G Model

PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS

Particuology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Particuology
journal homepage: www.elsevier.com/locate/partic

Chemical composition and source apportionment of PM2.5 and

PM2.5–10 in Trombay (Mumbai, India), a coastal industrial area
Sandeep Police a,b , Sanjay Kumar Sahu a,b , Mahesh Tiwari a,b , Gauri Girish Pandit a,b,∗
a
Environmental Monitoring and Assessment Section, Bhabha Atomic Research Centre, Mumbai 400085, India
b
Homi Bhabha National Institute, Anushakti Nagar, Mumbai 400094, India

a r t i c l e i n f o a b s t r a c t

Article history: PM2.5 and PM2.5–10 concentrations, elemental constituents, and sources in a densely populated coastal
Received 19 August 2016 industrial area (Trombay, Mumbai) were investigated in 2010 and 2011. The PM2.5 and PM2.5–10 concentra-
Received in revised form 8 September 2017 tions were 13.50–71.60 and 22.40–127.78 ␮g/m3 , respectively. The daily PM2.5 concentrations exceeded
Accepted 18 September 2017
the Indian Central Pollution Control Board limit (60 ␮g/m3 ) several days in winter. Of the elements ana-
Available online xxx
lyzed, Si then Al had the highest concentrations in PM2.5–10 , but black carbon then Si had the highest
concentrations in PM2.5 . The element concentrations varied widely by season. Al, Ca, Fe, Si, and Ti con-
Keywords:
centrations were highest in summer, Cl, Mg, and Na concentrations were highest in the monsoon season,
Trombay
PM2.5
and the other trace metal concentrations in both PM2.5 and PM2.5–10 were highest in winter. The PM2.5
PM2.5–10 and PM2.5–10 sources were apportioned by positive matrix factorization. PM2.5 and PM2.5–10 had six domi-
Black carbon nant sources, crustal material (8.7% and 25.3%, respectively), sea salt spray (6.1% and 15.0%, respectively),
Metals coal/biomass combustion (25.5% and 13.8%, respectively), fuel oil combustion (19.0% and 11.2%, respec-
Positive matrix factorization tively), road traffic (17.7% and 12.6%, respectively), and the metal industry (10.6% and 7.0%, respectively).
Anthropogenic sources clearly contributed most to PM2.5 but natural sources contributed most to PM2.5–10 .
© 2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of
Sciences. Published by Elsevier B.V. All rights reserved.

Introduction Yatkin & Bayram, 2008). Exposure to PM is associated with diseases

Air pollution in India is very severe. Accelerating industrializa-
tion, transport use, urbanization, construction, economic growth,
and energy demands have caused air quality to deteriorate severely,
especially in terms of particulate matter (PM) in Indian cities (CPCB,
2010; Tiwari et al., 2015). The International Agency for Research on
Cancer classes PM in outdoor air as a Group I carcinogen (Hamra
et al., 2014). Atmospheric PM pollution, especially PM2.5 (PM with
aerodynamic diameters ≤2.5 ␮m) and PM2.5–10 (PM with aero-
dynamic diameters of 2.5–10 ␮m), is a major current problem
because of the adverse effects PM has on human health and the
environment. PM2.5 and PM10 exceeding the Indian Central Pollu-
tion Control Board (CPCB) annual limits have been found in most
Indian cities (Banerjee, Murari, Kumar, & Raju, 2015). There is great
interest in PM because of increasing PM concentrations in the atmo-
sphere and the negative effects of PM on atmospheric chemistry,
visibility, coastal ecosystems (because of air–water exchange), and
the global climate (Ramanathan, Crutzen, Kiehl, & Rosenfeld, 2001;
∗ Corresponding author at: Environmental Monitoring and Assessment Section,
Bhabha Atomic Research Centre, Mumbai 400085, India.
E-mail address: ggp@barc.gov.in (G.G. Pandit).
ranging from lower respiratory tract infections to cerebrovascular
disease (Hamra et al., 2014; Mimura et al., 2014). Worldwide, about
8% of lung cancer deaths, 5% of cardiopulmonary deaths, and 3% of
respiratory infection deaths have been attributed to PM exposure
(Cohen et al., 2005). The World Health Organization uses PM as an
indicator when evaluating air pollution that damages health (An,
Hou, Li, & Zhai, 2013; Tao, Mi, Zhou, Wang, & Xie, 2014).
Chemically characterizing PM is very important to epidemiolog-
ical and source apportionment studies, and allows radiative forcing
associated with PM to be quantified in climate studies. PM is a
complex mixture of anions, cations, mineral dust, trace elements,
organic carbon, elemental carbon, and water. The relative abun-
dances of these components vary widely, spatially, and temporally,
because of the natures of the sources (Li, Wiedinmyer, & Hannigan,
2013). PM2.5 is mainly emitted during combustion, so is emitted in
engine exhaust and during fossil fuel combustion for power gener-
ation, biomass combustion, and metal smelting (Cheng, Liu, & Chen,
2010). PM2.5–10 has non-wind-suspended and wind-suspended
components. Anthropogenic sources, such as construction activ-
ities, open-cast mining, and mechanical processes (e.g., tire and
brake emissions), contribute to the non-wind-suspended com-
ponent. Natural sources, such as volcanic emissions, agricultural
tilling, sea spray, biological particles, surface soil, and dust on paved

https://doi.org/10.1016/j.partic.2017.09.006
1674-2001/© 2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
2 S. Police et al. / Particuology xxx (2017) xxx–xxx
areas, contribute to the wind-suspended component (Cheng et al.,
2015).
It is important to identify the contributions of the different
sources to decrease or control PM pollution. Source apportion-
ment studies are indispensable when attempting to identify PM
sources and quantify their contributions. Multivariate receptor
models are very popular tools for source apportionment studies.
Such models have been used to identify and quantify PM sources
around the world (Bove et al., 2014; Geng et al., 2013; Gummeneni,
Yusup, Chavali, & Samadi, 2011; Karnae & John, 2011; Minguillón
et al., 2014; Seneviratne et al., 2011; Sharma et al., 2014). Princi-
pal component analysis/absolute principal component scores, edge
analysis, chemical mass balance, and positive matrix factorization
(PMF) are examples of receptor models. PMF is a powerful model
capable of resolving PM sources without prior knowledge of the
sources (Cohen, Crawford, Stelcer, & Bac, 2010; Mooibroek, Schaap,
Weijers, & Hoogerbrugge, 2011). PMF has many advantages over
other models (Ancelet, Davy, Trompetter, & Markwitz, 2014).
The adverse effects of PM and increasing interest in PM studies
led us to perform the study presented here in Trombay, a densely
populated coastal industrial area in Mumbai, India. The Trombay
site is near various large and small industrial plants, port areas, and
areas affected by heavy traffic, which can all contribute to PM pol-
lution. Little information is available on particulate-bound metal
concentrations, temporal and seasonal variations in PM concentra-
tions, or the specific sources of PM at the site. The study lasted two
years, from January 2010 to December 2011, and the goals were to
assess temporal and seasonal variations in PM2.5 and PM2.5–10 and
particle-bound element concentrations and to identify and quantify
the sources of PM at the study site.
Materials and methods
Site description
Trombay (19.0◦ N, 72.9◦ E) is a northeastern suburb of Mumbai,
India, and covers about 25 km2 . The sampling site was about 20 km
from Mumbai center, with the Arabian Sea on one side and an indus-
trial area on the other. A detailed map showing the locations of
the site and the surrounding industrial plants is shown in Fig. 1.
The sampling site was near a coal- and oil-burning power plant,
oil refineries, a fertilizer complex, and automobile, metallurgical,
textile, chemical, paint, and small metal industry plants. The sam-
pling site was also near several other industrial areas (Maharashtra
Industrial Development Corporation, Taloja, Turbhe, Trans Thane
Creek, and others). Mumbai port and Jawaharlal Nehru (Nhava-
Sheva), the largest ports in India, handle liquid chemicals, crude
oil, petroleum products, and other commodities. The sampling site
was a few kilometers away from each port. Mumbai is densely
populated and has worse traffic problems than other Indian cities.
Mumbai has three distinct seasons, winter (October–February),
summer (March–June), and the monsoon (July–September). The
mean summer and winter maximum temperatures are 32 and
30 ◦ C, respectively, and the mean summer and winter minimum
temperatures are 25 and 20.5 ◦ C, respectively. Mumbai receives
heavy rainfall (300–800 mm) during the monsoon, and mean
annual rainfall is 2240 mm. Winter winds are generally light and
from the north and northeast. Monsoon winds are stronger and
are moist and from the southwest (from marine areas). The wind
speeds at the study site were 0.5–7.5 m/s in summer and winter
but as high as 17 m/s in the monsoon.
PM sampling and gravimetric analysis
Each PM sample was collected using a Gent’s dichotomous
sampler with a stacked filter unit containing two 47 mm diame-
Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
ter Nuclepore polycarbonate filters (one coarse, with 8 ␮m pores,
and one fine, with 0.4 ␮m pores) in series. The mean flow rate
was 16 L/min. The pre-impaction stage intercepted particles with
aerodynamic diameters >10 ␮m. The coarse filter with 8 ␮m pore
size collected PM2.5–10 , and the fine filter with 0.4 ␮m pore size
collected PM2.5 . Each sample was collected for 24 h, and a sam-
ple was collected twice a week from January 2010 to December
2011. Each filter was preconditioned (equilibrated for 24 h at room
temperature and 50% ± 5% relative humidity) and then weighed
using a Mettler AE240 electronic microbalance (Mettler Toledo,
Switzerland, error ±10 ␮g) at least three times before and after
sampling to determine the mass of particles collected and to ensure
the measurement variance was low. The net PM mass was cal-
culated by subtracting the pre-sampling filter weight from the
post-sampling filter weight. The net mass was divided by the air
volume to give the daily PM concentration in ␮g/m3 .
Chemical analyses of PM2.5 and PM2.5–10
Elemental analyses of the particles were performed using a
Xenemetrix EX-6600 energy dispersive X-ray fluorescence (EDXRF,
Xenemetrix, Israel) spectrometer with a Rh anode (the X-ray
source) with a 60 kV, 6 mA (400 W) power supply. The system had a
Si (Li) detector (resolution 150 eV at 5.9 keV Mn X-ray) and various
secondary targets. The primary X-ray beam spectrum was modified
using the secondary targets Ge, Mo, Si, and Ti and was then used to
excite the elements in a sample. The concentrations of 18 elements
(Al, As, Ca, Cl, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, S, Si, Ti, V, and Zn)
were determined. The best sensitivity was achieved using the opti-
mum voltages, analysis times, and currents shown in Table 1. The
instrument was calibrated using thin-film Micromatter Standards,
then the analyte concentrations in the samples were determined
using the calibrations. The EDXRF measurement method was val-
idated using NIST SRM 2783 (air particulate on filter media, NIST,
USA) reference material. The measured concentrations were within
6% of the NIST SRM 2783 certified values.
The black carbon (BC) concentrations in the PM2.5 samples were
determined by making reflectance measurement using an EEL 43D
smoke stain reflectometer (Evans Electroselenium Limited, USA),
in which light from a tungsten lamp was projected onto a well-
defined spot on the sample and the reflected light measured by
photocells in the black housing. Reflectances were measured at five
points on each filter, and the mean value was used to estimate the
BC concentration using a mass absorption coefficient of 9.7 m2 /g
and a formula presented by Salako et al. (2012). The reflectometer
was calibrated following the appropriate instructions using a blank
filter from the same batch as the filters used to collect the samples.
Source apportionment
The sources of the PM were apportioned using the US Envi-
ronmental Protection Agency (EPA) PMF model version 5.0. PMF
decomposed the data matrix X (n × m; n and m being the numbers
of samples and chemical species, respectively) into two matrices,
factor contributions G (n × p) and factor profiles F (p × m) with non-
negative constraints (p is the number of factors extracted) (Paatero,
Eberly, Brown, & Norris, 2014). PMF is described in more detail else-
where (Geng et al., 2013; Guinot, Gonzalez, De Faria, & Kedia, 2016;
Minguillón et al., 2014; Pakbin, Ning, Shafer, Schauer, & Sioutas,
2011; Rahman et al., 2011). Uncertainties for the data points could
affect the PMF results, and the measured uncertainty was an input
parameter for the PMF analysis. The measurement uncertainty was
calculated using Eq. (1) (Polissar, Hopke, Paatero, Malm, & Sisler,
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
S. Police et al. / Particuology xxx (2017) xxx–xxx 3

Fig. 1. Map showing the sampling site (red dot) and the major and minor industrial areas and plants surrounding the site (yellow markers). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1
Optimized energy dispersive X-ray fluorescence operating parameters for the analysis of the elements of interest in particulate matter.

Secondary target Elements Voltage (kV) Current (␮A) Analysis time (s) Atmosphere

Si and Ti Na, Mg, Al, S, Si, Cl, K, Ca 25 4800 600 Vacuum

Ge Ti, V, Cr, Mn, Fe, Ni, Cu, Zn 40 800 600 Air
Mo As, Pb 40 1000 600 Air

Table 2 of 60 ␮g/m3 . The PM2.5 concentration exceeded the CPCB limit on

Statistical summary of the PM2.5 , PM2.5–10 , and PM10 concentrations (in ␮g/m3 ) at
more days in 2010 (7.4%) than in 2011 (1.2%). It can be seen from
Trombay in 2010 and 2011.
Table 2 that the PM2.5 and PM2.5–10 concentrations were higher
Year PM2.5 PM2.5–10 PM10 in 2010 than in 2011. The PM10 concentration was calculated by
Min. 2010 14.30 22.70 38.64 adding the PM2.5 and PM2.5–10 daily concentrations together. The
Max. 71.60 127.78 182.26 annual mean PM10 concentrations in both years exceeded the CPCB
Avg. 31.80 67.90 99.73 annual limit of 60 ␮g/m3 , but by less in 2011 than in 2010. The PM10
Stdev. 14.90 28.10 41.09
concentration exceeded the daily limit of 100 ␮g/m3 on about 35%
Min. 2011 13.50 22.40 39.95 of the sampling days in both years, but on 44.4% of days in 2010.
Max. 70.00 110.00 157.83
The mean PM2.5 concentrations exceeded the annual limits set
Avg. 27.70 52.40 80.70
Stdev. 10.80 23.50 32.10 by the US EPA (15 ␮g/m3 ), World Health Organization (25 ␮g/m3 ),
and European Union (20 ␮g/m3 ). The PM10 concentrations were
1.8 times higher than the World Health Organization annual limit
1998). Two parameters (the method detection limit and error frac- (50 ␮g/m3 ).
tion) were used in the equation. The PM concentrations were lower than the mean PM con-
 centrations reported in Delhi (PM2.5 118.30 ␮g/m3 , PM2.5–10
2
U= (F × C) + (0.5 × MDL)2 , C > MDL, (1)
113.60 ␮g/m3 , PM10 232.10 ␮g/m3 ) (Tiwari et al., 2015), Hyder-
where U is the uncertainty, C is the component concentration, MDL abad (PM2.5 72.60 ␮g/m3 , PM10 105.20 ␮g/m3 ) (Guttikunda &
is the method detection limit, and F is the error fraction. Kopakka, 2014), Pune (PM2.5 72.30 ␮g/m3 , PM10 113.80 ␮g/m3 )
(Yadav & Satsangi, 2013), and China (PM2.5 130.47 ␮g/m3 , PM10
Results and discussion 206.65 ␮g/m3 ) (Shi et al., 2015) but higher than the mean PM con-
centrations reported in Central America (Herrera-Murillo et al.,
The PM2.5 and PM2.5–10 concentrations are summarized in 2013) and central Taiwan, China (Hsu et al., 2016).
Table 2. The mean annual PM2.5 concentrations were well below The PM2.5 /PM2.5–10 and PM2.5 /PM10 ratios allow the rela-
the CPCB limit of 40 ␮g/m3 in both study years, but the concentra- tive contributions of fine particulate sources (i.e., anthropogenic
tions in about 4.3% of the samples exceeded the CPCB daily limit sources) to PM10 concentrations to be assessed. The PM2.5 /PM2.5–10

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
4 S. Police et al. / Particuology xxx (2017) xxx–xxx

Table 3 Table 4
Statistical summary of the element concentrations (in ng/m3 ) in PM2.5–10 in 2010 Statistical summary of the element concentrations (in ng/m3 ) in PM2.5 in 2010 and
and 2011 on a yearly basis. 2011 on a yearly basis.

2010 2011 2010 2011

Min Max Avg. Min Max Avg. Min Max Avg. Min Max Avg.

Na 228.96 2954.40 1051.46 198.12 4980.63 1510.07 Na 78.15 2406.07 401.46 89.69 3797.32 531.69
Mg 16.46 1861.72 649.45 3.24 3492.48 842.60 Mg 13.51 1203.08 223.95 21.85 1493.84 333.02
Al 248.58 7936.02 2471.74 79.08 7165.69 1973.63 Al 63.73 3587.98 749.43 77.91 3132.81 666.56
Si 296.91 11506.14 3206.86 79.91 9490.27 2677.12 Si 90.12 3698.73 880.90 198.56 6518.37 1010.28
S 50.67 1676.82 490.97 69.91 5029.52 1007.56 S 138.46 1892.97 789.54 173.29 2723.73 964.33
Cl 385.95 3914.30 1449.13 250.10 7412.45 2121.39 Cl 109.56 3873.78 609.90 120.19 4784.63 736.17
K 68.77 1414.15 443.64 88.50 4542.96 858.68 K 38.01 697.14 264.14 39.51 1449.35 315.87
Ca 131.41 5270.75 1837.23 41.19 5301.85 1303.45 Ca 46.36 1730.95 366.89 46.36 2376.79 424.40
Ti 18.08 803.53 181.63 14.50 694.06 163.52 Ti 1.06 332.35 39.39 1.02 238.71 36.15
V 1.31 348.49 35.50 1.31 111.36 32.15 V 1.52 74.08 19.58 1.85 209.28 30.82
Cr 1.00 88.67 10.54 1.00 70.62 9.42 Cr 1.00 48.40 6.42 1.00 145.50 10.36
Mn 2.48 200.80 37.80 2.94 145.63 39.97 Mn 1.25 70.67 11.93 1.25 322.91 17.86
Fe 213.58 7202.02 1961.95 146.57 4313.30 1497.32 Fe 38.79 1822.93 370.22 44.79 2433.62 408.69
Ni 1.18 135.74 17.20 1.01 38.74 12.34 Ni 2.14 49.13 13.76 1.98 224.45 24.36
Cu 1.00 64.72 13.48 1.00 54.14 13.52 Cu 1.00 46.78 9.86 1.01 279.52 12.85
Zn 3.92 744.40 124.83 2.92 689.70 127.64 Zn 1.33 795.47 86.25 2.88 377.54 53.35
Pb 3.19 838.41 107.38 3.34 554.08 105.92 Pb 8.45 775.40 146.58 2.12 412.38 100.66
As 0.35 3.15 1.25 0.38 5.23 1.88 As 0.33 5.57 1.74 0.33 6.14 1.92
BC 1039.70 9649.50 3968.54 681.29 9650.00 3923.97

and PM2.5 /PM10 ratios were 0.22–1.87 (mean 0.53) and 0.18–0.65
(mean 0.34), respectively. The mean annual ratios for 2010 and
2011 indicated that PM2.5 sources were more important in 2011
than 2010. The PM2.5 /PM2.5–10 and PM2.5 /PM10 ratios were high-
est during the monsoon, next highest in winter (0.61 and 0.36,
respectively), and lowest in summer (0.48 and 0.32, respectively).
Monthly variations in PM2.5–10 and PM2.5
A box–whisker plot of the monthly PM2.5 and PM2.5–10 con-
centrations is shown in Fig. 2. Each box represents the difference
between 25th and 75th percentiles, the point within the box is the
mean, and the horizontal line within the box is the median. The
concentrations beyond the 75th and 25th percentiles are shown as
whiskers.
It can be seen from Fig. 2 that the monthly PM2.5 and PM2.5–10
concentrations varied similarly, the highest concentrations being
in December–January and the lowest in July–August. The monthly
mean PM2.5 and PM2.5–10 concentrations were higher in 2010
than in 2011. The highest monthly mean PM2.5 concentrations
were 60.38 and 44.57 ␮g/m3 , in January 2010 and December 2011,
respectively. The highest PM2.5–10 concentrations were 106.78 and
94.16 ␮g/m3 , in January 2010 and December 2011, respectively.
The concentrations in January and December could have been high
because of calm weather conditions and low mixing heights.
The lowest monthly mean PM2.5 concentrations were 18.07
and 16.09 ␮g/m3 , in July 2010 and August 2011, respectively. The
lowest monthly mean PM2.5–10 concentrations were 38.94 and
26.89 ␮g/m3 , in June 2010 and July 2011, respectively. Washout
of PM by rain could have caused the low concentrations in these
months. Similar monthly variations in PM concentrations have
been found in many previous studies (Deshmukh, Deb, & Mkoma,
2013; Liu, Li, Wu, & Xu, 2015; Yadav, Sahu, Jaaffrey, & Beig, 2014).
Element compositions of PM2.5 and PM2.5–10
The element concentrations in the PM2.5–10 and PM2.5 on a
yearly basis are shown in Tables 3 and 4. In both years, the annual
mean Si concentration in PM2.5–10 was highest, followed by the Al
and Cl concentrations, and the Ti concentration was lowest. Of the
trace metals As, Cr, Cu, Mn, Ni, Pb, V, and Zn in both years, Zn had
the highest annual mean concentration, followed by Mn and Pb,
and As had the lowest concentration. The Al, Ca, Fe, Si, and Ti con-
centrations in PM2.5–10 were higher in 2010 than 2011 but the Cl,
K, Mg, Na, and S concentrations were higher in 2011 than 2010.
The BC concentrations were higher than the concentrations of
the other analytes in PM2.5 in both years, but the BC concentra-
tions varied widely. The highest monthly mean BC concentration
in 2010 was in December but the highest concentration in 2011 was
in January. The lowest BC concentrations were in July and August.
Watson, Chow, and Houck (2001) found that high BC concentra-
tions are normally related to biomass burning and motor vehicle
exhausts. The BC concentrations in Trombay could be high because
there are many BC sources (a coal/oil-burning power plant, biomass
combustion sites, and many vehicles, including ships) nearby. We
did not measure BC concentrations in PM2.5–10 because coarse
white particles can mask BC, causing BC concentrations in PM2.5–10
to be underestimated. In both years, Si had the second highest
concentration in PM2.5 , followed by S, and Ti had the lowest concen-
tration of the major elements. Of the trace metals in PM2.5 , Pb had
the highest concentration, followed by Zn, and As had the lowest
concentration. It can be seen from Table 4 that the major elements
(Ca, Cl, Fe, K, Mg, Na, S, and Si) and some trace elements (Cr, Cu,
Mn, Ni, and V) had higher concentrations in PM2.5 in 2011 than
2010 but Al, Pb, and Zn had higher concentrations in 2010 than
2011. The annual mean concentrations in both size fractions may
have been different in 2010 and 2011 because of the emission rates
and meteorological conditions being different.
The CPCB has set annual concentration limits for various toxic
metals, including 6 ng/m3 for As, 20 ng/m3 for Ni, and 500 ng/m3 for
Pb. None of these metals (except Ni in PM2.5 in 2011) were found in
PM2.5 or PM2.5–10 at concentrations higher than the annual limits.
Malm, Sisler, Huffman, Eldred, and Cahill (1994) suggested that
a K to Fe ratio of 0.6 ± 0.2 indicates soil dust contributed K. We
found a ratio of 1.2 for both size fractions, indicating there were
anthropogenic sources of K. The annual K/Fe ratio for PM2.5–10 was
>0.6 except in 2010. The Ca/Al ratios for PM2.5 and PM2.5–10 were
0.77 and 0.67, respectively. These ratios were close to the soil refer-
ence ratio of 0.44, indicating that Ca and Al may have originated in
crustal material. The ratios may have been a little higher than the
reference ratio because of contributions from construction activi-
ties near the sampling site. The Cl/Na ratios for PM2.5 and PM2.5–10
were 1.11–1.74 (mean 1.47) and 1.01–1.75 (mean 1.41), respec-
tively. The ratios were close to the ratio for sea salt (1.9). The annual
mean Cl/Na ratios varied little. The V/Ni ratio is an indicator of var-
ious combustion sources. For example, a V/Ni ratio of 2.5–4.0 is

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
S. Police et al. / Particuology xxx (2017) xxx–xxx 5

Fig. 2. Box–whisker plots showing monthly variations in (a) PM2.5 and (b) PM2.5–10 concentrations during the study period.

Table 5
PM2.5–10 to PM2.5 element concentration ratios in 2010 and 2011 on a yearly basis.

Na Mg Al Si S Cl K Ca Ti V Cr Mn Fe Ni Cu Zn Pb As

2010 2.6 2.9 3.3 3.6 0.6 2.4 1.7 5.0 4.6 1.8 1.6 3.2 5.3 1.3 1.4 1.4 0.7 0.7
2011 2.8 2.5 3.0 2.6 1.0 2.9 2.7 3.1 4.5 1.0 0.9 2.2 3.7 0.5 1.1 2.4 1.1 1.0
Avg. 2.7 2.7 3.1 3.1 0.8 2.6 2.2 4.0 4.6 1.4 1.3 2.7 4.5 0.9 1.2 1.9 0.9 0.8

typical of ship emissions, and lower ratios (approaching 0.5) are Table 6
Percentage mass contribution of the elements to PM2.5 and PM2.5–10 in 2010 and
typical of diesel combustion (Horemans et al., 2011). The V/Ni ratios
2011.
for PM2.5 and PM2.5–10 were 1.93 and 2.54, respectively, indicating
that V and Ni were predominantly supplied by diesel combustion PM2.5 PM2.5–10
sources but some contribution from ships could not be neglected 2010 2011 2010 2011
(the ratios varied widely, from 0.3 to 11.82). The annual mean V/Ni
Si 2.77 3.64 4.72 5.11
ratio varied more for PM2.5–10 than for PM2.5 . S 2.48 3.47 0.72 1.92
The SO4 2− concentration was calculated from the S concentra- Al 2.36 2.40 3.64 3.77
tion using Eq. (2) (Manousakas et al., 2015). Cl 1.92 2.65 2.13 4.05
Na 1.26 1.92 1.55 2.88
[SO4 2− ] = 4.125[S] (2) Fe 1.17 1.47 2.89 2.86
Ca 1.15 1.53 2.70 2.49
The calculated SO4 2− concentrations in PM2.5 and PM2.5–10 were K 0.83 1.14 0.65 1.64
Mg 0.71 1.20 0.95 1.61
3.6 and 3.1 ␮g/m3 , respectively, corresponding to 12.1% and 5.1%, Pb 0.46 0.36 0.16 0.20
respectively, of the total PM2.5 and PM2.5–10 concentrations. Zn 0.27 0.19 0.18 0.24
The PM2.5–10 to PM2.5 element concentration ratios shown in Ti 0.12 0.13 0.27 0.31
Table 5 show that the major elements (e.g., Al, Ca, Cl, Fe, K, Mg, V 0.06 0.11 0.05 0.06
Ni 0.04 0.09 0.03 0.02
Mn, Na, Si, and Ti) contributed more to PM2.5–10 than to PM2.5 . The
Mn 0.04 0.06 0.06 0.07
ratios for the other elements (Cr, Cu, V, and Zn) show that they Cu 0.03 0.05 0.02 0.02
contributed more than the major elements to PM2.5 . The As, Ni, S, Cr 0.02 0.04 0.02 0.02
and Pb mass contributions to PM2.5 were high, indicating that these As 0.005 0.007 0.002 0.004
elements came predominantly from anthropogenic sources at the BC 12.50 14.10 – –

Trombay site.
Contributions of elements to PM2.5 and PM2.5–10
The mass contributions of various elements to PM2.5 and
PM2.5–10 in percent on an annual basis are shown in Table 6. Si made
the highest contribution to PM2.5–10 , followed by Al, but BC made
the highest contribution to PM2.5 (12.5% in 2010, 14.1% in 2011),
followed by Si (2.77% in 2010, 3.64% in 2011). The trace elements
As, Cr, Cu, Mn, Ni, and V made very low contributions, As making
the lowest contributions to both size fractions. Al, Ca, Cl, Fe, Na, Si,
and Ti made higher contributions to PM2.5–10 than PM2.5 , but As,
Cr, Cu, Ni, Pb, S, V, and Zn made higher contributions to PM2.5 than

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
6 S. Police et al. / Particuology xxx (2017) xxx–xxx

Fig. 3. Seasonal distributions of the element concentrations in (a) PM2.5 and (b) PM2.5–10 in 2010 and 2011.

PM2.5–10 . The mass contributions of most of the elements to PM
were higher in 2011 than 2010.
Seasonal variations in element concentrations
The seasonal trends in the element concentrations in PM2.5 and
PM2.5–10 are shown in Fig. 3(a) and (b). The Al, Ca, Fe, Si, and Ti con-
centrations were highest in summer and lowest in the monsoon in
both 2010 and 2011. The concentrations in summer were 2.2–2.9
and 2.7–3.6 times the concentrations in the monsoon in 2010 and
2011, respectively. The summer to monsoon concentration ratios
were higher for PM2.5–10 than PM2.5 . The high temperatures (and
therefore low relative humidity) and higher wind speeds in sum-
mer would have favored soil re-suspension, which is assumed to be

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
S. Police et al. / Particuology xxx (2017) xxx–xxx 7

Fig. 4. Element profiles (bars correspond to the left y-axis (in ng/m3 ) and red dots correspond to the right y-axis (in percentage)) of the factors determined by positive matrix
factorization for PM2.5 . (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

the dominant source of these metals (Khillare & Sarkar, 2012). The
Cl, Mg, and Na concentrations in PM2.5 were highest in the monsoon
in 2010 and 2011, as shown in Fig. 3(a). The Cl, Mg, and Na concen-
trations in PM2.5 were 1.5–1.9 times higher in the monsoon than
winter in 2010, and the concentrations were higher in the 2011
monsoon than in the 2010 monsoon. The Cl, Mg, and Na concen-
trations in PM2.5–10 were higher in summer than in the monsoon
and winter in 2010 but similar throughout 2011. Cl, Mg, and Na
generally originate in sea salt spray in coastal areas. A southwest
monsoon predominates in Mumbai, with the wind predominantly
coming from the sea, which could explain the high Cl, Mg, and Na
concentrations during the monsoon (Balakrishnaiah et al., 2011).
The K and S concentrations in PM2.5 and PM2.5–10 were highest in
winter in both 2010 and 2011, and next highest in summer in 2010
and in the monsoon in 2011.
The Cr, Cu, Mn, Ni, Pb, V, and Zn concentrations in PM2.5 and
PM2.5–10 were highest in winter, next highest in summer, and low-
est in the monsoon in both 2010 and 2011. The BC concentrations
in PM2.5 were highest in winter in both study years. The concen-
trations may have been highest in winter because of the effects
of temperature inversions and increased anthropogenic sources
(biomass burning, space heating, etc.) in winter (Khillare & Sarkar,
2012; Villalobos et al., 2015). The low wind speeds and low tem-
peratures in winter may have led to poor dispersion, increasing
the concentrations of these species in winter (Ancelet et al., 2014).
The concentrations may have been lowest in the monsoon because
water droplets will have scavenged PM and particle-bound ele-
ments from the atmosphere (Deshmukh et al., 2013). The high PM
and associated chemical species concentrations in winter indicate
that pollution in Trombay is worse in winter than in summer and
the monsoon.
PM2.5 and PM2.5–10 source apportionment
Source apportionment relies heavily on trial and error to reach
a meaningful solution. We performed source identification stud-
ies for 2010 and 2011 to determine the contributions of the same
sources in each year. Therefore, we presented PMF results con-
sidering two years (2010 and 2011) data. A wide range of factor
solutions (4–7) were examined, and the recommended approach

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
8 S. Police et al. / Particuology xxx (2017) xxx–xxx

Fig. 5. Element profiles (bars correspond to the left y-axis (in ng/m3 ) and red dots correspond to the right y-axis (in percentage)) of the factors determined by positive matrix
factorization for PM2.5–10 . (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(Reff, Eberly, & Bhave, 2007) was used to arrive at a suitable
solution. The six-factor solution produced realistic, meaningful
source profiles. Solutions using more than six factors gave unre-
alistic sources, but solutions using less than six factors gave mixed
sources. The name of each factor was determined from the marker
elements in the factor profile. The Q-value was lowest at F-peak = 0,
indicating the absence of rotational ambiguity, so the solution at
F-peak = 0 was used. The correlation coefficients (R2 ) for the rela-
tionships between the measured and modeled concentrations were
>0.80 for all the elements. It can be seen from Figs. 4 and 5 that PM2.5
and PM2.5–10 had contributions from six sources (crustal material,
sea salt spray, fuel oil combustion, coal/biomass combustion, road
traffic, and metal industries). The PM emitted by any source can
have a wide size distribution. Marker or signature elements emit-
ted by a source can be found in both PM2.5 and PM2.5–10 , which
could explain the similar elemental profiles we found for PM2.5 and
PM2.5–10 . The extracted factors are described in detail below.
The element profiles for factor 1 for PM2.5 and factor 2 for
PM2.5–10 were dominated by Ni and V. The factors explained >76%
of their total element concentrations in PM2.5–10 and PM2.5 and
the percentage contributions were higher for PM2.5 than PM2.5–10 .
The Ni and V concentrations in PM2.5 and PM2.5–10 correlated well
(R2 = 0.77), suggesting similar sources made contributions of these
metals. Masiol, Squizzato, Rampazzo, and Pavoni (2014) found that
Ni and V are present in crude oil and are key tracers of fossil
fuel/heavy residual oil combustion (Moreno et al., 2010), including
in petrochemical refineries, heavy fuel oil boilers, and ship engines.
The oil refineries, oil-fired power plant, and ports near the sampling
site could have contributed PM in the study area, so this source was
called fuel oil combustion.
The profiles for factor 2 for PM2.5 and factor 4 for PM2.5–10 were
dominated by As, K, and S. BC also contributed to the PM2.5 source
profile. As, K, and S contributed 91.8%, 60.2%, and 76.8%, respec-
tively, for the PM2.5 but 62.6%, 76.3%, and 80.9%, respectively for
the PM2.5–10 . These elements are the main tracers for coal com-
bustion (Rogula-Kozłovska, Klejnowski, Rogula-Kopiec, Mathews,
& Szopa, 2012). The coal combustion sources near the sampling site
could have contributed these metals to the PM. BC and K are also
markers for biomass burning (Kertész, Szoboszlai, Angyal, Dobos,
& Borbély-Kiss, 2010; Rahman et al., 2011). There are many slum

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
S. Police et al. / Particuology xxx (2017) xxx–xxx
Fig. 6. Pie charts showing the percentage contributions of the identified sources to PM2.5 and PM2.5–10 .
areas (in which biomass burning is common) near the sampling
site. The contributions of biomass burning to the presence of these
elements in PM cannot be neglected. Factor 2 for PM2.5 and factor
4 for PM2.5–10 were called coal/biomass combustion sources.
Factor 3 for PM2.5 and factor 5 for PM2.5–10 were dominated
by Cu, Pb, and Zn. BC also contributed 28.0% to PM2.5 in factor 3.
These elements are markers for traffic emissions (Liu et al., 2015;
Manousakas et al., 2015). Cu, Pb, and Zn may be found in PM because
of tire wear, lubricating oil combustion, and vehicle brake abra-
sion (Denier van der Gon, Hulskotte, Visschedijk, & Schaap, 2007;
Minguillón et al., 2014). Liu et al. (2015) stated that BC is the main
constituent of soot released by gasoline- and diesel-powered vehi-
cles. Mumbai is a large city with heavy traffic jams every day. This
source was called road traffic.
The main constituents of sea salt spray are Cl, Mg, and Na, and
there are traces of Br, Ca, K, and S. Factor 4 for PM2.5 and factor 1
for PM2.5–10 were dominated by Cl, Mg, and Na, which contributed
>77.0% of their total element concentrations. These elements were
therefore called sea salt spray. The Arabian Sea near the sampling
site could have contributed to the presence of these elements in
both PM2.5 and PM2.5–10 . As mentioned above, the Cl/Na ratio in
both size fractions was close to the ratio for seawater, and the Cl
and Na concentrations correlated well (R2 > 0.85), confirming the
contribution of sea salt spray.
Al, Ca, Fe, Si, and Ti made larger contributions to the PM2.5 fac-
tor 5 and PM2.5–10 factor 3 source profiles than to the other factor
profiles. These elements contributed >83.0% of their total mea-
sured element concentrations in PM2.5–10 but somewhat less to the
total measured element concentrations in PM2.5 . The concentra-
tions of these elements correlated well (R2 0.73–0.97 for PM2.5–10
and 0.83–0.91 for PM2.5 ), indicating that the elements had com-
mon sources. Crustal material could contribute these elements to
PM. Crustal material contributes the largest mass fraction of global
atmospheric aerosol and can be entrained by the wind, especially
when soil is mechanically disturbed (Cheng et al., 2015; Ivošević,
Stelcer, Orlić, Radović, & Cohen, 2016; Pipal, Jan, Satsangi, Tiwari,
& Taneja, 2014). Roads and subways were being repaired and con-
structed around the sampling area during the sampling period, and
these activities may have contributed these elements to PM. This
factor was called the crustal material source.
Factor 6 for both PM2.5 and PM2.5–10 had the highest contribu-
tions (>61.0%) of Cr and Mn. Cu and Zn also contributed to the PM2.5
factor 6 profile. The presence of these metals could be attributed to
metal industry activities (Shridhar, Khillare, Agarwal, & Ray, 2010;
Song et al., 2006). The many metal industry plants and storage facil-
ities near the sampling site could have contributed these metals to
the PM. These factors were called the metal industry source.
Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
9
Contributions of the sources to PM
The sources of PM2.5 and PM2.5–10 were the same, but the
contributions of the sources were very different, as shown in
Fig. 6. Crustal material made the largest contribution to PM2.5–10
(25.3%), followed by sea salt spray, coal/biomass combustion, and
fuel oil combustion. The metal industry source made the lowest
contribution (7.0%). Coal/biomass combustion made the largest
contribution to PM2.5 (25.5%), followed by road traffic, fuel oil com-
bustion, and the metal industry. Sea salt spray made the lowest
contribution (6.1%).
Natural sources, e.g., crustal material and sea salt spray, made
larger contributions to PM2.5–10 than PM2.5 , possibly because natu-
ral sources predominantly emit coarse particle (because emissions
are caused by mechanical disturbances). Anthropogenic sources,
e.g., coal/biomass combustion, fuel oil combustion, made larger
contributions to PM2.5 than PM2.5–10 . The predominantly fine par-
ticles produced during combustion processes could have been
responsible for these sources making larger contributions than
other sources to PM2.5 .
The source apportionment study results were compared with
results for other parts of India and other countries. Crustal mate-
rial, fuel oil combustion, and sea salt made larger contributions to
PM2.5–10 at Trombay than in other parts of the world, as shown
in Table 7. The contributions of road traffic to PM at Trombay and
elsewhere were similar.
The fuel oil combustion and coal/biomass combustion contribu-
tions to PM2.5 were higher at Trombay than elsewhere. Road traffic
contributed slightly less to PM2.5–10 at Trombay than elsewhere
(Table 7). Crustal material made similar contributions at Trombay
and most other places, but smaller contributions at Trombay than in
Delhi, Hyderabad, China, Europe, and Brazil. Industrial and sea salt
contributions to PM2.5 were comparable at Trombay and elsewhere.
Conclusions
The annual mean PM10 concentration exceeded the annual CPCB
limit in both 2010 and 2011, but the annual mean PM2.5 concentra-
tion was much lower than the CPCB limit. The PM concentrations
were higher in 2010 than in 2011. The PM2.5–10 fraction had the
highest mass contribution from Si, followed by Al, Fe, Na, Cl, Ca,
and the other elements. BC was the dominant contributor to PM2.5 ,
followed by Si, S, Al, and the other elements. The trace metal
contributions were higher for PM2.5 than for PM2.5–10 . The concen-
trations of the toxic elements As and Pb were much lower than the
CPCB annual limits, but the annual mean Ni concentration in PM2.5
(24.36 ng/m3 in 2011) was exceeded the CPCB limit (20 ng/m3 ). Sea-
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
10 S. Police et al. / Particuology xxx (2017) xxx–xxx

Table 7
Percentage contributions of the sources to PM2.5 and PM2.5–10 found in different studies around the world.

Location PM size Crustal Sea salt Oil combustion Road traffic Metal industry Coal/biomass References
fraction material combustion

5 sites in the Netherlands PM2.5 2 5 1 21 9 – Mooibroek et al. (2011)

Hong Kong, China PM2.5 1 6 10 29 4 – Cheng et al. (2015)
PM2.5–10 17 17 – 11 – –

Zhengzhou, China PM2.5 26 – 13 10 4 23 Geng et al. (2013)

Genoa, Italy PM2.5 6.7 15 13 22 – – Bove et al. (2014)
Rijeka, Croatia PM2.5 1.4 9.1 9.9 – 3.9 – Ivošević et al. (2016)
Changhua, Taiwan, China PM2.5 – – 18.8 24.2 – 34.7 Hsu et al. (2016)
Delhi, India PM2.5 20.5 4.3 13.7 – 6.2 – Sharma, Mandal, Jain,
Saraswati Sharma, and
Saxena (2016)
Hyderabad, India PM2.5 26 – 9 – 7 – Gummeneni et al. (2011)

Ile-Ife, Nigeria PM2.5 44 – – 12 – Owoade et al. (2016)

PM2.5–10 71 22 – 2 – –

Lisbon, Portugal PM2.5 16 8 8 14 – – Almeida, Pio, Freitas,

Reis, and Trancoso (2005)
PM2.5–10 20 47 5 13 – –

Tijuana, Mexico PM2.5 0.5 15.3 37.4 13.7 2.1 – Minguillón et al. (2014)
PM2.5–10 20.3 10.6 1.3 0.6 0 –

Kuala Lumpur, Malaysia PM2.5 8.3 – – 16.7 17.5 Rahman et al. (2011)
PM2.5–10 8.5 – – – 17.6 –

Colombo, Sri Lanka PM2.5 9 9 – – 66 – Seneviratne et al. (2011)

Corpus Christi, Texas, USA PM2.5 10.1 8.1 4.6 9.7 6 12.7 Karnae and John (2011)
Trombay, Mumbai, India PM2.5 8.7 6.1 19.0 17.7 10.6 25.5 Present study
PM2.5–10 25.3 15.0 11.2 12.6 7.0 13.8

sonal variations during the study indicated that the air in Trombay
is most polluted with PM and associated elements in winter, less
polluted in summer, and less polluted still in the monsoon season.
The source apportionment study indicated that the PM collected
came from six sources, soil dust, sea salt spray, coal/biomass com-
bustion sources, residual fuel oil combustion sources, road traffic,
and the metal industry. Natural sources such as crustal material
and sea salt spray made the largest contributions to PM2.5–10 , but
anthropogenic sources contributed most to PM2.5 , the largest con-
tributions coming from coal/biomass combustion (25.5%), fuel oil
combustion (19.0%), and road traffic (17.7%).
Acknowledgement
It is hereby declared that there is no financial or personal conflict
of interest.
References
Almeida, S. M., Pio, C. A., Freitas, M. C., Reis, M. A., & Trancoso, M. A. (2005). Source
apportionment of fine and coarse particulate matter in a sub-urban area at the
Western European Coast. Atmospheric Environment, 39, 3127–3138.
An, X., Hou, Q., Li, N., & Zhai, S. (2013). Assessment of human exposure level to PM10
in China. Atmospheric Environment, 70, 376–386.
Ancelet, T., Davy, P. K., Trompetter, W. J., & Markwitz, A. (2014). Sources of particulate
matter pollution in a small New Zealand city. Atmospheric Pollution Research, 5,
572–580.
Balakrishnaiah, G., Kumar, K. R., Reddy, B., Gopal, K. R., Reddy, R. R., Reddy, L. S. S.,
et al. (2011). Characterization of PM, PM10 and PM2.5 mass concentrations at a
tropical semi-arid station in Anantapur, India. Indian Journal of Radio & Space
Physics, 40, 95–104.
Banerjee, T., Murari, V., Kumar, M., & Raju, M. P. (2015). Source apportionment of
airborne particulates through receptor modeling: Indian scenario. Atmospheric
Research, 164, 167–187.
Bove, M. C., Brotto, P., Cassola, F., Cuccia, E., Massabò, D., Mazzino, A., et al. (2014).
An integrated PM2.5 source apportionment study: Positive matrix factorization
vs. the chemical transport model CAMx. Atmospheric Environment, 94, 274–286.
Central Pollution Control Board (CPCB). (2010). Air quality monitoring, emission inven-
tory and source apportionment study for Indian cities. New Delhi, India: Central
Pollution Control Board, Government of India.
Cheng, Y. H., Liu, Z. S., & Chen, C. C. (2010). On-road measurements of ultrafine
particle concentration profiles and their size distributions inside the longest
highway tunnel in Southeast Asia. Atmospheric Environment, 44, 763–772.
Cheng, Y., Lee, S., Gu, Z., Ho, K., Zhang, Y., Huang, Y., et al. (2015). PM2.5 and PM10–2.5
chemical composition and source apportionment near a Hong Kong roadway.
Particuology, 18, 96–104.
Cohen, A. J., Ross Alexander, H., Ostro, B., Pandey, K. D., Kryzanowski, M., Künzail, N.,
et al. (2005). The global burden of disease due to outdoor air pollution. Journal
of Toxicology and Environmental Health, Part A, 68(13–14), 1301–1307.
Cohen, D. D., Crawford, J., Stelcer, E., & Bac, V. T. (2010). Characterization and source
apportionment of fine particulate sources at Hanoi from 2001 to 2008. Atmo-
spheric Environment, 44, 320–328.
Denier van der Gon, H. A. C., Hulskotte, J. H. J., Visschedijk, A. J. H., & Schaap, M. (2007).
A revised estimate of copper emissions from road transport in UNECE-Europe
and its impact on predicted copper concentrations. Atmospheric Environment,
41, 8697–8710.
Deshmukh, D. K., Deb, M. K., & Mkoma, S. L. (2013). Size distribution and seasonal
variation of size-segregated particulate matter in the ambient air of Raipur city,
India. Air Quality, Atmosphere and Health, 6, 259–276.
Geng, N., Wang, J., Xu, Y., Zhang, W., Chen, C., & Zhang, R. (2013). PM2.5 in an industrial
district of Zhengzhou, China: Chemical composition and source apportionment.
Particuology, 11, 99–109.
Guinot, B., Gonzalez, B., De Faria, J. P., & Kedia, S. (2016). Particulate matter char-
acterization in a steel works using conventional sampling and innovative lidar
observations. Particuology, 28, 43–51.
Gummeneni, S., Yusup, Y. B., Chavali, M., & Samadi, S. Z. (2011). Source apportion-
ment of particulate matter in the ambient air of Hyderabad Hyderabad city,
India. Atmospheric Research, 101(3), 752–764.
Guttikunda, S., & Kopakka, R. (2014). Source emissions and health impacts of urban
air pollution in Hyderabad, India. Air Quality, Atmosphere and Health, 7, 195–207.
Hamra, B. G., Guha, N., Cohen, A., Laden, F., Raaschou-Nielsen, O., Samet, J. M., et al.
(2014). Outdoor particulate matter exposure and lung cancer: A systematic
review and meta-analysis. Environmental Health Perspectives, 122(9), 906–911.
Herrera-Murillo, J., Rodríguez-Román, S., Rojas-Marín, J. F., Campos-Ramos, A.,
Blanco-Jiménez, S., Cárdenas-González, B., et al. (2013). Chemical characteri-
zation and source apportionment of PM10 and PM2.5 in the metropolitan area of
Costa Rica, Central America. Atmospheric Pollution Research, 4, 181–190.
Horemans, B., Cardell, C., Bencs, L., Kontozova-Deutsch, V., De Wael, K., & Van
Grieken, R. (2011). Evaluation of airborne particles at the Alhambra monument
in Granada, Spain. Microchemical Journal, 99, 429–438.
Hsu, C. Y., Chiang, H. C., Lin, S. L., Chen, M. J., Lin, T. Y., & Chen, Y. C. (2016). Elemental
characterization and source apportionment of PM10 and PM2.5 in the western
coastal area of central Taiwan. Science of the Total Environment, 541, 1139–1150.
Ivošević, T., Stelcer, E., Orlić, I., Radović, I. B., & Cohen, D. (2016). Characterization and
source apportionment of fine particulate sources at Rijeka, Croatia from 2013 to
2015. Nuclear Instruments and Methods in Physics Research B, 371, 376–380.
Karnae, S., & John, K. (2011). Source apportionment of fine particulate matter
measured in an industrialized coastal urban area of South Texas. Atmospheric
Environment, 45, 3769–3776.
Kertész, Z., Szoboszlai, Z., Angyal, A., Dobos, E., & Borbély-Kiss, I. (2010). Identi-
fication and characterization of fine and coarse particulate matter sources in

Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
G Model
PARTIC-1076; No. of Pages 11 ARTICLE IN PRESS
S. Police et al. / Particuology xxx (2017) xxx–xxx
a middle-European urban environment. Nuclear Instruments and Methods in
Physics Research B, 268, 1924–1928.
Khillare, P. S., & Sarkar, S. (2012). Airborne inhalable metals in residential areas of
Delhi, India: Distribution, source apportionment and health risks. Atmospheric
Pollution Research, 3, 46–54.
Li, R., Wiedinmyer, C., & Hannigan, M. P. (2013). Contrast and correlations between
coarse and fine particulate matter in the United States. Science of the Total Envi-
ronment, 456, 346–358.
Liu, G., Li, J., Wu, D., & Xu, H. (2015). Chemical composition and source apportionment
of the ambient PM2.5 in Hangzhou, China. Particuology, 18, 135–143.
Malm, W. C., Sisler, J. F., Huffman, D., Eldred, R. A., & Cahill, T. A. (1994). Spatial and
seasonal trends in particle concentration and optical extinction in the United
States. Journal of Geophysical Research, 99(D1), 1347–1370.
Manousakas, M., Diapouli, E., Papaefthymiou, H., Migliori, A., Karydas, A. G.,
Padilla-Alvarez, R., et al. (2015). Source apportionment by PMF on elemental
concentrations obtained by PIXE analysis of PM10 samples collected at the vicin-
ity of lignite power plants and mines in Megalopolis, Greece. Nuclear Instruments
and Methods in Physics Research B, 349, 114–124.
Masiol, M., Squizzato, S., Rampazzo, G., & Pavoni, B. (2014). Source apportionment
of PM2.5 at multiple sites in Venice (Italy): Spatial variability and the role of
weather. Atmospheric Environment, 98, 78–88.
Mimura, T., Ichinose, T., Yamagami, S., Fujishima, H., Kamei, Y., Goto, M., et al. (2014).
Airborne particulate matter (PM2.5 ) and the prevalence of allergic conjunctivitis
in Japan. Science of the Total Environment, 487, 493–499.
Minguillón, M. C., Campos, A. A., Cárdenas, B., Blanco, S., Molina, L. T., & Querol,
X. (2014). Mass concentration, composition and sources of fine and coarse
particulate matter in Tijuana, Mexico, during Cal-Mex campaign. Atmospheric
Environment, 88, 320–329.
Mooibroek, D., Schaap, M., Weijers, E. P., & Hoogerbrugge, R. (2011). Source appor-
tionment and spatial variability of PM2.5 using measurements at five sites in the
Netherlands. Atmospheric Environment, 45, 4180–4191.
Moreno, T., Querol, X., Alastuey, A., de la Rosa, J., de la Campa, A. M. S., Minguillón,
M., et al. (2010). Variations in vanadium, nickel and lanthanoid element concen-
trations in urban air. Science of the Total Environment, 408(20), 4569–4579.
Owoade, K. O., Hopke, P. K., Olise, F. S., Adewole, O. O., Ogundele, L. T., & Fawole, O. G.
(2016). Source apportionment analyses for fine (PM2.5 ) and coarse (PM2.5–10 )
mode particulate matter (PM) measured in an urban area in southwestern
Nigeria. Atmospheric Pollution Research, 7, 843–857.
Paatero, P., Eberly, S., Brown, S. G., & Norris, G. A. (2014). Methods for estimating
uncertainty in factor analytic solutions. Atmospheric Measurement Techniques, 7,
781–797.
Pakbin, P., Ning, Z., Shafer, M. M., Schauer, J. J., & Sioutas, C. (2011). Seasonal and
spatial coarse particle elemental concentrations in the Los Angeles area. Aerosol
Science and Technology, 45, 949–963.
Pipal, A. S., Jan, R., Satsangi, P. G., Tiwari, S., & Taneja, A. (2014). Study of surface
morphology, elemental composition and origin of atmospheric aerosols (PM2.5
and PM10 ) over Agra, India. Aerosol and Air Quality Research, 14, 1685–1700.
Polissar, A. V., Hopke, P. K., Paatero, P., Malm, W. C., & Sisler, J. F. (1998). Atmospheric
aerosol over Alaska 2. Elemental composition and sources. Journal of Geophysical
Research, 103(D15), 19045–19057.
Rahman, S. A., Hamzah, M. S., Wood, A. K., Elias, M. S., Salim, N. A. A., & Sanuri, E.
(2011). Sources apportionment of fine and coarse aerosol in Klang Valley, Kuala
Lumpur using positive matrix factorization. Atmospheric Pollution Research, 2(2),
197–206.
Ramanathan, V., Crutzen, P. J., Kiehl, J. T., & Rosenfeld, D. (2001). Aerosols, climate,
and the hydrological cycle. Science, 294, 2119–2124.
Please cite this article in press as: Police, S., et al. Chemical composition and source apportionment of PM2.5 and PM2.5–10 in Trombay
(Mumbai, India), a coastal industrial area. Particuology (2017), https://doi.org/10.1016/j.partic.2017.09.006
11
Reff, A., Eberly, S. I., & Bhave, P. V. (2007). Receptor modeling of ambient particulate
matter data using positive matrix factorization: Review of existing methods.
Journal of the Air and Waste Management Association, 57, 146–154.
Rogula-Kozłovska, W., Klejnowski, K., Rogula-Kopiec, P., Mathews, B., & Szopa, S.
(2012). A study of the seasonal mass closure of ambient fine and coarse Dust
in Zabrze, Poland. Bulletin of Environmental Contamination and Toxicology, 88,
722–729.
Salako, O. G., Hopke, P. K., Cohen, D. D., Bilkis, A. B., Biswas, S. K., Pandit, G. G., et al.
(2012). Exploring the variation between EC and BC in a variety of locations.
Aerosol and Air Quality Research, 12, 1–7.
Seneviratne, M. C. S., Waduge, V. A., Hadagiripathira, L., Sanjeewani, S., Attanayake,
T., Jayaratne, N., et al. (2011). Characterization and source apportionment of
particulate pollution in Colombo, Sri Lanka. Atmospheric Pollution Research, 2,
207–212.
Sharma, S. K., Mandal, T. K., Jain, S., Saraswati Sharma, A., & Saxena, M. (2016). Source
apportionment of PM2.5 in Delhi, India using PMF model. Bulletin of Environmen-
tal Contamination and Toxicology, 97(2), 286–293.
Sharma, S. K., Mandal, T. K., Saxena, M., Rashmi, Rohtash, A. Sharma, et al. (2014).
Source apportionment of PM10 by using positive matrix factorization at an urban
site of Delhi, India. Urban Climate, 10, 656–670.
Shi, G.-L., Tian, Y.-Z., Ye, S., Peng, X., Xu, J., Wang, W., et al. (2015). Source apportion-
ment of synchronously size segregated fine and coarse particulate matter, using
an improved three-way factor analysis model. Science of the Total Environment,
505, 1182–1190.
Shridhar, V., Khillare, P. S., Agarwal, T., & Ray, S. (2010). Metallic species in ambient
particulate matter at rural and urban location of Delhi. Journal of Hazardous
Materials, 175, 600–607.
Song, Y., Zhang, Y., Xie, S., Zeng, L., Zheng, M., Salmon, L. G., et al. (2006). Source
apportionment of PM2.5 in Beijing by positive matrix factorization. Atmospheric
Environment, 40, 1526–1537.
Tao, Y., Mi, S., Zhou, S., Wang, S., & Xie, X. (2014). Air pollution and hospital admis-
sions for respiratory diseases in Lanzhou, China. Envirnmental Pollution, 185,
196–201.
Tiwari, S., Hopke, P. K., Pipal, A. S., Srivastava, A. K., Bisht, D. S., Tiwari, S., et al.
(2015). Intra-urban variability of particulate matter (PM2.5 and PM10 ) and its
relationship with optical properties of aerosols over Delhi, India. Atmospheric
Research, 166, 223–232.
Villalobos, A. M., Amonov, M. O., Shafer, M. M., Devi, J. J., Gupta, T., Tripathi, S. N., et al.
(2015). Source apportionment of carbonaceous fine particulate matter (PM2.5 )
in two contrasting cities across the Indo–Gangetic Plain. Atmospheric Pollution
Research, 6, 398–405.
Watson, J. G., Chow, J. C., & Houck, J. E. (2001). PM2.5 chemical source profiles for
vehicle exhaust, vegetative burning, geological material and coal burning in
Northwestern Colorado during 1995. Chemosphere, 43, 1141–1151.
Yadav, R., Sahu, L. K., Jaaffrey, S. N. A., & Beig, G. (2014). Temporal variation of partic-
ulate matter (PM) and potential sources at an urban site of Udaipur in western
India. Aerosol and Air Quality Research, 14(6), 1613–1629.
Yadav, S., & Satsangi, P. G. (2013). Characterization of particulate matter and its
related metal toxicity in an urban location in south west India. Environmental
Monitoring and Assessment, 185, 7365–7379.
Yatkin, S., & Bayram, A. (2008). Source apportionment of PM10 and PM2.5 using pos-
itive matrix factorization and chemical mass balance in Izmir, Turkey. Science of
The Total Environment, 390, 109–123.