Resuspended Dust and Vehicle Exhaust Emissions in Central India

Eight composite PM10-2.5 source profiles were developed for resuspended dust and vehicle exhaust emissions with 32 chemical species, including 21 elements (Al, As, Ca, Cd, Co, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, Pb, S, Sb, Se, V, and Zn), 9 water-soluble ions (Na, K, Mg, Ca, NH4, Cl, F, NO3, and SO4), and carbonaceous fractions (OC and EC). Dust samples were dominated by crustal elements (Al, Ca, Fe, and Mg) while exhaust emissions showed high abundances of carbonaceous aerosol (OC and EC). Crustal species (Al, Fe, Mg, and Na) were more enriched over native soils in PM10-2.5 as compared to PM2.5. The higher coefficients of divergence (COD) indicate that profiles differ from each other. Ca accounted for nearly 30% of PM10-2.5 mass in construction dust while Fe accounted for nearly 20% of PM10-2.5 mass in paved road dust. Threeand four-wheeler diesel exhaust profiles consisted of 5–7% EC, with 6–10 times higher Pb, Se, and S abundances than those in two-wheeler gasoline exhaust profile. The heavy-duty diesel exhaust profile consist of nearly 20% EC with abundant (> 0.5%) trace elements (e.g., Pb, Se, and Zn).


INTRODUCTION
Air pollution is of great concern in India, especially the high levels of particulate matter (PM) emitted from uncontrolled industrial processes, solid waste and biomass burning, vehicular exhaust, and resuspended road dust (Pant and Harrison, 2013;Pant et al., 2015).Real-world source characterizations are needed to obtain chemical source profiles for input to receptor models, such as the Chemical Mass Balance (CMB), to identify and quantify source contributions.The U.S. EPA SPECIATE (USEPA, 2013), European SPECIEUROPE (Pernigotti et al., 2016), and China Source Profile Shared Service (CSPSS) (Liu et al., 2017) databases have assembled many of these profiles.Gargava and Rajagopalan (2016) found that road dust and vehicular exhaust emissions account for ~30-70% and ~15-20% of the measured PM 10 mass, respectively, in India.Various studies have been conducted (Chow et al., 2003;Ho et al., 2003;Kong et al., 2011, Patil et al., 2013;Han et al., 2014;Kong et al., 2014;Matawle et al., 2015;Pant et al., 2015;Wang et al., 2015;Liu et al., 2016) to derive dust and motor vehicle exhaust profiles (Chow et al., 2004;Han et al., 2014;Matawle et al., 2015;Liu et al., 2017).This study reports additional PM 10-2.5 chemical source profiles for resuspended dust and vehicle exhaust emissions specific to India.

Source Sampling and Chemical Analysis
Source sampling was conducted in Raipur, the capital of Chhattisgarh, India (21°14′22.7′′N,81°38.1′′E),with a population of ~1.6 million (Census, 2011), as documented by Matawle et al. (2014Matawle et al. ( , 2015) ) for PM 2.5 .This paper describes the PM 10-2.5 chemical profiles for the eight resuspended dust and vehicle exhaust emissions tests.Source samples are summarized in Table 1.Geological samples typical of Central India include paved road and construction dust in Raipur City, as well as unpaved surface dust and non-agricultural soils outside of Raipur City.Sweeping and grab sampling methods were employed to obtain 0.5-1 kg of each dust which were air dried (~25°C), sieved (Tyler 400 mesh to 38 µm in geological diameter), and resuspended in a laboratory chamber through PM 2.5 and PM 10 inlets at 5 L min -1 following Chow et al. (1994) as applied in past studies (Watson and Chow, 2001;Watson et al., 2001;Chow et al., 2004).
Motor vehicle exhaust samples were acquired from four major vehicle categories that are common in India including: two-wheeler gasoline, three-and four-wheeler diesel, and heavy-duty diesel vehicles.Vehicles manufactured between 2000 and 2001 were selected for in-plume sampling through collocated PM 2.5 and PM 10 inlets on Minivol samplers (Airmetrics) at a flow rate of 5 L min -1 .Vehicles were operated under steady state conditions for 30-60 minutes to ensure adequate deposit on quartz-fiber filters (Whatman catalog No. 1851-047) for subsequent chemical analysis.Five sets of samples were collected from each source, for a total of 40 samples.
Detailed chemical analysis and quality assurance/quality control procedures are documented in Matawle et al. (2014Matawle et al. ( , 2015)).Laboratory filter blanks and field trip blanks were submitted to the same chemical analysis to assess background levels.One standard sample was analysed after each 10 samples to assure 80%-120% recovery.Triplicate analyses were performed for each sample to achieve ±10% reproducibility.The limits of detections (LODs) for each species were reported in Matawle et al. (2014).

PM 10-2.5 Chemical Source Profile
The four resuspended dust and four vehicle exhaust profiles are summarized in Tables 2 and 3, respectively.The sum of species accounted for 40-47% and 52-69% of PM 10-2.5 mass for dust and vehicle exhaust profiles, respectively.Crustal elements (Al, Ca, Fe, K, Mg, and Na) were the most abundant species in dust, contributing 31-45% of the PM 10-2.5 mass, whereas total carbon (TC = OC + EC) constituted 49-57% exhaust.The OC/TC ratios ranged from 0.65-0.98,comparable to 0.57-0.98 reported in India for PM 10 (CPCB, 2008b) and PM 2.5 (Matawle et al., 2015).The low sum of species for dust is mainly due to the lack of silicon (Si) in the profile.Si is often the most abundant element in crustal dust (Chow et al., 2003).The quartzfiber filter prohibits Si analysis and the use of Si/Al ratio as Table 1.Descriptions of source type, sampling location, and source sampling method.1. a source marker (Contini et al., 2016).Future studies should be conducted with parallel Teflon-membrane and quartz-fiber filters to accommodate complete chemical speciation (Chow et al., 1994;Watson et al., 2001).

Source Profile for Resuspended Dust
Fig. 1 shows four abundant crustal species: Ca, Fe, Mg, and Al.The most abundant species, Ca, varied two-fold among the four profiles, from 27.9 ± 7.3% in construction dust (CD) to 14.3 ± 22% in non-agricultural soils (SD).Ca is commonly found in construction dust (Yatkin and Bayram, 2008;Kong et al., 2011;Pant and Harrison, 2012;Shen et al., 2016) owing to its presence in concrete.Ca was not water soluble, with Ca 2+ /Ca values in the range of 0.14-0.18,with a lower ratio for construction dust (0.012).Fe was most abundant (17.5 ± 0.8%) in unpaved road dust (UPRD), compared to a lower abundance in construction dust (CD, 7.1 ± 0.7%).Al levels were low (0.8-0.9%) in paved and unpaved road dust, but they were highest at 2-3% in soil and construction dust.Mg levels were similar, in the range of 2-4% of PM 10-2.5 mass.These abundances are comparable to those from past studies for PM 2.5 , PM 10-2.5 , and PM 10 (Chow and Watson, 1994;Watson et al., 2001;Amato et al., 2009;Patil et al., 2013;Matawle et al., 2015;Wang et al., 2015;Samiksha et al., 2017).As expected, most of the soil-related K was not water soluble.K was 12 times higher than soluble K + with a K + /K ratio of 0.08; higher than 0.1-0.5 reported in past PM 10 (CPCB, 2008a; Kong et al., 2014) and PM 2.5 (Watson et al., 2001;Matawle et al., 2015) studies.This is in contrast to biomass burning profiles where the K + /K ratio is in the range of ~0.87-0.90(Watson et al., 2001;Chow et al., 2004).TC accounted for  1.
2-7% of PM 10-2.5 mass.The OC abundance in PRD was 5.6 ± 3.5% compared to that in UPRD at 2.1 ± 1.3%.Heavily travelled roads are subject to more vehicle exhaust deposition.Pb, V, and S abundances were 3-5 times higher in PRD as compared to other dusts, similar to abundances in other Indian cities for PM 10 (Samara, 2005;CPCB, 2008a).OC/TC ratios ranged from 0.56 in UPRD to 0.98 in CD, consistent with 0.64-0.99 reported in past PM 10 studies (Ho et al., 2003;Chow et al., 2004;Gupta et al., 2007).
Enrichment Factors (EF) were calculated relative to Ca in local soil as a reference element because: (1) The study region is located in a rock basin with high Ca abundances; (2) Ca correlates with other elements in the dust matrix (Quraishi, 1997); and (3) Past studies have used Ca as an EF reference element (Sharma and Pervez, 2003).The EF (Cao et al., 2008;Chakraborty and Gupta, 2009;Behera and Sharma, 2010) is: where (X i /Ca)sample and (X i /Ca) crust are ratios of the abundance of element X i and Ca in PM samples and in crustal materials, respectively.Fig. 2 shows elemental EFs and Fig. 3 compares EFs for PM 10-2.5 and PM 2.5 profiles (Matawle et al., 2015).EFs for both size fractions are detailed in Supplemental Table S1.As shown in Fig. 3, Cd had high EFs for all sources ranging from 9-17 for PM 2.5 and 2-8 for PM 10-2.5 , comparable to other studies (Han et al., 2014;Kong et al., 2014).For PRD and UPRD profiles, As, Cu, and Zn were enriched (EF > 5 for PM 2.5 and EF > 3 for PM 10-2.5 ), consistent with influences from traffic emissions such as tire and brake wear.Most of the crustal species (Al and Mg) were enriched in PM 10-2.5 as compared  to PM 2.5 , while most of the anthropogenic related elements were more abundant in PM 2.5.

Source Profiles for Vehicle Exhaust Emissions
The four exhaust profiles are shown in Fig. 4. TC was the most abundant species accounting for 49-57% of the measured mass.The OC abundance was highest for the two-wheeler gasoline exhaust (2WVG, 48% ± 6.6%) whereas the EC abundance was highest for the heavy-duty diesel (HDVD, 19.7% ± 1.5%).These levels were 3-26 times higher than two-to four-wheeler exhaust profiles, but comparable to PM 10 profiles reported by Han et al. (2014).The OC/TC ratios in the range of 0.65 to 0.98 were similar to the 0.55-0.95for the PM 10 profiles of Han et al. (2014) as well as 0.57-0.98 in Matawle et al. (2015) and 0.66-0.80 in Watson et al. (2001) for PM 2.5 .The largest difference were in the OC/EC ratios, ranging from 1.9 for HDVD to 63.8 for 2WVG, mainly due to the low EC levels (0.8 ± 0.2%) for two-wheeler gasoline exhaust profile.Other elemental abundances were low except for Na, ranging from 0.63 ± 1.5% in 3WVD to 6.7 ± 5.2% in HDVD.The heavy-duty diesel vehicle profile contained the highest Pb (0.77 ± 0.06%), Se (0.76 ± 0.09%), Cl -(1.1 ± 0.13%), and SO 4 2-(1.0± 0.2%), abundances.Three-and four-wheeler diesel exhaust profiles consisted of 5-7% of EC, with 6-10 times higher Pb, Se, and S than two-wheeler gasoline vehicles.
Approximately 91-93% of measured mass was achieved for exhaust profiles, with lower values (65-76%) for the dust profiles, mainly due to the lack of Si and Ti measurements.As expected, OM was the major fraction (72-95%) of exhaust emissions, whereas geological minerals (80-94%) dominated the dust profiles.

Coefficients of Divergence
To evaluate the similarities and differences among the profiles, coefficients of divergence (COD) were calculated, as described by Matawle et al. (2014Matawle et al. ( , 2015)).When COD values < 0.2, as suggested by Contini et al. (2012), the two sources are similar, and when the COD > 0.2 the two sources are considered different (Wongphatarakul et al., 1998;Wilson et al., 200;).Table 4 shows high COD values ranging from 0.48 between 3WVD and 4WVD to 0.84 between PRD and HDVD, indicating that the profiles are not collinear.

Implications for Source Apportionment Source Markers
Potential source markers are identified by the following equation (Yang et al., 2002;Kong et al., 2011): where X i is the i th species concentration; (X i /∑X) j is the abundance of i th species divided by the sum of the measured 32 species concentration (∑X) for source j; (X i /∑X) min is the minimum abundance of the i th individual species divided by ∑X (Yang et al., 2002;Chen et al., 2003).Individual species concentrations are further normalized by dividing the i th species concentration by the sum of the i th concentrations (Kong et al., 2011).Species with the six highest ratios are potential source markers.Similar approaches used by other studies are summarized in Table 5. Past studies (Mitra et al., 2002;Watson et al., 2008;Viana et al., 2008;Guttikunda, 2009;Kong et al., 2011;Matawle et al., 2015) showed that Al, Si, K, Ca, Mg, and Fe were commonly used as markers for dust sources, whereas OC, EC, S, or SO 4 2-, and Pb were markers for exhaust.As shown in Table 5, Pb and Se may be markers for paved road dust in PM 10-2.5 and for unpaved road dust in PM 2.5 (Matawle et al., 2015).

Diagnostic Ratios
Diagnostic ratios are used to distinguish among sources (Arditsoglou and Samara, 2005;Kong et al., 2011;Matawle et al., 2015).The V/Ni ratio was used to assess emissions from marine vessels and residual oil combustion and Cu/Sb and Cu/Zn ratios were used for traffic emissions (Pey et al., 2010).Arditsoglou and Samara (2005) used Zn/Pb ratios in the range of 0.3-0.4 as to infer exhaust emissions, and 1.2 for oil combustion.Mitra et al. (2002) suggested a Mn/V ratio << 1 for oil burning and >> 1 for coal burning emissions.

CONCLUSION
PM 10-2.5 source profiles from paved road and construction dust in Raipur, unpaved road dust and non-agricultural soil outside of Raipur, along with vehicle exhaust from gasoline two-wheelers, diesel three-and four-wheeler and heavy-duty diesel vehicles were acquired.In addition to gravimetric mass, these samples were analysed for 21 elemental species, 9 water soluble ions, and carbon (OC and EC).Crustal elements (Al, Ca, Fe and, Mg) dominated the resuspended dust while carbonaceous species (OC and EC) were more abundant in vehicle exhaust emissions.Ca was most abundant in construction dust (27.9 ± 7.3% of PM 10-2.5 mass) while the most abundant Fe (17.5 ± 0.8%) was found in unpaved road dust.Heavy-dusty diesel vehicles (HDVD) reported the highest EC abundance (19.7 ± 1.5%) with very low EC (0.75 ± 0.21%) found in gasoline two wheelers (2WVG).Elevated levels of Pb (0.77 ± 0.06%), Se (0.76 ± 0.09%), and Zn (0.91 ± 0.31%) were also apparent in HDVD.The coefficients of divergence (COD) ranged 0.48 to 0.84 suggesting profiles were significantly different Dust samples from a construction site located in the study area Chamber resuspension sampling PRD Paved Road Dust Dust samples from the surface of paved road of the study area Chamber resuspension sampling UPRD Unpaved Road Dust Dust samples from the surface of unpaved road outside the city of Raipur Chamber resuspension sampling 2WVG Two-Wheeler Vehicles (gasoline) Samples from exhaust pipes of petrol driven 2-wheelers In-plume sampling 3WVD Three-Wheeler Vehicles (diesel) Samples from exhaust pipes of diesel driven 3-wheelers passenger auto rickshaws In-plume sampling 4WVD Four-Wheeler Vehicles (diesel) Samples from exhaust pipes of diesel driven 4-wheelers personal cars In-plume sampling HDVD Heavy Duty Vehicles Samples from exhaust pipes of diesel driven heavy duty trucks In-plume sampling a Five samples were collected and composited to develop each source profile.

Table 2 .
PM 10-2.5 composite sources profiles (weight percent by mass) for resuspended dust inside and outside of Raipur City.
a See profile description in Table

Table 3 .
PM 10-2.5 composite sources profiles (weight percent by mass) for vehicle exhaust emissions.
a See profile description in Table

Table 4 .
Coefficients of Divergence (COD) for resuspended dust and vehicle exhaust emissions a See profile description in Table1.

Table 5 .
Source markers of PM 10-2.5 for resuspended dust and vehicle exhaust emissions.

. Four-Wheeler Vehicles (diesel) (4WVD) PM 10-2.5 S, EC, OC, SO 4 2-, Pb, and Zn Present study
from each other.Lower than usual mass reconstruction for resuspended dust (65-76%) reconfirm the importance to include Si and Ti in future studies.Source markers were identified as Al, Ca, and Fe for resuspended dust and OC, EC, and Pb for vehicle exhaust emissions.These regionspecific profiles are more representative of pollution source characteristics and can be used for future source apportionment studies.