Supplementary Material
Composition and effects of inhalable size fractions of atmospheric aerosols in the polluted atmosphere. Part I. PAHs, PCBs and OCPs and the matrix chemical composition
Linda Landlová1, Pavel Čupr1, Juraj Franců2, Jana Klánová1, Gerhard Lammel1,3
1 Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment, Kamenice 5, 625 00 Brno, Czech Republic
2 Czech Geological Survey, Leitnerova 22, 658 69 Brno, Czech Republic
3 Max Planck Institute for Chemistry, Hahn-Meitner-Weg 1, 55128 Mainz, Germany
S1 Methodology - Gas-particle partitioning models
S1.1 Input data
Temperature dependent Koa (PAH: Ma et al., 2010; Odabasi et al., 2006; PCB: Li et al., 2003; OCP: Shen and Wania. 2005; Xiao et al., 2004) and pL data (PAH: Paasivirta et al., 1999; Lei et al., 2002; PCB: Li et al., 2003; OCP: Paasivirta et al., 1999) are used. using the 7-day mean temperature.
The particulate phase BC content was estimated based on the long-term mean EC/OM ratio in various seasons and depending on air mass origin in the region. To this end, size-resolved (5 impactor stages) seasonally and air mass history categorized 24-h aerosol data from Melpitz (51°32'N/12°56'E), covering the years 2004-09 (Spindler et al., 2012) were used. The 6 size classes were correspondingly allocated to the 5 size classes used by Spindler et al., 2012: uppermost (1st) impactor stage (i.e. 3.5-10 µm) to the uppermost 2 stages (i.e. 1st and 2nd. 3.5-10 and 1.2-3 µm, respectively), 2nd to 3rd (i.e., 1.2-3.5 µm to 1.5-3 µm), 3rd to both 4th and 5th (i.e., 0.42-1.2 µm to 0.95-1.5 and 0.45-0.9 µm), and 4th and 5th according to their mass distribution combined into the 6th (i.e., 0.14-0.42 and 0.05-0.14 µm to < 0.45 µm). Daily values of EC/OM vary strongly. but the longer sampling time of our data set (weekly) reduces the related uncertainty significantly. to an estimated less than a factor of 2. Air mass history classification was done based on backward trajectory calculation (120 h) which in turn was based on analysed wind fields (HYSPLIT model (NOAA; Draxler and Rolph. 2003), hence, similar as done by Spindler et al., 2012, was unambiguous for 4 out of the 5 weekly samples, while all advection sectors were influencing 1 sample (13.-20.11.2007, urban site Kotlarska). The mean of the EC/OC ratios for westerly and easterly advection in winter (Spindler et al., 2012) was adopted to estimate EC content of this sample. Local sources might have been more influential at this site leading to EC/OM deviating from the background. This effect is considered to be small, as even at inner city sites in the region dominated by traffic EC/OM was found to be determined by the background (John and Kuhlbusch. 2004; LFUG. 2005).
S1.2 ppLFER method
Organic molecule's electron donor and acceptor as well as van der Waals interactions (hexadecane/air partition coefficient is taken as a measure for the latter) are considered for adsorption to aerosol surfaces. These three interactions and in addition a cavity formation term accounting for solvation (the molar volume is taken as a measure) are considered for absorption into the particulate matter organic phase (OM; Roth et al., 2005). Gas-particle partitioning is understood as the additive combination of adsorption and absorption.
Input data:
Particulate phase composition was categorized in 5 components (ammonium salts, NaCl, minerals, OM, BC). In lack of complete data substance parameters of (NH4)2SO4, quartz and 1-octanol (Goss, 1997; Goss and Schwarzenbach, 2002; Schwarzenbach et al., 2003) were taken as surrogate parameters for ammonium salts, minerals, and OM, respectively. Substance parameters of 1-octanol (Goss. 1997) and elemental carbon (Roth et al., 2005) were adopted for OM and BC, respectively. Size resolved individual components' SSAs were derived from the total specific surface (per impactor stage, m² g-1). The specific surfaces of the aerosol samples was calculated using typical specific surface areas (m² g-1) and PM composition (Section 2.3). Respective chosen values are each 10-5 m² g-1 for ammonium salts and NaCl and 255 m² g-1 for both OM and EC. 11 minerals were for simplicity lumped into 3 categories, i.e. minerals of low (1.5×10-5 m² g-1), medium (1.5 m² g-1) and high (114 m² g-1) surface.
The relationship and parameters describing the adsorptive interactions were adopted from Goss, 2001, using regression coefficients for the ppLFER suggested by Roth et al., 2002:
(S1) log Ksurf/air = 0.136 L Ög + 3.67 A HAsurf + 5.13 B HDsurf - 8.47
where L, A and B are the compound descriptors for the logarithm of the hexadecane-air partitioning coefficient, the effective hydrogen bond basicity. and the effective H-bond acidity, respectively. The complementary surface descriptors, Ög, HAsurf, HDsurf are the square root of the van der Waals surface free energy and the H-donor and H-acceptor activities of the surface, respectively.
The substance parameters L, A, and B were adopted from Abraham, 1993, and Atapattu and Poole, 2008. They are humidity dependent (Schwarzenbach et al., 2003; Goss. 2004). Regressions of the humidity dependence were taken from Götz et al., 2007. The measured 7-day mean humidity was taken.
As:
(S2) Ksurf/air((mol/m²)/(mol/m³)) = K*p( ) / (cTSP(µg/m³) S(m²/g) 10-6(g/µg))
We obtain for the gas-particle partitioning coefficient:
(S3) K*p( ) = 10-6(g/µg) Ksurf/air(m) cTSP(µg/m³) S(m²/g)
The relationship and parameters describing the absorptive interaction, i.e. absorption in particulate OM. were adopted from Roth et al., 2005:
(S4) log Kbulk/air(m³/g) = a Khexadecane/air + b HAbulk + c HDbulk + d V - const
Roth et al. 2005 found for Chur urban aerosol (15°C. rh=50%) a =1.2, b=0.77, c=2.36, d=-1.14, const = -6.26, and for Washington urban aerosol (15°C. rh=65%) a =1.14, b=1.44, c=2.64, d=-0.85, const = -6.34. The latter set was chosen as humidity in these measurements was closer to the observations tested here.
As:
(S6) Kbulk/air((mol/g)/(mol/m³)) = [106(µg/g) cp(ng/m³) 10-9(g/ng) / (cTSP(µg/m³) Mg(g/mol))]
/ [10-9(g/ng) ca(ng/m³) / Mg(g/mol)]
We obtain:
(S7) K*p( ) = 10-6(g/µg) Kbulk/air(m³/g) cTSP(µg/m³)
The gas-particle partitioning coefficient, K*p, which includes both the adsorptive and absorptive interactions is available by combination of the adsorptive and absorptive K*p and normalisation with bulk mass and SSA, respectively:
(S8) K*p( ) = S/V(m-1) Kisurf/air(m³/m²) + 10-6(g/µg) cTSP(µg/m³) Kiju(m³/g)
S2 Results
S2.1 Organic contaminants particulate levels
Table S1. Relative standard deviation of PAH, PCB and DDT concentrations in the atmospheric gaseous and particulate phases (PM10). 4 samples per site.
Location / Cement mill / Quarry / Small airport / Traffic junction / Village / TownSum of PAHs / Gaseous / 0% / 64% / 31% / 22% / 33% / 34%
Particulate / 25% / 71% / 31% / 23% / 21% / 19%
Total PCBs / Gaseous / 0% / 41% / 24% / 20% / 49% / 26%
Particulate / 70% / 96% / 67% / 55% / 20% / 31%
Sum of DDT compounds / Gaseous / 0% / 19% / 47% / 30% / 82% / 25%
Particulate / 54% / 57% / 36% / 25% / 30% / 17%
S2.2 Gas-particle partitioning measurements
Table S2. Mass fractions (%) of (a) PAHs and (b) PCBs and OCPs associated with particles <3 µm (PM3) and <0.95 µm (PM0.95) and particulate mass fraction q = cp / (cp + ca) in PM10.
/ Location / Cement mill / Quarry / Small airport / Traffic junction / Village / Town /Sum of PAHs / in PM0.95 (%) / 73 / 75 / 90 / 90 / 87 / 82
in PM3 (%) / 89 / 85 / 98 / 97 / 99 / 98
q / 0.15 / 0.23 / 0.23 / 0.27 / 0.27 / 0.35
2-ring PAHs / in PM0.95 (%) / 35 / 34 / 38 / 55 / 26 / 32
in PM3 (%) / 79 / 63 / 80 / 80 / 64 / 66
q / 0.39 / 0.12 / 0.04 / 0.02 / 0.03 / 0.02
2rPAHs/PAHs / 0.122 / 0.057 / 0.009 / 0.006 / 0.003 / 0.003
3-ring PAHs / in PM0.95 (%) / 53 / 57 / 81 / 84 / 80 / 69
in PM3 (%) / 76 / 77 / 94 / 95 / 98 / 95
q / 0.02 / 0.02 / 0.02 / 0.03 / 0.03 / 0.04
3rPAHs/PAHs / 0.072 / 0.049 / 0.040 / 0.049 / 0.055 / 0.044
4-ring PAHs / in PM0.95 (%) / 75 / 77 / 89 / 90 / 86 / 80
in PM3 (%) / 88 / 86 / 98 / 97 / 99 / 97
q / 0.13 / 0.26 / 0.27 / 0.36 / 0.38 / 0.42
4rPAHs/PAHs / 0.252 / 0.245 / 0.290 / 0.335 / 0.403 / 0.346
5-ring PAHs / in PM0.95 (%) / 82 / 80 / 91 / 91 / 88 / 85
in PM3 (%) / 93 / 88 / 99 / 98 / 99 / 98
q / 0.80 / 0.87 / 0.93 / 0.95 / 0.97 / 0.95
5rPAHs/PAHs / 0.378 / 0.446 / 0.446 / 0.409 / 0.389 / 0.431
6-ring PAHs / in PM0.95 (%) / 86 / 81 / 92 / 91 / 88 / 85
in PM3 (%) / 94 / 88 / 99 / 98 / 99 / 98
q / 0.98 / 0.98 / 1.00 / 0.99 / 1.00 / 0.98
6rPAHs/PAHs / 0.149 / 0.174 / 0.185 / 0.159 / 0.126 / 0.147
7-ring PAHs / in PM0.95 (%) / 86 / 81 / 94 / 91 / 87 / 86
in PM3 (%) / 92 / 87 / 99 / 98 / 99 / 98
q / 0.94 / 0.96 / 0.99 / 0.100 / 1.00 / 0.99
7rPAHs/PAHs / 0.019 / 0.024 / 0.030 / 0.041 / 0.024 / 0.030
b.
total PCBs / in PM0.95 (%) / 38 / 34 / 54 / 74 / 71 / 59
in PM3 (%) / 69 / 67 / 78 / 87 / 87 / 81
q / 0.05 / 0.06 / 0.08 / 0.11 / 0.26 / 0.02
Cl3PCBs / in PM0.95 (%) / 33 / 33 / 32 / 48 / 36 / 33
in PM3 (%) / 67 / 67 / 64 / 73 / 68 / 67
q / 0.02 / 0.02 / 0.02 / 0.01 / 0.05 / 0.02
Cl3PCBs/totPCB / 0.12 / 0.14 / 0.09 / 0.04 / 0.05 / 0.07
Cl4PCBs / in PM0.95 (%) / 36 / 35 / 35 / 52 / 56 / 42
in PM3 (%) / 71 / 68 / 65 / 78 / 81 / 71
q / 0.04 / 0.04 / 0.04 / 0.04 / 0.18 / 0.06
Cl4PCBs/totPCB / 0.15 / 0.14 / 0.09 / 0.06 / 0.12 / 0.12
Cl5PCBs / in PM0.95 (%) / 36 / 33 / 38 / 59 / 57 / 42
in PM3 (%) / 68 / 67 / 69 / 78 / 79 / 71
q / 0.06 / 0.12 / 0.10 / 0.19 / 0.24 / 0.15
Cl5PCBs/totPCB / 0.26 / 0.27 / 0.18 / 0.10 / 0.16 / 0.17
Cl6PCBs / in PM0.95 (%) / 40 / 33 / 59 / 74 / 77 / 64
in PM3 (%) / 68 / 66 / 82 / 86 / 91 / 84
q / 6 / 9 / 14 / 30 / 37 / 22
Cl6PCBs/totPCB / 0,313 / 0,306 / 0,373 / 0,391 / 0,410 / 0,347
Cl7PCBs / in PM0.95 (%) / 43 / 35 / 73 / 83 / 83 / 75
in PM3 (%) / 70 / 68 / 88 / 92 / 94 / 91
q / 0.12 / 0.19 / 0.36 / 0.69 / 0.71 / 0.54
Cl7PCBs/totPCB / 0.16 / 0.14 / 0.27 / 0.41 / 0.26 / 0.29
in PM0.95 (%) / 39 / 36 / 55 / 74 / 81 / 73
Sum of DDT compounds / in PM3 (%) / 69 / 67 / 78 / 87 / 93 / 89
q / 0.04 / 0.05 / 0.06 / 0.21 / 0.24 / 0.12
DDE/DDT / 1.30 / 1.44 / 1.82 / 0.61 / 0.97 / 0.68
DDE / in PM0.95 (%) / 40 / 39 / 59 / 63 / 79 / 71
in PM3 (%) / 69 / 67 / 79 / 82 / 94 / 89
q / 0.02 / 0.03 / 0.04 / 0.10 / 0.14 / 0.06
DDD / in PM0.95 (%) / 36 / 33 / 48 / 60 / 61 / 54
in PM3 (%) / 68 / 67 / 75 / 80 / 83 / 78
q / 0.07 / 0.15 / 0.18 / 0.31 / 0.35 / 0.24
DDT / in PM0.95 (%) / 41 / 34 / 56 / 84 / 86 / 77
in PM3 (%) / 70 / 67 / 78 / 92 / 94 / 91
q / 0.08 / 0.15 / 0.21 / 0.52 / 0.72 / 0.28
S2.2 Gas-particle partitioning modelling
Table S3. Total particulate mass fractions, q, and coefficients of correlation, R², between predicted (various models of gas-particle partitioning) and observed values of particle size fraction specific partitioning coefficients K*p i (sizes < 1.5 µm a.ed., i = 4…6. K*p = SK*p i) and particulate mass fractions (qi = cp i / (cp i + ca), in brackets) of (a) seven PAHs and (b) two PCBs and DDE. 5 sites, each 1 sample. JP = Junge-Pankow (Pankow, 1987), Koa = Harner and Bidleman (1998), LL = Lohmann and Lammel (2004), ppLFER = Goss and Schwarzenbach (2002). n.a. = not applicable.
Model / q / JP / Koa / LL / ppLFERFLN / 0.04 / 0.40 (-0.04) / 0.40 (-0.07) / 0.48 (-0.03) / 0.63 (0.33)
PHE / 0.10 / 0.14 (-0.06) / 0.14 (-0.10) / 0.22 (-0.08) / n.d.(1)
ANT / 0.03 / 0.38 (0.37) / 0.41 (0.18) / 0.50 (0.30) / 0.48 (0.58)
FLT / 0.30 / 0.19 (-0.02) / 0.19 (0.02) / 0.26 (0.05) / 0.32 (0.29)
PYR / 0.22 / 0.84 (0.90) / 0.83 (0.80) / 0.83 (0.85) / 0.39 (0.75)
BAA / 0.85 / 0.85 (0.78) / 0.86 (0.82) / 0.86 (0.80) (2) / 0.15 (-0.10)
CHR / 0.80 / 0.89 (0.68) / 0.88 (0.85) / 0.84 (0.84) / 0.16 (-0.11)
PCB138 / 0.39 / 0.89 (-0.41) / 0.91 (-0.05) / n.a. / 0.39 (0.10)
PCB153 / 0.25 / 0.87 (-0.28) / 0.89 (0.14) / n.a. / 0.32 (0.25)
DDE / 0.17 / 0.90 (-0.34) / 0.93 (-0.14) / n.a. / 0.27 (0.12)
(1) lack of input data
(2) no temperature dependence of Koa derived from Kow. pL(T) and s(T)
(3) Ksa estimated based on a regression for PAHs. log Ksa = -1.0667 log pL + 5.905 (Lohmann. personal communication)
Fig. S1: Predicted (various models of gas-particle partitioning) vs. observed values of particle size fraction specific partitioning coefficients K*p i (sizes < 1.5 µm a.e.d., i = 4…6, K*p = SK*p i) of semivolatile (a) PAHs and (b) PCBs and DDE, JP = Junge-Pankow, LL = Lohmann and Lammel. Data pairs of the total (PM10) partitioning coefficients, K*p, are included.
a.
b.