Five sampling sites (Fig. 1) are part of the passive air monitoring network in the AOSR. Details regarding site locations, sampling media preparation, and sample deployment have been previously described (Schuster et al., 2015). The samplers were mounted approximately 3 m above the ground. In brief, PUF disks were pre-cleaned with accelerated solvent extraction (Dionex ASE 350) using acetone, petroleum ether, and acetonitrile, prior to use. Since October 2015, PAS-DD have been deployed alongside PAS at a subset of five sites (Figs. 1 and S1). PAS-DD was previously demonstrated to collect gas-phase PAHs (i.e., dry gas-phase deposition) at similar rates as PAS (Eng et al., 2013). A sampling rate of about 5 m3 d-1 was derived previously using co-located PAS and high-volume samplers for the gas- and particle-phase PACs (Harner et al., 2013). Concentrations measured by PAS-DD were expressed as ng m-3 for comparison with PAS, using the PAS effective air sample volume, corrected for reduced exposure surface of PAS-DD (Shoeib and Harner, 2002). Two field blanks were included in the sampling. Oil sands ore was collected from an open-pit mine and provided through the cooperation of Canada's Oil Sands Innovation Alliance (COSIA) and samples of delayed and fluid petcoke were obtained from the University of Alberta and were the same samples previously characterized and reported by Zhang et al. (2016).

The monitored PACs, NPAHs, and OPAHs are listed in Table S1 in the Supplement. Standards for the target analytes were purchased from Cambridge Isotope Labs (Andover, MA), Chiron (Trondheim, Norway), and AccuStandard (New Haven, CT). Deuterium-labeled recovery and internal standards were purchased from Cambridge Isotope Labs (Andover, MA) and CDN Isotopes (Point-Claire, Québec, Canada). The deuterated recovery surrogates included 2,6-dimethylnaphthalene-d12, acenaphthene-d10, anthracene-d10, benzo(b)naphtho(2,1-d)-thiophene-d10, 1-nitronaphthalene-d7, 2-methyl-1-nitronaphthalene-d9, 5-nitroacenaphthene-d9, 9-nitroanthracene-d9, 3-nitrofluoranthene-d9,1-nitropyrene-d9, and 6-nitrochrysene-d11. The deuterated internal standards included fluorine-d10 and benz(a)anthracene-d12, 2-nitrobiphenyl-d9, and 2-nitrofluorene-d9. PUF disks (TE-1014, 14 cm diameter × 1.35 cm thick) were purchased from Tisch Environmental (Village of Cleves, OH).

The highest concentrations in air for ΣPACs_PAS and ΣPACs_PAS-DD were 58 and 150 ng m-3, respectively (Table S2), and observed at site AMS5, which is in close proximity to upgrading facilities. The lowest concentrations of ΣPACs_PAS and ΣPACs_PAS-DD were 14 and 24 ng m-3, respectively, and were observed at AMS14, which is ∼ 35 km southeast of Fort McMurray and adjacent to a near-lake residential settlement (Anzac) that is well removed from the open-pit mining (Table S2) but within several kilometers of a bitumen upgrading facility. For all the sites, the PAS and PAS-DD samples exhibited similar relative PAC compositions, with C1 phenanthrenes and anthracenes (C1-PHEs+ANTs) and PHE being the most abundant alk-PAH and parent PAH, respectively (Fig. S2). For both PAS and PAS-DD samplers, the PAC chemical compositions were dominated by the 2–3-ring PAHs and 2–3-ring alk-PAHs, which accounted for 77–87 % of the sum of all target compounds (Fig. 2). Residential sites (AMS6 and AMS14) were characterized by lower concentrations in air of dibenzothiophene (DBT) and alk-DBTs (compounds that are enriched in bitumen) but higher retene (RET) concentrations (a marker of wood combustion) (Ramdahl, 1983), compared to other sampling sites that are closer to the main oil sands activities. This suggests reduced influence of petrogenic sources and an increased contribution from wood combustion at sites AMS6 and AMS14, which is consistent with their greater distance from open mining areas and greater proximity to residential area where wood is burned for heating and recreational activities. However, it should be noted that RET does not solely originate from wood burning and has also been reported for stack emissions in the oil sands region during the winter period, which challenges the explanation above (Watson et al., 2013). It is also possible that the small (relative to main mining area) upgrading facility, which is several kilometers southeast of AMS14, may play a role.

For all the sites, ΣNPAHs_PAS and ΣNPAHs_PAS-DD were dominated by 2-ring NPAHs, (Table S2). Concentrations in air of ΣNPAHs_PAS and ΣNPAHs_PAS-DD were highest at AMS5 (152 and 170 pg m-3, respectively) and lowest at AMS14 (8.42 and 14.2 pg m-3, respectively). At AMS5, 1-methyl-5-nitronaphthalene/2-methyl-1-nitronaphthalene (1M5NN/2M1NN) was the most abundant NPAH in the PAS sample, whereas 4-nitrobiphenyl (4-NBP) was the most abundant NPAH in the PAS-DD sample. The nitromethylnaphthalenes are known products of the gas-phase OH- and NO3-radical-initiated reactions with methylnaphthalenes (Reisen and Arey, 2004), while 4-NBP emission sources are not well understood and are usually not included in air monitoring campaigns. Past studies reported that ambient 4-NBP concentrations were low or below detection even in urban areas (Crimmins and Baker, 2006; Wang et al., 2011). A study reported an average 4-NBP concentration in air of 45 pg m-3 in Texas, at a sampling site close to petrochemical manufacturing plants and oil refineries (Wilson et al., 1995). Previous chamber reaction experiments demonstrated that 3-nitrobiphenyl (3-NBP) was the only nitro product formed by the OH-radical-initiated chamber reaction of biphenyl and that no nitro products were observed from the NO3-radical-initiated chamber reaction (Atkinson and Arey, 1994). In this study, 3-NBP was below the detection limit at all the sites, while 4-NBP was detected only at AMS5 and AMS9. It should be noted that, in the same sampling period, the 4-NBP concentration measured in a PAS sample at site AMS11 (data not shown), designated as the main oil sands source region (Jariyasopit et al., 2016), was a factor of 5 higher than that observed at AMS5, suggesting that airborne 4-NBP is closely linked to oil sands industrial activities. However, according to our previous study, 4-NBP was below the detection limit at all 15 passive sampling sites during the reporting period April to May 2014 (Jariyasopit et al., 2016). Long-term monitoring of NPAHs could provide insight into seasonal and temporal trends of NPAHs in air in the AOSR and improve understanding of their formation and sources.

NPAHs have been previously measured in PM2.5 released from stacks emissions in the AOSR (Watson et al., 2013). The majority of NPAH target compounds were below detection limits in the PM2.5 collected from stacks in the oil sands region, with the exception of 9-nitroanthracene (9-NAN) (Watson et al., 2013). In general, 9-NAN was the dominant NPAH in urban areas and present at approximately as high concentrations as 2-nitrofluoranthene (2-NF), which is the major particle-associated NPAH known to be formed by atmospheric OH- and NO3-radical-initiated reactions (Wang et al., 2011; Bamford and Baker, 2003; Albinet et al., 2007). In this study site, AMS5, which is close to upgrading facilities, exhibited the highest ΣNPAHs concentration; however, 9-NAN was below detection limits. This result suggests that stack emissions were not a major PM source contributing to NPAH concentrations measured by passive samplers.

For OPAHs, the highest concentration was found at AMS6 which is located in the town of Fort McMurray, approximately 30 km south of the main mining area. This is consistent with our previous study. The elevated OPAHs at this site are due to local primary combustion sources (e.g., vehicular exhausts) as well as the enhanced atmospheric transformation process which is dependent on gaseous oxidants emitted from the combustion sources (Jariyasopit et al., 2016). The highest concentrations for ΣOPAHs_PAS and ΣOPAHs_PAS-DD were 753 and 1480 pg m-3. The lowest concentrations for ΣOPAHs_PAS and ΣOPAHs_PAS-DD were 213 and 352 pg m-3, respectively, and were observed at AMS13. The dominant contributor was 2–3-ring OPAHs (Table S2). For both sampler types 9-fluorenone (FLO) was the dominant OPAH at all sites. Different target OPAHs have been reported by various air monitoring campaigns, making it challenging to compare the OPAH air concentrations in the AOSR to data available in the literature. On average, the ΣOPAHs concentrations derived from PAS and PAS-DD samplers in this study were comparable to or lower than concentrations measured at urban sites (Wang et al., 2011; Ahmed et al., 2015; Albinet et al., 2008a).

A key aspect of the study was to compare the performance of PAS and PAS-DD to capture PACs, NPAHs, and OPAHs. Their relative performance can be illustrated using the enhancement ratio which is defined as the ratio of the concentration of an analyte in PAS-DD to that in PAS. A value close to 1 indicates comparable ability of the two sampler types for capturing an analyte. Figure 3 shows enhancement ratios for PACs for all the sampling sites. Average ratios for 2–3-ring, 4-ring, and 5–6-ring PACs were 2.3, 2.8, and 3.6 respectively. Similarly, the enhancement ratio increased with molecular weight for the NPAHs (Fig. S3) up to an average value of about 4. Comparison of the enhancement ratios of PACs among the sites indicates that the ratio is partly dependent on particle loadings. This is evident from higher enhancement ratios of higher-ring PACs, except for BghiP (Benzo(ghi)perylene), observed in AMS5, which is the near-source site where ΣPACs, ΣNPAHs, and ΣOPAHs concentrations were highest. This finding illustrates the enhanced ability of PAS-DD to capture more of the higher-molecular-weight target compounds that are associated with the larger depositing particles in air, whereas the PUF-PAS preferentially samples the smaller airborne particles.

The enhancement ratios for the NPAHs and OPAHs were considerably more variable among sites compared to the patterns observed for the PACs (Fig. S3). This may be due to multiple factors contributing to their presence in air, which can be by direct emission from primary sources as well as production in air through transformation processes. Transformation reaction rates will vary among NPAH and OPAH compounds and also spatially, depending on atmospheric conditions and oxidant concentrations.

The composition of PACs, NPAHs, and OPAHs in petcoke and oil sands ore samples was investigated in order to assess if these compositions are reflected in the passive air samplers, thereby indicating potential contributions. Results of residue analysis are discussed below and summarized in Table 1. Individual PAC concentrations and composition are given in Table S3 and Fig. S4.

The levels of ΣPACs (dry weight basis) were approximately 5 times higher in the delayed petcoke (85 300 ng g-1) compared to the fluid petcoke (16 700 ng g-1) (Table 1). The greater abundance of PACs in delayed petcoke (vs. fluid petcoke) is likely due to the lower temperatures involved in the delayed coking process (Anthony, 1995), resulting in reduced losses of PACs due to degradation and volatilization, leaving PACs more concentrated in the residue. Furthermore, the residues of 2–3-ring alk-PAHs and 4–6-ring alk-PAHs were ∼ 10 times and ∼ 2 times higher than their parent PACs (Table 1). DBT and alk-DBTs concentrations in the delayed petcoke were ∼ 5 times higher than the concentrations in the fluid petcoke. Despite the difference in absolute residue concentrations of PACs, the PAC profiles were similar in delayed and fluid petcoke: 4–6-ring compounds accounting for ∼ 60 % of the ΣPACs (Figs. 4 and S4), of which 4–6-ring alk-PAHs were the dominant group. It is also noteworthy that chrysene (CHR) and the sum of alk-benzo(a)anthracenes, alk-triphenylenes, and alk-chrysenes (alk-BTCs) were major components of petcoke (Table S3 and Fig. S4). The oil sands ore residue concentration of ΣPACs (dry weight basis) was 680 000 ng g-1, which was ∼ 40 and ∼ 8 times higher than ΣPACs determined in the fluid and delayed petcoke, respectively (Tables 1 and S3). The oil sands ore had the highest residue concentrations of 2–3-ring parent PAHs, alk-PAHs, RET, and alk-DBTs; however, the delayed petcoke had the greatest residue concentrations of 4–6-ring parent PAHs and DBT (Table 1). In contrast to the PAC compositions of the fluid and delayed petcoke, 2–3-ring PACs were the major component in the oil sands ore accounting for ∼ 70 % of the ΣPACs (Table 1 and Fig. 4). A past study reported percent contribution of 2–3-ring PACs in oil sands samples ranging from 79 to 94 % of the ΣPACs (Yang et al., 2011). Variability in percent contribution was caused by the degree of biodegradation in the samples, with greater biodegradation leading to reductions in 2–3-ring alkylated PAHs (Yang et al., 2011).

For the petcoke and oil sands ore samples, residues of the sum of NPAH and OPAH concentrations were 2 to 4 orders of magnitude lower than the PAC concentrations. Similar to the results for PACs in petcoke samples, the oil sands ore exhibited higher ΣNPAHs (332 ng g-1), which were 1.6 times and 178 times higher, respectively, than in the delayed (208 ng g-1) and fluid petcoke (1.91 ng g-1) (Table S3). However, the delayed petcoke was the most enriched in 4–6-ring NPAHs. The most abundant NPAHs in the delayed petcoke were 6-nitrobenzo(a)pyrene (6-NBaP) (138 ng g-1), whereas the most abundant in the ore was 4-NBP (228 ng g-1). The presence and absence of these compounds in the various source samples present an opportunity to use these as markers to evaluate emission sources to air. Figure 5 summarizes the occurrence of the various NPAH marker compounds by showing extracted ion chromatograms for samples of delayed and fluid petcoke, ore, and two passive air samples. In the following section of the discussion, we consider the implication of these profiles for indicating potential contributions to air of petcoke and ore particles.

The delayed petcoke exhibited higher residues of ΣOPAHs compared to fluid petcoke and ore – approximately 10 times and 3 times higher, respectively (Tables 1 and S3). The most abundant OPAHs in fluid petcoke, delayed petcoke, and ore were 9,10-anthraquinone (ANQ, 31 ng g-1), benzo(a)fluorenone (BaFL, 180 ng g-1), and FLO (113 ng g-1), respectively. All individual OPAH concentrations measured in the delayed petcoke were higher than those measured in the fluid petcoke and the oil sands ore except for FLO, which was highest in the ore. The fluid petcoke and ore were enriched in lower-ring OPAHs; however, the delayed petcoke was enriched in higher-ring OPAHs. OPAHs have been identified as transformation products of biological and chemical processes and reported to be abundant in soils from PAH-contaminated sites in Sweden and thought to be formed through PAH remediation processes including bioremediation and Fenton oxidation (Lundstedt et al., 2007).

Comparisons of PAC compositions for the delayed petcoke, fluid petcoke, and oil sands ore versus the PUF-PAS, and PAS-DD air samples revealed interesting differences as shown in Figs. 4 and S4. For instance, the petcoke particles exhibited enrichment in the higher-molecular-weight, semi-volatile, and particulate-associated parent PAHs and alk-PAHs, while the oil sands ore and passive air samples were dominated by lower-molecular-weight and more-volatile alk-PAHs. DBT makes up a small proportion (∼ 0.1–1 %) in all the samples. Alk-DBTs comprise 13 % of the ΣPACs in the delayed and fluid petcoke but were 17 % in the oil sands ore (Table S3), whereas small compositions of alk-DBTs were observed in the PAS and PAS-DD air samples (mean of 6–7 %) (Table S4). RET accounted for 1–3 % of the ΣPACs in the oil sands ore, PAS, and PAS-DD but it accounted for much smaller proportions in the fluid petcoke and delayed petcoke. Furthermore, as illustrated in Fig. S4, individual 2–3-ring PACs in the oil sands ore and passive air samples make up similar proportions of the ΣPAC burden. Given the relatively higher volatility of these 2–3-ring PACs, this suggests that oil sands ore is contributing substantially to these gas-phase PACs in air, most likely through volatilization from open mine faces.

To assess the potential contributions of particle-associated PACs in air, stemming potentially from either petcoke or the exposed ore from open-pit mines, we focus on the higher-molecular-weight compounds which exist primarily in the particle phase. These compounds dominate the PAC composition of petcoke (Fig. S4), since the more-volatile, lower-molecular-weight compounds are depleted during the high temperature coking process. If petcoke and/or ore particles represent an important contributor to PACs present in air, then their compositions should be reflected in PAS and especially the PAS-DD samples. The compositions of the 4–6-ring PACs in delayed petcoke, fluid petcoke, oil sands ore, and passive air samples are compared in Fig. S5. The parent 4–6-ring PAH composition of the passive air samples did not match the 4–6-ring PAH compositions of the petcoke and ore, suggesting a minimal contribution of these sources to parent PAH burdens in air. This implies that other sources of parent PAHs (e.g., combustion, vehicle emissions) were more dominant, whereas the 4–6-ring alk-PAH compositions for all samples were more similar, suggesting some contribution of petcoke and oil sands ore particles in the passive air samples. However, these findings for parent PAHs and alk-PAHs are somewhat contradictory. If petcoke particles had contributed substantially to alk-PAHs in air, then the parent PAHs contained in these same particles (and making up ∼ 25 % of the PAC residue, Table S3) should have also been important contributors, especially since concentrations in air of parent PAHs are much lower and therefore more sensitive to the petcoke contribution. The results are less contradictory in the case of ore since ore residues are dominated by alk-PAHs (∼ 95 %, Table S3). In other words, ore particles could be contributing substantially to the alk-PAHs present in air but not substantially to parent PAHs in air, which is consistent with Figs. S4 and S5. In fact, the relative proportions of alk-PAHs to parent PAHs in ore are consistent with the relative proportions observed in air, with a dominance of the alk-PAHs (Fig. 4, Table S2).

In the case of NPAHs and OPAHs, their relatively low residue concentrations in petcoke and ore samples complicate the assessment of potential contributions of these particles to air samples. As shown in Fig. 5, the high molecular weight and particle-associated NPAH marker compounds, 2-NP, 1,6-DNP, and 6-NBaP are detected in various petcoke and ore samples but not reflected in air. However, the gas-phase marker compound 4-NBP  which is present in the delayed petcoke and the oil sands ore, is also captured at site AMS5 and AMS9 (Table S2). These two sites are the closest to open-pit mines, which points to the potential importance of volatilization from open-pit mines as a source to air of 4-NBP and other volatile PACs. This finding is supported by the results of a simple ore–air partitioning experiment showing that 4-NBP was in fact detected in air that has equilibrated with ore. Details of the experimental setups which are based on Francisco et al. (2017) are provided in the Supplement.

We conclude that oil sands ore is contributing substantially to burdens of PACs in air near mining areas but not at sites further removed from open mines. This finding is consistent with air and snow monitoring studies that indicate that most of the deposition of mining-related particles and associated chemicals occurs within the first several kilometers of mining areas (Schuster et al., 2015; Kelly et al., 2009).

In this aspect of the study, passive air samples were screened for petcoke particles using scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS). Petcoke particles can be distinguished from ore and other particle types by their unique morphology (using SEM) and elemental composition (based on EDS spectra). An image and EDS spectrum of an authentic delayed petcoke particle is shown in Fig. 6a, demonstrating the unique relative elemental abundance (excluding carbon) as S (sulfur) > Si (silicon) ≈ Al (aluminum), whereas ore particles have different relative elemental abundance (i.e., Si > S > Al) (Zhang et al., 2016). Based on their morphology and elemental composition, petcoke particles were screened for and identified in all of the passive air samples. It should be noted that the elemental composition of delayed and fluid petcoke particles are indistinguishable; however, they can be distinguished from each other by morphology at 1000× magnification (Zhang et al., 2016). Figure 6b shows an image of petcoke-like particle in PAS-DD PUF deployed at AMS5, as well as the elemental profile with the dominance of sulfur. The results of the particle screening exercise carried out on all of the PAS and PAS-DD samples confirmed the presence of petcoke particles in PUFs collected at AMS5, AMS9, and AMS13, which are the sampling sites closest to mining activities (Fig. 1). This indicates that surface-derived particles such as petcoke and ore particles can be suspended in air and transported away from mining areas, albeit over relatively short distances. Because volatile PAC residues in ore particles are much greater than in petcoke particles (Table S3), it is also very likely that ore makes a larger contribution to air burdens of PACs, which is consistent with the earlier discussion related to marker compounds and chemical profiles.