Insights from the combined bulk chemical and surface characterization of airborne PM₁₀ on source contributions and health risk: the case of three Mexican cities

Abstract

The composition profiles of PM₁₀ were investigated using microscopy and spectroscopy techniques. A 24-h intensive, wintertime measurement campaign was conducted to collect filter samples using high-volume instruments. The samples were analyzed by X-ray photoelectron spectroscopy (XPS), inductively coupled plasma mass spectrometry (ICP-MS), scanning electron microscopy-energy dispersive X-ray spectroscopy (SEM-EDS), micro-Raman spectroscopy (MRS), and X-ray diffraction (XRD). The hazard quotient and incremental lifetime cancer risks were calculated to assess the health risks. The PM₁₀ concentrations ranged from 29.80 to 78.86 μg/m³. Fe, Zn, and Cu were the most abundant elements with average contents of 737.8 ± 258.69, 164.33 ± 64.73, and 83.31± 68.65 ng/m³, respectively. The highly toxic elements As, Cr, and Pb were found at concentrations that suggest potential health risks. The XPS revealed C, O, and Si to be the major elements in the PM₁₀ surface composition associated with aliphatic and aromatic hydrocarbons. The morphological analysis and PCA showed that the Fe-rich particles exhibited spheroidal morphologies, while the Pb-rich particles exhibited irregular and prismatic-type morphologies. These results suggest metal-work, lead-battery manufacturing, and the earth’s crust as emission sources. The cumulative risk of cancer suggests a moderate potential health risk that implies the need for preventive and corrective actions to improve air quality in urban cities. The combination of analytical techniques can be used to obtain better insights into the physical and chemical behavior of airborne particles in the atmosphere and potential tools to complement receptor models for source apportionment.

Introduction

Particulate matter (PM) is the most complex air pollutant and the one that is most strongly associated with mortality and morbidity events in the general population (Kumar et al. 2022, 2023). Recent studies have shown that chronic exposure to atmospheric particulate matter can have detrimental effects on lung tissue, vascular function, blood pressure, and survival rate in animal models (Xiao et al., 2020). Furthermore, PM can produce several toxic effects, depending on its size and chemical composition. Particles with aerodynamic diameters less than 10 μm (PM₁₀) can penetrate the tracheobronchial region, leading to various adverse effects on human health (Manigrasso et al., 2020). Moreover, these particles are associated with highly toxic species such as trace elements, elemental carbon (EC), and organic carbon (OC), including polycyclic aromatic hydrocarbons (PAHs). Trace elements (e.g., As, Pb, Cu, Cd, Ni, Cr, and Co) found in PM₁₀ have been widely studied because of their high toxicity, persistence, and biological accumulation (Bisht et al., 2022). Chemical elements such as Ni, As, and Cd and their compounds have been classified as carcinogenic to humans (class I) by the International Agency for Research on Cancer (IARC), while Pb, including inorganic lead compounds, and Co metal and soluble Co (II) salts have been classified as class 2A (IARC, 2012). Unfortunately, in many cases, the concentrations of heavy metals in the ambient air exceed the permissible limits, indicating environmental risk to human health. For instance, a global atmospheric arsenic emission inventory conducted with the Goddard Earth Observing System reported that the average airborne As concentrations over Chile and eastern China in 2005 were 8.34 and 5.63 ng/m³, respectively. In India, the airborne As concentrations in 2015 reached 4.57 ng/m³, which was higher than the yearly average level (Zhang et al., 2020). Thus, there are a clear needs to increase monitoring of metal pollutants in the ambient air, develop emission control technologies, establish strict regulations, increase emission inspections, and enforce regulatory compliance.

The elemental compositions of atmospheric particles in bulk and at the surface level differ because of the complex heterogeneous reactions that occur on the surface (Brostrøm et al. 2020). From a health perspective, this is relevant since it is the surface of the particle that interacts with body fluids and tissues following ingestion or inhalation (Atzei et al., 2016, 2014a). Several studies have shown the role of the particle surface composition in a variety of environmental processes, including atmospheric elimination, cloud condensation nuclei, and adsorption of potentially toxic substances, such as heavy metals and PAHs (Cubison et al., 2008; Dusek et al., 2006; Ochoa-Hueso et al., 2017). The chemical elements Fe, C, O, N, S, Al, Cu, Cd, Zn, Ca, Na, Si, Cl, and K have been detected on the surface of particulate matter using X-ray photoelectron spectroscopy (XPS) (Atzei et al., 2014a; González et al., 2017; Guascito et al., 2015; Xu et al., 2016). Huertas et al. (2012) indicated that C, O, K, and Si were the main elements found in samples of total suspended particles (TSP) collected in Madrid, with average percentages of 41.5, 34.7, 11.6, and 5.7%, respectively. The deconvolution of the elemental global signal of the chemical elements detected by XPS has been shown to allow the determination of the contribution of diverse functional groups (aromatic and aliphatic hydrocarbons, sulfates, sulfides, nitrates, and nitrites), adsorbed onto TSP surfaces, as the case of PM₁₀ air samples collected in Milan (Atzei et al., 2014a). Even though the relevance of characterizing the bulk and surface elemental properties of airborne PM is recognized, few studies have done so (e.g. Megido et al., 2016). Furthermore, potential health impacts and source contributions of airborne PM have rarely been assessed simultaneously.

It is essential, therefore, to carry out in-depth studies of the elemental composition of PM₁₀ at both the surface and bulk levels. Concerning emission sources, PM₁₀ and PM_10-bound metals have been found to originate from both natural processes and anthropogenic activities. Vehicular traffic plays an important role since both emissions and wear of vehicle components (brakes and tires) are closely related to the presence of heavy metals such as Pb, Cu, Cd, and Zn, as well as carbonaceous material and highly toxic organic compounds, including PAHs and their derivatives. The combustion of fossil fuels and industrial processes in general (e.g., industrial smelting and the petrochemical industry) has also been recognized as important contributors of these pollutants to urban air, especially in highly industrialized regions.

In Mexico, the air pollution problem in the northern states such as Nuevo Leon and Chihuahua increased because of rapid urban and industrial growth (SEMARNAT, 2012). The Monterrey metropolitan area (MMA) is located in the state of Nuevo Leon in the northeast of Mexico. This region has more than 5.6 million inhabitants (INEGI 2017) and is characterized by high levels of industrial and commercial activity, as the base of many national and international companies, which makes its economy the second largest in the country and one of the most important in Latin America. Various industries, such as cement, steel, glass, ceramics, chemical, textile, and petrochemicals, along with thousands of small- and medium-sized factories, are concentrated in the MMA. Furthermore, it has a fleet of approximately three million vehicles, 23% of which are diesel fuel-based vehicles (INEGI, 2017). The geographical characteristics of the MMA also contribute to high levels of air pollution, since the MMA is surrounded by mountain chains that prevent the dispersion of pollutants in the region. The poor quality of ambient air in the MMA has been a constant concern for local authorities and researchers. In this sense, various research groups have made an effort to characterize both chemically and morphologically the particles in suspension in this city, which has allowed the identification and quantification of different types of polluting species, such as PAHs, carboxylic acids, aldehydes, carbides, metals, elemental carbon, sulfates, quartz, nitrates, and carbonates (González et al. 2016, 2017; Mancilla et al. 2016, 2019, 2021; Longoria-Rodríguez et al. 2021). However, most of the studies have focused on coarse particles (TSP) and fine particles (PM_2.5). Furthermore, there have been no studies on the potential risks of metals linked to PM₁₀ particles, which is important given the level of toxicity of these pollutants for human health.

Interestingly, recent statistical health data obtained in 2020 shows that the state of Chihuahua, which includes both cities, CHI and CDJ, ranks third in cancer death rates in Mexico with 8.2 deaths per 10,000 inhabitants. The state of Nuevo Leon, where MAM is located, ranks sixth with 7.6 deaths, which is still above the average of 7.2 in Mexico (IARC, 2021; INEGI, 2022). In men aged 60 or older, records show the second leading cancer death rate of 4.79 among 10,000 men due to tracheal, lung, and bronchial cancers. Moreover, in women aged 60 or older, records show the first leading cancer death rate of 4.71 due to breast cancer among 10,000 women. This cancer death rate data becomes relevant as the annual number of days with exceeding permissible contaminant levels increases in the three cities studied herein.

Similarly, in cities such as Chihuahua (CHI) and Ciudad Juárez (CDJ), air pollution caused by PM₁₀ particles has become a serious problem, with concentrations exceeding current standards (SEMARNAT, 2012) during most of the year. Both cities are in the state of Chihuahua, in the north of Mexico, with CHI being the capital and CDJ the most important municipality. CHI has a population of 878,500 inhabitants and a vehicle fleet of 540,910 units and is greatly influenced by activities related to the mining-metallurgical industry, whereas CDJ, with a population of 1.4 million inhabitants and 640,900 vehicles, is characterized by the influence of the textile industry, in addition to mining activities. Many studies have shown that industrial emissions, particularly those from the metallurgical industry and mining activities, are among the main sources of metals in ambient air (Fernández-Camacho et al. 2012; Duan et al. 2021), so there is a need to investigate the elemental composition of PM₁₀ in both cities. Although several studies have been conducted in these two cities in the last two decades to characterize the chemical and morphological composition of PM₁₀ (Benin et al., 1999), (Porras et al., 2010), recent reports in this regard are lacking. Consequently, the environmental and health impacts of exposure to this pollutant are currently unknown.

Since knowing the chemical composition of atmospheric particles and their origin are key parameters for establishing mitigation strategies to control air quality and reduce adverse impacts, the main objectives of this research were to study the chemical characteristics of PM₁₀ samples collected in three cities in northern Mexico during the winter of 2020 using various state-of-the-art analytical techniques and identify the possible sources of emissions of metals linked to PM₁₀ by combining elemental profiles and sequential microanalysis (SEM-EDS/MRS) with statistical tools. The health risks (carcinogenic and non-carcinogenic) associated with inhalation of PM₁₀-bound heavy metals were also evaluated. It is important to highlight that in previous works, our research group has demonstrated the potential and scope of the sequential microanalysis method in the characterization of individual particles (Longoria et al. 2021). This methodology offers the advantage of relating the morphology with the elemental and molecular composition, allowing to establish with greater certainty the assignment of the emission sources of the particles. In the present work, we complement this methodology with a statistical analysis of the results obtained by ICP-MS to give greater support to the process of assigning emission sources. The results presented here will be useful to the respective government entities that deal with environmental pollution issues in implementing preventive and corrective actions to improve air quality in each region.

Materials and methods

Ambient PM₁₀ sampling

PM₁₀ sampling was conducted in three cities in México, MMA, CHI, and CDJ, which have distinct geographical and air pollution features in the northern region of Mexico. Samples were collected simultaneously for 24 h every 6 days in the three cities during the period from January to March 2020. For this purpose, high-volume samplers for PM₁₀ (model TE-310, TISH) and fiberglass filters (EPM 2000, 8″ × 10″) were employed. The filters were conditioned before and after sampling in a chamber under controlled humidity (<50%) and temperature (20–25 °C), in agreement with the Norm IO 2.1 USEPA (USEPA, 1999a). The mass of the PM₁₀ samples was determined according to the difference in weight before and after the collection using an analytical balance XS205 dual range from Mettler Toledo under controlled relative humidity. For each sample, the PM₁₀ concentration was calculated by dividing the collected material mass (g) by the total volume of sampled air (m³). In the MMA, PM₁₀ samples were collected at an urban site known as Santa Catarina (25°40′32.4″N 100°27′54.1″W). This region is characterized by a wide range of industrial activities, including ceramic plants, metal processing plants, and the chemical industry. It is also worth noting that Santa Catarina is located downwind of major emission sources located all around the MMA; thus, it receives both local emissions and emissions that were emitted elsewhere. In addition, Santa Catarina is considered an area with high vehicular traffic, especially heavy freight traffic. In CHI and CDJ, the samples were collected close to the Advanced Materials Research Center (28°42′57.5″N 106°07′59.5″W) and monitoring station Delfhic-sec (31°42′44″N 106°23′43″W), respectively. Both sampling sites are characterized as being urban areas with moderate industrial activity and light vehicular traffic. As previously mentioned, the main difference between these two cities is the type of industry that predominates. CHI is greatly influenced by the metallurgical and mining industry, while in CDJ, the textile and mining industries are most influential. The specific location of the monitoring site is shown in Fig. 1. Meteorological parameters such as temperature, relative humidity (RH), atmospheric pressure, solar radiation, precipitation, and wind speed recorded in the different sampling sites are shown in Table S1.

Fig. 1

Geographic location of the three study cities in Mexico

Determination of the elemental composition on the PM₁₀ surface

The surface composition of the sample PM₁₀ was analyzed by XPS using a Thermo Scientific Escalab 250Xi spectrometer with a six-channel (Channeltron) detector. A monochromatic Al Kα (1486.69 eV) X-ray source was used for the experiments. A piece of each filter (1 cm²) was cut and placed on the sample holder using carbon tape. Survey spectra were obtained over the 1100-eV range at a resolution of 1 eV per step, 100 ms per step, and 150-eV pass energy. All measurements were done under ultra-high vacuum conditions (10⁻¹⁰ Torr). High-resolution spectra were collected for species of interest at 0.1 eV per step, 100 ms per step, and 20.0-eV pass energy. The spectra were corrected using the C_1s signal (285 eV) as a reference. Then, the percentage of each element presented in the PM₁₀ was calculated with the corresponding software (Avantage Data System version 5.9, Thermo Scientific). In addition, the chemical speciation of C, S, N, and Ca was carried out by deconvolution of the signals from each element in the high-resolution XPS spectra. It is noteworthy that the atomic percentage of silicon was corrected using the method described by Atzei and Rossi (2004). The signal assignment was performed using information from the literature and the US National Institute of Standards and Technology (NIST) standards database (NIST, 2012) for reference.

Determination of elemental composition in the bulk of PM₁₀ using ICP-MS

All the chemical reagents used in the present investigation were of analytical grade. The solutions were prepared with Milli-Q water (resistivity of 18.2 MΩcm^-1), and all the materials used were carefully cleaned before use. The trace elements of toxicological interest in the PM₁₀ samples, such as Cd, As, Pb, Ni, Co, Cu, Cr, Fe, Mn, V, and Zn, were analyzed by ICP-MS following the US Environmental Protection Agency (USEPA) protocol, Compendium Method IO-3.5 (USEPA, 1999b). The concentrations of the 11 elements were determined using an inductively coupled plasma/mass spectrometer (ICP-MS, Agilent model 7800). The extraction of the metal was carried out following the methodology described by the USEPA, Compendium Method IO-3.1 (USEPA, 1999c). Briefly, strips of sample and filter blank, corresponding to 10% of the total area, were placed in a Teflon autoclave container with 20 mL of an acid solution (5.55% of HNO₃, 16.75% of HCl, and distilled water). Then, the samples and filter blank were digested at 180 °C for 40 min in a microwave digestion system, MARS (CEM). Both the samples and the filters were analyzed in duplicate. The concentration of the metals in the samples was determined by deducting the concentrations of metals from the filter blank and the reagent blank. Background correction was carried out according to the IO-3.1 and IO-3.5 methods of the USEPA. (U.S. EPA, 1999c; U.S. EPA, 1999b). Precision was determined by analyzing a standard metal solution 10 times, resulting in a relative average deviation of less than 5%. Excellent linear relationships were established from the calibration curves built in the selected range for each element, and correlation coefficients ≥ 0.9958 were obtained. To evaluate the accuracy of the method, samples were fortified by adding a known amount of metal to the filters with collected material. Recovery assays were carried out by fortifying filters containing the collected material with known concentrations of each target metal. The overall recoveries for the fortified samples ranged from 81.3 to 106.7%. The detection limits found were 0.70 ng/mL for Pb, 0.07 ng/mL for As, 0.34 ng/mL for Cr, 0.13 ng/mL for Cd, 0.19 ng/mL for Co, 0.2 ng/mL for V, 0.22 ng/mL for Mn, 0.95 ng/mL for Cu, 0.32 ng/mL for Ni, 5.40 ng/mL for Zn, and 0.27 ng/mL for Fe.

Morphological and sequential analysis of individual microparticles using SEM-EDS/MRS

The microparticles collected in the three cities were characterized individually using a sequential microanalysis methodology (Longoria-Rodríguez et al., 2021). The method comprises the sequential application of SEM-EDS and MRS on the same particle to establish relationships between morphology and the elemental and molecular composition of the particle. The study focused on the particles of some of the metals determined by ICP–MS. First, characterization of the morphology, particle size, and elemental composition was performed by SEM–EDS. Pools equivalent to four sampling weeks were made at each study site. The particles were mechanically removed from the filters without tearing off the glass fibers and then placed onto a copper grid divided into quadrants. Images were taken at low- and high-microscopy resolution. Low-resolution (<×5000) electronic scanning was performed using an SEM (JEOL JSM-6010PLUS/LA) under low-vacuum conditions at 30 Pa using an electron backscatter (JEOL EX94400T4L11 Dry SD detector). High magnification images (>×5000) were acquired using an SEM FEI Nova Nano SEM 200 at high vacuum conditions with a backscattered electron detector (INCA X-Sight EDS detector). Both instruments were operated at 15 kV. The atmospheric particles were then characterized using the MRS technique. The analysis was performed using a Horiba Scientific™ Lab RamH Evolution Raman microscope at 40 mW. The laser beam (532 nm) was focused on individual particles using an Olympus SLMPN ×50/0.35 NA microscope objective. Raman signals were acquired with 5 and 20% of the beam intensity using a coupled charge detector (CCD) at 220 K. Individual particle measurements were performed with an acquisition time of 5 s and 15 accumulations per spectrum. Calibration was conducted with a silicon standard, and spectra were processed with the LabSpec version 6 software (Horiba Scientific). The Raman spectra obtained from the samples were compared with the standards of the RRUFF database.

Statistical processing

Multivariate approaches were employed to facilitate the interpretation of elemental data obtained from the bulk of PM₁₀ in the three sampling cities for the following purposes: (i) detecting relationships between elemental compositions, (ii) understanding the distribution modes of the pollutants, and (iii) identifying possible emission sources. Pearson’s correlation coefficient (r) values were computed, and principal component analysis (PCA) was performed. The full dataset was divided into three matrix formats, in which the rows contained the metal profiles measured by ICP-MS (i.e., 8 elements for CDJ, 9 elements for CHI, and 10 elements for MMA), and the columns corresponded to 10, 11, and 12 PM₁₀ samples from CDJ, CHI, and MMA, respectively. Elemental concentration data were then mean-centered and autoscaled before PCA was conducted. Autoscaling made it possible to compensate for the variability ranges in calculating the loadings since the metals were determined at different concentration scales. All statistical tests were conducted using a 95% confidence interval (significance p < 0.05). All calculations for meaningful interpretations of the data were made using the Minitab® Statistical Software.

Health risk assessment

To assess the health risk by inhalation of PM_10-bound heavy metals, in this study, the daily dose for non-cancer risk (D_inh) was determined according to the method proposed by the USEPA. For this calculation, the following equation was used:

$${D}_{inh}=\left({C}_{\textrm{metal}}\times IR\times EF\times ED\right)/\left( BW\times AT\right)$$

(1)

In Eq. (1), C_metal is the metal concentration (mg/m³) in the PM₁₀ fraction, IR represents the volume of inhaled air by a person each day (unit: m³ day⁻¹), and ED is the exposure duration (6 years for children and 70 years for adults) (USEPA, 2001). EF is the exposure frequency (365 days/year). BW is the body weight (unit: kg), and AT is the average time in days (which is equal to EDx 365 days). In addition, the hazard quotient (HQ) was calculated for use in assessing the potential non-carcinogenic risks for individual metals (USEPA, 2009), using the following equation:

$$HQ={D}_{inh}/ RfD$$

(2)

where RfD is the reference dose of metal through inhalation (mg kg⁻¹ day⁻¹). An HQ less than 1 is considered to indicate a negligible risk, whereas an HQ greater than 1 is considered to indicate a potential health risk. The total hazard index (HI), which represents the potential health risks posed by multiple heavy metals, was calculated as the sum of the HQs of the individual trace elements for the whole observation period (HI ¼ Xn i¼1 HQ). Finally, the incremental lifetime cancer risk (ILCR) was calculated using the following equation:

$$ILCR= Dinh\times SF$$

(3)

where SF is the cancer slope factor of an individual trace element ([mg kg⁻¹ day⁻¹]⁻¹).

Analysis of the crystalline phases in PM₁₀

The characterization of the crystalline phases of PM₁₀ was carried out by XRD by employing a PANalytical Empyrean X-ray diffractometer with an X’Celerator ultra-fast detector based on real-time multiple-step technology with Bragg–Brentano geometry. The diffractometer was operated with Cu radiation (Kα = 1.5405 Å) at 45 kV and 40 mA. A representative amount of sample material was removed from the filter and deposited onto a zero-background holder made of amorphous silicon. The scans were performed in the 2Ɵ range from 5° to 90° with a step scan of 0.016° and 59 s per step. Semi-quantitative analyses of the crystalline phases in the samples were carried out by applying the Rietveld method, fitting the profile of the reflections obtained in the diffractograms with respect to the standards reported in the PDF+4 plus database of the International Center for Diffraction Data (ICDD). For data treatment, X’Pert High Score Plus software version 3.0.5 was used.

Results and discussion

PM₁₀ mass concentrations

Table S2 shows the statistical summary of PM₁₀ concentrations during the winter season in the three study sites. As expected, the average concentrations in the MMA (78.86 ± 25.11 μg/m³) were higher than those recorded in either CHI (43.76 ± 22.67 μg/m³) or in CDJ (29.80 ± 14.05 μg/m³). The daily concentration of the PM₁₀ in MMA exceeded the permissible limit of 75 μg/m³ established by the Mexican Official Norm NOM-025-SSA1-2014 (SSA, 2014) in six of the samples collected and was 1.57 times higher than the 24-h value permitted by the World Health Organization (WHO) and European Union (EU) standards (50 μg/m³) (EU, 2008; WHO, 2005). Although the concentration levels of PM₁₀ in CHI and CDJ did not exceed the maximum permissible limits, it is important to note that the concentrations recorded in CHI were very close to the maximum limit stipulated by the EU and WHO. The notable differences observed between the investigated cities could be attributed to the contribution of several factors, such as the higher vehicular traffic and intense anthropogenic activities that occur in MMA (e.g., industrial processes, extraction of minerals in quarries, burning of charcoal) and differences in the weather conditions and geographic characteristics of each site. As Table S1 shows, the meteorological parameters recorded in CHI and CDJ were similar; however, significant differences were observed with respect to MMA. This is important because several researchers have reported that weather factors have a certain impact on the concentration and composition of particles (Ghosh et al., 2018). In this sense, the wind speed in the MMA was 6.25 to 7.39 times higher than that in the other study sites, which possibly favored the resuspension of road dust and soil, contributing to an increase in the concentration of PM₁₀. On the other hand, it is possible that the high RH registered in the MMA favored the high levels of PM₁₀ in this region. Researchers have reported high concentrations of PM₁₀ under such conditions. For example, Ramirez et al. (2018) indicated that lower levels of solar radiation and higher humidity levels contributed to an increase in the concentration of PM₁₀ in samples collected in Bogota, Colombia. Similarly, the high concentrations of PM₁₀ in Beijing reported by Gao and Ji, (2018) were attributed to, among other factors, the highest RH levels occurring in the winter season. This could be related to a further increase in the formation of secondary species such as sulfates, nitrates, and some organic compounds (Lu et al., 2016). According to Yue et al. (2022), the concentration of sulfate in PM₁₀ was favored by RH values of greater than 40%. On the other hand, Rösch et al. (2017) reported that high RH correlates with increased particle formation due to nucleation processes. Furthermore, high RH favors reactions between reactive organic compounds and oxidizing species such as O₃ and hydroxyl radicals, leading to the formation of low-volatility species, which further condense to produce SOA (Pytel et al. 2022).

Determination of the elemental composition on the surface of PM₁₀

The XPS technique is a powerful tool for assessing the surface composition and chemical states of various elements in the matter (Atzei et al., 2014a). The XPS survey spectra of the particulate matter collected exhibited signals that could be assigned to C, O, Si, N, S, Na, Ca, Al, Zn, and Mg. Figure 2 presents the average atomic percentages of the elements identified in the PM₁₀ samples. The XPS analyses revealed that three elements, C, O, and Si, predominated in the samples from the three study sites; the elements N, S, Na, Ca, Al, Zn, and Mg were present in smaller proportions. The elements C, N, and S were also present at high concentrations in PM₁₀ samples collected in Milan (Italy) during wintertime (Atzei et al., 2014a). The average atomic percentage of C was notably higher in MMA, indicating a greater influence of carbonaceous material in this region. This is an essential issue to be addressed in researching particulate matter pollution because carbonaceous material plays a crucial role in atmospheric chemistry, radiative transfer, and precipitation, and it is therefore closely related to adverse health effects. Furthermore, the presence of carbonaceous species on the surface promotes the formation of SOA. For example, it has been shown that the surface carrying acidic species leads to enhanced SOA formation (Kumar et al., 2017). Biomass burning and industrial and vehicular emissions constitute the main anthropogenic sources of carbonaceous species in atmospheric particles. The high percentage of C in the MMA could be associated with vehicular and industrial emissions, as well as from charbroiling operations, an activity that takes place intensely in the area.

Fig. 2

Average atomic percentage of the elements identified on the surface of the PM₁₀ samples

To carry out the chemical speciation of some elements of interest (C, N, and Ca), high-resolution spectra were collected in the appropriate range of binding energy. Figure 3 shows, as an example, the high-resolution scan over the C_1s, N_1s, and Ca_2p regions. Multiple contributors to the C_1s signal were observed and were classified into eight types: carbide (283.6 eV), elemental (284 eV), aromatic (C=C, 284.7 eV), aliphatic (C-C, 285 eV), ether/alcohol (C-O, 286.5 eV), carbonyl (C=O, 288.3 eV), carbonate (289.5 eV), and aromatic (π-π*, 291.8 eV). XPS analysis was able to discriminate between elemental carbon and organic carbon species, making it possible to quantify the carbon-based compounds present on the PM₁₀ surface. These assignments can be verified with information from the literature (Atzei et al., 2014a, 2014b; Mustafi et al., 2010; Niu et al. 2020). Table 1 shows the area percentages for each component. In MMA and CDJ, aliphatic groups were the main components of the C_1s signal, followed by aromatic hydrocarbons, while in CHI, aliphatics and carbonates predominated. Aromatic signals suggest the presence of polycyclic aromatic hydrocarbons (PAHs) at the surface level in the particles, which are highly toxic compounds to human health and have been extensively linked to vehicle emissions and charcoal burning (Wang et al., 2012). It is important to highlight that food cooking operations, mainly those that use charcoal (e.g., meat charbroiling), which are conducted intensively in the MMA, could be contributing to the emissions of particles rich in C and compounds organic such as PAHs. The highest percentages of EC (16%) were recorded in the MMA, reflecting the impact of high vehicular traffic and the burning of charcoal in the region. In relation to carbides, the highest percentages were found in MMA and CDJ. Some authors have reported that carbides present in the atmosphere, particularly SiC, could be related to the wear of filters of diesel engines (Querol et al., 2012)

Fig. 3

Deconvolution of high-resolution spectra in the region of the a C_1s, b Ca_2p, and c N_1s signals from the PM₁₀ collected in the MMA and CHI

Table 1 Atomic percentage of the species contributing to the C_1s signal

Similar results for the N_1s signal were observed in all cities, with the amide/pyrrolic (399.7 eV) groups being the most abundant, followed by nitrates (405.8 eV) and nitrites (403.9 eV) (Table S3). Unlike the pyrrolic amide groups, which are generated by primary processes such as the burning of fossil fuels, the nitrates and nitrites in the particles are attributable mainly to the photochemical oxidation of gaseous precursors (NO_x) in the atmosphere. As shown in Table S3, nitrite ions predominated in MMA (8.8%). These ions are not stable under conditions of high solar radiation; therefore, the lower solar radiation recorded in MMA (Table S1) could explain the high percentages of these ions. However, in MMA, the deconvolution of the Ca_2p signal showed two contributors to the global signal, corresponding to carbonate (CO₃^2-) at a binding energy of 347 eV, which was the most abundant, and sulfate ions (SO₄^2-) at 348.3 eV. In contrast, in the cases of CHI and CDJ, the speciation of the Ca_2p signal showed a single component corresponding to carbonate.

Elemental composition of the PM₁₀ fraction, emission sources, and health risks

Trace elements in PM₁₀

The average concentrations of As, Cd, Pb, Ni, Co, Cu, Cr, Fe, Mn, V, and Zn in the PM₁₀ samples are shown in Table 2. Most of the elements were detected in the samples from all three monitoring cities, except for Co and Mn in CDJ and Mn in CHI. The most abundant elements in MMA were Fe, Zn, and Cu, with average contents of 737.8 ± 258.69, 164.33 ± 64.73, and 83.31± 68.65 ng/m³, respectively. As, Cd, and Co had the lowest concentrations in MMA. These results are consistent with those of other researchers in a study conducted in Asturias (Spain), where Fe and Cu were found to be the most abundant metals in PM₁₀ samples (Megido et al., 2017). Similar results were obtained for some metals in CHI and CDJ. For instance, Zn, Fe, and Cu were the most abundant elements, while Ni and Cd were present in lower concentrations in both cities. The concentrations of Zn in CHI (52,282.6 ± 6681.9 ng/m³) and CDJ (30,435.6 ± 28,035.15 ng/m³) were notably higher than that in MMA (164.3 ± 64.7 ng/m³). Notably, the state of Chihuahua has historically been recognized as an important mining center where metals such as Zn and Fe are intensively exploited, which could explain the concentrations of these metals in both cities. In addition, the state currently ranks second nationally in Zn production and fourth in Fe production, so these results were expected. The Zn concentrations in the three study cities were notably higher than those found by other researchers in Shiraz, Iran (Parvizimehr et al., 2020), Düzce, Turkey (Bozkurt et al., 2018), Elche, Spain (Galindo et al., 2018), and Xiamen, China (Wu et al., 2020), which ranged from 66.3 to 116.1 ng/m³.

Table 2 Average concentrations of trace metals (ng/m³) in samples PM₁₀ collected in the three sampling cities

As shown in Table 2, high concentrations of As and Cr were recorded in CHI and CDJ, contrasting with the low levels found in MMA. The average arsenic contents in CHI and CDJ were 4.25 ± 0.49 ng/m³ and 4.23 ± 0.48 ng/m³, respectively. Several government agencies have regulated the maximum allowable concentration of pollutants in ambient air to protect the population from its potentially harmful effects. In this regard, the EU has established a limit of 6 ng/m³ for As in ambient air. However, the allowable limit for avoiding a potential risk of cancer is stricter, with the WHO recommending that the mass concentration of carcinogenic metals in ambient air over any region not exceed values of 0.6 ng/m³ (Gandhi et al., 2021; WHO, 2016). The mean concentrations of As at the three sampling sites were below the limits established by the EU but exceeded the limits suggested by the WHO. The As levels were higher than those found in Guangxi, China (Guo et al., 2018), Sao Paulo, Brazil (Pereira et al., 2017), and Paldi, Ahmedabad, India (Gandhi and Khatri, 2021) and lower than those reported in Welgegund, South Africa (Venter et al., 2017), Agra, India (Sah et al., 2019) and Algiers, Algeria (Talbi et al., 2018). Regarding Cr, the total concentration of this highly toxic metal (contribution of all chromium species) recorded in CHI (10.93 ± 2.44 ng/m³) was twice that observed in CDJ (5.49 ± 4.31 ng/m³) and was below the typical ambient-air concentrations reported for urban centers in the US (20 ng/m³) (USEPA, 2009). Chromium species monitoring in ambient air is essential for health since its toxicity level depends on the oxidation state, with Cr (VI) being the most dangerous form for humans (Liu et al., 2022). Prolonged exposure to Cr, specifically to Cr (VI), is known to determine the impairment in blood homeostasis due to oxidative stress in the body, affecting iron (Fe) metabolism and promoting renal damage (Di Vaio et al., 2018; Nair et al., 2013). Consequently, this metal has been categorized as carcinogenic for humans, in addition to being considered a mutagenic substance via inhalation. Cr (VI) concentration determined in this work was measured according to the US EPA protocols (USEPA, 2013), i.e., the concentration of Cr (VI) was 1/7 of the chromium total concentration. As an outcome, the average concentrations of Cr (VI) in CHI, CDJ, and MMA were 1.56 ng/m³, 0.78 ng/m³, and 0.27 ng/m³, respectively. The Cr (VI) values in the three cities were below those reported by (Parvizimehr et al., 2020) (2.97 ±1.90 ng/m³) in a study conducted in Shiraz, Iran. The authors attributed the high concentrations of hexavalent chromium primarily to industrial sources, as well as the burning of fuels for heating purposes and vehicle emissions.

The average concentration of Pb in MMA (56.68 ng/m³) was significantly higher than those found in CHI (1.91 ng/m³) and CDJ (0.15 ng/m³). This heavy metal has been shown to be very toxic since it can affect most of the organs and systems of the human body and can damage the central nervous system, so the same effects are suffered whether it is ingested or inhaled. In fact, Pb is considered by the IARC (International Agency for Research on Cancer) as a Group 1 carcinogenic element for humans (Jaafari et al., 2020). The Pb concentrations obtained in this study were significantly higher than those reported by Pereira et al., (2017) for Sao Paulo, Brazil (42 ng/m³), by Siudek, (2021) for Gdynia, Poland (1.96 ng/m³), and by Contini et al., (2014) for Lecce, Italy (10 ng/m³). In contrast, they were considerably lower than those reported by Gandhi and Khatri (2021) for Bapunagar, Ahmedabad, India (218. ng/m³), by Fernández Espinosa and Rossini Oliva (2006) for Seville, Spain (63.7 ng/m³), and by Duong and Lee (2011) for Ulsan, Korea (82 ng/m³).

The levels of Mn, Ni, and V in CDJ and CHI were lower than those in MMA. Co was also detected in MMA at an average concentration of 0.20 ng/m³. The overall concentrations of the other metals were in the ranges of 0.35–1.88 ng/m³ for V, 0.04–1.44 ng/m³ for Ni, and 2.60–19.49 ng/m³ for Mn. Bozkurt et al. (2018) and Guo et al. (2018) reported higher Co levels in Düzce, Turkey, and Guangxi, China, respectively, than those obtained in this study. Mn values similar to those obtained in this study were measured in a semi-urban region (Bolpur) of West Bengal, India, but those values differed greatly from values reported for urban (Berhampore) and industrial (Durgapur) regions of West Bengal, India (Ghosh et al., 2018). The concentrations of Ni and V reported by Koukoulakis et al., (2019) for the city of Elefsis, a heavily industrialized area of Greece, were significantly higher than those obtained for all three cities in this study. V, Ni, Mn, and Co have been associated with anthropogenic sources such as petrochemical combustion and refining and natural gas extraction and processing (Gharib and Al Sarawi 2018; Liu et al. 2018; Koukoulakis et al. 2019). In this study, the results obtained from the trace element data statistical analyses (PCA and PC) were matched with those from the morphological analysis in an effort to identify the possible sources of metal emissions reliably.

Morphological analysis of individual microparticles using SEM-EDS/MRS

SEM-EDS analysis focused on particles rich in the most abundant metals found in the three urban centers. Figure 4 shows micrographs with their respective EDS spectra of particles rich in elements such as Fe, Cu, and Zn. One group presented regular morphologies with certain symmetry and cleavage, typical of crystalline materials, commonly associated with natural sources of crustal origin. Another group was characterized by its spherical or spheroidal morphology, which is related to the condensation of vapors from anthropogenic sources (Aragon-Piña, 2014). In PM₁₀, particles with irregular morphologies were also observed. These are possibly associated with emissions from industries such as metalworking or the wear of automobile parts.

Fig. 4

Micrographs: A, C, E Fe-rich particles collected in CDJ, CHI and MMA, respectively. B, D, F Zn- and Cu-rich particle collected in CDJ, CHI and MMA, respectively

In CHI and CDJ, Fe-rich particles with irregular morphologies predominated. EDS analysis performed here revealed that Fe was the most abundant element in this group of particles, while other elements such as Ni, Cr, C, and O (Fig. 4a, c) were found in smaller proportions. Based on its morphology and composition, this group of particles could be associated with the wear of metal parts. Furthermore, in both cities, Fe-rich particles with regular morphology were observed, which could be related to the resuspension of soil dust. In contrast, the morphologies that predominated for the Fe-rich particles in the MMA were spherical and spheroidal (Fig. 4e). These particles were composed of elements in smaller proportions such as Al, Si, and C. It has been reported that Fe particles with spherical morphologies can acquire different surface textures, depending on the type of impurities and the formation conditions. Particles with corrugated surfaces, such as those found in MMA have been mainly associated with emissions from the metalworking and steel industries (Aragon-Piña, 2014, Gonzalez et al., 2017).

It is interesting to observe that the particles rich in Zn and Cu found in CHI and CDJ presented similar characteristics, which suggests that they come from similar sources. These particles showed a variable content of Pb and Al (Fig. 4b, d) linked to regular morphologies. Based on studies carried out with fixed sources, these types of particles have been associated with mining–metallurgical processes in Zn-Cu-Pb refineries (Aragón 2014). Particularly for MMA, some Cu-rich particles with spherical morphologies were found that could have been formed by the condensation of vapors from emissions from the metal fabrication industry. Particles with similar characteristics were reported by Labrada et al. (2012) and were attributed to this type of source. However, some researchers have also attributed them to vehicle emissions (Fig. 4f).

Sequential microanalysis using SEM-EDS and MRS

To delve further into the possible sources of PM₁₀ emissions, a microanalysis assay was carried out on the same particles sequentially using the SEM-EDS and MRS techniques, following the procedure described in previous work (González et al., 2018; Longoria-Rodríguez et al., 2021). The most relevant results of this study are presented below for illustration.

Figure 5 shows the results of the sequential microanalysis performed on a Cu- and Zn-rich particle collected in CDJ. The particle presented an irregular morphology with a size of 4.1 μm (Fig. 5a). EDS analysis revealed that the particle contained major elements such as Cu and Zn, as well as minor elements such as C and O. C may be present in the particle or may represent a surface coating. González et al. (2018) reported that most metallic particles have a coating of C and O that can mask the nucleus of the particle. The bands of the Raman spectra coincided with the bands present in the RRUFF database standards No. R060027 and No. R060071 for ZnO and CuO, respectively (Fig. 5d). Based on their characteristics, this group of particles could be related to anthropogenic sources from the mining–metallurgical refinement of zinc. These results are consistent with the characteristics of the study site, which demonstrates the importance of microanalysis in assigning emission sources.

Fig. 5

Optical and electronic micrographs of a Cu and Zn-rich particle collected in CDJ with the respective Raman and EDS spectra. EDS, energy dispersive spectroscopy

In CHI, a group of Cu-rich particles presented irregular morphologies and sizes between 3 and 9 μm. Figure 6 shows the results of the sequential microanalysis of a particle rich in this metal. The elemental composition of the irregular Cu-rich particles found in CHI differs from that in CDJ. In CHI, the Cu-rich particles were characterized by containing elements of crustal origin, such as Al and Si. The optical micrograph and the Raman spectra derived from the MRS analysis of the particle are shown in Fig. 6c, d, respectively. The Raman spectrum revealed the presence of bands that coincide with those of the RRUFF standard no. R050168, corresponding to the mineral kinoite (Ca₂Cu₂Si₃O₁₀·2H₂O), which has been found in the southern USA and has been associated with other minerals such as calcite, quartz, and copper. Therefore, this group of particles could be related to the resuspension of soil dust or natural sources.

Fig. 6

Optical and electronic micrographs of a Cu-rich particle collected in CHI with the respective Raman and EDS spectra. EDS, energy dispersive spectroscopy

In relation to Pb, Figure S1 shows the micrograph obtained from the microanalysis performed for a particle found in MMA. As can be seen, the particle was characterized as having an irregular morphology and a size of 11 μm. In addition to Pb, it was constituted by Sb, C, and O (Figure S1b). The bands observed in the Raman spectrum coincided with those of the RRUFF standards No. R060959 and R050389, belonging to PbO and Sb₂O, respectively. In the literature, both oxides have been associated with anthropogenic sources such as the battery manufacturing industry (Schreck et al., 2012). Unlike MMA, the Pb-rich particles collected in CHI presented prismatic-type morphologies indicative of the influence of natural sources. An example of this group of particles is presented in Figure S2. The EDS spectrum showed the presence of other elements such as C and O, in addition to Pb. Additionally, the bands that appear in the Raman spectrum (Fig. S2d) coincided with those of the RRUFF standard No. R070501, corresponding to the mineral Pb₂O(CO₃). This mineral may be associated with minerals that are related to the decomposition of PbCO₃ to PbO. (Rivovichev, 2000). Finally, it is important to highlight that the characterization of PM₁₀ through the microanalysis methodology allowed the speciation of particles rich in different elements in the three urban centers, which is very useful in identifying emission sources.

Multivariate data analysis

Pearson’s correlation coefficient

Pearson’s correlation coefficient (r) was calculated to evaluate the strengths of the correlations between the elements identified in the PM₁₀ samples from the three sampling cities. The results shown in Table 3 were interpreted following guidelines established by Campbell (Younes et al., 2021). Absolute values of Pearson’s correlation coefficient are used to judge correlations as very weak (0–0.19), weak (0.2–0.39), moderate (0.40–0.59), strong (0.6–0.79), or very strong (0.8–1). In this study, only strong and very strong correlations were considered to facilitate an understanding of the statistical analysis.

Table 3 Pearson’s correlation matrix for the element concentrations found in the PM₁₀ samples

A high correlation between Cr and Zn (0.749) was found in CDJ, suggesting that they come from the same source. According to the literature, these elements could be associated with industrial processes like those carried out in metal-mechanical industries (Manoli et al., 2002; Mooibroek et al., 2011; Police et al., 2016). Nevertheless, other authors relate them to the burning of fossil fuels (Hedberg et al., 2005, Song et al., 2014). Although As showed a moderate correlation with V (0.522), this metal is a carcinogenic tracer recognized by the WHO that requires special attention. Thus, the main source of As could be the resuspension of contaminated soil dust. In this regard, Porrás et al. (2010) reported high concentrations of As in samples of urban dust of Ciudad Juárez, which were attributed to emissions from the smelter plant located in the Mexico-USA border area. Interestingly, this plant was closed a few decades ago but was operational for a long time (100 years), thus contributing to the continuous release of As into the atmosphere. On the other hand, the presence of As in the ambient air could also be due to mining activities (USEPA 2010).

In CHI, As was strongly correlated with Cr (0.620) and Zn (0.760), and Cd was strongly associated with Fe (0.703), Mn (0.718), Pb (0.639), and V (0.648). Very strong correlations between Fe and Mn (0.959), Fe and Pb (0.861), Fe and V (0.968), Pb and V (0.824), Mn and Pb (0.864), and Mn and V (0.963) were found. The findings of other studies suggest that As and Cr can come from industrial sources, including metal smelting, iron and steel product manufacturing, and coal burning and traffic exhaust (Chen et al., 2016; González-Castanedo et al., 2014). Interestingly, the state of Chihuahua has an important industrial complex where smelting activities predominate, which could be the main source of emissions of these metals in the ambient air. For instance, Fernandez et al. (2012) reported relatively high concentrations of As, Cr, Pb, and Zn in PM₁₀ samples collected from an industrial region in Huelva, Spain, and attributed them to emissions from a Cu smelter plant, oil refinery, and a petrochemical plant. With respect to Fe and Mn, their common source could be the resuspension of soil. In this sense, particles with regular morphologies rich in Fe observed in CHI showed the influence of this type of source; however, industrial and vehicular emissions could also contribute. On the other hand, it has been reported that Cr and Pb can be derived from industrial sources such as the metallurgical and ceramic industries (Nagar et al., 2014), whereas As, Cu, Ni, and Zn could be associated with the burning of coal and fuel oil (Manoli et al., 2002; Querol et al., 2009). As for V, this element has been recognized as a tracer of oil combustion processes ((Shafer et al., 2012). In this sense, the relatively high concentrations of V reported by (Li et al., 2018) in PM₁₀ samples from an industrialized area of southern Spain were associated with oil refining and combustion. Lastly, the anthropogenic Cd emissions have been predominantly derived from industrial activities. In this sense, (Yuan et al., 2019) stated that pigment production in southeast China was the largest source of Cd emissions in the atmosphere, contributing to a lesser extent to ambient air pollution by wastewater treatment, aquaculture, and animal breeding. According to the German Environment Agency (Umweltbundesamt, UBA), this heavy metal can also be emitted by combustion processes based on energy production or by refineries, as well as the combustion of fuels in traffic.

In the case of MMA, strong correlations of As with Pb, of Co with Fe and Mn, of Cr with Mn and Ni, and of Cu with Pb were obtained in the range 0.611–0.751, while relations between Co and Cr, Fe and Mn, Fe and V, and Mn and V showed very high correlation coefficients ranging from 0.803 to 0.924, indicating that these elements could have come from the same source. Hence, Cu and Fe could come from anthropogenic sources such as industrial processes (Chen et al., 2021; Koukoulakis et al., 2019), as well as non-exhaust (brake wear) and exhaust emissions (Gao et al., 2017; Jaafari et al., 2020). However, natural sources such as soil resuspension could also contribute to emissions of these elements (González et al., 2017; Siudek, 2021). Particularly, Pb can be associated with industries manufacturing lead batteries located around the sampling site. It is important to highlight that the morphological characteristics of the Pb-rich particles observed through the application of microanalysis (SEM-ED/SRM) evidenced the contribution of sources related to the manufacture of Pb batteries. Besides, the activities of the metallurgical industry also play an important role in the emissions of this heavy metal (Jaafari et al., 2020; Li et al., 2018; Megido et al., 2017; Siudek, 2021). In contrast, Co, Mn, Ni, and V have been mainly associated with anthropogenic sources, such as petrochemical combustion and refining and natural gas extraction and processing (Gharib and Al Sarawi, 2018; Koukoulakis et al., 2019; Querol et al., 2009, 2007). In fact, a major state-owned oil refinery (25°36′02.4″N 99°59′42.1″W, 323 m.a.s.l.) is located 20 km from the current outer boundary of the MMA. The emissions from this refinery could be transported to this region by wind action. It is worth noting that this refinery has long been under environmental scrutiny because of non-compliance with regulations on ambient air emissions.

Principal component analysis

PCA was carried out to assign the possible sources of metal emissions more accurately. Elemental concentrations obtained for the three sampling cities were treated as input variables. PCA transforms the input variables into a series of new variables, called principal components (PCs), which are reduced to a minimum, maintaining the information in the original data without loss of variance. Each PC is an orthogonal combination of the original variables, and the feature decomposition method is widely used to calculate them. The maximum variance from the dataset is represented by the first PC, while each successive PC represents the accumulated variance unaccounted for by the previous PCs (Bro and Smilde, 2014). In this study, the cumulative percentage (%) value of the total variance was used to describe the amount of information represented by the number of PCs chosen. The results obtained for PCA loadings are shown in Table S4. As the results show, 100% of the total variance was represented by the contributions of PC1–PC8 at all three sampling sites, and the first three PCs explained more than 75% of the variability of the data. Therefore, these PCs were selected to assess the emission sources.

In CDJ, Cr, V, and Zn were associated with PC1, indicating that these metals could be derived from the emissions generated by industrial processes such as the activities of the mining–metallurgical industry (Police et al., 2016). The results obtained from SEM-EDS/MRS for the Zn-rich particles in CDJ revealed the contribution of this type of source. PC2 can be attributed to the burning of fuel oil and coal (Manoli et al., 2002; Querol et al., 2009) because of its high loading with Ni. PC3 is strongly associated with Pb, followed by Cd and Cu. Thus, this component can be attributed mainly to the metallurgical industry (Jaajari et al. 2020, Siudek et al. 2021), industrial processes (Koloulekis et al. 2019; Chen et al. 2021), and non-exhaust sources (brake wear).

In CHI, it was observed that PC1 was characterized by Fe, Mn, and V, while Cr and Zn were related to PC2. These components could be attributed both to the refining and combustion processes (Li et al., 2018; Saraga et al., 2017) and natural sources like the resuspension of soil (Gonzalez et al., 2017; Siudek et al., 2021). They can also be associated with non-exhaust (brake wear) and exhaust emissions (Jaafari et al., 2020, Gao et al., 2017) and industrial processes (Koloulekis et al., 2019, Chen et al., 2021). In addition, mining activities could be the main source of Fe and Zn emissions in the state of Chihuahua. Instead, PC3 was strongly represented by Cu and to a lesser degree by Ni and Cd, suggesting that these metals could be mainly emitted by mining activities and burning of coal and fuel oil (Querol et al., 2009), as well as from industrial activities (Yuan et al., 2019).

In MMA, PC1 was associated with Cr, Mn, and Co, which could be attributed to the petrochemical industry. These results are in agreement with those obtained by Pearson’s correlation (Gharib and Al Sarawi 2018, Koukoulakis et al., 2019). As and Cu dominated in PC2. According to the Cu-rich particle morphological analysis findings, this element group is related to emissions from the metalworking industry or vehicular sources. PC3 was associated with Pb, indicating that Pb in emissions could arise from the ceramic and glass industries, as well as the manufacture of lead-based batteries that takes place in the region.

PCA was also applied to the entire trace element dataset to explore the natural distribution of PM₁₀ samples in a reduced dimensional space. The projection for PC2 versus PC1 is illustrated in Fig. 7, which shows a clear clustering trend among the samples from the different cities monitored. The first two extracted PCs explained 69.0% of the total data variance (57.0% by PC1 and 12.0% by PC2), which is remarkable, according to findings reported by other researchers (González et al., 2017, Ghosh et al., 2018, Koukoulakis et al., 2019). Samples from CDJ showed mostly positive scores on PC2. In contrast, samples from CHI exhibited mainly negative scores on PC2, so a relative grouping between some samples from both sites is evidenced to the left of the coordinate axis (0; 0). The latter indicates a similarity between the trace element compositions found in both cities. In turn, scores from MMA were scattered throughout the vector space, represented by positive values of PC1. This score group showed a clear separation from the remaining samples because the metal pollutants representing this region were probably emitted by one or more different emission sources, as detailed above.

Fig. 7

Scores projected in the vector space PC2–PC1 obtained by analyzing the metal concentration data

The importance of elemental predictors for differentiation of the three sampling cities according to emission sources is shown in Fig. 8. Considering the importance value greater than or equal to 0.05, Pb was the metal exerting greater influence on the identification of sources in MMA, followed by Cu, Fe, Cr, Zn, and Co. This last element was only detected in MMA, so it was also important for the estimation of emission sources. In turn, the most important elements to attribute the sources at the CDJ and CHI sites were Fe, Zn, Cu, and Cr. The presence of Cu, Fe, and Zn at greater concentrations than other metals could explain these findings. Mining activities conducted in the state of Chihuahua could be polluting the atmosphere by releasing particles of these elements during mineral extraction and refinement processes. Consequently, this could be its main source of emission. It is interesting to note that these elements were also found to be the most abundant in the three sampling cities according to the SEM-EDS/MRS analysis.

Fig. 8

Importance of the elemental predictors for discriminating the sampling cities

according to metal emission sources

Overall, the combination of the PCA results and the morphological study results made it possible to differentiate the three Mexican cities monitored according to the trace element concentrations detected in the PM₁₀ samples. This allowed us to identify the emission sources of these pollutants reliably and estimate the pollution level in the urban atmosphere of each city studied.

Health risk assessment for metals associated with PM₁₀

Non-carcinogenic risk

The D_inh, HQ, and HI results for elements in PM₁₀ via inhalation are shown in Table 4. The results show that Mn had the highest HQ value in MMA, with an average of 2.31 × 10⁻¹. Mn is an essential trace element required for various metabolic processes; however, excessive exposure to Mn can cause multiple types of systematic damage to the body. Its potential harmful effects on multiple organs include slower body weight gain, increased relative liver and spleen weight indices (Zhu et al. 2020), and damage to the cardiovascular, reproductive, immune, and central nervous systems (Roels et al. 2012). High levels of Mn can also decrease monoamine oxidase activity and inhibit the respiratory chain. Besides, Mn can be accumulated in mitochondria inhibiting the efflux of calcium, which affects mitochondrial function in both males and females (Zhang et al. 2003). The health effect of all elements decreased in the following order: (i) MMA: Mn > Cr > Co > Pb > Cu > Fe > Cd > Zn > As > Ni, (ii) CHI: Cr > Mn > Zn > Cu > As > Pb > Fe > Cd > Ni, (iii) CDJ: Cr > Zn > As >Fe > Cu > Cd > Pb > Ni. A comparison of the HQ values of Pb, Cd, and Ni for the three urban centers shows that the highest HQ values were recorded in the MMA, with values of 3.70 × 10⁻³, 1.62 × 10⁻⁴, and 1.22 × 10⁻⁵, respectively, suggesting that these highly toxic elements could represent a greater health risk (not carcinogenic) for the region inhabitants. HQ for Pb in MMA was similar to that reported in Shiraz, Iran (Parvizimehr et al. 2020), but significantly higher than that reported in localities from West Bengal, India (Goudarzi et al. 2018). In contrast, it was lower than that reported in Seoul, Korea (Roy et al. 2019). All of the HQ values obtained in this study were less than 1, which indicates that the health risks of exposure to non-carcinogenic PM₁₀-bound trace metals through inhalation were acceptable in the three cities monitored. The highest mean non-carcinogenic hazard index (HI) was calculated for MMA (0.252), followed by those for CHI (0.122) and CDJ (0.047), which were below the safe limit.

Table 4 Results of D_inh, HQ, and HI and excess cancer risk from the inhalation of metals associated with the PM₁₀ collected in the three cities studied

Incremental lifetime cancer risks (ILCR)

The results obtained for the ILCR of As, Cr, Cd, Ni, and Pb are shown in Table 4. The ILCR illustrates the probability of an individual developing cancer over his lifetime due to a potential carcinogen (Gandhi et al. 2021). Similar trends were observed for the three cities: the ILCR values for both MMA and CHI decreased in the following order: Cr > As > Cd > Pb > Ni, while for CDJ, the observed order was Cr >As > Cd > Ni > Pb. The mean ILCR values for Cr were 7.78 × 10⁻⁵, 3.91 × 10⁻⁵, and 1.35 × 10⁻⁵ for CHI, CDJ, and MMA, respectively. This element is an essential nutrient that promotes insulin action in tissues of the body so that sugars, proteins, and fats can be used in metabolic functions. However, like some other trace elements, it can be toxic when accumulated. Cr(VI) compounds have been declared a potent occupational carcinogen, especially for people working in chrome plating, stainless steel, and pigment industries, who are constantly overexposed to high chromium concentrations. The reduction of Cr (VI) to Cr (III) results in the formatting of reactive intermediates that contribute to the cytotoxicity, genotoxicity, and carcinogenicity of compounds containing Cr (VI) species by combining with oxidative stress tissue damage and cellular events such as the modulation of the apoptosis regulatory gene p53. Chromium also plays an important role in altering the immune response since it participates in immunostimulatory or immunosuppressive processes (Shrivastava et al. 2002). Thus, Cr levels must be strictly monitored to safeguard human health. The Cr levels measured in the present study exceeded the highest contents measured by Siudek et al. (2021) in both spring 2019 (mean 2.03 ng/m³) and spring 2020 (mean 1.11 ng/m³) in the southern Baltic Sea, Poland. Those authors did not detect a statistically significant increasing trend (p-value ≤ 0.05) in the concentrations of Cr in PM₁₀ samples. Jaafari et al. (2020) reported Cr levels varying from 8.32 to 41.35 ng/m³ during the cold season and from 6.82 to 27.14 ng/m³ during the warm season in an urban area of Tehran, Iran, which were comparable to those obtained in the present study.

The highest ICLR values for As were found in CHI (1.09 × 10⁻⁵) and CDJ (1.08 × 10⁻⁵), which was in accordance with the mass concentrations of this element found in PM₁₀ samples from both cities. As expected, the ILRC for Pb was significantly higher in MMA (4.04 × 10⁻⁷) than in CHI (1.36 × 10⁻⁸) and CDJ (1.05 × 10⁻⁹). According to USEPA regulations (USEPA 2011, 2013), risk levels can be classified as definite (potential) risk (ILCR ˃ 1 × 10⁻⁴), accepted risk (1 × 10⁻⁴ ≥ ILCR ≥ 1 × 10⁻⁶), or negligible risk (ILCR ˂ 1 × 10⁻⁶). These criteria are applied to the evaluation of individual pollutants. In this study, individual metal risks ranged from 1.35 × 10⁻⁵ to 4.0 × 10⁻⁶ for MMA, 7.78 × 10⁻⁵ to 6.14 × 10⁻⁹ for CHI, and 3.91 × 10⁻⁵ to 1.74 × 10⁻⁹ for CDJ, indicating an acceptable risk for Cr and As and a negligible risk for the remaining elements in the three cities. However, by calculating the cumulative cancer risk, it is observed that the value obtained in the three studied cities exceeded the maximum limit stipulated by the USEPA (1 × 10⁻⁵), suggesting a moderate potential risk for the health of the population exposed to PM₁₀-bound metals (Bozkurt et al. 2018; Romanazzi et al., 2014; USEPA 2011, 2013). However, to estimate with greater certainty the risk associated with the inhalation of PM₁₀-bound metals, we recommend carrying out this study using concentration data for longer periods. In addition, it is necessary to increase the monitoring of metals in the air, develop emission control technologies, establish strict regulations and enforce standards, in such a way that the potential impact of these pollutants on the population’s health is minimized.

Determination of crystalline compounds by XRD

Figure 9 shows the average percentages of crystalline compounds found in the three studied cities. The mineral phases found in the study sites correspond to calcite, quartz, aluminosilicates, gypsum, and halite. PM₁₀ material with similar compositions has been reported in other cities around the world such as London (England), Kemerovo (Russia), and Sharjah (United Arab Emirates). (Williamson et al., 2013; Hamdan et al., 2021; Larionov et al., 2022). The most abundant crystalline phase in PM₁₀ particles from the three cities was CaCO₃, with abundances ranging from 67 to 78%. CaCO₃ can come from natural sources, such as material from the Earth’s crust, which is consistent with the fact that the northern region of Mexico is characterized by predominantly calciosol- and leptosol-type soils, whereas limestone is the most abundant mineral phase in both types; therefore, soil resuspension can be an important source of CaCO₃ in PM₁₀ (Omar and Gaistardo 2014). The levels of calcite composition found in PM₁₀ in Mexican cities were above other cities such as Beijing (China), Peshawar (Pakistan), and Abadan (Iran). (Lu et al. 2007; Ahmady-birgani et al. 2015; Zeb et al. 2018). Moreover, limestone is a raw material for the ceramic and construction industries and thus cannot be ruled out as an anthropogenic source of CaCO₃ since activities in these industries are abundant in the three cities studied (Song et al., 2014).

Fig. 9

Contents of the most important crystalline mineral phases present in the PM₁₀ obtained by X- ray powder diffractometry

In the case of MMA, the second most abundant crystalline phase found in PM₁₀ was CaSO₄, while for CHI and CDJ, it was SiO₂. According to the data provided by the monitoring stations of each city, the SO₂ concentration was significantly higher in MMA (0.086 ppm) than in CHI (0.0038 ppm) and CJ (0.0037 ppm), which could explain the higher percentage of CaSO₄ in this city. Previous works have shown that the formation of secondary sulfate occurs by photochemical reactions that occur between its main precursor (SO₂) and some oxidizing species such as O₃ and NO_x under high humidity conditions (Zhang et al., 2018). In this sense, the higher humidity recorded in the MMA could be favoring the conversion of SO₂ to sulfates. To delve into the relationship between CaSO₄ concentration and meteorological parameters, a Pearson correlation analysis was performed (Table S5). This analysis was performed only in the MMA since, in the case of CHI, CaSO₄ was detected in a single sample. The results reveal a strong correlation (0.91) between relative humidity and the percentage of CaSO₄; in addition, a negative correlation (−0.95) between CaCO₃ and CaSO₄ is also observed. These results corroborate the secondary origin of CaSO₄. As expected, the inverse relationship between CaCO₃ and CaSO₄ shows that the percentage of calcium sulfate increases as CaCO₃ is consumed.

Regarding SiO₂ (quartz), the percentages found in the samples were 18.1, 12, and 10.8% for CHI, CDJ, and MMA, respectively. This crystalline phase is commonly associated with natural sources, generally due to soil resuspension (Adamo et al., 2008; Song et al., 2014). However, the ceramic industry that uses quartz as a raw material could also be contributing to the emission of this mineral phase into the atmosphere (Aragon-Piña, 2014). It should be noted that this mineral has provoked great interest due to its negative effects on human health (Shiraki and Holmén 2002). The IARC classified two polymorphs of SiO₂ (quartz and cristobalite) as belonging to group 1 of carcinogenic materials. Therefore, the percentages of SiO₂ (quartz) found in the samples could represent a potential health risk. The average percentages similar to those found in this research have been reported in other locations, such as Utah (16.1%) and California (10.6%) (Shiraki and Holmén 2002). On the other hand, the minor crystalline phases in MMA and CDJ were aluminosilicates (1.5 and 6.1%, respectively), although the percentage was much higher in Chihuahua; these compounds were the ones found in the lowest percentage for this city. In addition to the species mentioned, low percentages of NaCl were recorded in MMA. This may be related to the presence of marine aerosols that carry this phase to the city since it is located less than 300 km from the coast of the state of Tamaulipas.

Conclusions

Modern microscopy and spectroscopy techniques have provided a unique opportunity to characterize the elemental and molecular levels of PM₁₀ in three cities in northern Mexico. High concentrations of PM₁₀ were found in the urban environment of MMA, exceeding the maximum permissible limits established by the Mexican Official Norm (NOM-025-SSA1-2014), the WHO, and the EU. Even though the permitted levels were not exceeded at the other sites, special attention must be paid to PM₁₀ atmospheric pollution in these cities. The XPS technique revealed that C, O, and Si were the major elements of the surface composition of PM₁₀, and the XPS results, in combination with high-resolution spectra, suggest the presence of aliphatic and polycyclic aromatic hydrocarbons. As, Cr, and Pb were detected at levels exceeding the limits suggested by the WHO, indicating a potential risk of cancer for the inhabitants of the three cities studied. The SEM-EDS results identified regular and irregular morphologies associated with natural and anthropogenic sources, respectively. Fe-rich particles, along with other elements in smaller proportions, exhibited spherical or spheroidal morphologies with corrugated surfaces, as well as irregular morphologies, which could be associated with emissions derived from metal fabrication and steel industries and wear of metal parts, respectively. Mining–metallurgical activity was the main source assigned to the emissions of particles rich in Cu and Zn, while Pb-rich particles showing irregular morphology were related to emissions coming from the manufacture of lead-based batteries, and those that presented prismatic-type morphology were attributed to the influence of natural sources. Even though some of the sources identified are local, the regional distribution of PM₁₀ mass concentration in the three cities studied is significant and concerning. The study results underscore the importance of knowing the chemical composition of PM₁₀ in the ambient air to design policies to reduce the emissions of polluting particles into the air. In addition, the application of the analytical techniques in combination with receptor models can serve as a new approach to distinguishing among emission sources for similar locations.