On the Morphology and Composition of Particulate Matter in an Urban Environment

Particulate matter (PM) plays a vital role in altering air quality, human health, and climate change. There are sparse data relevant to PM characteristics in urban environments of the Middle East, including Peshawar city in Pakistan. This work reports on the morphology and composition of PM in two size fractions (PM2.5 and PM10) during November 2016 in Peshawar. The 24 hous mass concentration of PM2.5 varied from 72 μg m to 500 μg m with an average value of 286 μg m. The 24 hours PM10 concentration varied from 300 μg m to 1440 μg m with an average of 638 μg m. The morphology, size, and elemental composition of PM were measured using Fourier Transform Infra Red (FT-IR) Spectroscopy and Scanning Electron Microscopy (SEM) with Energy Dispersive X-ray (EDX) Spectroscopy. The size of the analyzed particles by EDX ranged from 916 nm to 22 μm. Particles were classified into the following groups based on their elemental composition and morphology: silica (12%), aluminosilicates (23%), calcium rich (3%), chloride (2%), Fe/Ti oxides (3%), carbonaceous (49%), sulfate (5%), biogenic (3%). The major identified sources of PM are vehicular emissions, biomass burning, soil and re-suspended road dust, biological emissions, and construction activities in and around the vicinity of the sampling site.


INTRODUCTION
The size, shape, and composition of ambient particulate matter (PM) affects their ability to interact with both solar radiation and water vapor, which in turn leads to varying effects on climate, cloud formation and precipitation, and public health.Coarse particulate matter (PM 10 ), fine particulate matter (PM 2.5 ), and ultra-fine particulate matter with diameter less than 0.5 µm have different characteristics, sources, and potential health effects.Natural sources of PM tend to lead to larger particles, such as with windblown dust, sea salt, crustal material, volcanic emissions, and biological particles.Smaller particles tend to be produced by secondary processes, such as by gas-to-particle conversion, stemming from both anthropogenic and biogenic emissions of precursor gases.Inorganic salts, such as ammonium nitrate and ammonium sulfate, and secondary organic aerosol (SOA) are examples of secondarily produced PM.Some air pollutants are locally generated while others are transported by wind and cover very long distances after emission from their source point.There is a clear association between the concentration of PM and health effects (Dockery et al., 1993;Wilson and Suh, 1997;Kunzly et al., 2000;Alam et al., 2011), particularly the fine PM fraction that can more easily penetrate into the lungs and cause respiratory diseases (Schwartz et al., 1996;Youn et al., 2016).Black carbon (BC) in particular can damage the cells of the human body and possibily carcinogenic to humans (Baan, 2007).
To understand the effects of PM on human health and climate, characterizing the composition, sources, transport, and fate of aerosol particles is of great importance.The morphological and chemical characterization of PM has received increasing attention in recent years because such data are needed to accurately constrain aerosol radiative properties (Adachi et al., 2010) and health impacts (Ghio and Devlin, 2001).Monitoring of both PM mass and chemical composition is important for identification of emissions sources, determination of compliance for air quality standards, bridging the knowledge gap between air quality and its health effects, and establishment of effective pollution control programs (Kgabi et al., 2008).PM contains a myriad of components such as inorganic salts (e.g., ammonium sulfate, ammonium nitrate, sodium chloride), elemental and organic carbon, biological materials, iron compounds, trace metals, and minerals resulting from rocks, soils and construction (Satsangi et al., 2014).Bulk aerosol composition measurements only provide information about the elements or ions present, whereas size-resolved and single particle analyses yield improved insight into sources and formation mechanisms for species, and their resulting influence on climate and human health (Bérubé et al., 1999;Li et al., 2010a).
Direct evidence of the composition and morphology of aerosol particles can be provided by single particle analysis (Utsunomiya et al., 2004;Ro et al., 2005;Geng et al., 2010).The morphology of atmospheric particles has received increasing attention in recent years due to effect of the particle shape on their radiative (Adachi et al., 2010) and chemical properties (Ghio and Devlin, 2001).Although many particle shapes exist in the atmosphere, such as isometric particles, platelets, and fibers, the difficulty to measure morphological properties often leads to the assumption of particle sphericity for most applications (Reist, 1993).An added complication is that particles with irregular morphology can collapse into a spherical shape upon humidification (Shingler et al., 2016).Scanning Electron Microscopy (SEM) combined with Energy Dispersive Spectrometry (EDX) provides useful information about the elemental composition, size and shape of PM, and thus is a useful technique in distinguishing particles originating from different sources (Salma et al., 2001).
The goal of this work is to report on the characterization of PM in an urban environment in the Middle East, specifically Peshawar city in Pakistan.Samples were collected during a dense haze event in November 2016.The objectives of the current work are to report on the morphology of PM, and frequency of different particle types based on their classification into categories according to their elemental composition and morphology.The present study will help to understand the air quality of the city and the climatic implications of the ambient aerosol.Furthermore, this work adds to the currently limited inventory of morphological data for ambient aerosols in urban environments, especially in the Middle East.

Site Descriptions
Peshawar is the historical and capital city of the Khyber Pahktunkhwa (KPK) Province, and is located in the north of Pakistan (34.03°N, 71.56°E; Fig. 1).It is 359 m above sea level and covers an area of 1,257 km 2 .The population of the city is about four million and increasing continuously owing to migration of people in search of jobs and education (Alam et al., 2015).The rapid urbanization has promoted more vehicular traffic due to a larger number of motor vehicles used for public and private transportation.There are various industries in this city including textiles, paper, furniture, pharmaceuticals, steel, cigarettes, cardboard, and food processing.Peshawar is also a major center for the steel industry in Pakistan (Alam et al., 2015).Peshawar is hot during the summer (May-August) with a mean maximum temperature of 40°C.It has cold winters (November-March) with a maximum mean temperature of 10°C.
Sampling was performed in the Pakistan Meteorological Department (PMD) building, which is close to a roadside with a height of 5 m above ground level.PMD lies in the core of the old city and is exposed to emissions from high vehicular traffic.

Formation of Haze and Meteorological Conditions
The meteorological conditions of Peshawar city are summarized in Fig. 2.There was no rainfall (RF) in Peshawar during November 2016 (i.e., 0 mm), and that is the only month throughout the year in which there was no RF.During the study period, the recorded minimum temperature varied from 8°C to 14°C with an average value of 10°C, while the maximum temperature ranged from 18°C to 31°C with an average of 26°C.The relative humidity (RH) ranged from 49 to 85% with an average value of 66%.The wind speed (WS) was almost calm during the study period with some minor WS of 0.3 knots.The wind direction (WD) over the study site was predominantly northwesterly (NW).
Every year, haze episodes occur between October and November in northern Pakistan, particularly at the southern slopes of Himalaya, which spread over ~100 km in the region.The haze period investigated in this study in November 2016 was captured by the NASA Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) satellite, as shown in the Fig. 3.The red dots in the image are the hotspot locations, where MODIS measured high surface temperature due to the agriculture burning, which consequently contributed to the formation of haze.During autumn and winter seasons, biomass burning is very common, which when combined  with industrial and vehicular emissions produces spatially extensive haze in the region.

Sample Collection
A Low Volume Sampler (LVS) was used with two heads for both PM 2.5 and PM 10 collection.PM 2.5 and PM 10 were sampled during November 2016 for 30 consecutive days, amounting to 60 total samples of PM 2.5 and PM 10 .The LVS was operated at a constant flow rate of 16 LPM.PM was collected on quartz fiber filter substrates with diameter of 47 mm.The LVS sampled daily for 24 hours from 8 am to 8 am.Pre-weighed and conditioned filters were placed in the filter holder and screwed properly before operating the sampler.After sampling, the PM filters were removed with forceps, stored in a Petri dish, conditioned, weighed, and stored in the refrigerator at 4°C to prevent thermal degradation and evaporation of volatile components prior to further analysis.The gravimetric PM masses were calculated by subtracting the initial average mass of the blank filter from the final average mass of the sampled filter by using a microbalance.The following steps were used for the calculation of gravimetric mass of pre-and post-sampling: 24 h to equilibrate filters before weighing at room temperature, followed by weighing during the following 24-48 h.Each filter substrate was weighed three times before and after sampling, and the mean value was calculated.

Sample Analysis Scanning Electron Microscopy (SEM)
Samples were analyzed by SEM-EDX at the Centralized Resource Laboratory in the Department of Physics at the University of Peshawar.SEM-EDX analysis was carried out with the help of SEM (JSM-5910, JEOL Japan) equipped with an energy dispersive X-ray system (INCA-200, UK).SEM combined with EDX gives information about surface morphology, size, and chemical composition of ambient PM 2.5 and PM 10 .On the basis of the chemical composition and morphology, the PM produced from different sources can be distinguished using this technique.Other researchers have also investigated morphology and chemical composition using SEM-EDX (Chabas and Lefevre, 2000;Ma et al., 2001;Liu et al., 2005;Li et al., 2011).SEM uses an electron beam and is a high-resolution surface imaging technique.
For SEM sample preparation, sections of 1 mm by 1 mm were cut with scissors from the center of each selected filter substrate, and mounted with double-sided tape on an aluminum SEM stub for analysis.With the help of a vacuum coating unit called Gold Sputter Coater (SPI-MODULE, USA), a very thin layer of gold (Au) was deposited on the surface of each sample, in order to achieve better conductivity and reduction of electron charge.The sputter can prepare six samples simultaneously.The samples were placed in the corner of the SEM-EDX chamber and two images of each sample were taken.To analyze coarse particles, a microscope magnification of 550 was used, allowing detection fields of 60 × 150 μm.Fine particles were analyzed with a magnification of 2000, which yielded a detection field of 60 × 50 μm.Ten random particles were selected on each field and about 20 fields per each selected filter were observed, giving approximately 200 manually characterized particles per filter.In total, 12,000 particles were manually analyzed for 60 samples.For the analysis of particle morphology and location, the back-scattered electron mode was used.For every signal there is an installed detector that detects its corresponding signal and rejects others.The background or unwanted signals were therefore blocked in this way.The working conditions of the equipment were as follows: accelerating voltage = 0.5-30 kV; probe current = 50-100 μA; Si (Li) detector 20 mm away from the samples to be analyzed; X-ray detection limit = ~0.1%;X-ray spectrum acquisition time of 60 s live time.All images acquired for each field and particle were used to manually measure the morphological parameters such as particle shape and physical diameter (minimum, maximum and mean).For each examined filter substrate, the results were taken from three randomly selected fields, which provided representative results to minimize subjectivity.
For the determination of individual elemental composition of PM particles, EDX spectra of individual particles were obtained after scanning an electron beam.The different peaks were identified and the quantifying function of the computer program was used to obtain the peak intensities, which were converted to percentage weight (Pipal et al., 2014).As the samples were Au-coated, the Au data of EDX could not be used for quantitative purposes.Consequently, the Au contribution was manually subtracted during the evaluation of the EDX spectra.

Fourier Transform Infrared (FT-IR) Spectroscopy
FT-IR spectra were acquired with Perkin Elmer Spectrum Two equipped with UATR (Universal Attenuated Total Reflectance).Before taking data for a sample, a background scan was conducted.Each spectrum is an average of 200 scans and quantifies the transmission of infrared radiation as a function of wave number.Spectra were recorded in dry purge air using a 4 cm -l resolution over the spectral range from 4000 to 450 cm -l .Spectra of the individual quartz filters were measured before air sampling to be used as background reference spectra.The filter spectra were measured again after air sampling.The ratio of the spectra to the open-beam spectra at the time of measurement was calculated and stored as absorbance data files.A FT-IR spectrometer was used to analyze samples directly on quartz filters without sample preparation.The spectrum of the loaded filter automatically subtracted from that of unloaded filter, yielding information about the PM on the filter.FT-IR spectroscopy was used to identify various types of functional groups and bonds in materials at molecular levels.This method has been used by the previous researchers for detection of organic functional groups such as carbonyls, aliphatic carbons, nitrates, and sulfates (Marley et al., 1993;Maria et al., 2002).

Particle Morphology and its Chemical Nature
In the current study, particle morphology and composition were correlated based on the method widely applied by others (Sharma and Srinivas, 2009;Cong et al., 2010;Pipal et al., 2011;Pachauri et al., 2013;Satsangi et al., 2014).The analyzed particles were further classified into three groups' i.e.Geogenic, Anthropogenic, and Biogenic.Geogenic (Natural) particles consist mostly of soil dust (minerals) and biogenic particles.Soil particles have irregular shapes and rough surfaces and sometimes form aggregates with irregular shapes and sizes.Anthropogenic particles emitted from combustion processes were predominantly spherical and rounded with smooth surfaces.Spherical particles of Fe/Ti comprised completely of Fe/Ti oxide were characterized by nearly perfect sphericity, indicating that their origin was from metallurgical activities.Biological particles (pollen, spores and plant fragments) exhibit a high content of carbon and oxygen and are often characterized by minor contributions of elements such as Na, Mg, P, K and Ca.They are characterized by regular and symmetrical shapes, ranging from spherical to elliptical shapes.

Element Percent Weight Calculation
EDX analysis of each individual particle provides the percent weight of individual elements.EDX analysis of blank quartz fiber filters was also performed and the spectra results were manually subtracted from those of the real individual aerosol particles.The EDX spectra of each individual particle give percent weight of different elements.An average weight percent of each element was calculated for both PM 2.5 and PM 10 .The total number of particles belonging to each group was determined and then the percentage of each group has been calculated.

Mass Concentration of PM 2.5 and PM 10
The quantification of PM mass concentration is the main criteria for the assessment of air quality.Both PM 2.5 and PM 10 were collected with the LVS from 1 November to 30 November 2016.Fig. 4 shows the daily variation of PM 2.5 and PM 10 during the study period.The value of PM 2.5 varied from 72 to 500 µg m -3 with an average value of 286 µg m -3 .PM 10 ranged from 300 to 1440 µg m -3 with an average value of 638 µg m -3 .The WHO limits for PM 2.5 and PM 10 are 25 µg m -3 and 50 µg m -3 , respectively (Alam et al., 2015).In the current research work it was found that PM 2.5 and PM 10 were 11 and 13 times greater than the WHO daily permissible limit, indicating poor air quality for Peshawar city.Meteorological conditions play a very important role in governing PM mass concentrations.High values of relative humidity (RH) and temperature, in addition to no rainfall (RF), contribute to the high PM concentrations.Wind from the NW direction, which is exposed to industrial sites, also contributes to high PM values.Reasons for high PM mass concentration include re-suspension of road dust, vehicular emissions, industrial emissions, brick kiln emissions, and residential combustion (Alam et al., 2011(Alam et al., , 2015)).In contrast, the following are PM values in other regions: PM 2.5 /PM 10 = 23.22 ± 4.72 µg m -3 /51.79 ± 12.63 µg m -3 in Mexico City between 2003-2015 (Mora et al., 2017), 28.4 ± 25.4/87.3 ± 47.3 µg m -3 in Jeddah (Saudi Arabia) (Khodeir et al., 2012), 32 ± 6/121 ± 12 μg m -3 in Udaipur, India in April 2010 (Yadav et al., 2014), 90/278 μg m -3 in Agra, India (Pipal et al., 2014), 21.82/39.45μg m -3 in Shinjung, Taiwan (Gugamsetty et al., 2012), and 38 ± 12/70 ± 31 μg m -3 in townships in South Africa (Hersey et al., 2015).Mass concentrations of PM 2.5 and PM 10 were observed to be higher during working days (Monday-Saturday) as compared to the weekend day (Sunday) as shown in Table 1.Fig. 4 also shows that PM concentrations were low on the four Sundays during November 2016 (i.e., 6, 13, 20, and 27 November).Similar results with reduced PM levels on weekend days have been observed in other parts of the Middle East such as Ahvaz, Iran (Maleki et al., 2016) and Tehran, Iran (Crosbie et al., 2014).This significant difference is due to enhanced anthropogenic activity during working days, including especially vehicular activity and industrial emissions.

Elemental Composition of PM 2.5 and PM 10
Fig. 5 summarizes the distribution of weight percent of fourteen elements (i.e., O, C, N, Si, Ca, Fe, Al, Cl, K, Mg, S, Na, Ti, Zn) in both PM 2.5 and PM 10 based on EDX spectra.Quartz fiber filters consist of Si and O 2 ; therefore, those two elements were subtracted from the loaded filters.The weight percent of O, C, N, and Si exceeded those of other elements.The high percentage of C is due to carbonaceous material coming from anthropogenic activities such as vehicular traffic, diesel generators, and industrial emissions (Shahid et al., 2013), in addition to biomasss burning (e.g., Duong et al., 2011).One of the major sources of Si is soil mineral particles (e.g., Prabhakar et al., 2014).Therefore, the presence of Si particles in our study is from some likely combination of transported windblown soil particles, resuspended dust from vehicular traffic or construction, and from industrial combustion (Houghton et al., 2001 , 1997).Since particulate C offers attractive adsorption sites for various volatile compounds, its higher amount can promote deleterious health effects (Begum et al., 2012;Pipal et al., 2014).

Major Groups of Particulate Matter
On the basis of the elemental composition and morphology of PM obtained through SEM-EDX results, sampled particles were classified into the following major groups: Geogenic, Anthropogenic, and Biogenic.A discussion of these groups follows subsequently.

Geogenic Particles
Geogenic particles are natural particles that are of crustal origin.They are comprised of quartz, aluminosilicates, calcium-rich particles, chloride-based particles, and Fe/Ti oxides.

Silica/Quartz
The most common name for quartz (SiO 2 ) particles is silica, and they consist of high amounts of Si and O (≈ 50% by weight) (Pachauri et al., 2013).Pure Si particles are produced naturally as well as artificially (Li et al., 2010b).In the Earth's crust, Si is the most abundant chemical constituent and it is a component of sandstone and granite.The basic origin of quartz is soil but it is also found in building materials such as cement, glass, ceramic, bricks, and clays.These particles represent 12.3% of the total particles collected (Fig. 6).The size range of silica is from 2 to 20 µm.In this group, particle types such as like pyrope, grossular, almandine, biotite, and fly ash were identified.A triangular shape associated with biotite is shown in Fig. 7(b).Fly ash has spherical shape (Fig. 8(d)).These particles were also identified at Agra (India) by Pachauri et al. (2013).

Aluminosilicates
The major source of aluminosilicates is soil sediments that  are present in the highest percentage of the total analyzed Geogenic particles.Therefore, dust is the dominant source of aluminosilicates, while other sources include agricultural activities, fuel combustion and biomass burning (Satsangi et al., 2014;Byeon et al., 2015).Aluminosilicates account for ~72% of the total chemical compounds in the Earth's crust (Van Malderen et al., 1996;Cong et al., 2009).It is evident from our study that soil-derived aluminosilicate particles are generally composed of Si and Al oxides with varying amounts of Na, K, Mg, Ca, Fe, and Ti.Aluminosilicates represent 23.1% of the total particles collected (Fig. 9).Their size ranges from 1. Calcium-rich Particles Particles in the calcium-rich group expectedly exhibit a high percentage of Ca, with sources including re-suspended dust, crustal materials from vehicular traffic and construction activities, and windblown dust.According to Lough et al. (2005), calcium particles are emitted from construction of roads, houses, and buildings.Calcium carbonate particles are a major source of Ca in ambient PM (Ro et al., 2001;Li et al., 2003).As shown in Fig. 8(a), pentagon shaped carbonate minerals, like Calcite (CaCO 3 ), were identified along with the traces of other soil-related particles.These particles account for 3.1% of the total analyzed particles (Fig. 6).

Chloride Particles
Chloride particles were present in the form of C-Cl and Na-Cl.Fig. 7(b) shows evidence of a particle with tablet-like morphology and a size of 5.6 μm, with traces of C, O, N, and Si.These particles account for 1.5% of the total particles (Fig. 6).There are other chloride particles such as KCl from biomass burning (Braun et al., 2017).Characterization of aerosols at Pune (India) indicated that Cl, Na, Mg and K were soil-derived (Kulshrestha et al., 1998;Parmar et al., 2001).Shrivastava et al. (2009) noted the presence of chloride particles in aerosols over Delhi (India).

Fe/Ti oxides Particles
This group of particles is assumed to stem from soil emissions.The sources of the high concentrations of Al, Ca, Fe, Mg, Mn, Sr, Ti, and Zn are re-suspended road and soil dust (Watson et al., 2001;Gugamsetty et al., 2012).The particles collected from this category were rich in Fe, Ti, and O with the combination of mineral dust elements such as C, Al, K, Si, Ca, Na, and Mg.This group of particles mostly corresponds to Fe oxides with irregular morphology.The particles of this group account for 3.1% of the total particles examined (Fig. 6).The size range of these particles is

Anthropogenic Particles
Carbonaceous and industrial particles are included in this group of particles.The origin of these anthropogenic particles is local emissions.

Carbonaceous Particles
Carbonaceous particles contribute significantly to the total mass of the analyzed samples.The major source of carbonaceous particles in the study site is vehicular emissions.The morphology of these particles ranged from soot to complex structures, and depended on different factors such as burning conditions and fuel type (Bérubé et al., 1999;Posfai et al., 2004;Cong et al., 2009;Posfai and Buseck, 2010;Tumolva et al., 2010).
Soot is the combination of small individual spheres and has various sources of production like burning of diesels, coal, oil, and biomass (Wang et al., 2013).A mixture of Cbearing particles and other constituents such as Si, Na, Mg, K, Cl, Ca, and Al were also observed during the present study.The percentage of these particles is 49% of the total particles, as shown in Fig. 6.The size of these particles varied from 216 nm to 22 µm, with variable morphologies.A single C particle with spherical morphology, in which the contribution of C and O exceeded 90%, is shown in Fig. 9(a).Earlier research showed that these spherical particles may both scatter and absorb solar radiation (Hand et al., 2005;Alexander et al., 2008;Cong et al., 2009).Similar to our results, a single carbon particle of completely spherical shape was found by Pachauri et al. (2013) in Agra (India).
Clusters of carbonaceous particles were also observed as shown in Fig. 9(b).The sources of these particles are the burning of fuels and biomass.These particles may also be produced due to agricultural burning and waste incineration (Li et al., 2010b).The excessive use of motorbikes and motorcycles in Peshawar largely produces Zn from the combustion of lubricating oil (Begum et al., 2013).The extensive use of diesel engines in Peshawar emits high concentrations of C. The congested, slow, and jammed traffic in Peshawar produces continuous smoke and soot (Alam et al., 2015).The major source of black carbon (BC) in the atmosphere is the incomplete combustion of fossil fuels and biomass burning (Sahu et al., 2012).

Sulfate Particles
In the study site, sulfate particles are produced secondarily from its precursor, SO 2 , emitted via combustion of fossil fuel and biomass burning.S-related particles are present in all samples of airborne PM (Buseck and Posfai, 1999;Posfai et al., 2003).SO 2 can be adsorbed on mineral particle surfaces and form secondary minerals (Li and Shao, 2009).Diesel combustion inside heavy duty vehicles increased the concentration of S and Pb in the atmosphere, which is highly hazardous to human health.The percentage of sulfate particles was 4.6% of the total particles, as shown in Fig. 5.The size range of the sulfate particles was from 2.5 µm to 8.4 µm.The sulfate particles also possess capsule-like morphology with sizes on the order of 8.4 µm (Fig. 9(c)).
Using EDX analysis, morphology and elemental composition were determined.Particles of biological nature, whether dead or alive, consisted of minor amounts of elements like Na, Mg, K, P, Si, Fe, Cl, Al, and Ca, which are tracers present in plants (Matthias-Maser and Jaenicke, 1994;Matthias-Maser et al., 2000a, b;Posfai and Buseck, 2010).For the analysis of each biogenic particle, the following clustering method was used as determined by Coz et al. (2010): Bioaerosol: (C + O) > 75% and 1% < P; K; Cl < 10% Such particles of biological nature were selected in this group by using the above clustering techniques.Viruses, bacteria, fungal spores, pollen, plant debris and animal matter are included in the biological particles.These particles represent 3.1% of the total particles (Fig. 6).Single biological particles having spherical shape and size of 4.2 µm are shown in the Fig. 8(e).In these particles of biological nature, C and O are present in major amounts and elements such as Na, Mg, K, Ca, and Cl were found in smaller amounts.Such types of particles have been reported by many researchers (Matthias-Maser, 1998;Yeo and Kim, 2002;Cong et al., 2009;Coz et al., 2010;Chen et al., 2012).

Fourier Transform Infra-Red Spectroscopy (FTIR)
FTIR spectra were collected in transmission mode for each filter by averaging 200 absorbance scans at wave numbers from 400 to 4500 cm -1 with a resolution of 4 cm -1 .Quartz filters were scanned prior to use, and the resulting spectra were subtracted from scans after sampling to obtain the absorbance of the sampled aerosol.Filter support rings  were etched to ensure that alignment was maintained during consecutive scans.
Four loaded samples from PM 2.5 and three from PM 10 , along with unloaded samples were selected for FTIR analysis from different days.FTIR spectra of both PM 2.5 and PM 10 were similar (Fig. 10).The absorption band at 671 cm -1 shows the presence of calcium sulfate (CaSO 4 ) and the absorption band at 713 cm -1 indicates the presence of nitrate ions (i.e., NH 4 NO 3 ) (Allen et al., 1994).Si-O bending is located at 800 cm -1 for all samples (Simensen et al., 2009).The band at 880 cm -1 is indicative of calcite (Bahadur et al., 2010).Silicate shows its presence at 1040 cm -1 (i.e., C-O stretching) (Maria et al., 2002).Peaks at 1320 cm -1 and 1420 cm -1 are indicative of NO 3 -and NO 4 + ions, respectively (Coury et al., 2008).The presence of C-C aromatic skeletal stretching is indicated by the peak at 1620 cm -1 and the PM also contains water, which exhibits absorbance at this wavelength (Simao et al., 2006).Aldehyde and ketones reveal their presence at the 1700 cm -1 band (Coury et al., 2008).The peaks at 671 cm -1 and 2111 cm -1 represent CO 2 and CO, respectively, which suggests that the evaporative fraction contains organic and elemental C ( Colthup et al., 1990;Simao et al., 2006).

Identification of Aerosol Sources
Based on SEM-EDXand FT-IR analysis, the following sources of aerosols were identified, which contributed to the aerosol loading over Peshawar:

Combustion sources (Vehicular Emissions/Biomass Burning)
In the current study, carbonaceous particles were observed to be the most abundant.Since sampling was carried out near a roadside and railway tracks, the main source of carbonaceous particles was vehicular emissions.Biomass burning is the other source because in autumn season biomass burning is common in the region.Soot particles originate from the combustion of burning oil, gasoline, diesel, fuel oil, paraffins, and butane; therefore, they are an important tracer for vehicular traffic.Sulfur particles (S) were also found in the current study.Sulfur mainly originated from the burning of fuel and deposited on the existing minerals due to sulfurization process (Satsangi et al., 2014).Similarly Fe/Ti oxides were also found in the current work.The particles contained high amounts of Fe in the form of oxides and alloys, accompanied by other metals such as Cu, Zn, K, Pb, Ni, and Cr, which originated from traffic and industrial sources (Moreno et al., 2003).As no significant industrial source was identified within the direct proximity of this site, traffic emissions were considered as the main source of these particles.

Soil Dust/Re-suspension of Road Dust
In the present study, aluminosilicates were found in highest amount after the carbonaceous particles.Likewise chloride particles were also found in the form of C-Cl and Na-Cl.Re-suspended road and soil dust are the main sources of the high concentrations of Al, Ca, Fe, Mg, Mn, Sr, Ti, and Zn (Watson et al., 2001;Gugamsetty et al., 2012).Kulshrestha et al. (1998) and Parmar et al. (2001) characterized aerosols at Pune (India) and reported that Cl, Na, Mg and K stemmed from soil in their analyzed samples.

Cement/Limestone
In the current study Ca-rich particles were present as CaCO 3 (limestone) with other elements like Al, K, Zn, Cu and Si.Limestone is a mineral found in crust, soils and road dust, and is widely used as a building material and in cement manufacturing.According to Lough et al. (2005), calcium particles are emitted from construction of roads, houses, and buildings.Due to construction activities and re-suspended dust in the vicinity of sampling site, Ca-rich particles were found in the present study.

Biological Particles
Biological particles were also investigated in the present study.The sources of these particles are viruses, bacteria, fungal spores, pollen, plant debris, and animal matter.The sampling site was surrounded by a large number of plants, trees, and agricultural land.

CONCLUSIONS
Due to increased urbanization, industrialization, transportation, and constructions of roads, buildings, and overhead bridges in Peshawar, Pakistan, PM 2.5 and PM 10 concentrations have increased significantly and are in excess of WHO limits.The aerosols emitted from regional sources have significantly harmed the air quality of the city.The following are the major findings of the present study, which included filter measurements during November 2016: • During the study period, the 24 hours average mass concentrations of PM 2.5 and PM 10 were 252.6 and 771.8 µm, respectively.The PM 2.5 and PM 10 concentrations measured in Peshawar were 10 to 16 times higher than the recommended WHO limits, respectively.• Morphological analysis of PM revealed that the following groups of particles were present in Peshawar: geogenic, anthropogenic, and biogenic.In geogenic particles, aluminosilicates were the main contributor.Among anthropogenic aerosols, carbonaceous particles were the most abundant due to extensive fuel combustion and biomass burning.Biological particles were very few in number.• Aerosols emitted from vehicular exhaust were the main contributor to the poor air quality.The remaining aerosols were mainly generated from road dust and fossil fuel combustion.• Bonds between different components were determined by FTIR Spectroscopy.

Fig. 1 .
Fig. 1.Map of Peshawar showing the sampling site.

Fig. 3 .
Fig. 3. MODIS-AQUA satellite images during the study period (November 2016) on various days.The circle in each map indicates the sampling location.

Fig. 5 .
Fig. 5.The average percent weight of the various elements in Peshawar city during November 2016.

Fig. 6 .
Fig. 6.Frequency of particle types observed during the study period.
from 5.2 to 17.5 µm.A single iron, titanium-oxide particle of spherical morphology having a size of 4.8 µm is shown in Fig.8(C).Similar iron titanium oxide particles of spherical morphology were reported byPachauri et al. (2013) in Agra (India).

Fig. 8 .
Fig. 8. SEM-EDX spectra: (a) Ca-rich particle having a size of 17.1 µm with a pentagon-like shape; (b) Chloride particles of size 5.6 µm and tablet-like shape; (c) an iron titanium oxide particle with traces of elements such as Ca, K, Mg, Na, Al, and Si having size of 4.8 µm and of spherical morphology.

Fig. 9 .
Fig. 9. SEM-EDX spectra: (a) A single carbon particle with completely spherical morphology dominated by C and O (> 90%) having a size of 6.4 µm; (b) A cluster of carbonaceous particle with traces of soil-related elements; (c) A sulfate particle of capsule-like shape having length of 14 µm and width of 2.8 µm; (d) A spherical particle with smooth texture made up of Al-Si-O (fly ash) of size 14 µm; (e) Spherical particles of biological nature having size of 4.2 µm.

Table 1 .
; Herman et 24 hours average PM concentrations during working (Monday-Saturday) and weekend (Sunday) days.