Chemical characterization and factor analysis of PM_2.5 in two sites of Monterrey, Mexico

Abstract

The Monterrey Metropolitan Area (MMA) has shown a high concentration of PM_2.5 in its atmosphere since 2003. The contribution of possible sources of primary PM_2.5 and its precursors is not known. In this paper we present the results of analyzing the chemical composition of sixty 24-hr samples of PM_2.5 to determine possible sources of PM_2.5 in the MMA. The samples were collected at the northeast and southeast of the MMA between November 22 and December 12, 2007, using low-volume devices. Teflon and quartz filters were used to collect the samples. The concentrations of 16 airborne trace elements were determined using x-ray fluorescence (XRF). Anions and cations were determined using ion chromatography. Organic carbon (OC) and elemental carbon (EC) were determined by thermal optical analysis. The results show that Ca had the maximum mean concentration of all elements studied, followed by S. Enrichment factors above 50 were calculated for S, Cl, Cu, Zn, Br, and Pb. This indicates that these elements may come from anthropogenic sources. Overall, the major average components of PM_2.5 were OC (41.7%), SO₄ ²⁻ (22.9%), EC (7.4%), crustal material (11.4%), and NO₃ ⁻ (12.6%), which altogether accounted for 96% of the mass. Statistically, we did not find any difference in SO₄ ²⁻ concentrations between the two sites. The fraction of secondary organic carbon was between 24% and 34%. The results of the factor analysis performed over 10 metals and OC and EC show that there are three main sources of PM_2.5: crustal material and vehicle exhaust; industrial activity; and fuel oil burning. The results show that SO₄ ²⁻, OC, and crustal material are important components of PM_2.5 in MMA. Further work is necessary to evaluate the proportion of secondary inorganic and organic aerosol in order to have a better understanding of the sources and precursors of aerosols in the MMA.

Implications:

The MMA has become one of the most air polluted areas in Mexico. High levels of PM_2.5 have been measured and effective actions need to be taken to reduce air pollution and the associated health risks. Several sources of primary PM_2.5 and precursors of secondary particles exist in the MMA. This study provides valuable information for the local environmental authorities to identify possible sources of primary PM_2.5 and its precursors. The effectiveness of the actions taken to improve air quality will lead to health benefits for the population, reducing their associated costs.

Introduction

Anthropogenic emissions of particulate matter (PM) and its precursors are typical in urban environments. In particular, PM with aerodynamic diameter less than 2.5 μm (PM_2.5) has been widely studied due to its impact in human health, such as cardiovascular diseases and respiratory disorders (Dockery et al., 1993; Pope et al., 1995; Schwartz, 1996). Ambient concentrations of PM_2.5 have been related with morbidity and mortality in urban environments with annual PM_2.5 concentrations in the range of 11.0 to 29.6 μg/m³ (Dockery et al., 1993). Airborne aerosols may contain different chemicals that come from different sources like combustion, mining and industrial activities, soil, and other sources. Due to their size, these particles remain suspended in the air and can easily be inhaled by the population exposed reaching the alveoli. The harmful effects of coarse and fine PM on the respiratory tract are well documented in the literature (Dockery et al., 1993; O'Neil et al., 2002; Pope et al., 1995; Schwartz, 1996).

In Mexico, the 24-hr average air quality standard for PM_2.5 is 65 μg/m³ and the annual average is 15 μg/m³. High concentrations of PM_2.5, well beyond these national air quality standards, have been measured in the Monterrey Metropolitan Area (MMA). The MMA is located in the State of Nuevo León, northeast of Mexico, as shown in Figure 1. Since 1993 the state environmental authorities have operated the air quality monitoring system in the MMA, known as SIMA. This system started with five air quality monitoring stations, and two other monitoring stations were integrated to the system in the year 2010. The stations measure criteria pollutants and meteorological parameters. The monitoring activities for PM_2.5 started at the end of 2002. The data reported in the MMA Air Quality Program show that the main pollutants are particulate matter (PM₁₀ and PM_2.5) and ozone (SEMARNAT y Gobierno del Estado de N. L., 2008). In the period 2004 to 2007, the program reports that 517 days were above the PM₁₀ 24-hr air quality standard of 120 μg/m³ and 69 days were above the PM_2.5 24-hr air quality standard of 65 μg/m³. It also reports that 109 days were above the ozone 1-hr air quality standard of 110 ppb in the same years. The data in the program also show that the annual PM₁₀ air quality standard of 50 μg/m³ has been exceeded by almost 60%, while the PM_2.5 air quality standard of 15 μg/m³ has been exceeded by a factor of almost 2. These data show the urgent need to adopt effective air quality policies to reduce the concentration of PM_2.5 in the atmosphere of the MMA.

Figure 1. Location of the State of Nuevo León, México, and the Monterrey Metropolitan Area.

Table 1 shows the annual concentration of PM_2.5 in the five monitoring stations in the period 2003 to 2007 (SIMA [Integrated System of Environmental Monitoring], 1993). The data show that the annual PM_2.5 air quality standard of 15 μg/m³ has been exceeded since 2003 in all the monitoring stations in the MMA. In particular, stations NW and SW report high levels of PM_2.5. This is expected, since predominant winds in the MMA are from the east and southeast (SEMARNAT y Gobierno del Estado de N. L., 2008). These two stations reach annual average values around twice the Mexican annual air quality standard for PM_2.5.

Table 1. Annual average PM_2.5 concentration of (μg/m³) measured at five automatic monitoring stations in the MMA during 2003 to 2007

Station	2003	2004	2005	2006	2007	Average
SE	24.910	25.160	NA	NA	32.230	27.430
NE	26.890	30.840	27.630	36.520	36.270	31.630
Center	25.450	29.110	26.560	27.800	28.330	27.450
NW	26.860	28.110	25.420	30.720	30.480	28.320
SW	34.170	34.610	32.520	46.330	36.080	36.740
Annual average	27.660	29.570	28.030	35.340	32.680
Note: Data obtained from Sistema de Monitoreo Ambiental (SIMA), 2008: http://www.nl.gob.mx/?P=med_amb_mej_amb_sima (accessed June 2008).

The high PM_2.5 concentrations measured in the MMA represent a high health population risk due to chronic exposure. A study shows that the exposure to PM₁₀ in the MMA may cause high health care costs (Villa et al., 2008). In the case of PM_2.5 the health costs have not been estimated yet. Although these costs are not known, actions need to be taken to reduce the high levels of PM_2.5 concentration. Different statistical analysis techniques can provide data to determine the source of primary PM_2.5 and its precursors. Among them, chemical composition of PM_2.5 samples and source profiles are used to determine the source apportionment. The results of this analysis are useful to provide information for establishing cost-effective strategies to reduce airborne concentration and, consequently, health risks of the population exposed.

Several sources of primary and precursors of secondary aerosols exist in the MMA. Important sources include a refinery, quarries, chemical and metallurgical industries, construction activities, and traffic resuspension in paved and unpaved roads. The emission inventory for the year 2005 reports that mobile sources contribute 47.5% of NOx and 96.5% of CO; point sources contribute 98% of SO₂ and 32.8% of NOx emissions and 13.5% of PM₁₀ and 27.8% of PM_2.5 primary emissions. Area sources contribute 68% of primary PM₁₀ and 51% of primary PM_2.5 (SEMARNAT y Gobierno del Estado de N. L., 2008). The fraction of primary and secondary PM_2.5 is not known in the MMA. Precursors of secondary aerosol include SO₂, NOx, and organic and elemental carbon. The emission inventory reports that an oil refinery located east of the MMA contributes about 50% the total emissions of SO₂ (SEMARNAT y Gobierno del Estado de N. L., 2008). Other important sources of this pollutant include power plants and fuel consumption by industries in the MMA. These data indicate that SO₂ may have a strong contribution to secondary inorganic aerosol. Mobile and area sources account for almost 92% of VOCs. This pollutant may have a contribution to secondary organic aerosol too. The data of the inventory show that different sources of primary PM_2.5 and its precursors exist in the MMA. It is important to know the chemical composition to detect the relative importance of the different species in the PM_2.5 mass concentration and their origin (primary or secondary).

The MMA is located in the northeast of Mexico. It is the third most populated metropolitan area in Mexico, with 3.5 million inhabitants. The estimated fleet is 1.7 million vehicles in 2008, and the MMA is also the center of intense industrial activities. Despite the severity of the PM_2.5 air pollution problem in the MMA, there are few studies that have addressed this issue. Most of these studies have focused on PM₁₀ and PM₁₅ chemical composition (Aldape et al., 1999; TRC Environmental Corporation, 1995), and few studies have addressed the impacts of PM_2.5 on the health of the MMA population (Peters et al., 2006). This paper provides data on the chemical composition of PM_2.5 in two sites downwind of the MMA, its possible sources, and the apportionment of these sources on PM_2.5 concentration levels. The methods used include statistical and multivariate analysis (factor analysis), and enrichment factor evaluation based on PM_2.5 chemical composition data. The results provide valuable information on source apportionment of PM_2.5 in the MMA.

Experimental Methods

Monitoring Campaign

Two sites downwind of the MMA were selected for sampling PM_2.5. The sites are identified as Escobedo (ESC) and Santa Catarina (SC) and they are approximately 21 km apart. The ESC site was located in the yard of a municipal office in an area surrounded by warehouses. Near this site industrial activities are scarce and there is a highway approximately 1 km north from the sampling site. Heavy-duty and light-duty vehicles circulate on this highway because it is an alternative road that connects the eastern with the western area of the MMA, forming an arc at the north area of MMA. The SC site was located in an area of administrative offices, in an urban area surrounded by manufacturing industry and some chemical industries. These two sites frequently report the highest PM_2.5 concentrations measured in the MMA by the SIMA monitoring stations. Figure 2 shows the location of the stations and the two sampling sites (ESC and SC). In the same figure are displayed the location of the refinery, east of the MMA, and the power plants located to the east, north, and center of the MMA.

Figure 2. Map of the MMA with the location of the sampling sites (ESC and SC), the monitoring stations, the power plants, and the oil refinery.

The samples were taken on different days between November 22 and December 12, 2007. We collected a total of 29 pairs of samples, 14 pairs at the SC site and 15 pairs at the ESC site. Two pairs of low-volume sampling devices by Minivol Airmetrics were used to collect 24-hr samples of PM_2.5 at each site. These samplers (∼5 L/min) have been used before to address mass and chemical composition of fine PM (Chow et al., 2002; Vega et al., 2007). These samplers have shown a high level of confidence in their operation reliability. They have yielded similar concentration levels when they are located with another Minivol or other reference particulate sampler device (Baldauf et al., 2001; Chow et al., 2008; Hill et al., 1999). The inlet flow was calibrated every day during the sampling period using a precalibrated digital manometer (Dwyer^TM).

Collection of PM_2.5 Samples

Samples of PM_2.5 at each site were collected using two different sets of filters. Trace elements and mass were analyzed using 47-mm Teflon filters. Anions, cations, elemental carbon (EC), and organic carbon (OC) were analized using 47-mm quartz filters. The filters were prepared and analyzed by an accredited laboratory in Tigard, OR (Chester LabNet). The filters were conditioned 48 hr at 20 ± 1°C and with a relative humidity of 30 to 40 ± 5%, according to the method of reference for gravimetric use. The field blank filters (Teflon and quartz) were conditioned under the same conditions. All quartz filters were prefired to eliminate residues that could bias the chemical analysis. The filters were analyzed and the mass of the chemicals was subtracted from those measured in the filters used for sampling. Trace elements were analyzed using x-ray fluorescence (XRF) (Watson et al., 1999). Sixteen elements were detected with significant concentrations after substracting the elemental concentrations found in the blank filter. The elements quantified were Al, Si, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Br, and Pb. Ions were analyzed by ion chromatography (IC). The anions analyzed were Cl⁻, nonvolatilized NO₃ ⁻, and SO₄ ²⁻. The cations analyzed were NH₄ ⁺ and K⁺ (Na⁺ was dismissed after subtracting the concentration measured in the blank). Organic carbon (OC) and elemental carbon (EC) were analyzed by thermal-optical transmittance (TOT).

Enrichment Factor

The enrichment factor (EF) is used for assessing the relative contribution of chemical elements different from a typical crust composition. The EF for element i in an aerosol sample is defined by eq (1):

(1)

Here, C_i is the concentration of the ith component and C_Si is the concentration of Si. Hence, the enrichment factor compares the ratio between the concentrations in the environment and crustal material. Elements Al and Fe can also be used as a reference besides Si. We decided to use Si as a reference since, as commented earlier, the emission inventory of the MMA reports that most of the primary particles come from natural sources and mainly from crustal sources with high Si content. We used the relative composition from an average sample of soil given by Mason (1966) as a typical soil composition. This composition has been used with good results as a reference crustal composition in other studies (Balasubramanian and Qian, 2004; Chan et al., 1997; Kang et al., 2004).

Statistical Analysis

Factor analysis (FA) is a multivariate statistical technique that helps to identify and quantify the impact of relevant pollution sources on the measurement (receptor) site. Factor analysis assumes that although many different parameters describing a system can be determined, such as elemental composition or particle size, there are fewer causal forces that result in the state of the system as it is observed. Factor analysis tries to identify these fewer causal forces and interpret their nature (Hopke, 1991). MinTab version 14 was used to perform factor analysis with Varimax rotation. This rotation maintains independence among the factors and produces a loading matrix (Balasubramanian and Qian, 2004). The chemical compositions were compared to detect possible influence of upwind and local sources. The FA technique has been used in several studies related with air particulate matter (Antony Chen et al., 2002; Marcazzan et al., 2001; Wang et al., 2005).

Results

Table 2 shows the measured mean PM_2.5 mass concentrations of all samples, their standard deviation, and their maximum values. This table also shows the same parameters for the different elements analyzed: anions, cations, OC, and EC. We observe that the average concentration of PM_2.5 is similar at both sites. Also, the concentrations of elements are similar except for Al, Cr, Br, Cl⁻, and K⁺, which show the largest differences. The concentrations of OC and EC are similar in both sites too.

Table 2. Average PM_2.5 concentrations (μg/m³) ± standard deviation and maximum

	Escobedo			Santa Catarina
	Average ± SD (μg/m^3)		Maximum	Average ± SD (μg/m^3)		Maximum
Mass	37.780	±12.610	59.290	36.150	±9.880	51.730
Al	0.251	±0.223	0.703	0.124	±0.086	0.262
Si	0.520	±0.374	1.176	0.443	±0.275	0.843
S	1.507	±0.837	3.239	1.674	±0.940	3.476
K	0.379	±0.223	0.775	0.317	±0.134	0.580
Ca	1.696	±1.156	3.577	1.542	±1.179	3.281
Ti	0.021	±0.015	0.052	0.018	±0.009	0.029
V	0.006	±0.008	0.030	0.005	±0.004	0.014
Cr	0.002	±0.003	0.012	0.004	±0.003	0.01
Mn	0.036	±0.030	0.102	0.027	±0.020	0.062
Fe	0.493	±0.359	1.187	0.466	±0.266	0.890
Ni	0.002	±0.002	0.005	0.001	±0.002	0.004
Cu	0.013	±0.009	0.029	0.016	±0.020	0.078
Zn	0.189	±0.106	0.411	0.130	±0.053	0.210
Br	0.026	±0.018	0.061	0.009	±0.005	0.019
Pb	0.072	±0.064	0.223	0.083	±0.058	0.213
Cl^−	0.853	±0.480	1.486	1.117	±0.492	1.932
NO3 ^=	3.619	±2.219	8.049	3.855	±2.528	8.473
SO4 ^=	6.395	±3.207	13.233	5.883	±2.842	12.362
NH4 ^+	0.987	±0.841	2.992	1.023	±0.892	2.765
K^+	0.322	±0.205	0.645	0.231	±0.118	0.399
OC	13.599	±6.658	24.067	12.081	±4.630	19.185
EC	2.735	±0.975	4.058	2.623	±1.398	5.361

Figure 3 shows a comparison of the PM_2.5 daily concentration at the two sites during the field campaign. Due to a failure in the sampling devices, the filters of the first day at the SC site were dismissed. We observe similar values of PM_2.5 concentration at both sites. The measured PM_2.5 concentration at the ESC site ranges from 14.7 to 59.3 μg/m³ and at the SC site from 17.8 to 51.2 μg/m³. The PM_2.5 mean concentration at ESC was 37.8 ± 12.6 μg/m³ while at SC it was 36.15 ± 9.9 μg/m³.

Figure 3. Concentration of PM_2.5 measured at the two sampling sites (ESC and SC).

Figure 4 shows the wind roses for the sampling period prepared with data obtained from SIMA. We observe a predominant pattern of winds with high speeds coming from the east and southeast. We would expect that the samples collected would include mainly primary PM_2.5 emitted from different sources east and within the MMA, as well as secondary PM_2.5 formed by precursors emitted from the same locations upwind. These sources include the refinery, power plants, and chemical plants in the MMA, as well as natural sources.

Figure 4. Wind roses during the sampling period: (a) northwest station and (b) southwest station.

Chemical Composition

We performed a correlation analysis to evaluate the quality of the data (Chow et al., 1996; Vega et al., 2004). Two field blanks were collected and their values were subtracted from the samples analyzed. As expected, the sum of individual chemical species was less or equal to the measured mass, as shown in Figure 5. The same figure shows that the correlation values between cations and anions was r²  = 0.81, between soluble potassium (K⁺) and K was r ² = 0.91, between chloride (Cl⁻) and Cl was r ² = 0.55, between calculated NH₄ ⁺ and measured NH₄ ⁺ was r ² = 0.94, and between SO₄ ²⁻ and S was r ² = 0.97. The calculated NH₄ ⁺ was estimated using the equation reported by Zhang et al., (2002). This equation assumes that the ammonium present in particulate matter exists mainly in the form of ammonium sulfate, (NH₄)₂SO₄, and ammonium nitrate (NH₄NO₃). In general, good quality of data was obtained since high correlation coefficients were found in the data validation process.

Figure 5. Determination coefficients of components in the PM_2.5 samples.

Aerosol samples were treated assuming that they are formed by seven components: crustal matter, salts, ammonium sulfate, ammonium nitrate, EC, organic matter (OM), and trace elements not included in crustal matter. This classification has been used in other studies and is based on the mass balance and typical PM_2.5 sources (Chan et al., 1997; Kovouras and Koutrakis, 2001; Zhang et al., 2002). Ammonium nitrate and sulfate are an estimation of inorganic secondary aerosol. The concentration of ammonium nitrate is estimated as 1.29 × NO₃ concentration and ammonium sulfate as 1.374 × SO₄ concentration. Organic matter is obtained as 1.2 × OC concentration in order to compensate for the mass of elements like oxygen, nitrogen, and hydrogen. Some authors suggest a value of 1.4 or even 1.6 (El-Zanan et al., 2009; Turpin and Lim, 2001). In this work we used a factor of 1.2 because it presented a better adjustment to the total daily mass concentrations obtained from the SIMA monitoring stations nearby the sampling sites. Crustal material was estimated as 1.89 × Al + 2.14 × Si + 1.4 × Ca + 1.43 × Fe concentrations. Trace elements were obtained adding all the elements from Na to Pb except Al, Si, Ca, Fe, Cl, K, Na, and S. Mineral salts were estimated as a function of ion Cl⁻ affected with a factor of 1.65 (Vega et al., 2004).

The contributions of these components in PM_2.5 mass are displayed in Figures 6 and 7. The average contributions considering all samples at both sites are displayed in Figure 6. In this figure we observe that the OC contributes the highest percentage, 42%, followed by secondary inorganic aerosol with 36% (including ammonium sulfate, 23%, and ammonium nitrate, 13%). Crustal material accounts for 11% of PM_2.5 mass and EC contributes 7%. Contributions from salt and trace elements were small, accounting for 3 and 1% of the measured PM_2.5 mass, respectively. Figure 7 shows the contribution in percentage of secondary aerosol, crustal matter, elemental carbon, and organic matter for each site. As can be observed, the chemical compositions at both sites are very similar. It is important to note the small contribution of crustal matter and the high contribution of secondary aerosol and organic matter. Part of this organic matter may be secondary, as discussed in the next sections. In the case of OM, the contribution was 43% and 41%, respectively, at the ESC and SC sites, while secondary aerosol accounted for 35 and 36 %. Contributions of crustal material were 13% and 11% and EC contributed 6 and 7% of PM_2.5 at the ESC and SC sites, respectively. Only salt was higher at the SC site (5%) and lower at the ESC site (1%). Trace elements accounted for 1% at each site.

Figure 6. Relative contribution of chemical species in the PM_2.5 samples, general average.

Figure 7. Relative contribution of the chemical species in the PM_2.5 samples at each site: (a) Escobedo (ESC) and (b) Santa Catarina (SC).

Enrichment Factor (EF)

Using eq 1, we calculated enrichment factors (EF) for the different components in the PM_2.5 samples. Figure 8 shows these factors for different elements for the ESC and SC sites. Three groups can be differentiated according to the calculated enrichment factors (Biegalski et al., 1998). Elements with EF > 50 are in the first group. In this group we found S, Cl, Zn, Br, and Pb. These elements are typically related to anthropogenic sources, that is, vehicle exhaust and industries (Chan et al., 1997). The second group is formed by elements with EF in the range 3 to 50. This group includes K, Ca, Ti, V, Cr, Mn, Fe, and Ni. These elements can have either an anthropogenic or crustal source. The third group is formed by elements that have a crustal source and EF < 3. In this group we found Al, Si, and Ti.

Figure 8. Enrichment factor (EF) at the ESC and SC sites.

Primary Organic Carbon (POC)

The EC trace method was used to estimate de primary and secondary contribution of OC in the samples collected in the MMA (Turpin and Huntzicker, 1995). All the data from the two sites were used on the calculation of the primary organic carbon (POC). The method is based in the fact that EC and OC tend to have the same emission sources. Hence, it is possible to estimate a primary ratio between OC and EC for a given area. This ratio is represented by (OC/EC)_p, where the subscript p stands for primary. In the case where the ambient ratio of OC/EC is greater than (OC/EC)_p, the additional OC is expected to be secondary (Strader et al., 1999). There are different approaches to obtaining the primary OC/EC ratio (Cabada et al., 2004; Castro et al., 1999; Gray et al., Turpin and Huntzicker, 1995; Turpin et al., 1991). Typically, the lowest values of the OC/EC ratio are used to estimate the POC. In this study we used the lowest 20% of ambient OC/EC values to obtain a Deming regression estimate for the POC (Baumann et al., 2008; Cabada et al., 2004; Castro et al., 1999; Gray et al., 1986; Strader et al., 1999; Turpin and Huntzicker, 1995; Turpin et al., 1991). The samples taken in days where rain was present were eliminated to avoid deviations in the data analysis (Cabada et al., 2004). The OC values used in this study were corrected using the blank data. The POC is estimated using eq (2) (Turpin and Huntzicker, 1995):

(2)

The contribution of secondary organic carbon (SOC) is obtained subtracting POC to the total measured OC. In eq (2), b represents the noncombustion contribution of OC to the POC and the possible contribution of sampling artifacts (Cabada et al., 2004).

We applied the Deming regression to the data in Table 2. The use of Deming regression has shown better results in estimating the slope and intercept when the set of data (variable x) is not a totally controlled variable (Turpin and Huntzicker, 1995). The result obtained using the data of the samples is shown in eq (3):

(3)

A negative value of the intercept b [−(1.07±3.13)] represents a correction of POC for other EC not associated with OC or noncombustion EC, including meat broiling operations and biogenic sources (Baumann et al., 2008, Cabada et al., 2004). The uncertainty in the estimated value of this parameter suggests that there might be EC from noncombustion sources. Future research is necessary to obtain a more accurate value of the possible contribution of noncombustion sources on POC.

Some studies have not included the parameter b in their analysis. This would suggest that OC from sources different than combustion are ignored. In this work we estimated the percentage of SOC formation for the two cases. If b is considered in the calculation of POC, the estimated SOC average contribution in the field campaign is 34% (calculated from total OC). If b is removed and SOC is recalculated, the average contribution of SOC is 24%. Therefore, we can expect that the range of POC in OC is between 24% (minimum 16% and maximum 39%) and 34% (minimum 0% and maximum 68%). In the MMA there are not previous studies of organic aerosol formation. The impact of the noncombustion sources in the formation of SOC is unclear. From our results we can say that the average daily concentrations of SOC formation are between 3.3 μg/m³ to 4.7 μg/m³ (24% to 34% of total OC). Table 3 shows a summary of the results for the lowest 20% of ambient OC/EC ratio measured from the field campaign data and the estimated OC and EC content.

Table 3. Data used for Deming regression (the lowest 20% of OC/EC values)

Date	Site	OC (μg/m^3)	EC (μg/m^3)	OC/EC
29/11/2007	SC	3.805	16.802	4.415
27/11/2007	ESC	3.612	15.928	4.409
06/12/2007	SC	4.411	18.690	4.236
10/12/2007	ESC	1.524	6.032	3.956
07/12/2007	SC	5.361	20.500	3.823
21/11/2007	SC	3.499	12.867	3.676
21/11/2007	ESC	3.208	11.747	3.661

Factor Analysis

The factor analysis was applied to a set of 14 variables selected from the 29 samples. After applying VARIMAX rotation, three principal components with eigenvalues > 1 were selected (Hopke, 1991). Variables with a commonality less than 0.5 were eliminated. Table 4 shows the loads after rotation. Factor 1 explains 45% of total variance. This factor groups elements from geological sources (Al and Si) and it also includes products of combustion as EC and OC. These may come from vehicle fuel combustion. Among others elements, Br and Zn have been associated with fine particles emitted in the exhaust gases of vehicles and are also included in Factor 1 (Kyotani and Iwatsuki, 2002). Soluble potassium is associated with biomass burning and is detected in this factor too. Factor 2 groups V, S, and SO₄ ⁻² and explains 20.1% of total variance. Vanadium is considered a tracer of fuel oil combustion (Cass and McRae, 1983). Sulfur is associated with emissions from electric power plants, fuel combustion in industries, and refinery activities (Chu, 2005). This is consistent with the emission inventory of the MMA as it was commented before. Cl and Cr are grouped in Factor 3. This factor explains 16.4% of total variance. We consider that sea salt is not a probable source due to the distance of the ocean. Therefore, we believe that chlorine and chromium are associated with industrial activities. The enrichment factor greater than 1000 that we found for chlorine confirms this hypothesis.

Table 4. Factor analysis for trace elements, ionic species, OC, and EC in the PM_2.5 samples

	Rotated component matrix
	Factors extracted
Variables	1	2	3
V	0.097	0.708	0.468
Cl	0.201	0.07	0.896
Cr	0.171	−0.142	0.88
Al	0.846	−0.25	0.054
Fe	0.839	−0.254	0.348
Si	0.846	−0.35	0.188
Zn	0.822	−0.063	0.216
Br	0.77	−0.011	−0.056
S	−0.188	0.94	−0.094
PM2.5 mass	0.842	0.306	0.338
SO4 ^2-	−0.19	0.949	−0.164
K +	0.824	0.043	0.064
OC	0.929	−0.116	0.201
EC	0.703	−0.379	0.313
Variance	6.298	2.817	2.297
Percent of variance	45.0%	20.1%	16.4%
Percent of cumulative variance	45.0%	65.1%	81.5%
Note: The numbers in boldface correspond to the variables grouped in Factor 1, 2, or 3.

Discussion

As reported in the air quality program of the MMA, predominant winds in the region are from the east and southeast (SEMARNAT y Gobierno del Estado de N. L., 2008). Winds coming from the southeast flow around the mountains, then change direction to the west and flow over the MMA. Being downwind, the two sampling sites capture a mixture of primary and secondary PM_2.5 as a result of emissions of different sources. Thus, gases and particles emitted from a refinery, located southeast of the MMA as shown in Figure 2, are transported and mixed with the emissions of power plants (located at the northeast, north, and center of the MMA), with the emissions from vehicles in the city, and with the emissions of industries located inside and surrounding the MMA. As a result, a mixture of NOx, SO₂, CO, VOCs, and other pollutants forms a haze over the city with an important content of primary and secondary PM_2.5, the latter formed from precursors emitted in the refinery east of the MMA and by sources within it.

Despite the high levels of PM_2.5, little is known about the contribution of the different sources. Important decisions to control air pollution depend on available information about the problem. The results of this study show that the refinery and industrial, mobile, and natural sources have an important contribution in the high levels of PM_2.5 found in the MMA (SEMARNAT y Gobierno del Estado de N. L., 2008). The chemical data and the enrichment factor analysis show that secondary inorganic aerosol is an important component of PM_2.5. Nitrates, sulfates, and organic matter are important components of the secondary aerosol. The factor analysis shows consistency of the chemicals identified and the related emission sources with influence in the MMA, as the refinery, power plants, and combustion processes. An important contribution of soil, a natural source, was identified. This material is a primary component of PM_2.5 and may come from natural sources and quarries located in the MMA, as well as resuspension of dust caused by the circulation of vehicles in the city.

Based on the results of this study, we can conclude that about 40% of PM_2.5 is formed by secondary species, mainly ammonium sulfate and ammonium nitrate, emitted by anthropogenic sources. Crustal matter and salt contribute 14% of total mass. The sources of these components are mainly natural and may be produced by quarries and dust resuspension by traffic. The results obtained are not totally conclusive and we cannot fully determine the source apportionment of PM_2.5. However, the results presented here provide an evaluation of primary and secondary aerosol contribution in PM_2.5 concentration levels in the MMA. This is the first study to estimate the contribution of different sources in PM_2.5 concentration in the MMA. More studies on chemical characterization, source apportionment, and airborne dynamics are necessary to have a better understanding of the impacts of emissions. Samples of PM_2.5 must be collected in other seasons to detect variations on PM_2.5 chemical composition and to detect the contribution of its possible sources under different meteorological conditions in the MMA.

Conclusions

Mass concentration and chemical composition of PM_2.5 were analyzed in samples taken at two sites of the Metropolitan Monterrey Area, Mexico. The samples were taken between November 22 and December 12, 2007. PM_2.5 average mass concentration and its standard deviation at the ESC site were 37.8 ± 12.61 μg/m³. The maximum value measured at this site was 59.29 μg/m³. The SC site had an average and standard deviation of PM_2.5 mass concentration of 36.5 ±9.88 μg/m³. The maximum value measured at this site was 51.73 μg/m³. Although the average values found were below the 24-hr air quality standard of 65 μg/m³, the annual average PM_2.5 concentration calculated with data measured by the monitoring stations has been above the annual air quality standard of 15 μg/m³ the last 5 years in all the monitoring stations.

Despite the sampling sites being separated by 21 km, the PM_2.5 chemical composition at both sites was similar. We found that the most abundant elements in PM_2.5 mass were S, Ca, Ni, Si, K, Zn, Al, and Pb. The OC was the most abundant component and constituted 42% of the average PM_2.5 mass, followed by ammonium sulfate and ammonium nitrate accounting for 36%. Crustal material constituted only 11% of the average PM_2.5 mass. SOC was estimated in the range of 24% to 34% of total OC. The factor analysis supports the hypothesis that emissions from a local refinery, traffic, power plants, and industrial activities have an important contribution in the formation of secondary aerosol. Natural sources are also an important source of primary PM_2.5. However, more studies are necessary to achieve more conclusive source apportionment results for PM_2.5. These include understanding PM_2.5 seasonal variations and composition in other sites at the MMA.

Acknowledgments

We gratefully acknowledge the scholarship of PROMEP for the development of this work, the Air Quality Research Program at ITESM for financial assistance, and the APMRN of the State of Nuevo León for assistance in the sampling sites.