family (Fig. 2c3-d3, c4-d4). In the low-temperature measurements, a continuous decay of magnetization was observed warming up to 300 K and no obvious magnetization change was identi ed around 120 K (Fig. 2a5-d5). Temperature dependence of in-phase magnetization depend on both temperature and frequency (Fig. 3a-d).
Coherently, all FORC diagrams have a central peak around the origin of the coercivity axis ( Fig. 4a-d) (Wang et al., 2019c). Here, SIRM/χLF of top-soils and roadside dusts was 12.25 ± 3.91 ⨯10 5 A m − 1 and 11.19 ± 2.13 ⨯10 5 A m − 1 , respectively. The B 1/2 of 47.9 to 67.6 mT and the DP of 0.41 to 0.42 in the IRM decomposition spectra (Fig. 2a3-d4) with signi cant magnetization loss around 580°C (Fig. 2a1-d1) suggested magnetic signing of the magnetite-like phase. During the aging of iron oxides, the magnetic signs of high-coercivity iron-oxides were overshadowed by low-coercivity components, especially the sample of low χLF (Yang et al., 2020). The presence of a highcoercivity hematite in the studied topsoils and roadside dusts was identi ed by the peak around 565 nm in the diffuse re ectance spectra (Fig. 3e-h) and a continuous magnetization loss above 610 •C in the temperature dependence of magnetization ( Fig. 2a1-d1). The peaks around 435 nm in the diffuse re ectance spectra ( Fig. 2e-h), the possibility of goethite cannot rule out. According to (Wang et al., 2019b), the irreversible heating curves and cooling curves are linked to the alteration of ferrimagnetic minerals warming up to 680°C. Here, magnetization in the cooling curve is higher than those in the heating curve at room temperature ( Fig. 2a1-d1) and re ect the thermal alteration of clay and organic matter during the heating processes. A minor magnetization loss near 300-450°C during the heating (Fig. 2a1-d1) and the presence of unstable maghemite (Wang et al., 2019b) was also identi ed.

The enhanced magnetic particles in surface roadside dust
Variations in LF, SIRM and ARM match with the proportion of ferrimagnetic phases, e.g. magnetite, maghemite and pyrrhotine in the total magnetic mineral assemblage. High values of LF and SIRM were related to the abundance of superparamagnetic (SP) and/or multi-domain magnetic phases. As observed in studies of mineralogical dust, the elevated dust LF, SIRM and ARM were associated with anthropogenic magnetic minerals, e.g. industrial activities and tra crelated sources (Cabanova et al., 2019). Aguilera et al. (2020) found that vehicle exhaust was the dominant magnetic source of street dusts in urban areas. The values of χLF varies among different types of dusts: 4603.73⨯10 −8 m 3 kg − 1 for industrial dusts (Szuszkiewicz et al., 2015), 1100 ± 320⨯10 −8 m 3 kg − 1 for subway dusts , and 35.6 ± 2.2 ⨯10 5 10 −8 m 3 kg − 1 for top-soils (Zhang et al., 2012). Here, the values of LF, SIRM and ARM in the roadside dust were much higher than those in the topsoil (Table 1). Similar magnetic signatures of street dust and topsoil have been recorded in Beijing, China (Zhang and Zhang, 2008), Ostrava, Czechia (Švédová et al., 2020) and West Bengal, India (Maity et al., 2020) and re ect the anthropogenic contribution of magnetic phases in the iron-containing minerals. A positive correlation with LF and SIRM (Fig. 3m) were linked to the ferrimagnets in the magnetic mineral assemblage. By combining temperature dependence of magnetization, isothermal remanent magnetization decomposition spectra and rst-order reversal curves, the magnetic phases of roadside dust were revealed to be mainly composed of oxidized magnetite of different magnetic grain size (Fig. 2). Little contribution of paramagnetic minerals was recorded in roadside dust as in-phase magnetization decreased upon cooling from 300 to 10 K (Fig. 3c-d). The presence of minor SP magnetic phases was evident as the dependence of in-phase magnetization on frequency and temperature ( Fig. 3c-d). Atmospheric particulates in pollution (Liu et al., 2019;Wang et al., 2019c), road-deposited sediments (Aguilera et al., 2020;Maity et al., 2020), street dusts (Wang et al., 2011;Wang, 2013;Zhu et al., 2013;Li et al., 2014;Wang et al., 2019b) and dust-loaded leaves (Cao et al., 2015) of low fd% have been recorded in other metropolitan sites. Here, χfd% was signi cantly lower in surface roadside dust than in park lawn top-soil and suggested the main contributions of the magnetic component of surface roadside dust were exogenous magnetic phases. Consistent with this, a weak relationship between LF with fd% (r = 0.05, P > 0.50) re ected a negligible contribution of ne pedogenic grains to the elevated LF. Across the studied area, all samples fall in the pseudo-single domain (PSD) range ( Fig. 4) with magnetic grains of ~ 1-5 µm (Fig. 3o). Such characteristics have been documented in Loudi (Zhang et al., 2012) and Beijing (Zhang and Zhang, 2008), China, which further suggests that enhanced LF, SIRM and ARM in roadside dust were linked to the local exogenous PSD magnetitelike minerals.

Environmental implications
In general, the coexistence of magnetic particles and heavy metals have been documented in atmospheric particulates, road-deposited sediments and urban street dusts (Aguilar Reyes et al., 2013;Leng et al., 2018;Aguilera et al., 2020). The hazard index of magnetic particles in pollution combined with iron and other heavy metals in Thessaloniki, Greece (Bourliva et al., 2016) and Yunnan, China (Tan et al., 2018) were either higher than or very close to the safe level. Evidently, Magnetic minerals in polluted dust enriched with heavy metals have been identi ed in Thessaloniki, Greece (Bourliva et al., 2016), Warsaw, Poland (Dytłow et al., 2019), Morelia, Mexico (Aguilar Reyes et al., 2013) and Ostrava, Czechia (O'Shea et al., 2020) using SEM-EDS. A signi cant amount of iron oxides with variable surface charge and high redox activity have been traced in the human brain (Dlháň et al., 2019). The elevated LF, SIRM and ARM in dust (Table 1) were related to the exogenous magnetic minerals and possibly to tra c-related activities. Abundance of magnetic minerals in dust were related to tra c ow, as evidenced by the strong correlation between tra c ow and the magnetic signals of deposited particulate matter (Muñoz et al., 2017). Magnetic particles in polluted dust including atmospheric particulate matter (Qiao et al., 2013;Warrier et al., 2014) and road dusts (Zhang and Zhang, 2008;Wang et al., 2019b), tend to be concentrated in the magnetic size of ~ 1-5 µm. Prolonged exposure to those magnetic particles could enhance the occurrence of respiratory ailments in sensitive organisms. Further work is needed to quantitate the occurrence of iron-bearing particles in polluted dust with elemental compositions between magnetic signatures and road type, tra c ow and population size in Xiamen Island.

Conclusions
Dust fd% was found between 0.57 and 4.10% and re ected a less-important contribution of lithogenic components in the total magnetic mineral assemblage. Analyses of temperature dependence of magnetization, isothermal remanent magnetization, and FORC suggested the elevated dust LF, SIRM and ARM were related to exogenous iron-containing magnetic particles. The type of magnetic minerals both in the topsoil and roadside dust is mainly magnetite-like minerals, with a minor amount of hematite. Consistent with the magnetic size range of ~ 1-5 µm in polluted dust from the coastal cities in Shanghai, China, prolonged exposure to these dusts could exacerbate respiratory ailments in the most sensitive organisms. These data will be useful to optimize environmental policies of air quality in the Xiamen Island, China. Table   Table 1 Statistical summary of magnetic properties LF (10-8 m3kg-1) ARM (10-8 m3kg-1) SIRM (10-6 Am2kg-1) fd% (Dimensionless) S-300 (Dimensionless) ARM/ LF (Dimensionless) ARM/SIRM (