Understanding sources and processes using chemistry of ne and coarse particles simultaneously collected from different windward locations in N-NW India

PM2.5and PM2.5-10 particles, simultaneously collected from Bikaner (BKR) and New Delhi (DEL) through Jhunjhunu (JHJ) located along S-SW wind pathduring summer in N-NW India were chemically characterized to understand their sourcesand related processes.Winds preferentially picked and trasnported PM2.5 compared to PM2.5-10 in windward direction. Ratios of mass concentrations and WSII in PM2.5 and PM2.5-10 were three or more at all sites. Thermal power plants, vehicles, and plastic burning were major contributors of SO42-, NO3- and Cl- ionsat DEL. Crustal materials, salt lakes/playas and contaminated aged particles were sources of WSIIs in PM2.5 over BKR and JHJ. In PM2.5, burning of wood and cow dung resulted in high OC/EC ratio (6.9) at BKR whereas EC emissions from vehicles lowered this ratio (3.5) at DEL. EC was dominated by char-EC compared to soot-EC. Major elements showed similar concentrations in both size particles but were depleted compared to Upper Continental Crustal (UCC) due to silica dilution effect. Ba and Sr, and Ba, Cr, Sr and Zn content showed site specic variations in PM2.5 and PM2.5-10, respectively. Trace elements in PM2.5 showed high enrichment compared to UCC at DEL due to re-suspension of roadside dust and vehicle related emissions.


Introduction
Arid regions over continents are potential sources of dust particles, which is transported to local, regional and continental scales by prevailing winds. 1,2,3,4,5 Wind-assisted transport of particles from desert regions in uences air quality 6,5 human health 7,8 atmospheric chemistry 9,10 , visibility 11 and nutrient dynamics 12 . Physical and chemical nature of atmospheric particles are variable with respect to their sources and size and are also in unced by metereological factors, secondary chemical processes and geographical location 13,14 . Fine particles (PM 2.5 ) are largely contributed by anthropogenic sources with high residence time and are inhaled while breathing and cause direct health effects compared to coarse particles 15,16 .
New Delhi, the capital of India, is facing severe air quality problems during summer season due to wind-assisted transport of particles from the Thar Desert region in upwind direction 17,18,19 . Rapid urbanization, industrialization and transportation system further amplify the problems of air pollution in urban area of downwind (New Delhi) region 20,13,18 .
Mixing of crustal particles with anthropogenic particles such as (sulphate, nitrate), organic carbon (OC), elemental carbon (EC) during long-range transport and deposition affects atmospheric chemistry 21,19 . Water-soluble Inorganic ions (WSIIs), OC, EC in particles are useful tool to understand anthropogenic sources and chemical interactions in ambient atmosphere whereas major ( Al, Fe, Ca, Ti, Mn ) and trace elements (Ba, Cd, Co, Pb, Cr, Cu, Zn) can be used as source signatures of crustal materials and anthropogenic sources such as fossil fuel combustion, abrasion of vehicle tire, and industrial emission 22,23,24,19 . Studies conducted in past over N-NW India remained focused on PM 10 and SPM in this region and have suggested signi cant in uence of dust storm on ambient air quality 25,26,14 . In this work, simultaneous sampling of PM 2.5 and PM 2.5-10 was done at Bikaner located in active desert in upwind region, Jhunjhunu in paleo-desert region and New Delhi in downwind region in sub-humid climate. All sampling sites are located along S-SW wind corridor from west to east in north-northwest India. Samples were analyzed for WSII, OC, EC and for elements in the residue lter of water-soluble fraction to understand sources, processes and impact of S-SW winds on air quality and particles concentrations.

STUDY AREA
Study area for this work extended from great Indian Thar desert in the west to New Delhi for PM 2.5 and PM 2.5-10 sample collection during summer season in N-NW India (Fig.1). This region experiences transition in climate, meteorological conditions, geomorphic features, altitude from sea level, urbanization and industrialization on moving from west ( Bikaner: 28.01ºN and 73ºE) ) to east ( New Delhi: 28.38ºN and 77.12ºE) through Jhunjhunu (28.06ºN and 75.25ºE) covering a distance of 600 km. The Geat Indian Thar Desert in the west acts as potential source of particles and high intensity winds picks particles and transport them in windward direction. Particle carrying capacity of S-SW winds decreases along windward direction due to increase in elevation from sea level, Aravalli Mountains, more vegetation cover, and sub-humid climate (Fig.1). This study domain represents a unique geographical region, good natural laboratory for the studies on particle generation, transportation and depositional processes along dominant wind path during summer season and their environmental implications. More details about the study area can be accessed from our previous work 25,26,18 . Meteorological data for the study period were taken from Resources Laboratory: National Oceanic and Atmospheric Administration (ARL-NOAA) (http://www.arl.noaa.gov/). Cluster mean trajectories of air backward trajectories ending at the sampling sites BKR, JHJ and DEL during the sampling period May-2012 were calculated at two level 500m and 1000m above ground level (AGL) (Fig.2). Meteorological data (average wind speed, relative humidity, planetary boundary layer) over BKR,JHJ and DEL are provided in Supplementary Table S1 online. Methodology PM 2.5 and PM 2.5-10 particles were simultaneously collected for the period (May 2012) using mass ow controlled high volume air sampler (Tisch Environmental Inc) tted with cascade impactors installed at rooftop of School of Environmental Sciences (SES) buildings in Jawaharlal Nehru University (JNU), New Delhi (DEL), and rooftop of house buildings located at Chakwas village near to Jhunjhunu (JHJ) city and at Udairamsar village near to Bikaner (BKR) city during summer season (Fig. 1). Sampling height at all sites vary between 8-10 m. Sampling sites were selected away from city to avoid any direct in uence of pollution sources. PM 2.5 and PM 2.5-10 particles were collected on quartz ber lters (QFFs) of 20×25 cm size and perforated QFFs of 13 ×18 cm size placed on impactor plates, respectively. Both types of QFFs were pre-combusted and pre-conditioned 18 . Each sample was collected for 24 h duration. A total of 32 samples of PM 2.5 and PM 2.5-10 each (11 at BKR, 9 at JHJ and 12 at DEL) were collected with a frequency of 2 or 3 samples per week during summer season. Some samples were rejected too in the eld itself due to electricity problems, breakdown of instrument and damage of lters under strong windy conditions. Field blanks and laboratory blanks were also collected and handled in a manner similar to that of actual sample 18,19 .
An aliquot of the sample lter was extracted in Milli Q water using ultra sonicator for major cations (Ca 2+ , Mg 2+ , Na + , K + and NH 4 + ) and anions (F − , Cl − , NO 3 − and SO 4 2− ) and analyzed using Metrohm Ion Chromatograph (IC) model 882 Compact IC plus1 pro1 equipped with conductivity channel installed at JNU, New Delhi. Field blank and laboratory blanks were also analyzed simultaneously with particle samples. More details on extraction and ion analysis can be accessed from our previous work 27,14,18 . Residue lter paper after extractions of WSII was digested following "B " solution method 28  Particle generation (in source region; the Thar desert) and particle transport processes by high intensity S-SW winds were size selective in nature. Winds preferentially picked ner particles as evident from high PM 2.5 concentrations at all sites. An alternative explanation could also t in here that coarse particles were also lifted but wind intensity could not sustain them in ambient atmosphere and they settled down under gravity effect and therefore, could not be captured in real time sampling. Relatively high PM 2.5-10 concentration (161 µgm -3 ) at DEL could be due to local contributions from resuspension activities 34 . Higher PM 2.5 /PM 2.5-10 mass ratio at JHJ can be a location manifestation and attributed to relatively higher altitude compared to BKR and DEL on either side. Air quality remained worst hit during the sampling period as PM 2.5 and PM 2.5-10 load exceeded the national ambient air quality standard (NAAQS) of India for PM 2.5 (60µg m -3 ). PM 2.5 load has intermittently exceeded the NAAQS by a factor of 10 and 31 at BKR and DEL, respectively. DEL could act as sink for particles transported from upwind region as the wind speed substantially decrease due to increased vegetation and elevation form sea level, Aravali mountains and the sub-humid climate. Transport of dust from desert region in west India and far west to New Delhi and Indo-Gangetic Plains (east to New Delhi) has also been observed through satellite data 35,16,36 . In addition to natural factors, New Delhi had own local sources such as vehicular and thermal power plant emissions.

Water-Soluble Inorganic Ions (WSIIs)
Average data on WSIIs in PM 2.5 and PM 2.5-10 collected from BKR, JHJ and DEL are provided in Table 1. Sea salt contributions to individual ions and ∑WSII in ne and coarse particles at all sites were very limited (Table 1) Fig. 4).Average ratio of ΣWSII in PM 2.5 and PM 2.5-10 (PM 2.5 /PM 2.5-10 ) were around three or more at all sites but individual ion ratios were even higher and showed sampling site-speci c variations (see Supplementary Table S4 and Fig. S1 online). Spatially, WSII in both size particles were present in nearly similar concentrations in at BKR and JHJ sites, suggesting homogenizing effects of winds over upwind sites. The SO 4 2-, NO 3 and Clions in PM 2.5 and PM 2.5-10 at DEL were substantially higher than those observed at BKR and JHJ sites. Other aspect was higher concentrations of Ca 2+ , K + , Na + and Mg 2+ in both size fractions over DEL compared to other two sites except that Mg 2+ was highest in PM 2.5 over JHJ and Na + was highest in PM 2.5-10 over BKR.
Average percentage contribution of each soluble ion to total ∑WSII in PM 2.5 and PM 2.5-10 are given in Supplementary Further, inter-ionic correlations were drawn to observe ionic associations in particles (  54,14 . Presence of VOCs, photochemical activity (solar radiation) and oxidizing agents such as O 3 , OH radical, NO X also help in formation of secondary organic particle (SOA) 55,56 . However, the observed SOC over BKR and JHJ could be a result of mineral particle mediated photo-oxidation of POC 14,18 .

Total Carbonaceous Matter (TCM)
TCM represents total sum of organic matter (OM) and elemental matter (EM). OC was multiplied with a factor of 1.6 for urban areas (DEL) and with 2.1 for rural areas (BKR and JHJ) to get OM while EM was calculated from EC using a factor of 1.1 for urban as well as rural areas 52 . Highest contribution of TCM was observed over DEL (13.3%) followed by BKR (9.4%) and JHJ (4.6%), respectively. The TCM was dominated by organic matter compared to elemental matter over all sites.

Carbon Fractions and Source Identi cation
Total eight carbon fractions (OC1, OC2, OC3, OC4, OP, EC1, EC2 and EC3) were analyzed and each fraction can be used to distinguish source(s) 57,58 . Concentration of each fraction and their percentage contribution in TC showed different trends over each sampling site (see Supplementary Table S7 and Fig. S3 online) and the percentage contribution of each fraction to TC is shown graphically in the Supplementary Fig. 3 online. The OC2, OC3 and OC4 contributed above 10% to TC and their concentrations increased from BKR to DEL while their percentage contribution to TC deceased from BKR to DEL. This was so because of increased EC compared to OC at DEL. OC1 was present in lowest amounts at all sites. Both concentrations and contributions of OP to TC increased along windward direction and were highest over DEL, possibly due to high contributions of water-soluble polar components in PM 2.5 over DEL 58,59,14 . The EC1 (char-EC) was more than soot-EC and was highest over DEL followed by BKR and JHJ due to smoldering combustion, biomass burning, coal combustion, vehicular emission 60,30 .
Principal component analysis (PCA) has been applied for source identi cation of eight carbon fractions in PM 2.5 51 .
Principle components having eigen value of greater than one were only considered(   Supplementary Table S8 online and graphically shown in (Fig. 6). Among two sizes, major elements were enriched in PM 2.5 compared to PM 2.5-10 at all sites, a observation similar to PM 2.5 and PM 2.5-10 concentrations. Al, Ti and Mn showed very similar concentrations at all sites in PM 2.5 and PM 2.5-10 except that Al showed site speci c variations in PM 2.5-10 only (Fig. 6). Fe and Ca showed variations in both ne and coarse particles with respect to sampling sites and were highest over DEL in PM 2.5 . In PM 2.5-10 , Al, Ca and Fe were highest over BKR and lowest over JHJ in both size of particles (Fig. 6). Higher concentrations in PM 2.5-10 at BKR were due to low silica dilution effect on concentrations of major elements. Quartz grains hosting silica being larger in size and heavier in mass were subjected to winnowing away effect and therefore, were not picked by winds from the source region at BKR and JHJ sites 25 . Re-suspension of sediments lying on famous Delhi quartzite could have added silica rich particles to PM 2.5-10 and caused silica dilution effect on major elements at DEL site 61,25 .
Similar to major elements, trace elements were also higher in PM 2.5 compared to PM 2.5-10 overall sampling sites (Fig. 6).
Among the sites, all trace elements were present in highest concentrations over in DEL in both size particles. Cd and Co were present in lowest amounts in both size fractions and showed similar concentrations at all sites. Concentrations of Cr, Cu and Pb were very similar over BKR and JHJ sites and were lower than that observed at DEL in both size particles. Ba and Sr in PM 2.5 and Ba, Cr, Sr and Zn in PM 2.5-10 showed sampling site speci c variations, although their concentrations were high in DEL samples (Fig. 6). Cr nds mixed origin in particle in this region 25 and could have been added by minerals in upwind region and by re-suspension of anthropogenically contaminated roadside dust at DEL. Ba and Zn are contributed by vehicular emissions, oil burning and wear and tear of tyres and vehicular parts 62 .
To better understand spatial variations in elemental composition of particles along dominant wind path, major and trace element concentrations in PM 2.5 and PM 2.5-10 of different sites are plotted in equiline plot (Fig. 7). Major element over BKR and JHJ in PM 2.5 are plotting on equiline while in PM 2.5-10 particles element enriched towards BKR due poor silica dilution effect over BKR as discussed before. On moving towards downwind (DEL), Ca and Fe get enrich on account of re-suspension of road dust and industrial emission (Fig. 7) 23 . Trace element distribution between BKR and JHJ are almost similar in PM 2.5 and PM 2.5-10 while with DEL enrichment in Cu, Cr, Pb Ba observe in both size particles suggesting in uence of vehicular, industrial emission and re-suspension of road dust 23,24 . Uniform composition over BKR and JHJ indicate common crustal sources and homogenization effect of winds on particle composition. This was also revealed through calculations of Coe cient of Divergence (CD), a dimensionless qualitative measure of homogeneity/heterogeneity in particle chemistry of different sampling sites (see Supplementary Table S9 online) 63  Conclusions PM 2.5 and PM 2.5-10 mass concentrations and spatial variations suggested that 1) prevailing S-SW winds are responsible for high particle concentrations and air quality remained worst hit, particularly for PM 2.5 and 2) wind action in the Thar desert (source) region contribute more PM 2.5 compared to PM 2.5-10 particles. Particles in upwind regions were largely crustally derived particles and had low WSIIs (nearly 5%) compared to nearly 13% over DEL due to more anthropogenic sources. Vehicular emissions, thermal power plants, plastic burning and brick kiln industry are dominantly responsible source for high SSIs over DEL. Burning of wood and cow dung at domestic levels were major reasons for high OC/EC ratio at BKR whereas EC contributed by vehicular emissions and other combustion processes have resulted in lower of OC/EC ratio at DEL. Mineral dust mediated photo-oxidation of POC emitted by biomass burning could be possible source of SOC at BKR and JHJ whereas high SOC over DEL could be linked to photochemical reactions among NO x and VOC, emitted from direct sources. Silica dilution effects resulted in lowering of major element concentrations compared to UCC. Trace elements found their crustal origin in upwind region while re-suspension of surface/road side dust and direct emission from vehicle and wear and tear of vehicle parts caused high ER of trace elements over DEL. Variations in watersoluble inorganic ions and elements were speci c to particle size, and sampling site and was caused by multiple sources and subsequent process in ambient atmosphere. Upwind sites (BKR and JHJ) experience homogenization due to mixing by winds, whereas heterogeneity was noticed between DEL-JHJ, in downwind region and among DEL-BKR sites.