Chemical Speciation and Source Assignment of Particulate ( PM 10 ) Phase Molecular Markers in Mumbai

Particulate matter (PM10) has emerged as the single most important pollutant across Indian cities, as its levels exceed the regulated standards at most places. PM10 was collected and analyzed at seven sites representing different land use patterns in Mumbai during 2007–08 for organic carbon (OC) and elemental carbon (EC). OC was further speciated for organic markers, which are useful for identifying sources. Average PAH concentration ranged from 47.84 ± 14.07 to 247.70 ± 163.19 ng/m. n-alkanes concentration varied from 157.20 ± 35.28 to 759.13 ± 451.65 ng/m. Hopanes and steranes showed the highest concentrations of 94.14 ± 28.66 and 32.64 ± 11.79 ng/m and lowest concentrations of 7.81 ± 2.00 ng/m and 2.83 ± 0.73 ng/m, respectively. The values for levoglucosan varied from 0.69 ± 0.46 to 3.23 ± 2.07 μg/m. The percentage contribution of the sum of the concentrations of these compounds to the total organic carbon varied from 1.7–5.1%. The sources contributing to particulate PAHs were identified using diagnostic ratios. Indicators like the Carbon Preference Index (CPI), Cmax and Plant Wax Number (%WNA) were used to identify the relative contributions to n-alkanes from anthropogenic and biogenic emissions. The source apportionment results and also the concentration trends for these molecular markers indicate that anthropogenic sources, especially vehicular exhaust, wood combustion and coal combustion, are mainly responsible for the organic fraction of particulate matter in Mumbai city. The sources identified qualitatively in this work can help in selecting the appropriate source profiles for estimating the quantitative contribution of these sources.


INTRODUCTION
Urbanization, mixed land use comprising of industries, commercial establishments and associated increase in energy demands have resulted in a profound deterioration of urban air quality in India.The increasing concern for particulate matter in urban air of Indian cities is evident from stricter PM 10 standard for industrial areas and promulgation of a new PM 2.5 standard by Government of India's Ministry of Environment and Forest in November 2009.The complexities of PM sources in Indian urban centers are due to highly variable sources and their locations.
The organic aerosols have received major attention as they play an important role in both direct and indirect aerosol forcing.Organic aerosols are also associated with a range of adverse health effects (Wolff and Klimisch, 1981).Organic carbon (OC) itself is made up of hundreds of individual compounds like alkanes, alkenes, carboxylic acids, hopanes, steranes and aromatic compounds like Polycyclic Aromatic Hydrocarbons (PAHs) (Schauer et al., 1996).Based on the consumption of fuels in different sectors in India, the emissions of carbonaceous aerosols amounts up to 2.8 Tg (Streets et al., 2003).The study of these individual organic compounds can provide in-sight into the sources of OC.Compounds associated with specific source classes are commonly referred to as Molecular Markers (Simoneit, 1984(Simoneit, , 1999)).PAHs are a major cause of concern as these are ubiquitously present in air, soil, sediments, water and food.According to US Department of Health and Human Service's Agency for Toxic Substances and Disease Registry (ATSDR), these compounds are carcinogenic and mutagenic to humans and animals (ATSDR, 1995) and are emitted by variety of sources, some of which are automobiles, resuspended soils, refineries and power plants.16 PAHs have been included in US EPA's list of 188 Hazardous Air Pollutants (Khaiwal et al., 2008).Based on global atmospheric emission inventory of 16 PAHs for the year 2004, India ranks 2 nd with emissions of 90 Gg/year and also has higher proportion (3.6%) of hazardous high molecular weight (HMW) PAHs emissions than global average (Zhang et al., 2009).Commonly used methodologies for identifying and characterizing the contribution of different sources include Diagnostic ratios and Principal Component Analysis (PCA) (Khaiwal et al., 2006).n-alkanes are another subgroup of the carbonaceous material which can originate from both man made sources like vehicular exhaust, natural gas combustion, coal combustion, wood burning, cooking emissions from meat or vegetables and natural sources like vegetative detritus, pollens and microbial spores.The indicators like Carbon Preference Index (CPI), C max and Plant Wax Number (%WNA) are commonly used to identify the relative contributions to n-alkanes from fossil fuel (e.g., traffic) and biogenic emissions (Abas and Simoneit, 1996;Kalaitzoglou et al., 2004).Hopanes and steranes along with other high molecular weight organic acids mainly exist in particulate phase (Zielinska et al., 2004).Hopanes are used as molecular markers for indicating contribution of vehicular emissions to the atmosphere as these compounds are comparatively involatile, not easily biodegradable, geologically mature, and relatively stable in the atmosphere (Maxwell et al., 1971;Simoneit et al., 1988).Hopanes indicate the contribution from vehicular exhaust as they are commonly found in motor lubricating oil (Simoneit, 1984;Rogge et al., 1993b).Levoglucosan (1,6-anhydro-b-Dglucopyranose) is a stable compound and its presence could be correlated to cellulosecontaining substances, and hence it could be suitably used as a marker for biomass burning (Simoneit et al., 1999;Nolte et al., 2001).
In India, there are limited studies related to measurement and characterization of OC.With a view to understand the role of OC and molecular marker with specific reference to sources, ambient air quality, was monitored at seven sites, representing different land use sites in the city of Mumbai.The sampling was conducted during March 2007 to March 2008 to get samples in summer, post monsoon and winter season.The major of the objectives of study were: i) to estimate PM 10 , OC and EC at varied sampling sites ii) to carry out speciation of particulate carbon in terms of "Molecular Markers" e.g., PAHs, n-alkanes, Hopanes, Steranes and Levoglucosan and iii) qualitative source apportionment of PAHs and n-alkanes to assign sources.

Sampling Sites
Mumbai, formerly known as Bombay, is the capital of the state of Maharashtra, and the most populous city of India, with an estimated population of about 16 million confined in an area of 437 km 2 .It is the commercial and entertainment capital of India.It has attracted migrants from all over India because of the immense business opportunities, making the city a potpourri of various communities and cultures.The vehicular population in the last four decades has increased from 150000 in 1971 to about 1100000 in 2001.The current population of vehicles is about 1500000.Mumbai was known as major industrial city till 1990.However, over last two decades, many industries have closed down or moved out paving way for intense commercial activities.There are still 40 air polluting industries in Mumbai.Other sources contributing to air pollution in Mumbai include bakeries, hotels/restaurants, crematories, construction activity, garages, domestic cooking, open eat outs, paved/unpaved road dust, refuse burning, ports/dock, aircrafts and railways.Seven representative sampling sites were selected within the city.Of these, Colaba was designated as control site, Dadar as commercial site, Mahul as industrial site, Andheri and Mulund as kerb sites, Khar as residential site of upper income group and Dharavi as slum residential site.The sampling sites are shown in Fig. 1 Sample Collection and Analysis Sample Collection for Particulate Matter and Organic Carbon Particulate Matter (PM 10 ) was collected on Teflon Filters of 47 mm diameter and 1 μ pore size (Whatman PTFE) and the gravimetric estimation was carried out after equilibrating the filters for 24 hours after sampling.The charge on the Teflon filters was neutralized by keeping it on Antistatic Device having Polonium-210 for 30 seconds before weighing.The collected mass was then measured using calibrated Sartorius ME5 Microbalance in a Clean Room Chamber maintained at 40% (± 5%) relative humidity and 20°C (± 3°C) temperature.Samples for organic carbon content estimation were collected on pre fired (900°C for 3 hours) Tissue Quartz Filters of 47 mm diameter and 1 μ pore size (PALL Life Science).All sampling was carried out using Partisol® Model 2300 Speciation Sampler.The duration for sampling was 24 hours for 30 days in each season at all the sites.The flow rate was maintained at 16.7 L/m.Samples for OC/EC analysis were stored at -20°C to prevent loss due to volatilization.

Analysis of Carbon Content and Molecular Markers
The particulate carbon contents of all the samples were estimated by Desert Research Institute's Thermal/Reflectance Optical Carbon Analyzer, Model 2001 A, Protocol Improve A (DRI, 2000).The composite samples for each season were then analyzed for individual non-polar particulate organic compounds at Desert Research Institute (DRI), Nevada, USA by DRI's In-Injection Port Thermal Desorption Method followed by Gas Chromatography/Mass Spectrometry.This method is capable of both qualitative and quantitative analysis of non-polar organic compounds on aerosol loaded filters.The target compounds include n-alkanes, iso/anteisoalkanes, hopanes, steranes, phthalates, other alkanes, alkenes, cyclohexanes, and polycyclic aromatic hydrocarbons (PAHs).Small strips of aerosol-laden filter materials were packed into a gas chromatography (GC) split/splitless injector liner.The organic compounds on the filter were thermally desorbed in the injection port and focused onto the head of a GC column for subsequent separation and mass spectrometric detection.Using data collected with the mass spectrometer (MS), the peak area of ions known to be present in the analytes was measured and the quantification was done using internal standards.For QA/QC, replicate analysis was performed at a rate of one every ten samples to ensure good instrument reproducibility.Also, certified standard solutions were used to check six-point calibration curve prepared from standards mixed in-house.The lower detection limits for PAHs, n-alkanes were 0.1 pg/mm 2 and 0.3 pg/mm 2 respectively.Whereas, using this method, hopanes and steranes could be detected even at concentration as low as 0.1 pg/mm 2 (Ding et al., 2009).
Levoglucosan was analyzed by Ion Chromatography.In this method, filtered aliquot of aerosol extract was passed through an ion exchange column (Carbopac MA1, 4 × 250 mm, Dionex, Sunnyvale, CA, USA).The eluent gradient used for the separation was 17 mM-600 mM NaOH.From the separator column, the species were detected using pulsed amperometric detection with waveform "A" using disposable gold electrodes on the Dionex ED50A detector module.The quantitative estimation of Levoglucosan was done from amperometric peak areas.The calibration was performed using minimum of five standards.In terms of QA/QC, replicate analysis and a standard run as an unknown was done after every 10 samples to evaluate the instrument and detector's performance.The detection limit for levoglucosan was 1 µg/L (Saarnio et al., 2010)

Particulate Matter, Organic and Elemental Carbon Levels
Site wise variations in average concentration of PM 10 , OC and EC levels are presented in Fig. 2. The highest average concentration of PM 10 was observed at Dharavi, 231.4 ± 49.2 µg/m 3 apparently due to poor road conditions, large number of small scale industries operating in this area along with other sources such as refuse burning, vehicular exhaust etc. Whereas, at control site located in Colaba, PM 10 level of 138.4 ± 46.4 µg/m 3 was recorded.The highest concentration of organic (47.3 ± 13.1 µg/m 3 ) and elemental carbon (12.0 ± 3.1 µg/m 3 ) was found at kerb site in Mulund.The next highest concentrations of OC (46.6 ± 14.6 µg/m 3 ) and EC (9.9 ± 2.1 µg/m 3 ) were at Dharavi and followed by Dadar where OC concentration was 33.9 ± 15.4 µg/m 3 whereas EC values were reported to be 9.8 ± 3.5 µg/m 3 .Average contribution of OC and EC to total PM 10 mass was 18.8% and 4.7% respectively (NEERI, 2009).

Concentration of Molecular Markers
In particulate samples collected during this study, 43 PAHs, 26 n-alkanes, 18 Hopanes, 12 Steranes and Levoglucosan were quantified.At Colaba, these molecular markers constituted 0.3% of total PM mass followed by 0.4% at Dadar, 0.6% each at Mahul and Mulund, 0.7% each at Khar and Andheri and 1.2% at Dharavi.The percent contribution of the sum of the concentrations of these compounds accounted for 1.7% of total organic at Colaba (lowest) whereas its contribution was highest (5.1%) at Dharavi.

PAH
During this study, the highest concentration of PAHs was found to be at Dharavi (247.70 ± 163.19 ng/m 3 ) and the lowest was observed at the background site, Colaba (47.84 ± 14.07 ng/m 3 ).The site wise variation of PAHs and its distribution in different groups is presented in Fig. 3.The carcinogenic PAH compounds accounted for 23.3% of total PAHs at Mulund, whereas highest contribution of these species was observed at Colaba (29.2%).This is probably because it is in downwind direction of Mahul (industrial site).Mass concentrations of PAHs have been reported for Indian cities including Mumbai, Ahmedabad, and Nagpur (Sharma et al., 2003).During previous study in Mumbai by Kulkarni et al. (2000), the total concentration of 7 PAHs at Saki Naka (slum residential area near to a busy road) and IIT Powai (residential, small scale industrial belt) was found to be 38.8 ng/m 3 and 24.5 ng/m 3 respectively.The study by Raiyani et al. (1993), at an industrial site in Ahmedabad, reported 11 PAHs with the ambient concentration range of 90-195 ng/m 3 , which is almost in the same range of Mahul site as an industrial site in this study.Outside India, studies carried out in Mexico (11 PAHs) by Marr et al. (2004) and in Chicago (16 PAHs) by Li et al. (2005) showed the concentration range of 60-910 ng/m 3 and 13-1865 ng/m 3 respectively.The PAHs levels found in Mumbai show slightly higher concentrations compared to previous studies in Mumbai, Ahmedabad and Nagpur.Amongst the different PAHs detected, presence of tracers like Picene indicates coal combustion sources.

n-alkanes
n-alkane compounds ranging from C 15 to C 40 carbon atoms were detected in the samples collected at all the sites.Total average n-alkane concentration during the study period ranged from 157.20 ± 35.28 ng/m 3 at Colaba to 759.13 ± 451.65 ng/m 3 at Dharavi.At Khar, upper income residential site, the concentration was 450.47 ± 204.82 ng/m 3 .At kerb sites, Andheri and Mulund, ambient concentration of n-alkane was 392.01 ± 146.91 and 372.74 ± 132.11 ng/m 3 respectively.The site wise variation of n-alkanes and its different categories based on its volatility is presented in Fig. 4. The long chain alkanes constituted 22.9% of total nalkanes at Andheri and its contribution was only 3.9% at Mulund.In general, n-alkane concentration during the study period was higher than those measured in Delhi (C 15 to C 40 ), 137.9 to 598.0 ng/m 3 (Sharma et al., 2003).Whereas in study conducted in China, total n-alkane concentration varied in the range of 233.2 to 1037.0 ng/m 3 (Bi et al., 2005).The presence of long chain alkanes (> C37) indicate open burning of plastic (Simoneit et al., 2005) and other municipal waste which contributes significantly to n-alkanes in India (Fu et al., 2010).
The values of steranes vary between 2.83 ± 0.73 ng/m 3 at Colaba to 32.64 ± 11.79 ng/m 3 at Mulund.Higher values close to vehicular sources show that steranes maximum contribution is from vehicular sector.The concentration for individual Sterane compounds at different sites is presented in Fig. 5(b).In a rural area in Portugal, the concentration of sterane was less than 1 ng/m 3 (Pio et al., 2001).
An earlier study by Chowdhury et al. (2007) in Mumbai, sum of Hopanes and Sterane at Worli (residential site) was 23 ± 5 ng/m 3 which is much lower than average concentration observed at both residential sites during this study.
According to earlier study in Mumbai by Chowdhury et al. (2007), the concentration was 0.91 ± 0.18 µg/m 3 .In other Indian cities like Chennai, the values were in the range of 0.004 to 0.36 µg/m 3 (Fu et al., 2010) whereas in Delhi, the ambient concentration was found to be varying between 0.04 to 1.58 µg/m 3 (Sharma et al., 2003).The highest concentration at Dharavi was due to high prevalence of biomass burning for cooking by slum dwellers.Whereas, next higher concentration at Khar shows contribution due to higher level of garden waste burning.

Qualitative Source Apportionment Diagnostic Ratios
Qualitative source identification for PAHs was carried out by calculating Diagnostic ratios.This method was first used by Daisey et al. (1979) and later followed by many others (Galarneau, 2008).This is based on the assumption that different sources release PAH in characteristic proportions, which remain unaffected by partitioning between vapour and particle phase and hence these ratios can be used to predict PAH origin/sources (Allen et al., 2008).Diagnostic ratios were calculated only for: i) PAHs which always and Benzo(e)pyrene (BeP).The site wise variation of diagnostic ratio is depicted in Fig. 6.

Carbon Number Maximum (C max ) and Carbon Preference Index (CPI)
C max is the carbon number with the highest peak in the chromatogram.Though, plant waxes, abrasion particles and fossil fuel combustion aerosols contain n-alkanes, plant waxes have high percentage of odd number of carbon atoms alkanes, but this is not the case for petroleum alkanes (Fraser et al., 2002).Colaba, Dadar and Mulund showed maximum value at C 31 indicating that one of the sources contributing to n-alkanes at these sites is biogenic in origin.
On the other hand, at remaining sites the predominant nalkane had even number of carbon atoms (C 40 ).The term, Carbon Preference Index (CPI), was introduced by Cooper and Bray in 1963 (Kotianová et al., 2008).It indicates the contribution of odd or even carbon number homologues within a sample.It is the ratio of the sum of odd carbon number n-alkanes to the sum of even carbon number n-alkanes (Abas and Simoneit, 1996).Alkanes with less than 20 Carbon atoms are considered to be too volatile (Kotianová et al., 2008) and hence CPI was calculated for n-alkanes ranging from C 21 to C 40 .CPI values of 1.17 to 2.34 were observed at other coastal cities, like Chennai, India (Fu et al., 2010).In Taipei city CPI values were in the range of 0.9 to 1.9 (Young et al., 2002).According to study carried out by Kotianová et al. (2008), at all the monitoring locations in Vienna, CPI values were less than 3, thus indicating the contribution of anthropogenic sources to particulate n-alkanes.

Plant Wax Number (%WNA)
The quantitative source apportionment of n-alkanes was carried out by calculating Plant Wax Number (%WNA).This term was first used by Simoneit et al. (1991) (Kotianová et al., 2008).It helps in quantitative estimation of the relative importance of biogenic and anthropogenic sources.It is based on the assumption that the wax n-alkanes are directly emitted from the vegetation and not from the resuspended soil detritus.It is calculated as: here, the negative values of the numerator are taken as zero (Simoneit et al., 1991).The relative contribution from biogenic and anthropogenic sources in terms of Plant Wax Number for all the monitoring locations is presented in Table 1.
Based on the qualitative source apportionment techniques used, following sources were identified for ambient PAHs and n-alkanes in the city of Mumbai.
1. Vehicular Exhaust: (BbF + BkF)/BghiP ratio indicates vehicular exhaust contribution if it is in the range of 0.20 to 1.72 (Daisey et al., 1979;Ohura et al., 2004).Dharavi (1.65), Mulund (1.53) and Khar (1.52) showed maximum influence of vehicles.BghiP/IND ratio ranging from 1.0 to 2.7 indicates vehicular emission as the contributing source (Daisey et al., 1979;Ohura et al., 2004).This is true for kerb site like Mulund (0.94) and busy commercial site like Dadar (0.88).Dharavi, with ratio of 1.01, also indicates the influence of vehicular emissions due to operation of number of small scale industries.
Based on this ratio, Dharavi and kerb site like Andheri show gasoline exhaust as a major contributing source.
The contribution of diesel vehicles to airborne particulate PAHs is indicated by number of ratios like IND/(IND + BghiP), BbF/BkF, IND/BghiP, (BbF + BkF)/BghiP, BaA/ (BaA + CHR) and CHR/BeP.Contribution from diesel vehicles is indicated if IND/(IND + BghiP) ratio ranges from 0.35-0.70(Grimmer et al., 1983;Rogge et al., 1993;Kavouras et al., 1999;Khaiwal et al., 2006).High ratio at Mahul (0.56) could be possibly due to movement of heavyduty diesel vehicles plying in this industrial area.Ships in Dockyard area near Colaba could be the reason for high ratio at this site.Andheri and Mulund, kerb sites characterized by continuous vehicular movement show ratio of 0.55 and 0.52 respectively.BbF/BkF ratio > 0.5 indicates diesel emissions (Pandey et al., 1999;Park et al., 2002).Probably due to intercity bus stop at Dadar where the buses mainly ply on diesel, ratio as high as 1.35 is observed.Andheri being a kerb site on Western Express Highway also shows high ratio (1.31) implying the contribution of diesel vehicles plying in this area.If IND/BghiP ratio approaches 1, it indicates influence of diesel vehicles (Caricchia et al., 1999).Ratio of 1.26 at Mahul indicates continuous movement of heavy-duty diesel vehicles plying in this industrial area.(BbF + BkF)/BghiP ratio of 1.6 indicates contribution of diesel vehicle exhaust (Westerholm et al., 1991;Ströher et al., 2007) and it is true in case of Dharavi (ratio = 1.65).Contribution from diesel exhaust is indicated if BaA/(BaA + CHR) ratio ranges between 0.38-0.64(Sicre et al., 1987;Kavouras et al., 2001) and highest ratio is observed at kerb site, Andheri (0.55).Also, ratios at Dadar and Colaba are indicative of movement of diesel vehicles at intercity bus stop and emissions from ships at Dockyard respectively at these sites.Ratio at Mahul is possibly the indicator of industrial and heavy truck emissions in this area.At all the sites, CHR/BeP ratios were greater than 0.6 indicating diesel vehicles as the major contributing source (Li and Kamens, 1993).Andheri (0.95) and Mulund (0.78) showed the maximum influence of diesel exhaust, being kerb sites.Dharavi has ratio of 0.93, probably due to movement of diesel vehicles carrying the raw material and finished products from huge number of small-scale industrial units operating in this area.
CPI values close to 1 indicate that the presence of n-alkanes in the ambient atmosphere is mainly due to petroleum products or due to incomplete combustion of petroleum (Simoneit, 1984;Mazurek et al., 1989).CPI values during the study period ranged from 0.97 at Dharavi to 1.12 at Mulund.At sites like Colaba, Dadar, Khar, Andheri and Mahul CPI values varied from 1.08, 1.07, 1.01, 1.03 and 1.0 respectively indicating the contribution of vehicular exhaust to particulate n-alkanes.

Wood Combustion:
The ratios indicating the contribution from wood combustion are IND/(IND + BghiP), BghiP/IND, BaA/(BaA + CHR), BeP/BaP, BaA/BaP and BkF/IND.Yunker et al. (2002) and Luo et al. (2005) stated that IND/(IND + BghiP) ratio > 0.5 where as BghiP/IND ratio of 0.8 (Li and Kamens, 1993) can be interpreted as use of wood as fuel.According to Gschwend and Hites (1981) and Ströher et al. (2007), contribution of wood combustion is indicated for BaA/(BaA + CHR) ratio of 0.48-0.54.This is mainly true for Khar and Andheri, which have large number of bakeries using wood as fuel.The ratio around 0.44 indicates wood combustion (Simcik et al., 1999).BeP/BaP ratio of 0.4 at Colaba and Dadar is indicative of use of wood as fuel in bakeries operating in these areas.BaA/BaP ratio of around 0.1 indicates wood combustion.Colaba, Mulund, Mahul, Dadar and Khar show the predominance of wood combustion.BkF/IND ratio of around 0.6 indicates wood burning (Li and Kamens, 1993).Dharavi followed by Colaba has ratio of 0.77 and 0.76 respectively.These ratios are indicators of use of wood as a fuel for bakeries as well as crematoria.Similarly, in case of Andheri, Dadar and Khar, ratio varies between 0.73, 0.72 and 0.63 acting as tracers for wood burning in area sources viz.bakery and crematoria.At Mahul and Mulund, people residing in the slum area nearby the site use wood as domestic fuel and hence has a ratio of 0.71 and 0.70 respectively.Grimmer et al. (1983), Pio et al. (2001) and Khaiwal et al. (2006), IND/(IND + BghiP) ratio of 0.56 indicates coal combustion and it holds true for Mahul which is having coal fired thermal power plant.BaP/BghiP ratio 2.42 at Andheri and 2.17 at Khar is indicative of use of coal (Daisey et al., 1979;Masclet et al., 1987) as fuel in hotels in these areas.IND/BghiP ratio of 1.89 in Mahul is due to coal-fired power plant.IND/BghiP ratio ranging from 1.06-1.12indicates PAH emissions due to use of coal/coke (Caricchia et al., 1999).At sites like Dharavi and Mulund, ratios fall in this range indicating use of coal/coke as fuel in this area.Coal combustion is said to be the probable source if BaA/(BaA + CHR) ratio is > 0.5 (Gaga et al., 2004) and it can be seen as a major influencing factor in Dharavi and Mahul, probably due to number of open eat outs and thermal power plant respectively at these sites.

Coal
4. Industrial and Heavy Duty Diesel Vehicles: Ratios like BaA/(BaA + CHR) and BeP/(BaP + BeP) indicate contribution from Industrial and Heavy Duty Diesel Vehicles Emission.BaA/(BaA + CHR) ratio at Mahul is the indicator of industrial and heavy truck emissions (Kavouras et al., 2001) in this area.Presence of industrial furnaces results in BeP/(BaP + BeP) ratio ranging from 0.17 to 0.48 (Yang et al., 1998).Mahul with 2 refineries and a power plant, Dadar having number of garages, Dharavi with number of small scale industrial units and Mulund having small industrial area in the vicinity show ratio in the same range.
5. Petroleum Refining: BaP/BghiP ratio ranging from 0.65 to 1.7 is used to identify the emissions from petroleum refining (Daisey et al., 1979) and Mahul site having 2 major refineries has average ratio of 1.89.According to Kavouras et al. (1999), BeP/(BaP + BeP) ratio in the range of 0.45 ± 0.27 indicates petroleum combustion residues.Amongst all the sites, the highest ratio was observed at Colaba (0.28) and Mahul (0.27).The petroleum refineries at Mahul could be the causative factor for high ratio there and Colaba being at the downwind site of Mahul, also has high ratio.
6. Road Dust: The contribution of road dust to particulate PAHs is identified by calculating ratio of PAHs viz.BbF, BkF and BghiP.Ratio of (BbF + BkF)/BghiP indicate contribution of road dust if it ranges from 1.0-2.45(Yang et al., 1998).Andheri being a kerbsite has maximum influence of vehicular movement resulting in resuspension of road dust and this is corroborated by highest ratio of 2.03 at this site.Mahul being an industrial area with continuous movement of heavy-duty vehicles has resulted in poor road conditions.This results in high resuspension of road dust.Colaba being on the downwind side of Mahul shows ratio of 1.95.Along with these sites, Dadar also shows high ratio (1.92) due to the fact that it is a busy commercial area with constant vehicular movement and thus resulting in road dust resuspension.
7. Domestic Soot: If (BbF + BkF)/BghiP ratio ranges between 1.5-14.0, it is indicative of domestic soot contributing to the above-mentioned particulate PAHs (Cretney et al., 1985).Based on the ratio, all the sites show influence of domestic soot.But, it appears to contribute maximum at Andheri (2.03) followed by Colaba (1.95) and Dadar (1.92), probably due to proximity of residential area to the sampling site.
8. Oil Burning: According to Simcik et al. (1999), BghiP/BaP ratio of around 0.5 indicates oil burning.Ratios of 0.61, 0.60 and 0.58 were observed at Mulund, Dharavi and Mahul respectively.These ratios indicate number of small scale industrial units using oil, at and around sampling location of Mulund and Dharavi whereas ratio at Mahul signifies its selection as an industrial site having significant oil burning activity.Khar being residential area, domestic oil combustion is one of the main reasons for ratio of 0.52.In case of Andheri, Dadar and Colaba, along with emissions from residential area, oil burning in hotels and restaurants contribute significantly to ambient concentration of these PAHs.9. Smelting: IND/BghiP ratio ranging from 0.88 to 1.18 is indicative of smelting operation (Caricchia et al., 1999).Dadar area having number of garages, Dharavi with number of metal scrapping units and Bhandup Sonapur area which falls in 2 × 2 km area of Mulund site, with number of small scale industrial units show same range of ratio.

CONCLUSIONS
Annual Particulate matter (PM 10 ) concentration in Mumbai ranged between 138.4 ± 46.4 µg/m 3 at Colaba and 231.4 ± 49.2 µg/m 3 at Dharavi, which are much higher than newly notified annual average standard of 60 µg/m 3 for India.The PM organic fractions in terms of OC, EC, PAHs etc. do provide useful information when combined with local knowledge of sources present around the site.
The ambient concentrations of PAHs, n-alkanes and Levoglucosan were maximum at Dharavi indicating that the combination of sources was responsible for such a high concentration levels at that site.Petroleum Biomarkers (hopanes and steranes), as expected, were in highest amount at Kerb site, Mulund.Based on the source apportionment methods like diagnostic ratios for PAHs and C max , CPI and %WNA used in case of n-alkanes, it was evident that anthropogenic input from sources viz.vehicular exhaust, wood combustion and coal combustion along with some site specific sources were significant contributors to the organic fraction of ambient PM 10 in Mumbai city.
Besides the qualitative source apportionment carried out in this study, there is a need to have quantitative estimation for carcinogenic compounds such as PAHs.In the present study, only non polar organic compounds were measured.Therefore, a comprehensive study inclusive of measurement of polar fraction of organic compounds should also be carried out to get further insights of particulate matter composition.These studies along with appropriate legislation, short term and long term control strategies for various sectors and public awareness raising campaigns will ensure "Cleaner air in urban areas."