Scavenging ratios of polycyclic aromatic compounds in rain and snow at the Athabasca oil sands region

. The polycyclic aromatic compounds (PACs). Monitored PACs, including polycyclic aromatic hydrocarbons (PAHs), alkylated PAHs, and dibenzothiophenes (DBTs), in precipitation and in 3 air at three near-source sites in the Fort MacKay and Fort McMurray area during January 2011 to May 2012 were used to generate a database of scavenging ratios (W t ) for PACs scavenged by 5 both snow and rain. Higher concentrations in precipitation and air were observed for alkylated 6 PAHs and DBTs. The sum of the median precipitation concentrations over the period of data 7 analyzed was 0.48 µg/L for the eighteen PAHs, 3.38 µg/L for the twenty alkylated PAHs, and 8 0.94 µg/L for the five DBTs. The sum of the median air concentrations for parent PAHs, 9 alkylated PAHs, and DBTs were 8.37 ng/m 3 , 67.26 ng/m 3 , and 11.83 ng/m 3 , respectively.

SI: The concentrations of some compounds were 0 while some were blank. Do the authors mean that blank was compounds cannot be quantified (no peak) and 0 is below method detection limits, or the concentration was below 0 after blank correction? Blank correction will influence the results of the compounds with high background (or blank) concentrations such as NAP, which will finally underestimate the values.
Response: The zero concentrations were instrument detection limits (IDL). The blank entries in the tables referred to weekly samples with surrogate recoveries below 50% or above 150% within a month, which were excluded from the total scavenging ratio calculation. No samples were collected in Jan 2011 at AMS5; therefore, the blank entries were replaced by a dash. In the revised paper, the zeroes will be replaced by IDL and we will indicate what the blank entries represent.
The data were blank corrected using an average blank from a batch of samples. The blanks from the air samples (PUF + filter) for naphthalene and alkylated naphthalene were typically higher than other PACs. For naphthalene, the average ± standard deviation of the blanks was 0.45 ± 0.26 ng/m 3 in this study, which were slightly higher than those reported by He and Balasubramanian (2009) (0.15 ± 0.07 ng/m 3 ). The air and precipitation concentrations and total scavenging ratios in this study were compared with those from previous studies, which had applied blank corrections to their samples as well. Applying similar quality control procedures ensure that the comparison of the results is consistent.
constant of gas-phase dominant PACs explained, to a large extent, the different W t values among 23 the different PACs and precipitation types. The trend in W t with increasing alkyl substitutions 24 may be attributed to their physico-chemical properties, such as octanol-air and particle partition coefficients and subcooled vapour pressure, which increases gas-particle partitioning and 26 subsequently the particulate mass fraction. This study verified findings from a previous study of 27 Wang et al. (2014) which suggested that snow scavenging is more efficient than rain scavenging 28 of particles for equivalent precipitation amount, and also provided new knowledge on the 29 scavenging of gas-phase PACs and alkylated PACs by snow and rain.  The probable human PAH carcinogens according to the USEPA are benz(a)anthracene, 44 benzo(a)pyrene, benzo(b)fluoranthene, benzo(k)fluoranthene, chrysene, dibenz(a,h)anthracene, 45 and indeno(1,2,3-cd)pyrene (ATSDR, 2008). In Canada, benzo(a)pyrene, benzo(b)fluoranthene, 46 benzo(j)fluoranthene, benzo(k)fluoranthene, and indeno(1,2,3-cd)pyrene are the PAHs that may constitute a danger to human life or health under the Canadian Environmental Protection Act 48 (CCME, 2010). Abnormal physical and neurological development in infants has also been 49 linked to prenatal exposure of PAHs (Perera et al., 2009). Less is known about the cancer and other health risks of alkylated PAHs because of the limited toxicity data (Baird et al., 2007;51 Wickliffe et al., 2014). Due to the potential toxic effects on animals and humans when exposed 52 to PACs, it is necessary to quantify the deposition budget of PACs to terrestrial and aquatic 53 ecosystems at local to regional scales. 54 The Athabasca oil sands industry in northern Alberta, Canada is known to release versus rain) for in-cloud and below-cloud scavenging processes. Previous below-cloud aerosol 65 scavenging studies found limited evidence that snow scavenging is likely more efficient than rain 66 scavenging based on equivalent water content (Wang et al., 2010(Wang et al., , 2014Zhang et al., 2013).

67
These findings can be verified with field measurements from the Athabasca oil sands region by 68 determining scavenging ratios for PACs, including PAHs, alkylated PAHs, and DBTs.

69
Furthermore, the investigation of precipitation scavenging efficiencies can now be extended to 70 gaseous pollutants.

71
The scavenging ratio is a simplified measure for analyzing wet deposition and is defined 72 as the ratio of the concentration of a chemical in precipitation to that in air (Cousins et al., 1999).

73
In literature, scavenging ratios were determined for radioactive particles, water vapor, sea salt,  Wania et al., 1999). Gas and particulate scavenging ratios for PACs have also been used to 85 estimate the relative contributions or importance of gas and particle scavenging to total wet  The objective of the study is to compare wet scavenging of PACs at the oil sands sites to 91 other locations using scavenging ratios and examine differences between snow and rain 92 scavenging and between gas phase and particulate phase scavenging. This involves generating a 93 database of scavenging ratios for PACs. The database could potentially be used in future wet 94 deposition estimation using monitored surface air concentrations at locations where wet 95 deposition is not monitored and will be detailed in a separate paper.  PACs are collected on the XAD-2 column. The wet sampler is also equipped with a U-shaped 120 siphon on the XAD-2 column outlet that maintains water in the column at all times; thus preventing air exchange. This is also minimized by covering the funnel where precipitation is   The scavenging of gaseous and particulate PACs by rain and snow have been studied 197 using scavenging ratios, which is a simplified approach at examining the overall wet deposition 198 process based only on the concentration of a chemical in precipitation to that in air. Tables S1 and S2), which is the period of data analyzed in this study. Ideally, monthly average 204 concentrations should be obtained from daily air concentrations in a month; however, due to the 205 extensive costs, air sampling was only performed once every six days. The uncertainties from 206 this averaging approach should not be larger than the uncertainties caused by other sources (e.g., 207 measurements itself and/or laboratory analysis). In addition to a lack of data to link individual 208 precipitation samples to individual air samples, scavenging ratios were not determined for every 209 precipitation event because the shorter time scale can lead to large variability in the scavenging 210 ratios (Barrie, 1985;Galloway et al., 1993). For instance, the monthly average scavenging ratios 211 can vary by a factor of 2-5, whereas the variability increases to an order of magnitude for daily 212 scavenging ratios (Galloway et al., 1993). In this study, the median scavenging ratios are based 213 on the monthly scavenging ratios over the snowfall and rainfall periods (about 5-7 months) and and particulate concentration in air (C air, p ), while W g is based on the dissolved PAC 227 concentration in precipitation (C prec, d ) and gas-phase concentration in air (C air, g ).

228
The total scavenging ratios were categorized into snow and rain samples. Snow samples were presented for snow and rain cases separately. The scavenging ratio calculation excluded 232 low precipitation samples (< 1.5 mm).

233
Using the particulate PAC fractions in air measured every sixth day at the AMS5 site, the 234 PACs were categorized into predominantly gas-phase (i.e. > 0.7 gas fraction) and particulate-phase (> 0.7 particle fraction) PACs in order to analyze differences in the precipitation   higher than associated with rain samples (Fig. 3c).

292
The median total scavenging ratios of parent PAHs for snow and rain scavenging were Turkey, the average total scavenging ratios for the gas and particle phases ranged from 8.52-8.97 302 x 10 5 (Birgül et al., 2011). Only the acenaphthylene snow scavenging ratio was 2 to 7 times 303 higher at the oil sands sites than the snow scavenging ratios at other locations.

304
When the individual snow and rain samples were analyzed (Supplementary Table S3), The scavenging ratios for snow samples were larger for particulate-phase than gas-phase 317 PACs at the oil sands sites. The median total snow scavenging ratios were 8.0 x 10 5 for 318 particulate-phase PAHs and 6.7 x 10 4 for gas-phase PAHs, which were within those from 319 previous studies (Fig. 4). These results are were in agreement with the strong relationship   Gas-phase dominant PACs, like acenaphthylene, have lower molecular weight and higher 331 vapor pressures and therefore are more volatile. However, a small mass fraction in particulate-332 phase could increase its overall scavenging ratio (W t ) dramatically compared to the pure gas-333 phase scavenging ratio (W g ) due to the much higher value of W p than W g in literature (Fig. 5a).

349
Similar to the snow scavenging results, the scavenging by rain was greater for particulate-350 phase than gas-phase PACs by an order of magnitude. The median W t was 1.8 x 10 5 for 351 particulate-phase PAHs and 1.1 x 10 4 for gas-phase PAHs. The scavenging ratios were within 352 those reported in literature, which ranged from 5100-1.2 x 10 6 for particulate-phase PAHs and 353 450-2.8 x 10 5 for gas-phase PAHs (Fig. 4). The larger W t of particulate-phase PACs compared 354 to gas-phase PACs is was also consistent with the empirical relationship between LogW t and the  Although the gas-phase PACs have very low particulate mass fraction in air, Fig. 5b 358 illustrates that the particulate scavenging ratios (W p ) can be 1-4 orders of magnitude larger than the theoretical and measured gas scavenging ratios (W g ) in literature. Similar to snow 360 scavenging, rain scavenging of particles containing gas-phase PACs contributed more it has led 361 to larger particulate scavenging contribution of gas-phase PACs to rain than gas scavenging. For 362 particulate-phase PACs, W p and W g for both rain and snow scavenging were more comparable 363 in the literature (up to 1 order of magnitude difference, Fig. 5). Thus, the particle scavenging 364 contribution to snow and rain will dominate the gas scavenging contribution because of the 365 larger particle fraction. The data needed to determine W p and W g were not available at the oil 366 sands sites to confirm literature findings and estimate the relative gas and particle scavenging 367 contributions to rain. In previous studies, the contributions of particle scavenging to rain were  The higher scavenging ratio for naphthalene in the individual rain sample at the oil sands 379 sites compared to literature must be attributed to gas scavenging, since φ = 0 for naphthalene 380 resulting in W t = W g . Gas scavenging can occur by dissolution of gaseous PACs to the surface 381 of raindrops. The gas scavenging ratio from the dissolution process (W g , diss ) depends on 382 temperature-corrected Henry's Law constant, temperature, and the universal gas constant (Franz 383 and Eisenreich, 1998). Another theory is thefor gas scavenging by is vapor adsorption to the surface of raindrops. This scavenging ratio (W g,ads ) can be determined from the air-water 385 interface coefficient and diameter of raindrops (Simcik, 2004;He and Balasubramanian, 2009).

386
However, the theoretical W g for naphthalene reported in literature was only 24.5 (Fig. 5b), which 387 is 10 5 times lower than the measured W g in the individual rain sample at the oil sands sites. The 388 differences between measured W g at the oil sands site and theoretical W g in literature for 389 naphthalene may be attributed to the different cloud and precipitation characteristics and are 390 considered the major sources of uncertainties for precipitation scavenging (Galloway, 1993;391 Franz and Eisenreich, 1998). scavenging ratios was observed for rain samples, and might be due to a lack of temperature-420 corrected Henry's Law constants (Ligocki et al., 1985a) . A moderate correlation coefficient of 421 0.56 was found between temperature-corrected water solubility of gas-phase PACs and total rain 422 scavenging ratios; however, no relationship was found for total snow scavenging ratios. The

423
water solubility of the gas-phase PACs had almost no effect on the total scavenging ratios in 424 snow and rain (r < 0.02) likely because organic compounds are only slightly water soluble. This 425 further implies that there is another mechanism involved in the gas scavenging of gas-phase 426 PACs besides the dissolution process or that particle scavenging makes a larger contribution to 427 the total wet deposition of gas-phase PACs.

428
Snow scavenging of gaseous PACs may be better modelled by surface or interfacial 429 adsorption (W g,ads ). Measured Log(W g ) were strongly correlated with theoretical Log(W g,ads ) 430 for snow events, but not correlated in the rain events. This indicates that interfacial adsorption 431 likely occurred in the snow events (Franz and Eisenreich, 1998). Wania et al. (1999) proposed 432 that the scavenging of gas phase PACs likely occurred by adsorption to the air-ice interface because W g was a strong function of both the partition coefficient for the air-ice interface and 434 vapor pressure of the supercooled liquid. It suggests that snow scavenging of gas-phase PACs is 435 potentially the dominant scavenging process for lower molecular weight or predominantly gas-436 phase PACs (Wania et al., 1999). Compared to snow scavenging, rain scavenging of gas-phase 437 PACs yielded much lower scavenging ratios in field and theoretical studies. For gas phase 438 PACs, W g derived from field measurements ranged from 160-3300, while the ranges for W g due 439 to dissolution and surface adsorption scavenging were only 24.3-710 and 0.2-21.4, respectively 440 (Fig. 5b). Thus, the scavenging of gas-phase PACs by surface adsorption is evidently much 441 lower for rain than snow and may explain the difference in snow and rain scavenging ratios at 442 the oil sands sites.

443
Besides gas scavenging, gas-phase PACs typically have very large particulate scavenging 444 ratios (W p ) because they are more likely to partition to particles. For snow events, W p of gas-445 phase PACs can be 15-385 times larger than measured W g in literature for gas-phase PACs. For 446 rain events, W p ranged from 5.5 x 10 3 -2.7 x 10 7 , while theoretical and measured W g ranged from 447 25-3300 in literature. Therefore, even though gas-phase PACs have very low particulate mass 448 fraction in air, the particle scavenging contribution to snow and rain can still be important 449 because W p can be much greater than the W g . Furthermore, the snow samples are associated 450 with lower air temperatures, which may increase the partitioning of gas-phase PACs to the 451 particulate phase (Pankow, 1991;Cousins et al., 1999). The average and standard deviation of 452 the air temperatures corresponding to snow and rain samples in this study were -8.6±7.2 ⁰C and 453 8.9±8.6 ⁰C, respectively, with an average temperature difference of 17.5 ⁰C. As shown in Fig. 2,   454 the average particulate mass fraction corresponding to snow samples were 2.5-4 times higher 455 than those of rain samples for phenanthrene and anthracene. The combination of higher W p and 456 particulate mass fraction would yield higher total snow scavenging ratios. Particulate mass fractions for alkylated fluorenes, DBT, and C1/C2-DBTs during cold temperatures were also 458 higher. Almost no differences in the particulate mass fractions between snow and rain samples 459 were observed for the other parent gas-phase dominant PAHs.

460
Particle scavenging ratios for PACs in snow events were also observed to be larger than 461 rain events. Average W p ranged from 10 5 -10 6 for the snow events and 10 3 -10 4 for the rain event 462 for particulate-phase PACs (Franz and Eisenreich, 1998). Snow is more efficient than rain at 463 scavenging particles because of its larger surface area (Franz and Eisenreich, 1998). The relative 464 scavenging efficienciesy between snow and rain depends on particle sizes and precipitation increased with the degree of alkylation (Fig. 6). These trends appeared to have some relationship 480 with their physico-chemical properties, such as Henry's Law constant, octanol-air partition 481 coefficient (K oa ), subcooled vapor pressure (p 0 L ), water solubility, and particulate mass fraction (φ) and gas-particle partition coefficient (K p ), obtained from Reid et al. (2013). For naphthalene, 483 the sSnow and rain W t for parent naphthalene was were much higher than those of alkylated 484 napthalenes. With an increase in alkylation, snow and rain W t decreased. This trend is 485 consistent with the increase in Henry's Law constant (Pa m 3 /mol), which leads to lower gas 486 scavenging by dissolution (Franz and Eisenreich, 1998). This is further supported by the 487 decrease in water solubility with increase alkyl substitutions. The large decrease in water 488 solubility from parent naphthalene to C1 naphthalene is also reflected in the W t .

489
The other physico-chemical properties play a more prominent role in the precipitation 490 scavenging of other alkylated PACs. For fluorenes and alkylated phenanthrenes/anthracenes, 491 snow and rain W t increased with alkylation. This trend is consistent with an increase in K oa , φ, 492 and K p as the number of alkyl substitutions increase. These physico-chemical properties are 493 related to gas-particle partitioning (Pankow, 1994;Harner and Bidleman, 1998), which leads to a 494 larger particulate mass fraction and thus, higher W t . The large increase in K oa from C3 to C4 495 fluorene corresponds to the large increase in W t . p 0 L is another physical property affecting gas-496 particle partitioning (Pankow, 1994), but is also related to snow scavenging of gas-phase PACs 497 by interfacial adsorption. As p 0 L decreases with alkyl substitutions, the interfacial adsorption 498 coefficient (K ia ) increases (Franz and Eisenreich, 1998). This results in higher contribution of 499 gas scavenging by snow to the total scavenging ratio. The physico-chemical properties that have 500 the most influence on the precipitation scavenging on of alkylated fluoranthenes/pyrenes and 501 benz(a)anthracenes/triphenylenes/chrysenes are φ and K p because these are particulate-phase 502 dominant PACs. Thus, particle scavenging contributions will dominate the total wet deposition 503 of these PACs. φ and K p are predicted to increase with alkylation. The increase in W t with 504 alkyl substitutions is likely attributed to the increase in particulate mass fraction. For DBTs, the 505 degree of alkylation increased with the rain W t , but did not have a large effect on snow W t . The 506 trend in the rain W t for rain is consistent with the increase in K oa , φ, and K p , which are properties affecting gas-particle partitioning. The decrease in p 0 L with increasing alkyl 508 substitutions should result in an increase in gas scavenging by snow (due to adsorption on the 509 air-ice interface); however, this was not reflected in the snow W t . 510 511 3.6 Uncertainties in snow and rain scavenging 512 The ratio of the maximum to minimum W t for snow and rain scavenging of the gas-phase 513 dominant PACs ranged from 2.4-14.6 and 1.4-10.8, respectively, based on was used to estimate 514 the uncertainties for snow and rain events with similar amounts of precipitation (Fig. 7a). The 515 median W t uncertainties among gas-phase PACs were a factor of 3.6 for snow scavenging and 516 1.9 for rain scavenging (Fig. 7a). The uncertainties from gas scavenging by snow can be very 517 large as shown in the estimated W g for the interfacial adsorption process (10 3 to 10 10 , Franz and 518 Eisenreich, 1998). Field measurements of W g can also be 0.43-20 times greater than the 519 theoretical W g for snow scavenging (Franz and Eisenreich, 1998 could also be enhanced by the presence of an organic layer (Franz and Eisenreich, 1998).

529
Differences in the snow and rain properties are also contributing factors to the gas scavenging 530 uncertainties. These data were not available in this study to assess their effects on the total 531 scavenging ratio uncertainties.
Gas-phase PACs partitions to particulate matter depending on temperature, which affects 533 the particulate mass fraction and subsequently, the total scavenging ratio. The snow events in 534 Fig. 7a not only have similar precipitation amounts, but are also associated with similar air 535 temperatures in order to minimize the temperature and gas-particle partitioning effects on the 536 uncertainties of W t . Rain events associated with similar air temperatures were also used in Fig.   537 7a. Despite analyzing precipitation events with similar air temperatures, particulate mass 538 fraction differed by a factor of 1.6 and 1.5 (median among gas-phase PACs) for the snow and 539 rain events, respectively. This may be due to other factors affecting the particle partitioning of 540 gas-phase PACs (e.g., aerosol water content and chemical composition). When precipitation 541 events with equivalent precipitation amounts but different air temperatures were examined, 542 uncertainties in the total scavenging ratio increased (Fig. 7b). The median W t uncertainties 543 among gas-phase PACs rose to a factor of 6.6 for snow events and 2.5 for rain events (Fig. 7a   544 and 7b). This temperature effect on the uncertainties of the W t is also supported by the larger 545 difference in the particulate mass fraction.

546
The uncertainties in snow scavenging of particulate-phase PACs were larger than those 547 for rain scavenging as shown in the range of scavenging ratios for snow and rain events with 548 similar precipitation rates (Fig. 7b7c). The W t uncertainties forratio of the maximum to 549 minimum snow and rain scavenging ratiosevents were up to 10 and 4.37.7, respectively, which snowmelt would be greater than that in snow. Wania et al. (1999) re-analyzed the data used in 574 Franz and Eisenreich (1998) and obtained much higher gas scavenging ratios. The large 575 variability in the total scavenging ratios (e.g., 4-5 order of magnitude range for PACs) may be 576 attributed to numerous factors that could not be accounted for in the scavenging ratios, such as 577 particle size distribution, droplet sizes, cloud and precipitation type, and air mass trajectories were consistent with the trends in their physico-chemical properties, such as subcooled vapour 598 pressure and octanol-air and particle partition coefficients. Henry's Law constant and water 599 solubility might play a role in the decrease in snow and rain scavenging ratios for naphthalene 600 with increase alkyl substitutions.

601
The database of scavenging ratios was also separated into subgroups to investigate the 602 relative importance of gas and particle scavenging by snow and rain. It was found that snow 603 scavenging is around 10 times more efficient (in terms of the scavenging ratio values) than rain 604 scavenging for both particulate-phase dominant and gas-phase dominant PACs. It was also 605 found that scavenging of particulate-phase dominant PACs is 5 to 10 times more efficient than scavenging of gas-phase dominant PACs under both rain and snow conditions. These findings 607 suggest that snow scavenging of particulate-phase PACs should contribute significantly to the 608 total wet deposition of PACs in this region.

609
The database of PAC scavenging ratios (Supplementary Table S3