3.1 Levels and comparison with other studies
PM2.5
The overall range of ∑16 PAHs in PM 2.5 during pre-monsoon, monsoon and post-monsoon were 17–89 ng/m3, 2–40 ng/m3 and 74–282 ng/m3 respectively. The mean concentration of total carcinogenic PAHs in Jharsuguda and Angul was 52 ng/m3 and 40 ng/m3 respectively. The mean concentration of ∑16PAHs in Jharsuguda (102 ng/m3) was two folds higher than Angul (66 ng/m3) (Table 1). In both sites, the overall trend of PAHs pollution was observed as post monsoon > pre monsoon > monsoon (SI Table S4 and S5). Concentration of ∑16PAHs in Odisha during post monsoon was 3 folds higher than pre monsoon and 10 folds higher than monsoon. PM2.5 bound PAHs were significantly different (p < 0.05) between three seasons (SI Table S8 ). Furthermore, significant difference of individual PAHs between different seasons were observed in all the stations (p < 0.05).The overall mean concentration for all the seasons for ∑16PAHs in PM2.5 in industrial regions of Odisha was slightly lower than Zhengzhou (111 ng/m3) (Wang et al., 2014). Mean ∑16PAHs concentrations in industrial sites of Odisha were comparable with Beijing (244 ng/m3) (Wang et al., 2008), Nanjing (125 ng/m3) (Wang et al., 2007), Xian (140 ng/m3) (Bandowe et al., 2014) in China, New Delhi in India (105 ng/m3), (Sarkar and Khillare, 2013), Zonguldak in Turkey (94 ng/m3) (Akyüz and Çabuk, 2009) but higher than Atlanta in USA (3 ng/m3) (Li et al.2009), Norway (8.60 ng/m3) and United Kingdom (11.20 ng/m3) (Eeftens et al., 2012) and Madrid in Spain (15 ng/m3) (Barrado et al., 2013) (Table S6).
Table 1
Concentration of PAHs in the two industrial region in Orissa, India
Concentration
in
ng m− 3
|
JHARSUGUDA
|
ANGUL
|
PM 2.5
|
PM 10
|
PM 2.5
|
PM 10
|
RANGE
|
AVG ± SD
|
RANGE
|
AVG ± SD
|
RANGE
|
AVG ± SD
|
RANGE
|
AVG ± SD
|
Naphthalene
|
nd − 92
|
24 ± 33
|
nd − 153
|
40 ± 54
|
nd -41
|
6 ± 9
|
nd -68
|
10 ± 14
|
2 Rings
|
nd − 92
|
24 ± 33
|
nd − 153
|
40 ± 54
|
nd -41
|
6 ± 9
|
nd -68
|
10 ± 14
|
Acenaphthylene
|
nd − 20
|
6 ± 7
|
nd − 35
|
9 ± 13
|
nd -20
|
3 ± 6
|
nd -33
|
6 ± 9
|
Acenaphthene
|
nd − 18
|
4 ± 5
|
nd − 32
|
7 ± 9
|
nd -10
|
1 ± 2
|
nd -17
|
2 ± 4
|
Fluorene
|
nd − 7
|
3 ± 2
|
1–11
|
5 ± 4
|
nd -64
|
6 ± 14
|
nd -104
|
11 ± 23
|
Phenanthrene
|
nd − 8
|
3 ± 3
|
1–14
|
4 ± 5
|
nd -27
|
3 ± 6
|
nd -43
|
4 ± 10
|
Anthracene
|
nd − 2
|
1 ± 1
|
nd − 4
|
1 ± 1
|
nd -4
|
1 ± 1
|
nd -6
|
2 ± 2
|
3 Rings
|
1–49
|
16 ± 16
|
2–86
|
27 ± 28
|
nd -98
|
14 ± 24
|
nd -158
|
24 ± 40
|
Fluoranthene
|
nd − 7
|
3 ± 3
|
nd − 13
|
6 ± 5
|
nd -20
|
2 ± 4
|
nd -33
|
4 ± 7
|
Pyrene
|
nd − 3
|
1 ± 1
|
nd − 5
|
2 ± 2
|
nd -4
|
1 ± 1
|
nd -8
|
1 ± 2
|
Benzo(a)anthracene
|
nd − 24
|
6 ± 9
|
nd − 41
|
11 ± 15
|
nd -13
|
3 ± 3
|
nd -21
|
4 ± 5
|
Chrysene
|
nd − 69
|
17 ± 23
|
nd − 114
|
29 ± 36
|
nd -36
|
9 ± 10
|
nd -59
|
15 ± 16
|
4 Rings
|
nd − 100
|
27 ± 31
|
nd − 170
|
47 ± 53
|
nd -53
|
15 ± 16
|
nd -88
|
25 ± 26
|
Benzo(b)fluoranthene
|
nd − 15
|
5 ± 5
|
nd − 53
|
10 ± 15
|
nd -40
|
13 ± 11
|
nd -66
|
21 ± 18
|
Benzo(k)fluoranthene
|
nd − 19
|
4 ± 7
|
nd − 32
|
7 ± 12
|
nd
|
-
|
nd
|
-
|
Benzo(a)pyrene
|
1–38
|
15 ± 10
|
2–63
|
27 ± 17
|
nd -35
|
13 ± 11
|
nd -59
|
22 ± 18
|
5 Rings
|
1–57
|
24 ± 18
|
3–129
|
44 ± 36
|
nd -64
|
26 ± 21
|
nd -106
|
43 ± 35
|
Indeno(1,2,3-cd)pyrene
|
nd − 7
|
2 ± 2
|
nd − 12
|
3 ± 4
|
nd -13
|
2 ± 3
|
nd -21
|
3 ± 4
|
Dibenzo(a,h)anthracene
|
nd − 12
|
2 ± 4
|
nd − 19
|
4 ± 6
|
nd -2
|
0 ± 1
|
nd -3
|
1 ± 1
|
Benzo(ghi) Perylene
|
nd − 23
|
7 ± 8
|
nd − 39
|
12 ± 14
|
nd -6
|
2 ± 2
|
nd -11
|
4 ± 3
|
6 Rings
|
nd − 36
|
11 ± 13
|
nd − 62
|
19 ± 22
|
nd -16
|
4 ± 4
|
1–26
|
7 ± 6
|
∑16 PAHs
|
2–282
|
102 ± 101
|
6–479
|
176 ± 172
|
2 -234
|
66 ± 66
|
3 -387
|
109 ± 108
|
Carcenogenic PAHs
|
1–133
|
52 ± 44
|
3–229
|
90 ± 77
|
1 -108
|
40 ± 34
|
1 -180
|
66 ± 55
|
PM10
Overall range of ∑16 PAHs in PM10 during pre-monsoon, monsoon and post-monsoon were 34–145 ng/m3, 3–65 ng/m3 and 116–479 ng/m3, respectively. Similar to the trend of PM2.5, the mean concentrations of total and carcinogenic PAHs were highest during post monsoon season. The mean concentration of ∑16 PAHs for all the seasons in Jharsuguda (176 ng/m3) was higher than Angul site (109 ng/m3) (Table S4 and S5). PAHs pollution was observed in the order post monsoon > pre monsoon > monsoon which is in line with the observation in PM2.5. PM10 bound PAHs were significantly different (p < 0.05) between three seasons (SI Table S9). Furthermore, significant difference of individual PAHs between different seasons were observed in all the stations (p < 0.05). Higher PAHs level in the post-monsoonal season can be reasoned with lower temperature, photochemical composition and radiation thereby reducing the evaporation from particulate to the vapour phase and stronger photochemical composition of PAHs (Feng et al., 2005; Odabasi et al., 1999). However, lower concentrations of PAHs during monsoon can be associated with washing out effect for particulates (Karar and Gupta, 2006).
The mean concentration of total PAHs in PM10 (131 ng/m3) was higher than Agra (43 ng/m3) (Masih et al., 2010), Visakhapatnam (57 ng/m3) (Kulkarni et al., 2014) but lower than Kanpur (616 ng/m3) (Singh et al., 2015), Tiruchirapalli (259 ng/m3) (Mohanraj et al., 2011) and Amritsar (154 ng/m3) (Kaur et al., 2013) in India. The mean concentration of total PAHs were higher than urban Malaysia (6 ng/m3) (Omar et al., 2002) and Xiamen in China (15 ng/m3) (Hong et al., 2007) (Table S7).
3.2. Seasonality and back trajectory analysis
HYSPLIT back trajectories analysis were performed over both Angul and Jharsuguda (Fig. 2). The air mass five days back trajectories for each hour (00 to 23 Z) at height of 500 m above ground level (AGL) of 144 trajectories (6 Days x 24 hours = 144 trajectories) were clustered into 3 clusters. Each site had different 6 day samples and the Reanalysis data resolution was 2.5o x 2.5o, for Angul, single plot was made for 4 locations. Radiosonde data over Bhubaneswar from University of Wyoming (Department of Atmospheric Science, University of Wyoming, http://weather.uwyo.edu/upperair/sounding.html) was used for the 6 days corresponding to sample collection dates.
Figure 2a shows the three major clusters at Angul site. The three cluster types were studied as per direction, location traced and seasons. About 50 % of the trajectories (n = 72) were during the 22nd March 2017 and winter time (23rd November 2017 and 15th January 2018) originating from the North-West location from a landmass which was a major dust source. Cluster 1 was traversing through the industrial locations of Rajasthan, Uttar Pradesh and Chhattisgarh for 4 days and settling down at 500 m height from 2000 m. Air masses in Cluster 1 constituted more than 85 % of total and carcinogenic PAHs in both PM 2.5 and PM 10. Cluster 2 (mainly during monsoon season) which was 33 % of total trajectories traversed through the south Indian states, Karnataka and Telangana and were below 500 m Above Ground Level (AGL) and finally crossed throughout Odisha in the last 2 days before reaching Angul. This cluster had more impact from south–west monsoon seasonal winds. Cluster 3 (17 %) arrived from the Bay of Bengal via Indian Ocean during the pre- monsoon on dated 24th May and originated from Indian Ocean and Arabian Sea. Interestingly, Cluster 2 and 3 together constituted less than 10 % of the total PAHs and carcinogenic PAHs in both PM 2.5 and PM 10. Thus, this cluster was affected mostly by localized emission sources from 12 hours journey over Odisha. This hypothesis has been further supported by wind rose showing the exact wind direction (Fig. 2b). Between 4 to 9 % frequency of high-speed wind were coming between North-West and North. Winds with a speed of 10–15 m/s were coming from the South and South-West directions. Figure 2c shows Cluster 1, which contributed about one third of total back trajectories that came from Arabian Sea via Maharashtra and Chhattisgarh during the winter season and traversed from west and from a height below 500 m AGL.
Over Jharsuguda, wind rose showed 10 m/s to > 15 m/s winds (with 1.6–6.4 % frequency) from the west (Fig. 2d). Over Jharsuguda, most of the trajectories (Cluster 2) arose from the industrial regions of Indo-Gangetic plain and we suspect the possibility of long range atmospheric transport from such emission regions. Thus, this cluster can be associated with a mixer of long-range transport and local pollution sources mainly from vehicular and industrial sources. Cluster 2 (50 %) was traversing during specific dates (21st March, 21st November, 21st December) during winter and pre-monsoon seasons of 2017 originating from north and North-West along the Indo - Gangetic plain (Fig. 2e). These months constituted nearly 80 to 85 % of total PAHs and carcinogenic PAHs in PM2.5 and PM10. It originated from the Arabian Sea and crossed over the states, Maharashtra and Chhattisgarh for 3 days before ending in this sampling site. Wind rose indicates high wind speed coming from North and North-West directions (Fig. 2b). In 2017, on 26th May, the 3rd cluster (17 %) originating from south and south-west i.e. the Bay of Bengal and can be used to track local sources pollution especially from vehicles and local industries. The maximum transfer of PAHs air parcel was observed during the winter season (November, December) followed by pre-monsoon (March) over both Angul and Jharsuguda sites. This winter season atmospheric transport might be caused by western disturbance, which brings the pollutant from industrial areas of Indo-Gangetic plains to the study area.
3.3. Source apportionment
PAHs can be classified by the number of aromatic rings such as two rings (Nap), three rings (Acy, Ace, Flu, Phe, Ant), four rings (Flt, Pyr, BaA, Chry), five rings (BbF, BkF, BaP, DBA) and six rings (InP, BghiP) and this was used to construct the distribution pattern of the PAHs. Some studies have reported that the low molecular weight PAHs (LMW, 2–3 ring) mainly exist in the coarse part of the PM (Tan et al., 2011). In this study, both the high molecular weight PAHs (HMW, 4–6 rings) and LMW PAHs mainly existed in the fine particles. During all the seasons 5 ring PAHs was dominant in all the locations (Table S10 & S11). LMW/HMW ratios of PAHs were in the range of 0.13–0.33, 0.02–0.76, and 0.24–0.9 during summer, monsoon and post monsoonal seasons respectively. Box-whisker plots representing ring wise distribution of PAHs in PM2.5 and PM10 among the three seasons in the two major industrial regions of Odisha are shown in Fig. 3.
PAHs viz. Flu, Pyr ,Chry, BbF, BkF, BaA, BaP, InP and BghiP were classified as combustion derived PAHs (COMPAHs). The ∑COMPAHs contributed between 54 to 97 % of the ∑16PAHs in PM 10 and 14 to 100 % in PM2.5 respectively. There was a clear difference in post-monsoonal samples from the other seasons and some deviations in each sampling site. In monsoon and post- monsoon season, combustion derived PAHs constituted more than half of the ∑16PAHs. Ratio of COMPAH/∑16PAH ratios were determined for non-catalyst (0.4) and catalyst equipped (0.5) automobiles and for heavy-duty diesel trucks (0.3) (Rogge et al., 1993). In this study the values of COMPAH/∑16PAH varied between 0.74–0.85, 0.53–0.97 and 0.52–0.79 during pre-monsoon, monsoon post-monsoonal seasons. Ratio of 0.5 might be from catalyst equipped automobiles mainly observed in winter season, however, the ratios in other seasons varying between 0.74 to 0.97 might indicate a complex source. Seven carcinogenic PAHs (∑7carcPAHs) are another important category for monitoring pollutants in the atmosphere to assess the carcinogenic potential of PAHs to humans. Mean concentration of ∑7carc PAHs followed the trend as: post-monsoon > pre-monsoon > monsoon in both PM2.5 and PM10 across industrial regions of Jharsuguda (Table S4) and Angul (Table S5).
PAHs have been used as tracers to distinguish between diverse sources (Lodovici et al., 2003; Vasconcellos et al., 2003). For the source apportionment, various diagnostic ratios were combined with principal component analysis to arrive at a suitable source type for a specific group of PAHs. Diagnostic ratios of PAHs are usually an effective way to identify sources because they exhibit the characteristics of specific sources, but they should be used carefully because some of them are variable in different ambient conditions due to the reactivity of some PAH species, such as the photolytic decomposition of BaP (Feng et al., 2005; Odabasi et al., 1999). Diagnostic ratio of Flt/(Flt + pyr) (Yunker et al., 2002), BbF/BkF(Agarwal et al., 2009), BaA/(BaA + Chry)(Ping et al., 2007), BaA/Chry (Wang et al., 2010), BaP/BgP (Wang et al., 2007) and ∑7carcPAHs /∑16PAHs (Zhang et al., 2006) were used in this study to understand the specific source type.
PC-1: First component represented 52 % of the total variance and was strongly weighted with LMW-PAHs, such as Nap, Ace and Acy and HMW-PAHs viz. BaA, Chry, BkF, DBA and BgP with best fit > 0.7. Ace, Chry, BgP(SI Table 16) are reported to be sourced mainly from traffic exhausts emission in India (Cheng et al., 2013). BaA/(BaA + Chry) ratio has been used to evaluate the contribution of vehicular emissions. Across all the seasons in both the sites, the mean ratio was observed less than 0.3 indicating catalyst equipped vehicles (Gogou et al., 1996) as the main vehicular emission source. Nap and Acy might have resulted from the coal combustion emission. Likewise, BaP and Ace might have resulted from coal and gasoline emission. This component can therefore represent PAHs sources mainly from vehicular emission.
PC-2: With 17 % variance, this component was mainly loaded with Flu, Ant, Phe, BbF and BaP. HMW PAHs like four to five rings can be significantly emitted from light vehicles and Fluoranthene is an indicator for heavy-duty diesel combustion (Marr et al., 1999). Pyr, BaA, Chry and Fl are markers for coal combustion (Tavakoly Sany et al., 2014) and BaP for biomass burning (Belis et al., 2011). Flu, Phe, Flt, and Ant could be from diesel-powered vehicle emissions. Phe/(Phe + Ant) ratio has been used to identify the importance of petrogenic hydrocarbons in relation to emission from biomass burning. Mean value of Phe/(Phe + Ant) ratio in both PM2.5 and PM10 in Jharsuguda and Angul across all the seasons were greater than 0.1 indicating petrogenic sources (Table S12 and S13). Mean Flt/(Flt + Pyr) ratio in all the sites across all the seasons were greater than 0.5 indicating diesel combustion as a major source type. This component can therefore represent mixed source types due to biomass combustion and petrogenic emission.
PC-3: With 7 % variance, this component was weighted only with InP. Mean value for the ratio of InP/(InP + BgP) was 0.5 in both PM2.5 and PM10 in Angul during all the seasons. However in Jharsuguda the values for this ratio during pre -monsoon was > 0.7 in both PM2.5 and PM10 but in other seasons the ratio was < 0.3.The ratio between 0.37–0.70 indicate diesel emissions (Kavouras et al., 1999),(Alves et al., 2017; Pio et al., 2001) and > 0.5 indicates coal, wood combustion (Gogou et al., 1996; Pio et al., 2001).
3.3 Intra-site variation
At different sampling sites in Angul, In PM10, ∑16 PAHs during pre-monsoon followed the trend: A1(Nalco Township) > A3(TTPS) > A4(MCL) > A2(Hakimpada). During monsoon the trend was as: A1 > A2 > A3 > A4 and during post-monsoon it was A1 > A4 > A3 > A2 in Angul. Similar trend was observed for ∑16 PAHs in PM 2.5 across all the seasons. This results strongly suggested that site A1 was the most polluted area and this may be due to the impact of vehicular emission of highway and the township near Nalco Smelter and its captive power plant and proximity to Talcher Thermal power plant and coal mine area. It is to be noted that this location is impacted by heavy vehicles and coal mining of Mahanadi coal field and Hakimpada (small industrial area) is the commercial area comprising of small-scale industries and office areas. But in post monsoon A1 (Nalco township) was found to be the most polluted area followed by MCL, coal mining area, TTPS township near Talcher Thermal power plant area and Hakimpada. Two-way ANOVA among compounds and sites during all the seasons showed significant difference both with respect to compounds and sites (p < 0.05). Two-way ANOVA among compounds and seasons at Hakimpada, Angul showed significant difference both with respect to compounds and seasons (p < 0.05).
At two different sampling sites in Jharsuguda, ∑16 PAHs in PM10 during pre-monsoonal season followed the trend Cox colony (B1) > TRL Colony (B2). Higher concentration of PAHs in Cox colony may be due to vehicular emission from the nearby Highways and emissions from the surrounding industrial region. In monsoon the trend was same i.e., Cox colony (B1) > TRL Colony (B2), may be due to vehicular emission and industrial emissions and during post-monsoon the trend was like TRL Colony (B2) > Cox Colony (B1) may be due to mining activities in winter season by the local mines at close proximity of the study area.
Two way ANOVA among sixteen PAH compounds across all the seasons at Cox colony showed significant difference with respect to compounds (p < 0.05) and with respect to seasons (p > 0.05). Two-way ANOVA among compounds and across all the seasons at Belpahar showed significant difference with respect to compounds (p < 0.05,) and significant difference with respect to seasons (p < 0.05). Higher concentration was observed during post-monsoon than the pre-monsoon and monsoonal seasons and might be possibly related to the lower temperature, weaker radiation strength and more emission sources (Karar and Gupta, 2006; Valavanidis et al., 2006).
3.4 Risk assessment
3.4.1 TEQs
Toxic equivalency factors (TEFs) of the individual PAHs have been popularly used to calculate their carcinogenic potential or Benzo(a) pyrene equivalence (BEQ). Comparing the ∑7carcPAHs weight and TEFs, low and higher molecular weight PAHs (BaP, DBa, BaA, BbF, BkF, Ind, Chry and Ant) were the main carcinogenic components of the 16 priority PAHs. The concentration of those components were quite low and their ratios were 0–30 %, suggesting the relatively lower human exposure health risk in Angul when comparing with other cities like 69.4 % of San Paulo city (Bourotte et al., 2005), 51.0 % at Las Condes during spring, 54.6 % at Providencia and 56.3 % at Las Condes (del Rosario Sienra et al., 2005).
The potential toxicity of the contaminated air samples was evaluated using the toxic BaP equivalent quotient (TEQ) for seven carcinogenic PAHs. Viz., BaP, BaA, Chysene, BbF, BkF, DBA and InP. Range of TEQs for total PAHs and carcinogenic PAHs in Jharsuguda (Table S14) and Angul (Table S15) for different seasons. Figure 4 shows distribution of total PAHs and carcinogenic PAHs for different seasons.
3.4.2 Inhalation Risk
The estimated LADD values of carcinogenic PAHs in PM2.5 and PM 10 for different age groups are presented in the Fig. 5. It can be observed from the figure that during the post monsoon season ILCR was higher in both PM 2.5 and PM 10 (Fig. 5) across all age groups. Across different seasons and age groups, ILCR values for daily inhalation and ingestion exposure to PAHs were higher than the values for daily exposure through dermal contact. This observation was in line with earlier study in South Africa (Morakinyo et al., 2019a). The risk was highest in children in the age group of 2–4 years. Due to the higher concentration of PM10, the ILCR were higher compared with PM2.5. In our study ILCR ranged between 10− 5 and 10− 3 representing potential cancer risk to significant cancer risk (Wang et al., 2011)