PM_2.5 Characterization and Source Apportionment Using Positive Matrix Factorization at San Luis Potosi City, Mexico, during the Years 2017–2018

Abstract

In growing Mexican cities, there are few studies on air pollution, especially on the topic of characterization for the chemical composition of Particulate Matter (PM). This work presents an X-ray Fluorescence (XRF) analysis and Total Carbon analysis of PM_2.5 in a two-year monitoring campaign from 20 May 2017 to 30 July 2018, collecting 96 daily samples in the northeast area of San Luis Potosi city to reconstruct the gravimetric mass and perform a source apportionment study using the Positive Matrix Factorization model (PMF). Concentration differences were due to different yearly seasons. In the year 2017, there was a major influence on heavy metals (V, Cr, Mn, Ni, Cu, Zn, Pb), and for the year 2018, there was a major crustal elements concentration (Na, Al, Si, P). Heavy metal concentrations are higher than any worldwide regulation limit. After applying these methods to the 49 samples collected for the year 2017, the mass reconstruction was nearly 70% of the gravimetric mass. XRF analysis quantified 17 elements (Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn) in addition to Total Carbon (Organic Carbon (OC) + Elemental Carbon (EC)). PMF receptor model was applied to identify possible contribution sources and resolved seven physically interpretable factors that contributed to the ambient particulate pollution at the sampling site: Urban Dust (24.2%), Mobile Sources (22.2%), Chemical industry (18.2%), Oil combustion (16.3%), Smelting Industry (12.4%), Fuel Oil + Ceramic Industry (4.4%), and Braking (2.3%). However, the brick kiln’s emissions may be present in at least four of the emission sources due to several types of combustible employed.

1. Introduction

Air pollution has become a major concern in the main cities throughout the world. This is due to its negative impacts on environmental quality and human health, assuming that nearly 80% of the world population lives in urban areas and due to the constant increment of the number of emission sources that enhance the growth concentrations of ambient particulate matter (PM). This term is used to describe solid or liquid particles with dimensions smaller than 50 µm (<50 µm), which are directly emitted or formed by precursor gases and dispersed in the atmosphere. PM is created by different mechanisms, principally nucleation, a subsequent condensation, growth by coagulation, or due to multi-phase chemical reactions [1], that can modify the number of particles, surface, and volume/mass size distribution, and can still grow with different gases and biological compounds, acting as a carrier of viruses, bacteria or pollen through the human respiratory system. PM can be mainly divided into inorganic and carbonaceous particles, which can also react and transform depending on thermodynamic properties, (natural or anthropogenic) emission sources, and geographical and meteorological conditions [2].

PM_2.5 are particles with an aerodynamic diameter of less than 2.5 µm and can also be identified as a fine fraction. They are mainly produced as secondary particles due to anthropogenic sources, especially in the cities, and present a daily or weekly residence life in the local atmosphere and can be transported for long distances, going from hundreds to thousands of miles. Because of this, they can produce regional or global impacts, such as climate forcing and global warming. Their influence on atmospheric processes includes cloud condensation nuclei (CCN) [3], acidification of clouds, snow, and ice [4], induced and dissolved acid rain induction or dissolved [5], reduction in visibility [6], and general disturbances to terrestrial and marine ecosystems, as well as their impact on the cultural heritage, all of which are well documented [7]. But the most important aspect associated with PM_2.5 and its chemical composition is related to different negative effects on human health, affecting the respiratory and cardiovascular system and enhancing mortality and morbidity ranges in both acute and chronic effects [8,9,10,11,12]. PM effects have been found in the inorganic fraction; heavy metals (As, Hg, Sb, Pb, Cd, Cr, and Be) have been designated by the United States Environmental Protection Agency (EPA) as ‘‘toxic air pollutants’’ due to their effects as metabolic poisons and/or carcinogens, as well as many compounds co-issued in the carbonaceous PM fraction. Adverse effects of high concentrations of metal (loid)s in blood, such as Pb and Cd, As in China’s children population, have been detected in their nervous systems, causing a subsequent decrease in their intelligence quotient (IQ), particularly in those who lives in industrialized towns in comparison with a reference town [13]. Therefore, they are possible culprits of health risks in zinc smelting districts in northeast China [14]. Also, high Mn concentrations exposure in the State of Hidalgo, Mexico, reported damage to the neurocognitive, neurological, reproductive, and pulmonary affections [15,16].

Furthermore, there exists a major concern about the PM carbonaceous fraction nowadays because it covers thousands of individual compounds from secondary and primary origins; on the other hand, OC contains PAHs and polyacids, while EC is an aggregate of smaller spheres, highly polymerized, refractory, and insoluble in water and common organic solvents [4]; it is mainly produced or derived for anthropogenic sources like chemical industry and fossil fuel burning, as well natural sources like biomass burning and biogenic sources.

Thus, Mexico has adopted PM as an air quality standard and has its own national limits since 1993, which has been readjusted for PM₁₀ and included for the first time PM_2.5 in 2014. Nevertheless, it has always been a discussion if these daily limits or even the annual limits represent a true indicator to evaluate air quality. In Mexico, just a few city governments have the resources to measure air pollutants such as PM₁₀ and PM_2.5 in all the monitoring sites. Apart from the foregoing, Mexican government authorities have focused on reducing particle concentrations in the PM₁₀ and PM_2.5 sizes; recently, they have proposed in the Mexican official standard (NOM-025-SSA1-2021) [17] a gradual reduction in the forthcoming years, even though only a few cities are required to measure these pollutants according to NOM-156-SEMARNAT-2012 [18] and only half of them comply with monitoring the two size fractions. Another recent effort has been the NOM-172-SEMARNAT-2019 [19], which provides a risk communication daily signal for the criteria pollutants. Nonetheless, not much progress has been made regarding the particulate matter characterization in the main cities of the country.

San Luis Potosi Metropolitan Area (SLPMA) has been growing in the last decade due to rapid industrialization and wide urbanization. Nowadays, SLPMA is an urban area with large metallurgical and mining activities with high concentrations of fluorite, and calcium sulfate has been documented, as well as a large industrial zone with more than 250 manufacturing industries in the south of the city. At NW, there is one of the biggest zinc processing plants in Latin America, which owns a sulfuric acid plant and is responsible for the generation of other metals as subproducts. Meanwhile, in the East, there are more than 150 brick kilns with minimum control. Consequently, they burn all kinds of waste, such as plastics, wood, coke, or even electronic waste. However, it is necessary to identify the real contribution of the main emission sources in the city to provide scientific support for emissions control in order to reduce particulate matter, establish mitigation strategies, and make effective corrective decisions with respect to environmental management policies and majorly to improve air quality. This work presents a PM_2.5 in a two-year monitoring campaign in the northeast area of San Luis Potosi city to reconstruct the gravimetric mass with X-ray Fluorescence (XRF) analysis and Total Carbon Analysis and perform a source apportionment study using the Positive Matrix Factorization model.

2. Materials and Methods

San Luis Potosi City Metropolitan Area is located in the central zone of the country, being the 11th metropolitan area, most populated in Mexico, covering San Luis Potosi City and its conurbation area (Soledad de Graciano Sánchez). It covers an area of nearly 1776 km² and has about 1.25 million inhabitants. Population density is 700 inhab•km⁻² for SLPMA [20]. It is located at an average altitude of 1864 m above sea level (a.s.l), at the central portion of two mountain ranges: on the Northwest side, there is the Sierra de San Miguelito and, on the East, the Sierra de Alvarez, generating a partial air basin in a well-documented semi-arid climate.

By and large, the wind flows transport predominantly occurs from the south to the north zone, but also local effects like convergence and divergence processes produced by mountain-valley circulation cause regional easterly winds and recirculation through the urban area that is unfavorable for the dispersion and transport of atmospheric PM beyond the urban area [21]. SLPMA monitoring site was the “Estación Biblioteca”, from the Department of Ecology and Environmental Management (SEGAM, for its acronym in Spanish), located at 22°10′34″ N and 100°59′22″ W, in the northeast part of the city. This monitoring station is situated at a suburban site with low vehicular density and moderate diesel bus density. In the proximities, there are also more than 100 brick kilns, some old chemical industries, and a one-rail train impact zone. Figure 1a shows the monitoring zone location at SLP city, and Figure 1b shows the principal industries in the city according to the National Statistical Directory of Economic Units (DENUE, by its acronym in Spanish) from 2021.

2.1. Sampling

PM₂.₅ samples (from midnight to midnight) were collected on a 24 h sampling base for two years divided into two periods, the first one from 20 May 2017 to 30 July 2018 and the second period from 20 May 2018 to 25 July 2019. Two MiniVol Air samplers (MiniVol TAS, Airmetrics, Springfield, OR, USA) were used every two days, operating at a low volume of 5 L min⁻¹, with a type B uncertainty of 0.5 L min⁻¹ and equipped with a 47 mm diameter inlet to collect into a Teflon filter (Pall Corp., New York, NY, USA, 1.0 µm pore size) for inorganic aerosols and in another one to collect with a Quartz filter (Whatman Int. Ltd., Maidstone, UK, 47 mm) for carbonaceous aerosols. Subsequently, air samplers were located at the roof level (between 5 m to 8 m) of the “Estación Biblioteca” site from the local monitoring network, from SEGAM to obtain their meteorological data. Meteorological data were obtained from the Anderson weather station. All sensors are calibrated annually.

2.2. Measurements

Gravimetric mass concentrations were determined by weighing the Teflon filters using an electronic microbalance (BA2105, Sartorius, Goettingen, Germany) with a resolution of 0.01 mg in controlled conditions of temperature (20 to 25 ± 1 °C) and relative humidity (35 to 45 ± 5%) both previously and after monitoring. Carbonaceous samples were collected on a 47 mm quartz fiber filter and pre-combusted for 3 h in a 450 °C condition. Samples were analyzed in a laboratory-based thermal-optical EC-OC analyzer (Sunset Laboratory Model-4, Tigard, OR, USA) using the EPA-NIOSH protocol to measure the concentrations of Total Carbon (TC), Organic Carbon (OC), and Elemental Carbon (EC). Detection limits (DLs) were in a low (1 ng/m³ to 50 ng/m³) range. This PM_2.5 carbonaceous analysis was determined at the “Universidad Autónoma Metropolitana, Azcapotzalco”, and the analytical procedure was recently explained [22].

Inorganic elemental analysis was performed in the custom-built XRF spectrometer for environmental applications at the “Instituto de Física, Universidad Nacional Autónoma de México”, equipped with an X-ray tube with an Rh anode (Oxford Instruments, Mountain View, CA, USA), operated at 50 keV and 500 µA, with a detection system composed by an Amptek X-123SDD spectrometer with a resolution of 120 eV at 5.9 keV. The filters were then placed in the analysis chamber at a high vacuum (10^–6 torr). Each sample was irradiated for 900 s. A filter scan was carried out to analyze most of the deposit filter area to measure the concentrations of all the elements with atomic numbers Z > 10. XRF spectra were fitted using the QXAS-AXIL computer code [23], and elemental concentrations were obtained by comparison with a calibration curve from a set of Micromatter standards of known areal density.

XRF Limits of Detection (LOD) were calculated on the basis of background radiation for the Teflon filter, and also a standard NIST SRM 8785 (National Institute of Standards and Technology, Gaithersburg, MD, USA) was irradiated to measure the detection system efficiency for XRF analytical procedure; these steps were described previously [22]. Uncertainties were determined according to the method described by Espinosa et al. [24].

2.3. PMF Analysis and Procedure

Receptor Modelling techniques are a widely used approach applied to chemical speciation data in aerosols. The fundamental receptor model is that mass conservation can be assumed, and a mass balance analysis can be used to identify and apport sources of atmospheric particulate matter [25]. PMF is one of the most widely used receptor models based on a weighted least square fit approach that employs realistic error estimates to weigh data values and imposes non-negativity constraints in the factor computational process [26].

This PMF model may be written as X = G • F + E, where X (matrix elements X_ij) is a known n by m matrix of the m measured chemical species in n samples; G (matrix elements g_ik) is a matrix of source contributions to the samples (i.e., time variations of the p factor scores) and F (matrix elements f_jk) is a matrix of chemical factors composition (often called source profiles). G and F are factoring matrices to be determined, and E (matrix elements e_ij) is defined as a residual error matrix or the difference between measured and modeled values. The PMF equation of a bilinear problem is defined as follows.

$$X_{ij }={\displaystyle \sum }_{k=1}^p{g}_{ik}•f_{jk}+e_{ij}$$

(1)

X_ij concentration (x) for a species (j) in a sample (i).

g_ik contribution (g) of the source (k) in a sample (i).

f_jk fraction (f) of a species (j) at the source (k).

e_ij residual error.

p is the number of factors.

Factor analysis can give a number of possible solutions, all mathematically correct. However, the best solution in PMF analysis should be supported by quantitative indicators, especially the number of factors [27]. One method to select the correct number of factors is by the evaluation of the Q-value, which can provide useful indications when the data-point uncertainties are well determined. A theoretical Qvalue determined by the product of (Qtrue/Qrobust), both output by each run of the PMF [28,29], can indicate consistent modeling data if this value is approximate to 1 and does not exceed 1.5, where Qrobust is calculated by excluding all outliers and Qtrue includes all points [30], and the near approach to the theoretical Qvalue can suggest a better fit.

PMF was applied to 49 samples from the year 2017. These concentrations were prepared using the input data procedure, including uncertainties and data below minimum detection limits suggested by Polissar et al. [31], using the EPA PMFv5 software. To obtain absolute source profiles and contributions, matrices G and F were normalized by setting the PM_2.5 mass concentration as a “total variable” (with 400% uncertainty); in this way, the PM mass is apportioned, based on its similarities with the identified sources, without influencing the fitting process [31,32,33].

To determine the results, a diverse number of factors (from 4 to 9) was systematically explored to find out the most reasonable solution (25 pseudorandom initializations were run for each test), looking for the best Qvalue and an explained contribution for all the factors in the PM_2.5 mass.

After defining the most appropriate number of factors, interpreted as the main emission sources considering different aspects, such as geographic location, meteorological data, as well as the type of industrial zone, mobile sources, and other local emission sources in the area (local database for the study site). Each factor was analyzed on the basis of the chemical species that compose it to determine the known source(s) to which it is associated in the literature and verified using the monitoring database. Each factor was then also modeled with the Open-air program from R-studio [34] to corroborate the origin of the different emission sources, and the ArcGIS computer code [35] to georeference the factors in the monitoring zone by using the meteorological data from the local network and locate a diameter distance of 5 km from the monitoring site ubication. Finally, results were overviewed with the reconstruction of back-trajectories, using the open access model HYSPLIT [36] at different end-point altitudes (100 m, 500 m, 1000 m, 1500 m a.g.l) and at different hours of the day (0, 6, 12, 18, UTC) for all the monitoring periods, to determine the possible origin of episodes with high concentrations or to identify local emission sources contribution.

3. Results

3.1. PM_2.5 Gravimetric Mass

Nearly 100 valid PM_2.5 samples were collected for the total sampling period at the Biblioteca-SEGAM site, with a 30.7 (6.6) µg m⁻³ 24 h average gravimetric mass, not exceeding the national standard PM₂.₅ daily mean value of 45 µg m⁻³ (NOM-025-SSA1-2014), when the monitoring campaign was made [37] but surpassing the 25 µg m⁻³ WHO recommendation (Figure 2).

The campaign was then divided into the 2017 period from 20 May 2017 to 22 November 2017, which covers the summer and autumn seasons, and the 2018 period from 27 January 2018 to 30 July 2018, covering winter, spring, and summer seasons. During the 2017 campaign, 47 valid samples were collected at the site with a gravimetric mass average of 33.2 (7.0) µg m⁻³, while in the year 2018, the 49 samples collected had a mass average of 28.2 (6.1) µg m⁻³, with the daily mean value exceeded in the first campaign for 7 days and the second a total of 4 days (Figure 2). However, if the WHO recommendation for the PM_2.5 concentration is applied, almost 60% of the total samples will exceed this regulation.

Comparing both years, concentration mass data were analyzed to contrast the hypothesis of normality with the Shapiro–Wilk statistical test (α = 0.05) and the Bartlett of homogeneity of variances test, and after concluded that all data presented a non-parametric distribution; the Mann-Whitney statistical test was used to compare the sampling years for gravimetric mass. Therefore, the PM_2.5 concentration mass for the year 2018, compared with the one from the year 2017, is significant (p = 0.0028).

Finally, the maximum PM₂.₅ concentration values were registered on 16 July 2017 and 25 March 2018, with 102.4 (20.6) µg m⁻³ and 104.2 (20.9) µg m⁻³, respectively. Back-trajectory analysis was made to check the meteorology and the connections between these high concentrations and air mass trajectories. HYSPLIT showed for July 16th a strong influence from the northeast emissions where the brick-kiln zones are located; also, calm wind conditions produced the accumulation of the PM_2.5 concentrations during this day. Moreover, on March 25th, the main emissions came from the Southeast, where the industrial zone is located (Figure 3).

A natural barrier is composed due to the presence of mountains in the western zone and another one in the southeastern zone. Because of that, the most important winds came mostly from the eastern zone of the city, which is partially arid and in its vicinity to the monitoring site is surrounded by different semi-industrial zones, train tracks, automobile road axes, and mainly the brick zone. Autumn 2017, Winter 2017–2018, and Spring 2018 also showed a secondary wind pattern: the SE winds are associated with its entrance through the south, where the main industrial zone of the ZMSLP is located (Figure 4).

The highest wind speeds are generated during the spring and summer seasons (wind speed > 2.2 m s⁻¹), as can be seen in both years, while the least wind speed occurs during the winter, preventing the dispersion of air pollutants. Even the calm percentage value for this period is reported at 21% with an average wind speed of 1.3 m s⁻¹. Autumn 2017 registered an 8.5% of calm values and an average wind speed of 1.8 m s⁻¹. Finally, the calm percentage value for spring and summer in 2017 was lesser than 1.6%, and for the year 2018, lesser than 3.9%. The data were obtained through the SEGAM network at the Biblioteca site (SINAICA, 2018) [38].

The SLPMA local network has just one PM₁₀ and one PM_2.5 monitoring equipment, not covering all of the Metropolitan Area, but the main problem is that it does not generate the minimum data concentrations for a valid year PM evaluation according to the NOM-025-SSA1-2014, neither the actualization of this regulation adjusted during the year 2021, because of that it was not possible to compare the PM concentration data from this study to the data showed by SEGAM.

3.2. PM_2.5 Chemical Composition

The contribution of main aerosol components to PM_2.5 was estimated based on the measured chemical composition according to the two analysis methods already mentioned.

Table 1. PM_2.5 Chemical composition for the total monitoring campaign (µg m⁻³).

Species	Year 2017					Year 2018
	N	Mean	Inc	Median	Std Dev	N	Mean	Inc	Median	Std Dev
Na	42	0.283	0.022	0.297	0.108	38	0.342	0.0325	0.328	0.210
Mg	<LOD	<LOD	<LOD	<LOD	<LOD	4	0.0538	0.0005	0.0538	0.025
Al	27	0.0769	0.0015	0.0541	0.0589	32	0.222	0.0133	0.118	0.308
Si	46	0.305	0.0112	0.277	0.194	37	0.530	0.0453	0.370	0.576
P	40	0.0198	0.0001	0.0155	0.0204	35	0.0450	0.0004	0.0383	0.0330
S	46	1.65	0.259	1.44	1.05	38	1.86	0.306	1.65	1.01
Cl	46	0.449	0.0174	0.455	0.115	38	0.324	0.0105	0.296	0.135
K	46	0.388	0.0130	0.388	0.119	38	0.249	0.0070	0.212	0.146
Ca	46	0.603	0.0381	0.576	0.189	38	0.408	0.0207	0.360	0.175
Ti	42	0.114	0.0023	0.113	0.0542	30	0.0636	0.0010	0.0496	0.0467
V	45	0.149	0.0034	0.143	0.0655	27	0.0739	0.0013	0.0665	0.0622
Cr	46	0.390	0.0153	0.384	0.0807	38	0.217	0.0066	0.170	0.119
Mn	46	0.550	0.0568	0.419	0.682	38	0.267	0.0117	0.153	0.244
Fe	46	1.02	0.0835	1.01	0.131	38	1.0034	0.1026	0.8262	0.5932
Ni	46	0.569	0.0285	0.569	0.080	38	0.387	0.0151	0.365	0.103
Cu	46	0.242	0.0067	0.229	0.0611	38	0.129	0.0027	0.0963	0.0777
Zn	46	0.378	0.0145	0.368	0.103	38	0.258	0.0082	0.238	0.110
Pb	46	2.35	0.444	2.28	0.290	38	1.16	0.184	0.916	0.412
OC	44	10.7	1.85	10.0	3.04	ND	ND	ND	ND	ND
EC	44	1.89	0.328	1.46	1.42	ND	ND	ND	ND	ND
POM	44	17.2	2.98	16.0	4.86	ND	ND	ND	ND	ND
PM2.5	47	33.2	6.99	31.2	14.7	49	28.2	6.11	22.6	15.8

ND = Not Determined. <LOD = Bellow Limit of Detection.

shows the differences in chemical compositions in the two years due to different monitoring seasons and emission sources. Also, the differences between the average concentrations of the elements caused by anthropogenic sources and those generated by the Earth’s crust are easily observed. For the year 2017, heavy metals (V, Cr, Mn, Cu, and Pb) were at least 50% higher than for the year 2018 due to the dryer season (Spring and Summer); however, crustal elements such as Na, Al, and Si, presented higher values during 2018.

Also, comparing both years, the same period from 20 May to 30 July, gravimetric mass and the %XRF measured, anthropogenic elements for the year 2017 were higher than for the year 2018, even V and Pb were the double concentration value, and for the year 2018 just the crustal element as Si, P, and Al were higher an at least 70% than those from the year 2017. The Mann-Whitney non-parametric test was performed for the comparison of independent samples (α = 0.05), which indicates that the variation in the average gravimetric mass concentration of PM₂.₅ in 2018 (27.46 µg m⁻³) compared to the year 2017 (27.46 µg m⁻³) was significant (p = 0.019). And for the XRF concentration mass measured, the year 2018 (7.77 µg m⁻³) sized up with the year 2017 (9.52 µg m⁻³) was also significant (p = 0.024). Meteorological data also showed cooler and dryer conditions for the same period in the year 2018 than the data from the year 2017 [38].

On average, the total XRF mass represents nearly 30% of the gravimetric mass in the site during a two-year monitoring period. It was possible to identify at least 18 elements (Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, and Pb). There are high concentrations of elements emitted mainly by anthropogenic sources, such as those from fossil fuels, like S, V, Ni, and TC, as well as heavy metals from mineral extraction processes (Cr, Fe, Ni, Mn, Cu, Zn, and Pb). There are no Mexican regulations, and only a few international agencies present air quality standards for some heavy metals, such as annual limit values for Ni = 20 ng m⁻³ (EU, directive 2004/107/CE for 2013) and for Pb = 500 ng m⁻³ (EU, Directive 1999/30/CE for 2005; also, the WHO recommended the same concentrations values). For a daily period, just the Indian air quality standards present a limit value for Pb = 1000 ng m⁻³ [39].

The concentration values obtained during the total sampling of this monitoring exceeded all limit values concentrations, especially for the metals elements, and compared with previous PM_2.5 studies from Mexico City, as showed by Hernández-López et al., the elemental concentration for Cr, Fe, was at least twice the ones collected from that urban area, while for elements as Ca, V, Mn, Ni, Cu, and Zn, were more than five times the concentrations values. As for Ca, the concentration values are attributed to the fluorite mining zone near the city. Other geological or crustal elements such as Al, Si, and Cl are in the same range as the PM_2.5 collected in the south zone from Mexico City. S and K concentrations also require special consideration, thinking about the different possible sources.

After applying the Total Carbon analysis, the carbonaceous fractions in PM_2.5 for the year 2017 consisted of 32.3% OC and 5.7% EC. Particulate Organic Matter (POM), i.e., the sum of all the carbonaceous compounds, was estimated by multiplying the OC concentration by a specific POM/OC factor to consider the contribution of the other light elements contained in these compounds (like H, N, and O), that could not be measured by this analytical technique. Different POM/OC factors can be applied according to the type of monitoring site, with increasing values from traffic to rural sites, taking into account the high quantity of natural VOC emissions in rural areas. However, for this semi-urban site, the factor used in agreement with the literature was 1.6 [40]. For the year 2017, the POM concentration average was 17.2 µg m⁻³ plus 1.9 µg m⁻³ due to EC, adding up to 60% of the total contribution to PM_2.5, which is consistent with this type of semi-rural locations, for example, Tepeji and Jaso, Hidalgo, reported by Vega et al. [41]. Nevertheless, a difference in concentrations between the seasons of the sampling year could not be distinguished. The OC/EC average ratio was 7.36, indicating that the carbonaceous aerosol is mainly organic and suggests the presence of various combustion emission sources, such as biomass burning, possible fires in the area, the presence of a large chemical industry or even the brick kiln zone but also considering the low traffic emission zone near the monitoring site.

Total Carbon seasonal concentrations in the year 2017 were 14.99 µg m⁻³, 11.65 µg m⁻³, and 12.22 µg m⁻³, and OC/EC was 5.86, 6.13, and 4.87 for spring, summer, and autumn, respectively, which indicates the highest EC in spring due to the biomass burning present in the fires at the city, the lowest EC in summer caused by the rainy season and the lowest OC/EC ratio in autumn produced by the low wind speed movement. Elemental Carbon concentrations were for the spring 2.19 µg m⁻³, for the summer 1.63 µg m⁻³, and 2.08 µg m⁻³ for the autumn.

3.3. PMF Source Apportionment Model Results

For the year 2017, a total of 49 PM_2.5 samples were used as the input data for the PMF model to analyze 20 variables, which represented nearly 70% of the total PM_2.5 concentration mass, and the only data with the carbonaceous fraction. The chemical species selected as inputs were previously tested and validated, considering those with more than 50% of the data set below the detection limit of the method (MDL) were excluded (Mn), and those species with a significant correlation with PM_2.5 were retained and subsequently classified according to the signal/noise ratio (S/N) criteria [42].

Then, PMF was used to find out the results for the different number of factors (from 4 to 9), looking for the best Qvalue to determine the most reasonable solution (25 pseudorandom initializations were run for each test). After that, 18 chemical species (Na, Si, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Pb, OC, and EC) were classified as “strong”, according to the method previously described and with a presence in more than 90% of total samples and only two elements, (Al, P) as “bad” variables (i.e., species with a signal to noise ratio lower than 0.2 and/or below the detection limit in more than 90% of the cases). This combination of chemical speciation and source apportionment by PMF allowed the identification of seven factors or main emission sources as the best fit and reasonable solution for the studied site. Qvalue was 1.3. A good conceptual understanding of the total emission sources in the monitoring area is important for establishing aerosol sources (Figure 5).

Table 2. PMF source apportionment for the Biblioteca-SEGAM monitoring site.

Monitoring Site	Fit	% Contribution	Emission Sources	Main Chemical Species
BIBLIOTECA SITE, 2017.	18 Species 7 Sources.	24.2	(1) Urban Dust	(OC, EC, Si, S, Pb, Fe, Ca, K)
		16.3	(2) Oil Combustion	(S, OC, Fe, Pb, EC, Ca, Zn, Ni)
		2.30	(3) Braking	(Ti, Pb, Mn, K, Ni, Cr, Na, Ca, Cu, Fe)
		22.2	(4) Mobile Sources	(OC, EC, S, Pb, Fe, Cl, Na, K, Ni, Cr)
		4.40	(5) Ceramic Industry + Biomass Burning	(OC, Ca, Na, V, Fe, S, Cu, Cl)
		12.4	(6) Smelting	(Zn, OC, EC, Pb, Fe, Ca, Mn)
		18.2	(7) Chemical Industry	(OC, Cl, Fe, Pb, Ni, Cr, Mn, EC)

shows the PMF source apportionment for the “Biblioteca” monitoring site.

Source 1 was identified as Urban Dust. It was traced by typical crustal species, mineral dust or also called soil as composed of Si, Fe, Ca, and K [43,44,45] added with a high contribution by S [46] and another significant influence of resuspended dust composed of OC, EC, and Pb (produced by anthropogenic fuels and asphalt, brakes, and tire wear); because of this combination, it was called Urban Dust. This source may be the result of local resuspension by wind and convective processes and contributed to 24.2% of the total PM_2.5 composition. Similar profiles have been found in previous works in the north of Mexico City in PM_2.5 [45] and PM₁₀ [46].

The second source was associated with Oil Combustion. Majorly composed by S [43] and OC, which is highly marked by OC [47], with a contribution of Ni, Fe, Ca, and Zn [44,48] related to industrial emissions [49] and a possible impact of the brick kiln zone. Additionally, EC was identified in this source. Anthropogenic fuels and residual oils with a high quantity of sulfur are included in this emission source. Figure 6b shows this source at the NE from the monitoring site, and according to DENUE 2021, there are some industries and the main brick kiln zone.

Source 3 was associated with the brakes of mobile sources, mainly the train brakes near the monitoring zone. This source was composed of Ti, Mn, K, Ni, Cr, Na, and Pb. This is the first time this source has been identified in a Mexican city, even though its contribution is very small, nearly 2%. Figure 6c shows, in color red, the main emissions for this source near the railroads. Some studies have found this source in railroads or subways and their resuspended emissions, composed of Mn, Cr, and Fe, being generated by the abrasion of the railroad tracks and brakes in New York City [50], Barcelona [51] and Seoul [52,53].

Source 4 is considered as vehicular emission-related, containing S, Ni, [46] traced by OC, EC [48,54] and a contribution of other elements (Na, K, Cl, Fe, Cr, and Pb) produced not just by combustion, but also by tires and brakes [22,45,49,55,56,57]. Also, this source can contain diesel or a fuel combustion mixture due to a nearby car racing circuit in “Parque Tangamanga” and a bus station at the zone. Figure 6d shows the principal contributions in color red across the main avenue of Rio Santiago but also at Hernan Cortes and Damian Carmona Avenues.

The fifth one was a mixture emission source. On the one hand, the presence of Cl, S, and Ca suggest the brick-making process for a specific kind of regional clay. Gonzalez [58] showed that elemental composition could vary on the kind of raw material used for the bricks, firing temperature, time, and kiln used. Also, this composition could be particular of a kind of brick with some glaze to make it in this zone, and on the other hand, the presence of S, V, and Cu suggested a mix of Fuel Oil with other trace elements marked with biomass burning like Na, Ca, Cl, Fe, and V [33,41,46], especially for episodes reported on 27 June, 8 August, and 17 October, for the year 2017. Vega et al. [41] also identified OC and K were not found in this source, as other studies have shown (explained later).

Source 6 is associated with a smelting industry source and has high correlations among Zn, Ca, Fe, Pb, and OC, with an enrichment with Mn and EC. Figure 6f shows the main high concentrations at three different zones, but also DENUE locates some smelting industries there. This source has been widely identified in zinc and lead smelters [59,60]. Zn, Pb, and Mn are referred to as steel processes and zinc metallurgy [61], but also Zn and Pb can be related to municipal solid waste incinerators and cement production [62]. Johnson [63] also observed this source with a high Zn and Mn correlation in three different stages contained in PM_2.5 in Mexico City during the MILAGRO Campaign.

The seventh source corresponds to the Chemical Industry composed of OC, Cl, and EC with high industrial markers contribution as Mn, Cr, Fe, Ni, and Pb. This source is identified in Mexican industrial zones, in Mexico City in the north zone [22,46], and in Queretaro [64].

Figure 6 displays the main emission zones for each factor for the year 2017, revealing that these areas had highly active hotspot disturbances; further studies are in progress to corroborate their existence. Finally, dots in colors represent the same type of industries shown in DENUE and with the same classification as Figure 1b.

4. Discussion

In Mexico, air pollution local networks only evaluate the gravimetric mass. Thus, there is a lot to know about the chemical composition of the PM pollutant in large Mexican cities, such as San Luis Potosi; predominantly in the fine fraction of PM_2.5 in order to create new air pollutants indicators and provide a new understanding of the interactions and processes of the PM fractions in their different sizes and growing processes to decrease the emission of source precursors, but mainly to reduce the health risks in the population.

Although San Luis Potosí is a growing city of just over 1.2 million inhabitants, the PM_2.5 mass concentrations do not frequently exceed the official regulations for this pollutant [37]; the obtained PM₂.₅ composition for this northern area of the city of San Luis Potosi showed a strong chemical reactivity and a high toxic composition due to the presence of heavy metals (Pb, As, Zn, and Cu) which have been reported since decades [65], and linked to cancer and non-cancer health risk [2]. Heavy metals concentrations surpassing for other studies carried out in larger cities such as Mexico City [22,49], Monterrey [66], Queretaro [64], and Hidalgo [15]. Also, strong chemical reactivity is due to high concentrations of carbonaceous compounds, a high amount of S due to fuel combustion, and a high concentration of Ca due to the soil resuspension towards the city and the main fluorite industry in Mexico produced at Villa de Zaragoza (near 40 km from SLP City), which is widely used in the metal-mechanic, chemical and construction industries. Also, K was very abundant during the two years, and a deeper analysis is necessary because this element has multiple emission sources, such as biomass burning, soil dust, waste incineration, coal fire, industry, meat cooking, and smelting [67]. In this study, K was also associated with at least five emission sources.

The application of PMF to these PM_2.5 samples showed the main emission sources at this monitoring zone. Mobile sources represent a percentage (≈20%) because there are only a few high avenues with moderate vehicular traffic, and high traffic happens just in the mornings from 7 to 9 A.M. near the site. The urban dust source represents a considerable emission percentage (≈25%) because, in this area, there are still many unpaved streets, producing the resuspension of the soil with other particles resulting from other pollution processes, for example, the braking of train tracks around the study area. This braking source was founded for the first time in Mexico cities, and it was verified with other international studies in cities with railways.

Finally, the northern zone of the city has suffered a serious problem in terms of contamination and effects on public health for decades due to the presence of a brick-making area, which has about 100 kilns that work intermittently. This study showed the emission of highly toxic air pollutants such as Cu, Fe, Zn, Mn, V, and Pb, while the carbonaceous fraction presents high emissions with OC and EC that could be an indicator for black carbon (BC) and associated with other carbonaceous compounds like Polycyclic Aromatic Hydrocarbons (PAHs), generally correlated, as well as a high amount of S. This can be associated with the incomplete combustion of different anthropogenic compounds, and biomass burning, but also with the presence of all kinds of waste (municipal, industrial, electric, and/or electronic). Because of that, four emission sources were associated: Chemical Industry, Oil Combustion, Smelting, and Fuel Oil + Ceramic Industry, corresponding to the brick manufacturing zone, as can be seen in Figure 6. As such, there is no brick kiln characteristic chemical composition source since it depends on the fuel used for the combustion. Therefore, it was possible to georeference and verify the presence of other industrial sources that produce the chemical associations proposed by the PMF analysis (Figure 7) through the DENUE. Also, these emissions can expand quickly from the north zone to downtown in the morning and may dissipate at night when the winds blow through the East.

The most similar source determined for PM_2.5 in Mexico was reported in Hidalgo by [41] Vega (2021), using ICP/MS and identified the Cement Kilns source through the presence of Cr, Co, Zn, Cd, Pb, Hg, OC, and EC. Also, Aldape et al. [68], in a PM_2.5 study, detected and determined the presence of an incinerator and coal-burning source (Mn, Zn, Pb) in Mexico City. Herrera-Murillo et al. [69] identified, in a PM_2.5 study in Salamanca, Mexico, the presence of a ceramic and brick manufacturing source composed of Al, Si, Ca, Fe, Zn, OC, EC, and Sulfates, plus a garbage-burning source; however, heavy metals were majorly associated by others anthropogenic sources in the PMF analysis. And recently, Quiroz-Carranza [70] reported a compilation of the contribution of some elements such as (Mg, Fe, Cu, Zn, Cd, As, Ba, Se, Sr, and Hg) and compounds such as (CO, CO₂, NO, NO₂, SO₂, dioxins, furans, and polychlorinated biphenyls, etc.), emitted according to different fuels used in a brick kilns zone in Queretaro, Mexico.

In other countries, the waste incinerator source is distinguished by the combustion of municipal solid and chemical industry waste. Therefore, many studies showed the presence of Cl, Cu, Zn, Cd, Hg, OC, and EC associated with this source [54,62,71,72,73,74]. As it has been identified at the zone, the brick kilns industry uses all kinds of waste, as well as incinerators, and because of that, there is a similar correlation in the chemistry species.

Meteorological conditions, rainfall, natural forest fires, and the emissions of precursors are associated with PM_2.5 concentrations in different seasons. The highest PM_2.5 concentrations were registered in the summer season due mainly to the brick kiln zone. During winter, stationary conditions accumulate particulate matter, mostly in the city (Figure 4d).

It is necessary to pay more attention to these growing Mexican cities (at least there are more than 10 with nearly the same population), which also present similar emission sources like chemical, construction, and mining industries, not overpassing the national particulate standards frequently, making local authorities not to take care of the pollution issues.

5. Conclusions

Legislation of PM concentration mass limits has been established worldwide in the last fifty years as an indicator of air quality. As observed in SLP city, PM_2.5 can overpass the national standards for a few days during the year. However, it is necessary to pay special attention to the particulate matter chemical composition and generate new air pollutants indicators.

In two years of PM_2.5 monitoring, concentrations of trace elements and total organic compounds were analyzed by different techniques and divided into two data sets. Concentrations differences in the two years were due to different yearly seasons when the contributions of different emission sources showed in the year 2017 a major influence on heavy metals (V, Cr, Mn, Ni, Cu, Zn, Pb), a significant decrement in the next year also due to a dry weather year, and a major crustal elements concentration (Na, Al, Si, P) in the year 2018. Heavy metal concentrations are higher than any worldwide regulation limit. Other elements like Ca, K, and Cl varied according to their main emission sources. The combination of an X-ray spectrometric technique and the Total Carbon method allowed the identification of possible sources of PM_2.5 in the SLPMA during the year 2017.

With the above, it was possible to determine the seven main emission sources as well as their contribution percentages to the sample area. This work carried out a PM_2.5 campaign in the city of San Luis Potosí, which is mainly influenced by a brick-making center and by some other chemical industries near the test point that impact the entire city.

The recognition of the contributing sources was supported by similar studies from other urban areas, especially from other countries. However, it is necessary to conduct more studies in different cities in Mexico to determine the concentrations of pollutants, source contributions patterns for anthropogenic fuels (industries, mobile sources), and natural sources (Biomass Burning, Resuspended Soils) to create a national source database. The most significant contributions to air pollution were caused by resuspended urban dust, mobile sources, and the main combination of oil combustion and chemical industry, caused by the brick kiln zone and other local industries were explained by the application of positive matrix factorization. Some specific and area sources at the north of SLP impact multiple areas of the city; however, geography and meteorology generally cause air pollution to disperse throughout it gradually. Finally, it is necessary to pay special attention to the brick kiln zone to prohibit the use of some of the materials in the combustion, and a major program to relocate not just the brick kiln industry but also some industries located inside the common suburbs to improve the air quality all over the city.

Future works on this issue must include another PMF analysis for different zones of the San Luis Potosi metropolitan zone to determine the influences of the main industrial zones in the south and the emissions produced downtown in order to know the main mix of chemicals emissions in the city, but also to expand the database in this kind of growing Mexican urban zones.