Introduction

Particulate matter (PM) is one of the most concerning threats among the air pollutants, as it leads to majority of premature deaths globally1. India is one of the worst affected countries in terms of PM exposure. Alarmingly, on 96% of the days in a year, India with her 1.42 billion people is exposed to PM2.5 levels higher than the WHO prescribed average 24 h limit of 15 µg/m3 2. In terms of fine particulate pollution, India experienced a staggering death toll of over 9.72 million in 2019, placing it in second position after China3. Several anthropogenic sources such as coal combustion, biomass burning, vehicular fuel combustion, industrial processes etc. contribute significantly towards this elevated level of PM emission in India (Fig. 1a). Comparative estimates of PM emission from these sources reveal that open burning of biomass has been the largest emitter of PM10 and PM2.5 for the past one and a half decade4. Coal combustion emission is steadily decreasing due to implementation of stricter emission standards that was put forth in 20155. In contrast, PM emissions from industrial processes is projected to surpass that of coal combustion in the present decade. Apart from these sources a large collective contribution is present from other miscellaneous sources (Fig. 1a).

Fig. 1: Comparative estimates of PM emission and emission factors of various anthropogenic processes.
Fig. 1: Comparative estimates of PM emission and emission factors of various anthropogenic processes.The alternative text for this image may have been generated using AI.
Full size image

a Stacked bars representing emissions of PM10 and PM2.5 (shaded bars) from different anthropogenic sources as compiled form “GAINS South Asia online,” 2020. For the past one and a half decades, open biomass burning has been the largest emitter of PM10 and PM2.5, while coal combustion emissions have steadily decreased due to the stricter emission standards implemented in 2015; b Particulate matter (PM) emission factors of open burning coal combustion, non-ferrous smelters and vehicular fuel. The emission factors for open burning are orders of magnitude higher than other processes. The emission factors have been compiled from Akbari et al.12; Raparthi and Phuleria13; Saud et al.14; Zhang et al.10; Zhu et al.15.

India, ranks second in coal consumption, and is home to 295 coal-fired power plants6,7,8,9. Coal fired plants and other non-ferrous industries have emission control systems such as cyclone separators, bag filters, electrostatic precipitators (ESP) to reduce the PM emission and metals associated with it10. On the other hand, the practice of open burning of biomass and solid wastes (SW) at open dumpsites, landfills or even individual residences, leads to uncontrolled release of PM. India ranks among the top countries globally in both waste production and open burning of wastes11. The mass of pollutants emitted from open burning per mass of waste (i.e., the emission factor) is an order of magnitude higher than other controlled processes10,12,13,14,15 (Fig. 1b).

These emitted PMs are associated with toxic elements like Arsenic (As), Cadmium (Cd), Chromium (Cr), Cobalt (Co), Mercury (Hg), Nickel (Ni), Lead (Pb) etc that can pose significant environmental and health risks. The elements released from coal that are of highest concern, due to their high concentrations, include As, boron (B), Cd, Hg, molybdenum (Mo), Pb, and selenium (Se). Elements of moderate concern include Cr, copper (Cu), Ni, vanadium (V), and zinc (Zn)16 Non-ferrous industries and other small-scale industries emit elements such as Cr, Mn, Fe, Cu, Zn, Ni, Cd and Pb17. On the other hand, SW incineration ash consists of varying amount of trace metals of which the major concerns are Al, Cr, Cu, Zn, Cd, Pb and Hg18. These elements are often enriched in bottom and fly ashes and pose environmental risks19. Fly ashes from SW incineration primarily consists of particles smaller than 5.5 µm, posing risk of the associated toxic metals being transferred to the human bloodstream through inhalation, if burnt openly20. The concentrations of these elements in SW ash vary across different countries21,22,23,24. However, in general the trend shows very high concentrations of Mn, Cu, Zn and Pb followed by moderate concentrations of Cr, Ni and Cd. A detailed country wise comparative analysis of the elemental concentration of toxic metals in SW incineration fly ash21,22,23,24,25 is presented in Supplementary Table 1.

These heavy metals are mostly bound to fine PM (PM < 2.5 µm) and can be transported over long distances across countries by atmospheric circulation26. Evidences of pollutants from Europe and North America reaching the Arctic have been reported in the past century27 When leaded gasoline was in use, Pb compounds emitted with vehicular exhaust were transported to remote regions, including Antarctica, Greenland, Alaska, and the open ocean. This long-range transport of Pb was traced with the use of stable Pb isotopes.28. Fine particles containing Pb, Hg, and Cd persist in the atmosphere for extended periods, allowing them to be deposited in areas far from their sources. Particularly, in case of Pb, the atmospheric residence time is typically 5–10 days after its emission until it gets deposited29.

To accurately quantify these emissions, one approach is to utilize heavy metal concentrations and isotopes. These isotopes, such as Pb, can serve as tracers of PM emissions, allowing for the identification and quantification of specific sources30. As there is no isotopic fractionation of Pb during physicochemical processes the isotopic signature of the source is preserved31. Metal concentrations are also crucial factors that supports the implications derived from isotopic compositions32,33. In the past decade PM bound Pb was found to be significantly enriched in Indian aerosols. This Pb is primarily sourced from coal combustion, metal ores utilized in industrial purposes and open burning of biomass and wastes30,32,33,34,35. The concentration of Pb in the PM emitting sources such as ore, Indian coal and biomass typically ranges from 1.6 to 5.5%, 20 to 144 ppm, and 0.2 to 28 ppm, respectively32,36,37,38. Thus, although emission factors for coal combustion, ore smelting, and vehicular fuel combustion are less than that of open burning, the former sources have orders of magnitude higher concentrations.

The major sources of PM in the Indian atmosphere have been a subject of investigation and study. As coal combustion emission is steadily decreasing, it is crucial to investigate whether the emission norms are effective in controlling emission of heavy metals from coal combustion. Being the largest power-supplying plant using lignite, Neyveli can emit significant amount of heavy metals. This makes it an ideal site for assessing how well current emission norms are controlling the release of these heavy metals to the atmosphere. A detailed country wise comparative analysis of the elemental concentration of toxic metals in lignite39,40,41,42,43,44,45,46,47,48,49,50,51 is presented in Supplementary Table 1. Thus, this study aims to investigate the primary sources of PM2.5 in an area housing the major lignite-fired power plant of South India using Pb isotopic compositions and elemental composition. The Indian atmosphere is additionally influenced by emissions from open burning and other large and small-scale industries. The ultimate objective is to quantify the contribution of each source towards atmospheric PM2.5 levels by utilizing the Pb isotope and elemental concentration data of PM2.5 so that better regulatory actions can be taken to improve air quality. In a densely populated developing country like India, where multiple sources of emission coexist, source apportionment becomes crucial for targeted mitigation strategies and implementation of stricter emission controls in specific emitting sectors.

Results

Elemental concentration and PM2.5 load

The concentration of the 13 (Mg, Al, Ca, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Cd and Pb) elements analyzed in the PM2.5 aerosols varied across months (Fig. 2a). Across all months, the elements can be divided into three distinct groups, following the general order: (Al, Ca, Fe, Mg, Zn) > (Cr, Cu, Mn, Ni, Pb) > (Cd, Co, V). All the months witnessed orders of magnitude higher mean concentrations of Al (951–3525 ng/m3), Ca (66–345 ng/m3), Fe (94–512 ng/m3), Mg (38–228 ng/m3) and Zn (33–481 ng/m3) compared to Cd (0.08–2.53 ng/m3), Co (0.05–0.25 ng/m3), Cr (0.90–13.5 ng/m3), Cu (2.02–44 ng/m3), Mn (1.96–68 ng/m3), Ni (0.91–13.5 ng/m3), Pb (2.32–79.89 ng/m3) and V (0.41–2.25 ng/m3).

Fig. 2: Temporal variations in PM2.5 load, elemental concentrations and enrichment factors.
Fig. 2: Temporal variations in PM2.5 load, elemental concentrations and enrichment factors.The alternative text for this image may have been generated using AI.
Full size image

a Temporal variation in concentrations of PM2.5 (top panel) and elements. The PM2.5 loads increased in the months of April–June compared to October–December. Concentrations of all the elements witnessed a sharp decrease in the month of May. b The top panel depicts the mean enrichment factors (EF) of each element in PM2.5 aerosols across the sampling period with significant enrichment in Cu, Zn, Cd and Pb (EFs > 10). The lower panels illustrate the temporal variation in EFs. All the enriched elements exhibited lowest EFs in the months of April–June, indicating a reduced anthropogenic contribution during these months.

Concentrations of all the elements witnessed a sharp decrease in the month of May (Fig. 2a). Anthropogenically sourced elements such as Cd, Pb and Zn demonstrated a decreasing trend through the months of April–June 2021, followed by an increase in the last three months of the year. Simultaneously, there was a sharp increase in PM2.5 load in the months of April–June compared to October–December of 2020, 2021 and February 2022 (Fig. 2a). Aluminum recorded highest concentration in April.

In the end member samples such as soil, open waste burning ashes, coal and fly ash the general order of concentrations followed were (Al, Ca, Mg, Fe) > (Cd, Co, Cr, Cu, Mn, Ni, V, Pb, Zn). The individual concentrations of the components are presented in Supplementary Table 2.

Enrichment factors and airmass back trajectories

Enrichment factors reveal that PM2.5 aerosols were enriched in Cu, Zn, Cd and Pb (mean EF > 10) (Fig. 2b), indicating anthropogenic sources. Zinc, Cd and Pb were observed to have mean EFs >10 in all the months. Copper exhibited mean EFs <10 in the months of April–June 2021, December 2021 and February 2022. In fact, all the enriched metals along with Mn and Co exhibited lowest EFs in the months of April–June.

The 72 h airmass back trajectories for an entire year reveal that the primary origin of the airmasses are oceanic (Supplementary Fig. 1). From October to February, part of the airmasses originate from the land while the rest originate from the ocean. From April to June all airmasses originate from the ocean (Fig. 3).

Fig. 3: Influence of airmass back trajectories, local fire events, and rainfall over Pb enrichments in PM2.5 aerosols across the sampling period.
Fig. 3: Influence of airmass back trajectories, local fire events, and rainfall over Pb enrichments in PM2.5 aerosols across the sampling period.The alternative text for this image may have been generated using AI.
Full size image

The 72 h airmass back trajectories for the sampling days coupled with their corresponding EFs of Pb for (a) October 2020 (b) November 2020 (c) December 2020 (d) April 2021 (e) May 2021 (f) June 2021 (g) October 2021 (h) November 2021 (i) December 2021 (j) February 2022. The trajectories have been computed using geographic information system (GIS) based software TrajStat (version: 1.5.3). The airmasses with EFs < 32 are marked in blue while higher EFs are marked in red. k The monthly average rainfall in India (green bars) along with observed local fire events (yellow circles) around NLCIL and EFs of Pb (red diamonds) in PM2.5 aerosols sampled at NLCIL. The red dotted line represents EF = 10. l Spearman Correlation between the number of local fire events versus elemental EFs in PM2.5 aerosols. The data for the fire events has been acquired from MODIS Terra and Aqua satellites available at NASA Fire Information for Resource Management System (FIRMS).

Pb isotopic compositions as source tracers

The mean Pb isotopic composition of the PM2.5 samples and end members are presented in Table 1. The individual composition of each sample is presented in Supplementary Table 3. The lignite and fly ash of the region are significantly different from coal samples found in other parts of India32. The Neyveli lignite is characterized by low 206Pb/207Pb ratios and higher 208Pb/206Pb ratios compared to other coals, indicating high Thorium contents.

Table 1 Mean Pb isotopic composition of NLCIL coal & fly ash, local solid waste & biomass burning ashes, NLCIL topsoil, ore and PM2.5 aerosols sampled in the years 2020, 2021 and 2022
Full size table

Quantification of end member contributions

The MixSIAR model results reveal a seasonal change in contribution percentages of the sources. The mean Pb isotopic composition of the considered sources along with their standard deviations and concentrations are presented in Supplementary Table 4.

Discussion

The higher mean concentrations of Al, Ca, Fe and Mg in the months of April–May can be due to their higher crustal abundances. Previous studies have also reported high concentrations of these elements in both coarse and fine aerosols in different parts of India and attributed them to crustal sources52,53,54,55. The range of elemental concentrations observed in this study is similar to those reported in PM2.5 aerosols from other regions of India53,56,57,58,59,60 (Supplementary Fig. 2). However, when comparing median elemental concentrations specifically in PM2.5 aerosols, Neyveli records lower median concentrations for all elements except Al and Zn. The elevated median concentration of Al, with simultaneous lower median concentrations of other elements, in the months of April–May indicates a higher contribution of crustal dust to the aerosol composition. Further insights into whether the elements have a crustal or anthropogenic origin can be inferred from the EFs, discussed later. Zinc, despite having significantly less crustal abundance compared to Ca, Fe, and Mg61, demonstrates similar concentrations to these elements. This suggests additional anthropogenic inputs of Zn other than crust. The concentration of Pb in PM2.5 observed in this study is lower than values reported in aerosols (TSP, PM10 and PM2.5) from urban and industrial regions of India in previous investigations32,33,53,55,58,59,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96. Previous studies have reported Pb concentrations exceeding the national ambient air quality standard (NAAQS) annual average of 500 ng/m³ in industrial and major Indian cities even after the complete phase-out of leaded gasoline in the year 2000 (Supplementary Fig. 3). In contrast, the maximum Pb concentration recorded in this study is 79.9 ng/m³, suggesting a lower level of anthropogenic Pb pollution in the studied region compared to other parts of India.

This decrease in concentration of Cr, Mn, Zn, Cd and Pb with increasing PM2.5 load and simultaneous increase in Al concentration in the months of April–May indicate a greater influence of geogenic sources in the months of higher PM2.5 loads. As the amount of coal combustion at the thermal power plant remained constant97, it is most likely, that sources other than coal influenced the atmospheric metal concentrations of the region. To gain a better insight about the sources (geogenic or anthropogenic), of the metals the crustal EFs have been determined.

Lowest EFs in the months of April–June indicate that the atmospheric aerosols experienced a greater geogenic influence in these months. Moderate to strongly negative Spearman correlations were observed between PM2.5 load versus the concentration of the enriched elements Cd, Pb and Zn (R = –0.56 to –0.72 and p = 0.02–0.09) and their EFs (R = –0.64 to –0.72 and p = 0.01–0.05) (Supplementary Fig. 4). As the operation of the lignite-fired power plant was consistent throughout the sampling period97, this variation in EFs must be due to other local and long-range sources.

By coupling back trajectories with EFPb, it can be observed that the airmasses, despite having oceanic origins in the months of October–February demonstrate very high EFPb (mean = 101 ± 38). While, similar oceanic origin airmasses in the months of April–June, demonstrate EFPb ≤ 28 (mean = 17 ± 7) in PM2.5 (Fig. 3). In India, the months of October to December mark the post-monsoon season, while April to June corresponds to the summer season. During the post-monsoon period, temperatures begin to fall, especially during the night across India, and wind patterns shift, signaling the onset of colder weather98. The changing wind patterns lead to wind blowing from the land to the sea, transporting pollutants from the land surface to over the ocean. The northern Bay of Bengal (BoB) remains under the influence of airmasses from Indo Gangetic plain while the southeastern BoB receives air masses from southeast Asia99,100. Similarly, the southern Arabian Sea remains under the influence of western and southern India101. Thus, the high EFPb in airmasses originating from the ocean during post-monsoon period can be attributed to the transport of pollutants from land to sea due to prevailing wind patterns. Further, the mean temperature is lower and as a result, temperature inversion traps cooler air beneath a layer of warmer air. This cooler air confines within it the ground-level pollution and thus, increases the pollution levels in the air masses just above the ground surface102. Long-term (2010–2018) LIDAR observations and radiosonde data from Gadanki (a rural site ~280 km north of Neyveli), indicates that thermal inversion can fully control the vertical extent of aerosols103. Further research across various locations in India confirmed that thermal inversion, rather than the atmospheric boundary layer, predominantly restricts aerosol vertical distribution. In contrast to the post-monsoon months, the lower EFPb during April to June can be due to a reduced land-based pollution influence over the ocean. In fact, during the summer months there is a decrease in the EF of all the heavy metals compared to the post-monsoon period. (Fig. 2b) There was also an increase in local fire events in the months of April and May compared to October to December (Fig. 3a–k), that could have resuspended a higher amount of desiccated crustal dust resulting in an increased PM2.5 load and simultaneous decreased EF104,105. A Spearman correlation between the number of fire events versus EFs reveals strong negative correlation with anthropogenic elements and positive correlation with PM2.5 load. (Fig. 3l). Thus, the post-monsoon months may have been influenced by both long range as well as local sources, while local sources were predominant in the summer months. A better understanding of the sources can be obtained from elemental ratios.

Elemental ratios are useful tracers for understanding pollution sources. Ratio between V and Pb (V/Pb) are tracers for coal combustion, vehicular and industrial emissions. Coal combustion typically has V/Pb ratios of >1, in contrast to vehicle exhaust emissions, where V/Pb < 1, and industrial emissions, where V/Pb ≪ 130,32,33,35. None of the aerosol samples exhibited V/Pb > 1. Instead, all the V/Pb ratios for the aerosols varied in the region of industrial and traffic emission source signatures. A bivariate plot between V/Pb and enriched elements such as Cu/Pb can provide insights about the various sources of metals in PM2.5. The plot (Fig. 4a), reveals that the aerosols do not align with the ratios of the coal and fly ash, despite the aerosols being sampled from the vicinity of a coal-fired power plant. In Cu/Pb vs V/Pb space, the crust-coal-fly ash and base metal ore (galena) depict two extreme end members with the signature of open SW & biomass burning lying in between (Fig. 4a). The PM2.5 samples fall in between these three end members suggesting a mixture of these three sources. However, a seasonal difference in these ratios can be observed. In the months of October–February, majority of the aerosols have V/Pb ≪ 1. In contrast, from April to June the aerosols shift towards SW & biomass burning with V/Pb ratios typical of traffic emissions. Moreover, previously sampled aerosols from Delhi’s waste dumping site Okhla also overlaps with April–June aerosols from Neyveli, indicating a strong presence of metals generated from SW & biomass burning34.

Fig. 4: Elemental ratios in PM2.5 aerosols depicting influence of different sources.
Fig. 4: Elemental ratios in PM2.5 aerosols depicting influence of different sources.The alternative text for this image may have been generated using AI.
Full size image

Elemental ratios plotted as (a) V/Pb vs Cu/Pb (b) Pb/Al vs Pb/Fe indicate that the PM2.5 aerosols are a mixture of crust, coal & fly ash, solid waste & biomass burning and industrial smelting (depicted as ore in the figure). The characteristic V/Pb ratios for industrial, traffic and coal/crustal sources (Das et al.32) are marked as shaded regions for reference. The October-February aerosols have V/Pb ≪ 1 and hence plot in the region of industrial emissions. The open waste burning aerosols from Delhi’s Okhla landfill site (Kumar et al.34) overlapping with April–June aerosols is plotted in V/Pb vs Cu/Pb space for reference. The April-June aerosols seem to be more influenced by solid waste & biomass burning and crustal sources with decreased Pb/Fe and Pb/Al ratios.

During the months of April–June a lower Pb/Al as well as Pb/Fe ratios were observed compared to the post-monsoon months. This suggests a greater crustal contribution during the summer months, likely due to increased pollutant dispersion resulting from absence of temperature inversion, as well as a stronger influence from local biomass burning events. Local evidences of sugarcane residue burning was observed on site. In Tamil Nadu, the primary crops are rice and sugarcane, with sugarcane being the dominant crop, produced at nearly three times the quantity of rice106. Of the total crop yield, ~13% is burnt as residue. Notably, certain areas in Tamil Nadu are also recognized as hotspots for PM emissions resulting from the burning of crop residues generated from rice, wheat, maize, and sugarcane cultivation106. The concentration of Pb in biomasses is typically much less than other industrial processes, that may have resulted into a lower Pb/Al and Pb/Fe ratios from April–June compared to the post-monsoon months. A bivariate plot between Pb/Al and Pb/Fe (Fig. 4b) provides similar inferences as the Cu/Pb vs V/Pb plot, with all the aerosols falling between crust, coal & fly ash, SW & biomass burning and ore. The summer aerosols lie closer to the open waste burning signatures and crust compared to post-monsoon aerosols. This suggests that there can be a greater influence of crust in the summer aerosols compared to the post-monsoon ones. Moreover, open burning of biomass and SW, together with emitting signatures from vegetation and SW also dehydrates the soil, increases local atmospheric turbulence due to fire-related convective currents, and subsequently mobilises the dehydrated soil particles in the air105. Thus, the elemental compositions during fire events may include signatures from vegetation and SW combustion, as well as contributions from crustal dust due to soil resuspension.

To identify the sources, aerosol samples and potential end members are plotted in a three-isotope space using the ratios of 208Pb/206Pb vs 206Pb/207Pb (Fig. 5). Thus, in the three-isotope space two distinct mixing lines can be formed, each with the natural background crust at one end. One of the mixing lines extends to Neyveli lignite (blue dotted line), while the other extends to Indian ore (black dotted lines) at their respective ends (Fig. 5). The Pb isotopic compositions of the PMs lie along the mixing line between Indian ore and crust37,107,108, instead of the local lignite and fly ashes (Fig. 5). In fact, the PM2.5 samples demonstrate significantly dissimilar compositions from coal and fly ash, indicating minimal influence from them. The Pb isotopic composition of the aerosols lie in the vicinity of vehicular exhaust emissions and seawater30,109. Local crust being significantly less radiogenic also lies close to the aerosols compared to upper continental crust (UCC)110,111. Notably, there is a considerable overlap of the aerosols with that of Indian fuelwood108 and local SW & biomass burning signatures. Isotopic signatures of previously sampled aerosols from Delhi’s Okhla landfill, where burning is prevalent also overlap with the Neyveli aerosols34. In the post-monsoon months as some of the airmasses originate from the coasts of southeast Asia, influence of transboundary aerosols can be present (Fig. 3h–j). In fact, the Neyveli PM2.5, aerosols have substantial overlap with aerosols from Thailand, indicating presence of transboundary sources in the post-monsoon months (Fig. 5). The mean Pb isotopic composition of the summer aerosols is about 6‰ more radiogenic in terms of 206Pb/207Pb and 208Pb/206Pb ratios than the post-monsoon aerosols indicating a seasonal shift in sources (Table 1). A higher radiogenic signature can be indicative of higher natural contributions (crust and seawater). This increase in crustal component can be due to airmasses originating from ocean (Fig. 3) and soil resuspension due to burning. This increased resuspension during burning, also may mobilise the Pb deposited in the past century from leaded gasoline use. Lead emitted from leaded gasoline in the 20th century was eventually deposited onto soils, where it accumulated over time. Due to its long residence time of ~200 years in soil112, Pb remains bound to soil particles. However, during fire events, the intense heat and turbulence cause the desiccated soil to become airborne, resuspending the previously deposited Pb back into the atmosphere. Recent studies have found evidences of such resuspended Pb contributing towards atmospheric Pb load113,114 Consequently, leaded gasoline also comes into play as a potential source. There are evidences from across the globe, demonstrating that leaded gasoline is still recirculating in the atmosphere30,114,115. Previous study in the Indo-Gangetic plain finds evidence of high contribution of legacy lead in biomass samples108. Thus, the aerosols are primarily a mixture of open waste burning, industrial smelting, traffic emissions, natural sources (crust and seawater), and leaded gasoline, with their respective contribution varying in summer and post-monsoon period. The regional soil appears to have been contaminated by the coal combustion process.

Fig. 5: The Pb isotopic compositions of the PM2.5 aerosols and relevant end members illustrated in three-isotope space (208Pb/206Pb vs 206Pb/207Pb).
Fig. 5: The Pb isotopic compositions of the PM2.5 aerosols and relevant end members illustrated in three-isotope space (208Pb/206Pb vs 206Pb/207Pb).The alternative text for this image may have been generated using AI.
Full size image

Two distinct mixing lines are observed in the tree-isotope plot, each with crust at one end: one extends to Neyveli lignite (blue dotted line), while the other extends to Indian ore (black dotted lines). The Pb isotopic compositions of the PMs lie along the mixing line between Indian ore and crust, rather than local lignite and fly ashes. The aerosols lie close to vehicular exhaust emissions (traffic) and seawater, with less radiogenic local crust also plotting close to the aerosols compared to UCC. There is a considerable overlap between the aerosols and Indian fuelwood, local solid waste & biomass burning signatures. Isotopic signatures of previously sampled aerosols from Delhi’s Okhla landfill, overlap with the Neyveli aerosols. The enclosure formed by the red dotted line signifies the mixing envelope formed using the relevant end members. The data for Indian fuelwood and OWB aerosols have been adopted from Mitra et al.108 and Kumar et al.34 respectively. Data for traffic emissions, Indian ore, crust, seawater, and local crust have been compiled from Ray and Das113, Deb et al.37, Lee et al.109 and Chakraborty et al.110 respectively.

Over the past decade, only five studies have measured the Pb isotopic compositions of Indian aerosols, primarily in PM10 samples from diverse settings, such as urban, rural, industrial, and waste-burning sites32,33,34,35,108. As Pb mostly concentrates in the finer fraction of the aerosols, Pb isotopic composition can be considered a reliable tracer of anthropogenic emissions. Thus, a comparison between Pb isotope ratios can provide insights into the distinct influences on aerosol isotopic composition in different environments. In 208Pb/206Pb space (Supplementary Fig. 5), aerosols linked to industrial and urban sources exhibit higher 208Pb/206Pb ratios compared to Neyveli PM2.5 aerosols32,33,34. Dust storm aerosols show significantly different median ratios due to the influence of dust from the Middle East (Mood’s median test; p < 0.0003)35. The study conducted in Delhi’s open-waste burning (OWB) site and urban area records lower median 208Pb/206Pb ratios (median = 2.1313) in the OWB aerosols compared to the urban ones (median = 2.1374)34. Thus, the median 208Pb/206Pb ratios of the OWB aerosols plot closer to the median 208Pb/206Pb ratios of Neyveli aerosols. The closest median 208Pb/206Pb ratios to the Neyveli PM2.5 aerosols (medianpost-monsoon = 2.1248; mediansummer = 2.1211) are observed in rural aerosols (median = 2.1222) and high-altitude aerosols (median = 2.1222). Rural aerosols can be under the influence of biomass burning as a source of fuelwood, while the high altitude aerosols are proven to have been influenced by wood charcoal burning108. In fact, the Pb isotopic compositions of the wood charcoal have significant overlap with the Neyveli aerosols in the triple isotope space (Fig. 5). During the post-monsoon season the Neyveli aerosols exhibit higher 208Pb/206Pb ratios compared to the summer samples indicating increased influence from the urban and industrial sources (Supplementary Fig. 5).

In the 3D mixing polygon simulation, the median probabilities of all the aerosols falling inside the mixing envelope formed by sources indicate whether all sources have been considered or some are missing116. In this simulation, polygons, that is, convex hulls (mixing envelops) are generated (1500 iterations) based on the distribution of the considered sources and their SDs. The mixing polygon simulation is based on the principle that the isotopic composition of the mixture (aerosols for this study) can be explained by the isotopic compositions of the sources forming the mixing envelope only if the mixture doesn’t fall outside 5% of the generated polygons. The sample(s) with less than 5% probabilities has/ve sources other than the considered end members and hence the selected sources cannot explain the composition of that particular mixture. From the elemental ratios and the mixing line in the triple isotope plot, it is evident that the relevant sources are: crust, vehicle exhaust, SW & biomass burning and industrial smelting. In the post-monsoon months, significant amounts of airmass originate from the coasts of southeast Asia (Fig. 3), thus, making it necessary to consider the transboundary aerosol as a source in the months of October–February. Incorporating the mean Pb isotopic composition of Thailand aerosols117,118 as a source significantly increases the median probability by 11% (ANOVA; p < 0.0001). In the summer months, the airmasses primarily originate from the ocean, making it a crucial component of the mixing envelope. Including seawater as a source increased the probability by 4.3%. Leaded gasoline, though phased out in 2000, can still contribute to PM2.5 aerosols through soil resuspension30,108,114,115. In the triple isotope space, leaded gasoline plots close to the less radiogenic PM2.5 aerosols. Including leaded gasoline as a source significantly increases the median probability (ANOVA; p < 0.0001), validating it as a necessary end member. Finally, as the local crust of southern India, is significantly different from UCC, it is also considered as a source110,111. Figure 6a illustrates all the changes in the median probabilities as each source is successively added. None of the aerosol samples demonstrated probability <5%. By incorporating all these sources to define the mixing envelope, the total median probability reaches 0.68. This indicates that all aerosols are a mixture of the accounted sources with varying contributions from respective sources. Despite differing in isotopic and elemental compositions, coal and fly ash must be considered as sources, as the aerosol samples are from a lignite-fired power plant. However, the inclusion of coal & fly ash as a source does not increase the probability at all (Fig. 6a), indicating limited influence of the source on PM2.5 aerosols. Thus, the mixing envelope (marked in red dotted line in Fig. 5), comprises of UCC, local crust, seawater, vehicular fuel, SW & biomass burning, Indian fuelwood, transboundary aerosols and Indian ore. The mean Pb isotopic compositions and Pb concentrations of all the sources fed to the simulation along with their references are listed in Supplementary Table 4.

Fig. 6: Relevance of the sources considered and the contribution of individual sources to atmospheric Pb along with field evidence of open burning.
Fig. 6: Relevance of the sources considered and the contribution of individual sources to atmospheric Pb along with field evidence of open burning.The alternative text for this image may have been generated using AI.
Full size image

a Bar plots depicting the median probability of the aerosol mixtures to fall inside the mixing envelope constructed by the initial and additional sources with each source added successively. The probabilities are determined with the help of 3 – dimensional mixing polygon simulation in 206Pb/207Pb, 208Pb/207Pb and 208Pb/206Pb space. Same letters above the bars indicate that the probabilities are not significantly different, as determined by Tukey’s method. Addition of transboundary aerosols and leaded gasoline significantly increased the median probability. Exclusion of Neyveli coal & fly ash does not affect the results. b Median contribution of individual sources towards atmospheric Pb determined from MixSIAR simulation. The percentages over/beside the bars represent the range of contributions from each source. MixSIAR results indicate negligible influence of coal and fly ash throughout the year in the vicinity of lignite fired powerplant. Open burning of solid waste and biomass emerges as the major anthropogenic contributor, even in a region of coal combustion. c Field photographs of solid waste & biomass burning at Neyveli.

The quantification of the contribution of these sources is performed using Bayesian statistical model MixSIAR in R, presented in Fig. 6b. Transboundary aerosols are only relevant in the post-monsoon months due to long range transport during the season and contributes up to 36%. However, consistent contribution from SW and biomass burning was observed across the seasons (up to 26% in post-monsoon season and up to 22% in summer), indicating their significant role as a pollution source throughout the year. In both post-monsoon and summer months there are significant contributions from background natural sources. (crust + seawater). This contribution from natural sources increases in the summer months, with maximum contribution reaching up to 49%, which is 27% in post-monsoon. As this natural contribution increases in the summer months, so does the contribution from leaded gasoline. This increase in contribution of natural component can be explained by airmass back trajectories and their associated consequences (Fig. 3). In summer, airmass back trajectories demonstrate no influence from transboundary aerosols, originating predominantly from short ranges and subsequently making the impacts of local burning such as dehydrated soil resuspension prominent. Thus, an increase in the relative crustal contribution and, consequently, the contribution of leaded gasoline from soil resuspension is observed in the summer months. Similar persistence of legacy Pb in the atmosphere has been recorded worldwide. Recent studies, utilizing Pb isotopic analyses, of aerosols in London observed long-term persistence of Pb from historically deposited leaded petrol in London’s atmosphere, despite its ban in the year 2000114,119. Even though Pb levels in the air have significantly dropped since the ban in London, about 32 ± 10 to 43 ± 9% of airborne Pb still originates from the resuspension of Pb-contaminated dust and soil, which was initially deposited in the 20th century114. Similar findings have been reported from South-East Asian countries, where Pb isotopic composition of aerosols prove substantial contributions (up to 39%) of legacy Pb113. In India, similar Pb isotopic analysis revealed that over 50% of the Pb in Himalayan aerosols still comes from leaded gasoline, even 15 years after its phase-out108.

In post-monsoon period, transboundary aerosols become a dominating factor, reducing the relative contribution of crustal sources and subsequently that of leaded gasoline. In both seasons a modest contribution form vehicular emissions (<10%) can be observed. Although, the aerosols lie on the mixing line joining crust and Indian ore, negligible contribution from ore is observed from MixSIAR analysis. Most importantly, MixSIAR results indicate negligible influence of coal and fly ash throughout the year in the vicinity of lignite-fired powerplant. This also confirms previous qualitative observations from elemental ratios and Pb isotopic compositions.

In the study area, the lignite storage yards, lignite transferring regions from mine to power plant and connecting roads with heavy vehicles are equipped with water sprinklers to control fugitive emissions120. The power plants have water sprinklers and dust extraction systems in lignite handling and conveying systems to minimize dust generation. Boilers are equipped with ESP in order to regulate stack emissions. The stacks are 220 m in height in 3 of the 5 power plants, while the new plant, “Neyveli New Thermal Power Plant (NNTPP),” that obtained environment clearance in 2010, has a stack of height 270 m for ensuring better pollutant dispersal. Additionally, lignite crusher houses and fly ash storages are also equipped with bag filters. Metal composition and Pb isotopes of PM2.5 suggests minimal influence of coal emissions as a result of effective regulatory measures in the coal combustion processes, leading to reduced heavy metal contamination in the atmosphere.

The SIPCOT industrial complex in the region covers a total area of 709.31 acres, housing 21 highly polluting red, 6 orange and 3 green category industries. Emissions from chemical and petrochemical facilities include pollutants such as PM, inorganic and volatile organic compounds, while pharmaceutical and textile dye industries additionally release various organic and inorganic compounds. There are several wood and coal-fired boilers associated with these industries emitting PM on regular basis. Wood-fired boilers contribute to emissions similar to those generated by biomass burning. The industries on average have a stack height of only 24.9 m and releases a total of around 179 kg/ day of PM121. Majority of these industries employ dust collectors, cyclone separators, and bag filters for controlling PM emissions. The Neyveli PM2.5 does not show influence of ore indicating industrial emission controls are effective.

Apart from thermal power plants and other industries, the region experiences emissions from open burning of crop residues, residential fuelwood burning and SW burning. The dumpsites face spontaneous or deliberate burning of municipal SW resulting in generation of organic as well as metal pollutants in the atmosphere122. Majority of the rural households in Tamil Nadu use biofuels such as fuelwood, wood-chips and agricultural waste for cooking purposes releasing respirable PMs123. In the district of Cuddalore alone, there are 905 villages that shows dependence on biomass burning as their source of energy124,125. As per 2018, Tamil Nadu generated 31.62 MT of crop residue out of which 4.11 MT was burnt106. Open burning of SW and biomass have been observed at Neyveli during the sampling period (Fig. 6c).

In India, the top five states with the highest power generation capacity from coal are Uttar Pradesh, Maharashtra, Chhattisgarh, Madhya Pradesh, and Gujarat, with the largest individual coal-fired power plants located in Madhya Pradesh, Gujarat, Maharashtra, Uttar Pradesh, and Odisha9. Among these regions, Gujarat lacks recent literature on the elemental composition of aerosols. In the absence of Pb isotopic data, the V/Pb ratio serves as a reliable alternative tracer for coal combustion. Given the high V content in coal, the V/Pb ratio from coal combustion is consistently greater than 1. In contrast, other anthropogenic sources, such as vehicular exhaust and industrial emissions, typically exhibit V/Pb < 1 and ≪1 respectively32,33,35,113. Thus, V/Pb ratios from the literature were compiled over the last decade to gain a national perspective on whether atmospheric Pb is sourced from coal combustion. V/Pb ratios from several key locations were compiled having coal-fired power plants nearby: Madhya Pradesh (Singrauli), Jharkhand (Jharia), Assam (Jorhat), West Bengal (Bandel and Kolaghat), Odisha (Bhubaneshwar), Andhra Pradesh (Vizag), Telangana (Hyderabad), Maharashtra (Trombay), Chhattisgarh (Raipur), Delhi, and Uttar Pradesh (Kanpur)32,33,34,55,92,95,126,127,128,129,130,131. Figure 7 illustrates the location of the aerosol sampling sites along with the nearby thermal power plants. Majority of the locations, except Jharia, Raipur, and Hyderabad, record V/Pb < 1128,129,131. Even in Jharia, Raipur, and Hyderabad, a significant proportion of aerosols demonstrate V/Pb < 1128,129,131. Notably, in Singrauli, home to India’s largest individual coal-fired power plant, all aerosols exhibit V/Pb ratios ≪1126. Similarly, aerosols from Odisha, Uttar Pradesh, West Bengal, Andhra Pradesh, Maharashtra, and Delhi consistently record V/Pb ratios <132,33,34,55,92,95,130. In 2008 at Jorhat V/Pb ratios in aerosols were recorded to be >1, which decreased to <1 in 2020, indicating a reduced coal combustion emission over the years in the region (Fig. 7). Despite India having 295 coal-fired power plants across the country9, atmospheric aerosols in most regions demonstrate V/Pb ratios <1, suggesting that coal combustion may not be the dominant source of atmospheric Pb in the country. Among these studies, the aerosols from Jorhat, Raipur, Trombay, and Vizag were found to have contributions from biomass burning despite the presence of coal-fired power plants nearby95,127,129,130. Moreover, the only Pb isotopic study conducted near a coal-fired power plant, in Bandel, West Bengal, also identifies high-temperature industrial processes, rather than coal combustion, as the primary source of atmospheric Pb in the region32. Another study conducted at an industrial site in Delhi, just 56 km away from a major thermal power plant (NTPC Dadri) also finds no significant contribution of coal combustion towards atmospheric Pb34. Previous studies utilizing Pb isotopic compositions and elemental concentrations have highlighted the contribution of biomass and SW burning at specific locations34,95,108,126,127,129,130. Our findings also align with these results, emphasizing the need for stricter controls on open burning of SW and biomass to reduce the emission of heavy metals into the atmosphere. The dominance of open burning as a pollution source concurs with the GAINS model and highlights the pressing necessity for better waste management strategies to reduce its environmental impact. Thus, India needs to redefine her pollutant priorities and act accordingly to mitigate their effect.

Fig. 7: Ratios of V/Pb at various locations illustrating the influence of coal combustion towards atmospheric aerosols across India.
Fig. 7: Ratios of V/Pb at various locations illustrating the influence of coal combustion towards atmospheric aerosols across India.The alternative text for this image may have been generated using AI.
Full size image

Map illustrating the aerosol sampling sites and nearby coal-fired thermal power plants. The available V/Pb ratios were compiled from key locations, including Madhya Pradesh (Singrauli)9, Jharkhand (Jharia)11, Assam (Jorhat)10, West Bengal (Bandel and Kolaghat)1,8, Odisha (Bhubaneshwar)12, Andhra Pradesh (Vizag)16, Telangana (Hyderabad)15, Maharashtra (Trombay)14, Chhattisgarh (Raipur)13, Delhi4, and Uttar Pradesh (Kanpur)2. The red dotted line indicates V/Pb > 1. Aerosols demonstrating V/Pb > 1 can be sourced from coal combustion while those having V/Pb < 1 and ≪ 1 can be sourced from vehicular exhaust and high temperature industrial processes1. The base map has been adopted from the open-source online repository Opendatasoft (https://public.opendatasoft.com/explore/dataset/world-administrative-boundaries/export/) and the locations have been illustrated using ArcGIS Pro 10.8. The V/Pb ratios were plotted using Origin Pro 2024a and all the components were compiled in Adobe Illustrator 2021, to form the final illustration.

Methods

Study area

The study was conducted in and around the largest power supplying lignite fired plant of Southern India located in the state of Tamil Nadu (Cuddalore district) and operated by Neyveli Lignite Corporation India Limited (NLCIL). Neyveli is located roughly 197 km south of Chennai and 52 km inland from the BoB. It covers an area of 835 km2. The Ponnaiyar River borders the town to the north, the Coleroon River to the south, and the BoB to the east. At Neyveli there are 3 operational open cast lignite mines (Mine I, Mine IA and Mine II) with a lignite reserve of about 1357 million tons, that belongs to the tertiary geologic era (~25 MYA). The combined yearly production capacity of the mines stands at 28.5 million tonnes. At present, NLCIL operates five pit-head thermal power generation stations with a total installed capacity of 3390 MW, all fueled by the mined lignite. The total area covered by the mines and power plant is around 206 km2 132. In addition to the lignite mines and power plants, the region also houses several small, medium and large-scale industries ranging from chemical, pharmaceutical, textile, fertilizers, automotive etc within the State Industries Promotion Corporation of Tamil Nadu (SIPCOT) industrial complex, established in 1984 (Supplementary Fig. 6).

In the study area, Tamil Nadu pollution control board report reveals exceedance of several air pollutants such as suspended particulate matter (SPM), PM10, Gaseous Hg, SOx and NOx from the NAAQS120.

Sample collection

Sampling of PM2.5 and the end members (lignite, fly ash, topsoil, crop ash and SW burning ash) took place in and around NLCIL between October 2020 and April 2022. The PM2.5 sampling was performed with Envirotech APM 550 MFC sampler at a constant flow rate of 1 m3/h. The samples were collected in pre-weighed 46.2 mm Whatman PTFE filters with a pore size of 2 μm. Before sampling, the filter papers were acid-washed in 1 N double-distilled HCl at 140 °C followed by multiple rinses with ultrapure water (resistivity>18.2 MΩ-cm). Finally, the filter papers were dried at 60 °C133. Following sampling, the filters were transferred to a desiccator and kept for 24 h, following which the final weights were recorded. Subsequently, they were carefully packed into resealable zip-lock bags for subsequent chemical analysis.

Concentration and Pb isotope determination

Sample processing was performed in clean chemistry laboratory using Milli-Q water (resistivity >18.2 MΩ-cm) and acids purified by sub-boiling distillation of concentrated reagent grade acids. The aerosol samples were subjected to hot plate digestion in screw cap Teflon beakers (SavillexTM) for 24 h at 120 °C in a mixture of concentrated 3:1 HNO3 and HF. About 20 mg of the top soil, SW burning ash, biomass burning ash, fly ash and coal samples were weighed and subjected to the same digestion procedure. Prior to acid digestion, the coals were ashed in a muffle furnace at 300 °C for 1 h and 500 °C for the next two hours134. The digested samples were evaporated to dryness. The dried samples were redissolved and refluxed twice in concentrated HNO3 to remove traces of HF. The solutions were dried down and redissolved in 2 mL of 2% HNO3, half of which was utilized for determination of elemental concentration and the other half was stored for performing anion exchange chromatography to separate Pb. For separation of Pb, the samples were dried down and redissolved in 1.1 N HBr and loaded in columns packed with anion exchange resin (Eichrom AG-1×8 chloride form, 200–400 mesh). Finally Pb was eluted in 6 N HCl135.

The Thermo Scientific iCAP Q ICP-MS at Earth Observatory of Singapore, Nanyang Technological University, Singapore, was utilized to analyse the elemental concentrations of the samples. The sample introduction setup included a Peltier-cooled quartz cyclonic spray chamber, a PFA concentric nebulizer (100 µL/min), and a demountable quartz torch with a quartz injector. Platinum sampler and skimmer cones were employed at the quadrupole–torch interface. All the samples and standards were diluted in 2% HNO3. The instrument was operated in a single collision cell mode with kinetic energy discrimination (KED), employing pure helium (He) as the collision cell gas to remove polyatomic interferences. The instrument was tuned in iCAP Q tuning solution comprising of 1 µg/L of Ba, Bi, Ce, Co, In, Li, and U. Multi element calibration standards ranging from 0.1 ppb to 500 ppb were freshly prepared gravimetrically and analyzed to obtain linear calibration curves. A multi-element standard of 10 ppb was used for standard–sample bracketing. Drift corrections and normalization of the obtained raw counts per second (cps) were performed with respect to the bracketing standard. The instrumental operating conditions are presented in Supplementary Table 5.

To determine the extraction efficiency and analytical precision for the measured elemental concentrations, NIST SRM-2783 (urban PM2.5) and NIST SRM-2704 (Buffalo River sediment). were digested and analyzed using the methods similar to the corresponding samples. Recoveries of the elements were found to be >90% (Supplementary Table 2).

Pb isotopes were analyzed by Multi-collector ICP–MS (Neptune Plus MC ICP-MS, Thermo-Fisher, Germany) at Earth Observatory of Singapore, Nanyang Technological University, Singapore, equipped with Aridus II desolvating nebulizer (100 µL/min). Instrumental mass fractionation correction involved Thalium (Tl) (NIST SRM 997) internal normalization (203Tl/205Tl ratio of 0.418911). Bracketing with NIST SRM-981 was adopted to normalize the obtained ratios with respect to the true value136,137.

All the samples were concentration matched with standard and anaysed at comparable Pb/Tl ratios, to minimize the mass bias generated from counting statistics. The precision and accuracy of the measurements were ensured by analysing NIST SRM-981 multiple times between samples as well as at the beginning and end of sessions. The contribution of the procedural blank (median blank = 0.07%) to the total measured concentration were determined as [Pb]procedural blank/[Pb]Total. Consequently, the blank subtracted 206Pb/207Pb and 208Pb/207Pb values were determined by mass balance represented by the following equation138:

$${\left(\frac{20x}{20y}\right)}_{{sample}}=\left[{n}_{T}{\left(\frac{20x}{20y}\right)}_{{total}}-\,{n}_{B}{\left(\frac{20x}{20y}\right)}_{{blank}}\right]/({n}_{T}-\,{n}_{B)}\,$$

Where, x = 6, 8; y = 7, 6; n is the fractional amount of substance and the subscripts T, and B refers to total and blank respectively.

Determination of enrichment factor and air mass back trajectories

To distinguish between crustal and anthropogenic origins, the crustal enrichment factor (EF) defined as EF = \(\left(\frac{{\left(\frac{{Element}}{{Reference\; element}}\right)}_{{Sample}}}{{\left(\frac{{Element}}{{Reference\; element}}\right)}_{{Crust}}}\right)\) for all elements were compared with Al as reference element. The factor \({\left(\frac{{Element}}{{Reference\; element}}\right)}_{{Crust}}\) was calculated from the composition of UCC61. In aerosols, the elements typically have EF ≤ 10, when originating primarily from crustal sources139. In this study, both the elements Al, Fe, which are commonly used for calculation of EF were present in considerably high concentrations. However, the concentrations of Fe can have anthropogenic inputs from small-scale industries present in the vicinity. Furthermore, Fe undergoes redox reactions and can be mobilized under various environmental conditions. This mobility can skew the EF calculation, as the concentration of Fe might not accurately reflect its crustal origin due to such geochemical processes. Hence Al was selected as the reference material for calculating EF.

Air-mass back trajectory statistics were obtained using meteorological data for the sampling dates, extending up to 72- h prior. The meteorology data, that included wind speed, wind direction, temperature, humidity, and precipitation, was compiled using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model through its graphical user interface140,141. This meteorological data was provided as an input into the geographic information system (GIS) based software TrajStat to generate 72-h airmass back trajectories at 500 m above sea level. These trajectories were then converted into a polylineZ shapefile format for spatial representation142.

Statistical analysis

To assess whether majority of the aerosol mixtures fall within the mixing envelope of the considered end members, a 3–dimensional (3D) mixing polygon simulation (Monte Carlo) was performed in R 4.2.2 (http://www.famer.unsw.edu.au/downloads.html). The simulation was performed in 206Pb/207Pb, 208Pb/207Pb, and 208Pb/206Pb space where the means and 1 standard deviations (SDs) of the sources were taken into account. The iterations generate mixing envelopes and the number of such envelopes that include within them the aerosols, represent the probability of the aerosols for lying inside the mixing envelope116.

To quantify the potential source contributions towards aerosol Pb Bayesian statistics-based modeling package, “MixSIAR” available in R was utilized. The model incorporates mean and SDs along with concentration of the end members to produce model fits for the aerosol mixtures through Markov Chain Monte Carlo simulations143.