Source Apportionment and Characterization of Particulate Matter ( PM 10 ) in Urban Environment of Lahore

The present study describes the characterization of carbonaceous species including elemental carbon (EC), organic carbon (OC), total carbon (TC), crustal (Al, Ca, Mg, Fe, S, and Ti) and trace metals (As, Ba, Cd, Cr, Cu, Mn, Ni, Pb, Sn and Zn) in PM10 samples collected from an urban site in Lahore, Pakistan. Sources of various pollutants and their characterization positive matrix factorization (PMF) model. The carbonaceous species (TC, OC, EC) and metals were measured in PM10 samples by NIOSH protocol using Sunset lab instruments and ICP-OES respectively. PM10 concentrations varied from 254 to 555 μg/m with an average of 406 ± 87 μg/m. The elemental carbon (EC) concentration varied from 3 to 56 μg/m with an average of 21 ± 15 μg/m. While the organic carbon (OC) concentrations varied from 21 to 212 μg/m with an average of 63 ± 42 μg/m. The OC/EC ratio varied from 1.5–7.6 with an average of 3.9 ± 1.6, indicating a contribution of both biogenic and secondary aerosol formation. A good correlation was also observed between EC and OC (R = 0.81) indicating their common origin. PMF has identified industrial dust (18.2% of PM10), vehicular emission (26.5% of PM10), bio mass fuel (24.3% of PM10) and re-suspended dust (4.6% of PM10) as major sources of PM10 in urban environment of Lahore.


INTRODUCTION
Particulate Matter (PM) has been recognized as a major factor in determining global climate change (IPCC, 2007).As an important component of air pollution, PM has been linked to various adverse health effects (e.g., Dockery et al., 1993;Khan et al., 1993;Gwynn et al., 2000;Pope et al., 2002;Husain et al., 2007;Lodhi et al., 2009;Khan et al., 2010;Stone et al., 2010;Kim et al., 2011) as well as having general environmental effects.
Airborne particulate matter (APM) with an aerodynamic diameter less than or equal to 10µm (PM 10 ), have been studied intensively over the past few decades (Saleem et al., 2008).The sources of PM 10 are different, which include a wide range of natural phenomena and human activities.These particles (PM) mainly originate from sea salt, soil dust, non-exhaust vehicle emissions, construction, and industrial fugitives etc.The particles with greater diameter (> PM 10 ) are settled quickly as compared to those ranging from some nm to tens of µm which remain buoyant in the atmosphere for days.Thus such particles can be transported over a long distance from the original source resulting in an enhanced level of ambient PM concentrations even at rural or background sites.Agriculture crops residue burning (ACRB) is serious issue for both health and environment perspectives, due to ACRB there is significant increase in the concentration of PM in North West India (Awasthi et al., 2010).
There has been growing concern on almost all levels of APM in Pakistan in the recent years.According to (Husain et al., 2007;Lodhi et al., 2009;Stone et al., 2010), industrialization, transportation and agriculture are among the key anthropogenic activities in Pakistan which have resulted in rapid deterioration of air quality due to growing levels of PM, heavy metals, and other air pollutants in the ambient environment.Moreover, it is of great interest that sources of particulate matter vary greatly with season, place and weather conditions (Ghosh, 2002;Zheng et al., 2005;Joksic et al., 2009;Roy et al., 2012).In particular, the decreasing trend in global temperature of recent years has been attributed mainly to the rising concentration of air-borne particles (PM) in the atmosphere.Furthermore, during stable meteorological conditions such as inversion, with low wind speeds, the highest PM concentrations are often reported (Pohjola et al., 2004).
Generally, it has been observed that the total particulate mass has a lower temporal variability than any of the major components of aerosol.Some of the authors hypothesized that this finding may be associated with opposite seasonal concentration variations in some of the major PM components (Kao and Friedlander, 1995).
Paved and unpaved roads have been proposed to be an essential contributor to the total PM 10 in many urban and sub-urban areas.A wide range of studies have shown that traffic-induced resuspension is the predominant source of coarse particles (Pakkanen et al., 2001b;Ruellan et al., 2001;Manoli et al., 2002;Sternbeck et al., 2002).Although the level of air pollutants are high in densely populated cities and industrial areas, their effect is widespread rather than localized (Gurjar et al., 2008).Kupiainen et al. (2003 and2005) investigated the effects of road sanding on the formation and concentrations of urban suspended road dust.They found that concentration of suspended PM 10 increased subsequent to the application of traction sand.In a compiled study of seven European cities, Querol et al. (2004) reported that mineral dust, combustion and secondary aerosols were important PM 10 sources.
Generally in Asia and specifically in South Asia, anthropogenic emissions of air pollutants which are linked with rapid urbanization and industrialization severely affect the environment and human health (Lee et al., 2006).Consequently, this leads to increase health problems, high mortality rates (Ostro et al., 2001;Pope et al., 2002;Chen et al., 2005;Oftedal et al., 2008;Gan et al., 2011) and adverse changes in climatic conditions (Sun et al., 2004;Ahmed et al., 2006).Some of these effects are explained in detail by (Hussain et al., 2011).Although vehicles and industrial emissions are the two main contributors of the ambient urban PM level, however, contributions from other pollution sources such as roadside dust, power plants, solid waste, and trans-boundary migrations can't be ruled out.Tyagi et al. (2012), have recently reported that PM 10 concentration was in the range of 959 µg/m 3 and 422 µg/m 3 in urban and sub-urban areas of Northern India whereas among metals, the highest concentrations were observed for Cr, Fe, Mn, Zn, and Al at rural sites as compared to industrial sites.Some of the recent studies cited in the literature (Hameed et al., 2000;Rattigan et al., 2002;Husain et al., 2007) reported about the air quality in the city of Lahore.Very high concentrations of aerosols associated with OC, EC and sulfate were found; moreover their results suggested that a large component of the carbonaceous aerosols in Lahore originated mainly from fossil fuel combustion and biomass burning.Previous studies of ambient air quality in the same region have reported total suspended particle concentrations of 900 µg/m³ (Ghauri et al., 2007).Furthermore black carbon (BC) values of 110 µg/m³ (Husain et al., 2007), 24-h maximum springtime PM 10 concentrations of above 460 µg/m³ (Zhang et al., 2008a), and wintertime PM 2.5 concentrations of 209 µg/m³ (Biswas et al., 2008) were reported.Additionally, an early study by (Smith et al., 1996) concluded that the PM in Lahore contains high levels of toxic and carcinogenic components such as polyaromatic hydrocarbons (PAH) and lead (Pb).The present study is focused on the characterization of carbonaceous aerosol and their source apportionment on the basis of PM 10 in the metropolitan Lahore.

Site Description and Collection of PM 10 Samples
Lahore (31.320°N; 74.220°E) is the second largest city of Pakistan, with a population of approximately 12 million.It is situated along the Ravi River, close to the Indian border (Fig. 1).The climate in Lahore is hot and semi-arid, with relatively wet and extremely hot summers and dry, warm winters.The mean maximum temperatures in summer (April June) range between 33 and 39°C and in winter from 17 to 22°C (months).The major industries in Lahore include the manufacturing of motor cars, motorcycles, steel, chemicals, pharmaceuticals, engineering products, and construction materials.The aerosols over this sampling site derive mainly from soil or road dust, industrial emissions, and vehicular emissions, or are secondary aerosols.Other anthropogenic sources include emissions from main highways, coal combustion and biomass burning (Biswas et al., 2008).The total suspended particles (TSP) samples were collected at an urban site in Lahore, shown in Fig. 1 during the month of March 2010.This site is largely influenced by road transport.
A set of 25 particulate matter (PM 10 ) samples were collected from an urban site in the center of Lahore city during March 2010.Quartz fiber filters with diameter 147 mm were used as sampling substrate.The detailed procedure of sample collection has been described elsewhere (Mukhtar and Limbeck, 2011).The collected PM 10 samples were stored in Petri dishes with Para film seals until analysis to avoid any kind of contamination and moisture absorption.

Analysis of Carbonaceous Species and Metals
Carbonaceous species (TC, OC, EC) were measured by NIOSH protocol using Sunset lab instruments, originally described by M. E. Birch and R. A. Cary (1996).An iCAP 6500 series ICP-OES spectrometer (Thermo Scientific, USA) was used for simultaneous multi-element analysis (Al, As, Ba, Ca, Cd, Co, Cr, Cu, Fe, Mg, Mn, Ni, P, Pb, S, Sb, Se, Sn, Sr, Ti, Zn).All used chemicals and reagents were of analytical reagent grade and were procured from Merck (Darmstadt, Germany).Metal standard solutions were prepared by diluting custom assurance multi element standard solutions from Spex Certiprep (NJ, USA) and Merck.The preparation of PM 10 samples and measurement of selected crustal and trace metal analysis were carried out according to Mukhtar and Limbeck, 2011.

Positive Matrix Factorization (PMF)
Positive Matrix Factorization (PMF), a multivariate receptor based model developed by Paatero and Tapper (1994) was applied to the measured results of the PM 10 samples, collected during March, 2010, in order to identify sources and their contributions to the receptor's site of PM 10 .The analysis of PM data by means of PMF receptor models is a well-established and Widely used technique for source identification and apportionment (Watson et al., 2005;Hopke et al., 2006).PMF allows effective use of information in the data and provides a flexible modeling approach.It has been applied successfully worldwide in receptor modeling and has proven useful in targeting sources for emission reduction programs (Paatero et al., 1997;Reff et al., 2007;Ulbrich et al., 2009;Chen et al., 2010) and such as that Unmix model (Henryet al., 2003;Song et al., 2007;Henry et al., 2010) have been used widely.
In general, the PMF receptor model assumes there are p sources contributing to a receptor site.This can be mathematically stated as where, X pq = Concentration of species q in p th sample g pl = Contribution of l th factor to the p th sample f lq = Fraction of l th factor that is species q e pq = Residual for the q th species in p th sample To estimate the contributions (g pl ) and source profiles (f lq ), PMF uses a constrained, weighted, least squares method.The task of EPA PMF is to minimize the sum of squares of standardized (residual divided by corresponding uncertainty value) residuals (Q) where, a = Total No. of samples.b = Total No. of species.s pq = Uncertainty for q th species in p th sample.
Eq. ( 1) can also be written in matrix form as where, M = Matrix of Measured data with dimension "no. of samples" M "no. of species".g = Contributions Matrix with dimension "no. of samples" M "no. of factors".f = Source profiles Matrix with dimension "no. of species" M "no. of factors".e = Matrix of residuals with dimension "no. of samples" M "no. of species".Matrix of the measured concentrations 'M' and uncertainty matrix 's' are inputs for the PMF model whereas the matrices 'g', 'f' and 'e' are obtained as output data.
The source contribution matrix "g" was further utilized for source apportionment by taking into account the measured PM 10 mass.To perform the quantitative source apportionment, a scaling coefficient, y k , is introduced in the model Eq. ( 1) so that y k was be determined by multi-linear regression of computed source contribution against measured PM 2.5 mass.Constant of linear regression was assumed to be zero.
After determination of y k values, the final scaled source contributions were determined.PM 10 mass was again calculated from scaled source contribution.
EPA-PMF version 3.1.1,a software tool based on the PMF model, was used in the current study.After several runs, EPA-PMF 3.1.1revealed the five most interpretable sources with a minimum Q value.
The dataset used in this study consisted of 25 samples.Data quality categorization was based on the signal to noise ratio (S/N).Out of 25 samples, 22 samples which have S/N > 2.5 were categorized as having strong data quality, while the remaining 3 samples with S/N between 0.39 to 1.63 were categorized as having weak data quality.

Carbonaceous Species and Metals
Carbonaceous species, crustal and trace metals were analyzed in the collected set of aerosol samples.PM 10 concentrations varied from 254 to 555 µg/m 3 with an average of 406 µg/m 3 .These concentrations are relatively higher than in other cities in the regions like Hangzhou-China (119 µg/m 3 ), New Delhi-India (219 µg/m 3 ), Kolkatta-India (197 µg/m 3 ), Punjab-India (116 µg/m 3 ) (Gupta et al., 2007;Cao et al., 2009;Tiwari et al., 2009;Awasthi et al., 2011).There are a large number of sources for EC e.g., two stroke vehicles, diesel engines, fossil fuel combustion, low burning efficiency, coal fired power plants and biomass burning.EC is also used as tracer for vehicular exhaust (Goetschi et al., 2002).According to Ghauri et al. (2007) and Zhang et al. (2008) traffic emissions, gasoline and diesel motor vehicles, are a major source of EC, followed by biomass combustion.OC can be emitted directly from sources known as primary carbon as result of fossil and biomass combustion, or can be produced as a result of a chemical reaction known as secondary organic carbon (Seinfeld and Pandis, 2006;Shen et al., 2007).The EC concentration varied from 3 to 56 µg/m 3 with an average of 21 µg/m 3 .While the OC concentrations varied from 21 to 212 µg/m 3 with an average of 63 µg/m 3 .Organic mass (OM) was calculated OM = OC  1.6 (Turpin and Lim, 2001).The OC/EC ratio may be used as an indicator of changes in emission sources, source processes or source regions.A very good correlation was observed between EC and OC (R 2 = 0.81) indicating their common origin.All these concentrations have been given in the Table 1.
Several studies have used OC/EC ratio exceeding 2 to identify secondary organic aerosol (Gray et al., 1986;Turpin , 1990;Hildemann et al., 1991;Chow et al., 1996).But higher OC/EC values from 4-11 give indication that a certain fraction of OC is derived from biogenic sources like biomass burning and wood combustion.The OC/EC ratio varied from 1.5-7.6 with an average of 3.9 ± 1.6, indicating contribution of both biogenic and secondary aerosol formation.
The concentrations of crustal and trace metal ranged from few ng/m 3 to some µg/m 3 in the investigated set of PM 10 samples.Among the crustal elements (Al, Ca, Mg, Fe, S, and Ti), the lowest concentrations were observed for Ti, varying from 0.1 µg/m 3 to 0.3 µg/m 3 with an average value of 0.2 ± 0.1 µg/m 3 , whereas the highest concentrations were found for Ca, varying from 11.5 µg/m 3 to 32.4 µg/m 3 with a mean value of 18.5 ± 5.8 µg/m 3 .Among the trace elements (As, Ba, Cd, Cr, Cu, Mn, Ni, Pb, Sn and Zn), As was found to have the lowest concentrations, ranging from 2.0 ng/m 3 to 8.0 ng/m 3 with an average value of 5.0 ± 3.0 ng/m 3 .The highest concentrations were observed for Zn varying from 1.0 µg/m 3 to 5.7 µg/m 3 with a mean value of 3.0 ± 1.9 µg/m 3 .The results reported in the current study were in accordance to literature findings from some mega South Asian Cities.For example, Schneidemesser et al. (2010) have also reported the highest concentrations of the crustal element Ca in PM 10 samples collected from an urban site in Lahore with annual mean of 9.1 ± 2.5 µg/m 3 , whereas the lowest concentrations were reported for trace element Co, with an annual mean of 3 ± 1 ng/m 3 .Moreover, the annual mean concentration of Zn was reported as 11 ± 8 µg/m 3 which is significantly higher than the one reported in current study.Similarly, Venkataraman et al. (2002) have found that the concentrations of crustal and trace elements in PM 10 samples collected from Mumbai, India ranged from 0.77 ± 0.43 µg/m 3 (Zn) to 3.33 ± 0.83 µg/m 3 (Al).

Source Apportionment Using PMF Model
PMF version 3.1.1was used to identify the optimal number of sources that contributed to the PM 10 mass in Lahore.Different numbers of factors were examined and the optimum number was found to be five ( 05), based on the results that sufficiently fit to the elemental concentrations.Sources identified are characterized as Industrial dust, Vehicular emission, Biomass burning, Coal combustion, and Re-suspended dust that contributed 18.2%, 26.5%, 24.3%, 26.3% and 4.6% respectively to PM 10 mass (Fig. 2).
Fig. 3 shows comparison between measured (gravimetrically measured mass collected on filter) and predicted PM 10 mass.The predicted mass of PM 10 was calculated from the sum of scaled source contribution values for each sample.The PMF model result appears to be reasonably good for the elemental data of particulate matter utilized.The correlation coefficient (R) between the measured and model predicted PM 10 mass was 0.80 with a slope of 0.97.This indicate that the resolved factors effectively accounted for most of the variations in PM 10 mass concentration.

SOURCE 1
The first source is identified as industrial dust, which has a major contribution from OC, EC, Cu, Cr, Cd, Fe, Mn and Ni, which results in 18.2% average mass of PM 10 .These heavy metals contribute greatly to industrial emissions in the urban and sub-urban areas (Gomiscek et al., 2004;Cloquet et al., 2006;Konarski et al., 2006;Lee et al., 2006).Furthermore, the rubber, steel and leather industries are assumed to be a substantial particulate contributor to the Present site.One of the reasons for this high concentration was the existence of some metal manufacturing plants and heavy industries, located near the sampling site.Iron and steel industry exhaust contributes a significant amount of PM 10 to ambient air (Mansha et al., 2012).Querol et al. (2007) recently performed a study in Spain and found the highest ambient concentrations of Cu in urban backgrounds (PM 10 ) with respect to industrial sites (ceramic and petrochemical industries).
According to the results of source apportionment, the PM 10 concentration related to Fe and Cd was 39.5% and 59.5% respectively and little periodic variation was observed for the rest of the profiles.Possible reasons could be tire abrasion, incineration, and combustion of fuels and lubricants (Fergusson et al., 1991;Jaradat et al., 1998;Figueroa et al., 2006).We could observe, a significant contribution from As and Ni as well in this Profile, referring to oil burning in vehicle and thermal power plants (Mansha et al., 2012).Ni may be associated with their wide use for the combustion of heating fuel (Vallius et al., 2005).Moreover, a small concentration was observed from Zn, Ec, Sb, Pb, Sn, As, Co and Ba.

SOURCE 2
Vehicular emission is identified by high loading of EC, OC, As, Pb, Cd, S and Fe, and contributed 26.5% to the PM 10 mass.Vehicle exhaust is an important particulate contributor to ambient air in urban areas (Mansha et al., 2012).Elemental carbon (EC) and organic carbon (OC) are major constituents of smoke from incomplete combustion processes, in urban areas mostly from road traffic and are responsible for the soiling characteristics of particles.These elements possess very good adsorptive properties and are produced as a semi-volatile compounds, or formed through atmospheric chemical reactions (Bowman et al., 1997).The authors hypothesized that this finding was mainly related to about 60-80% of the total carbon (Harrison et al., 1995), but their compositions and concentrations can vary greatly due to local geology, meteorology, surface conditions and, human activities, like traffic and construction etc. (Castro et al., 1999).
One of the reasons for a higher concentration of OC/EC is the continuous use of Diesel fuel in Pakistan, which contains high amounts of sulphur and Lead, resulting in the direct emission of sulphur and Lead compounds to the atmosphere, along with higher concentrations of OC/EC.Moreover, in rural areas of Punjab or around Lahore, biomass fuels are commonly used for cooking purposes for many years.Due to the shortage of natural gas supply in the country for the last two years, the use of biomass as fuel has also increased, ultimately resulting in an increase of S and Pb concentrations.However, our results are in line with the findings by (Gugamsetty et al., 2012), who reported a 24.92% average 'vehicle emission' contribution to PM 10 at Shinjung station, Taiwan.The Zn concentration in this profile is assumed to be much higher than that in the other sources, the reason is that Zn possibly comes from the tire wear or fuel burning (Pacyna et al., 1986;Klimaszewska et al., 2007) and contribute as the fingerprint of vehicular emission sources (Li et al., 2004;Hailin et al., 2008) in recent PMF analysis.

SOURCE 3
This source was characterized by high abundances of several metals Ti, Ca, Zn and Al with major contributions of EC and OC, Biomass fuel contributed to about 24.3% of PM 10 mass on average.It includes most likely firewood, coal and wood burning, household combustion of agricultural residues (Hailin et al., 2008), emissions from sugar mills, resuspended dust, use of fertilizers in farming, agricultural activity and emissions from brick furnaces, along with naturally occurring suspended dust.Among all, wood burning was identified as one of the most significant sources of bio mass burning emission at the sub urban sites.Overall mean contributions to PM 10 at the present sites were 56-60% (Zn), 26-34% (OC), 20-24% (Al) and 30-38% (Cd and Cu), while small contributions from S and Cd are also present.Moreover, no major contributions were anticipated from the "other", (mainly As, Co, Fe,) particle fraction.Al may come from fluid leakage combustion from diesel (as fuel) in vehicles, as well as brake wear and road construction near the site (Klimaszewska et al., 2007).Presence of Cd may be due to the wood and coal combustion at the present site.Moreover, the sampling was carried out during the month of March, where low temperature and wind speed also favored the accumulation of pollutants, resulting in high concentrations profiles during the investigated period.This shows the influence of meteorological conditions on the PM 10 concentration.Hailin et al. (2008) also reported a similar factor for the Beijing, (China) PM source apportionment.

SOURCE 4
The fourth source is characterized as coal combustion, identified by a high loading of EC, OC, TC, Ca, Zn along with Al and Mg.These metals mainly have an important contribution in the coarse particle fraction (Harrison et al. 1997;Lough et al., 2005).Coal combustion is the main source of As in Pakistan, but the decreasing trend was observed at the present site, which may be associated with the reduction of coal consumption in the urban areas.These results are consistent with the findings of (Hailin et al., 2008) for Beijing, (China), whereby they reported that it may be due to some coal consumption in the suburban areas and in nearby cities of the investigated site.Coal combustion as a result of electricity generation, industrial processing and residential use is a significant particulate source category for ambient PM 10 .This source seems to be dominated by Al (18.9%) and TC (39%) as shown in Fig. 4. Xue et al. (2010) also reported a similar factor for Panzhihua, China at six different sites for the PM source apportionment.
Moreover, Coal is mainly used in coal fired power plants and in brick kiln industries for manufacturing bricks in this region.A large number of coal fired power plants in India are along the border with Pakistan, while in Pakistan such plants are near to Faisalabad.Coal is also used for cooking purposes by a small number of population.Coal combustion accounted about 26.3% of total PM 10 mass on average.

SOURCE 5
Source 5 includes high loadings of EC, OC, Tc, Pb, Ca, Al, Ba, which refer to soil and road-side dust (Fig. 4).The re-suspended soil dust contribution (4.6%) of the total PM 10 mass on average may be from un-metalled roads, road pavement erosion and grass free belts.Ca may probably be from intensive construction activities in Lahore for several years.The Pb enrichment in this source indicates their presence in lower graded fuel production (Mansha et al., 2012) by most of the oil refineries in Pakistan.Enrichment of crust elements /soil trace elements i.e.Al, Ca, Fe, Mg and Ti can be attributed to being major constituents of air born soil and fugitive dust.These metals mainly have an important contribution in the coarse particle fraction (Harrison et al., 1997;Lough et al., 2005).Konarski et al. (2006) recently investigated the concentrations of Pb and found them to be almost 20 times higher at the urban site, in comparison to the rural environment.Presence of such sources and their association with increased Ca and Zn concentrations comply with the findings of Amato et al. (2009), Figueroa et al. (2006), and Li et al. (2001).Furthermore, a minor contribution from Ni, Cu, Pb, As and Cd was also found in the present source.

CONCLUSION
The current study reveals that the average PM 10 , EC and OC concentrations were 406 µg/m 3 , 21 µg/m 3 and 63 µg/m 3 respectively in urban environment of Lahore.A good correlation was observed between EC and OC (R 2 = 0.81) indicating their common origin.The OC/EC ratio varied from 1.5-7.6 with an average of 3.9 indicating a contribution of both biogenic and secondary aerosol formation.The contributors of EC and OC in urban areas are mostly from road traffic.The use of Diesel fuel in Pakistan with high sulphur and Lead, results in direct emission of sulphur and lead compounds to the atmosphere.
Main sources identified during the study are industrial dust (18.2% of PM 10 ), vehicular emission (26.5% of PM 10 ), bio-mass fuel (24.3% of PM 10 ) and re-suspended dust (4.6% of PM 10 ).The rubber, steel and leather industries are assumed to be a substantial particulate contributor to the present site.Whereas, wood burning has been identified as one of the significant biomass contributors in PM 10 at sub urban sites.Coal combustion for electricity generation, industrial processing and residential use is an important source of coarse particle contribution.Other coarse mode PM is largely attributed to crustal sources like un-metalled roads, road pavement erosion and grass free belts.

Fig. 1 .
Fig. 1.Map of Pakistan showing the sampling site of Lahore.

Fig. 4 .
Fig. 4. Sources profiles for the five factors resolved by PMF.

Table 1 .
Range and Mean Values for PM 10 and different metals.