Modeling of Gas-Wall Partitioning of Organic Compounds Using a Quantitative Structure-Activity Relationship

This study streamlines modeling of the gas–wall process (GWP) of semivolatile organic compounds (SVOC) by predicting gas–wall equilibrium partitioning constant (Kw,i ) and accommodation coefficient (αw,i ) of SVOC(i) using a quantitative structure–activity relationship. PaDEL-Descriptor, software that calculates molecular descriptors, is employed to obtain physicochemical parameters (i.e., hydrogen bond acidity (Hd,i), hydrogen bond basicity (Ha,i), dipolarity/polarizability (Si), and polarizability (αi)) of SVOC(i). For the prediction of Kw,i, activity coefficients (γw,i) of SVOC(i) to the chamber wall 10 are semiempirically predicted using chamber data in the form of a polynomial equation coupled with the physicochemical parameters. γw,i of various SVOCs differ in functionalities and molecular sizes ranging from 10 0 to 10. We conclude that the estimation of γw,i is essential to improve the prediction of Kw,i. To predict the impact of relative humidity (RH) on GWP, each coefficient in the polynomial equation for ln(Kw,i) was correlated to RH. Increasing RH enhanced GWP significantly for all polar SVOCs. For example, the predicted Kw,i of 1-heptanoic acid increased more than three times (from 0.58 to 1.96) 15 by increasing RH from 0.4 to 0.75 due to the reduction in γw,i. The characteristic time for GWP are estimated using Kw,i and αw,i to evaluate the effect of GWP on secondary organic aerosol (SOA) mass. It might be significant in the absence of inorganic aerosol, but insignificant in the presence of electrolytic salts, where aqueous reactions dominate SOA growth.

This process was repeated three times. The extracted sample was concentrated by blowing the solvent down to ~6 mL with rotary evaporation (Heidolph Rotary Evaporator Laborota 4001, Schwabach, Germany) at 80 ºC and farther to ~0.7 mL using a six-port evaporator (Sigma-Aldrich, St. Louis, MO, USA). Th magnesium sulfate powders (50 mg) were added to the concentrated extracts to remove the water, after which the samples were split into two GC vials.
A 2 μL denuder-extracted sample was injected in the on-column mode to GC/MS. The column oven temperature was held at 65 °C for 0.5 min, and then ramped to 100 °C at a 15 °C min -1 gradient and held for 2.5 min, and ramped finally to 280 °C at a 12 °C min -1 gradient and held for 8 min. The individual compounds in the particle extracts were identified using an external standard and analyzed tentatively using the National Institute of Standards and Technology (NIST) library.

Section S2. Absorbing organic matter (OMwall) and hygroscopicity of OMwall
Chemical composition of organic composition of ( − ) In Fig. S2, the two different FTIR spectra obtained in different time shows the similar compositions, suggesting that the organic composition of ( − ) can be fixed in a certain functional composition.

Hygroscopicity of OMwall
The hygroscopicity of solvent extracted OMwall was determined using an FTIR spectrometer (Nicolet Magma 560, Madison, WI, USA) interfaced with a specifically fabricated optical flow chamber by controlling the relative humidity in the range from 0.1 to 0.8. Humidity was controlled by combining humid air from a water bubbler and dry air from a dry air tank (Jang et al., 2010;Beardsley et al., 2013;Zhong and Jang, 2014). Calibration curves for water content (Mwall-water, mg m -3 ) of OMwall were obtained by the same method as Jang et al. (2010). The FTIR absorbance of the water peak (Abswater, 3350 cm -1 ) at different RH and the water mass fraction predicted from an inorganic thermodynamic model were related to determine the calibration curve. The water content estimated with the calibration curve was converted to the Mwall-water of this study with the volume, surface area of the chamber, and the surface area of the Teflon film (0.4 m 2 ) used to extract OMwall. Fig. S3 illustrates hygroscopicity of OMwall. Figure S3. The water content of extracted Mwall-OM from the surface of the FEP Teflon film (20 cm x 20 cm) was measured using the FTIR spectrometer interfaced with the specially designed flow tube to control relative humidity between 0.05 to 0.80. The Teflon film's surface area (0.4 m 2 ), the UF-APHOR chamber's surface area (86 m 2 ), and volume (52 m 2 ) were applied as parameters to obtain the water content of the dry organic matter. The error bar associated with the water mass was estimated with the uncertainty in FTIR absorbance and aerosol mass.

Mass concentration of Mwall-OM and its molecualr weight
Mwall-OM was measured in two different ways; extracted from the Teflon film with organic solvent and extracted from the chamber inside by wiping the wall with acetone. The Teflon film was extracted with organic solvents (acetone and dichloromethane) and the mass was obtained from the difference in mass before and after extraction. Mwall-OM mass was fit at 18.52 mg m -3 , which was obtained by averaging the mass of extracted Mwall-OM measured (11.81 mg m -3 and 25.22 mg m -3 ). Mwall-OM was collected by wiping the Teflon film over the surface area (20 cm  20 cm) with an acetonedrenched nylon filter. Extracted OMwall was concentrated onto a FTIR silicone disk using clean air and the mass was obtained from the difference weight of before and after placing the sample on the FTIR disc using an analytical balance (Mettler Toledo MX5 microbalance, Columbus, OH, USA). The Mwall-OM mass (6.9 mg m -3 ) extracted via wiping the Teflon film may be biased negatively because the wiping method can be imperfect. Thus, the Mwall-OM used in this study was set at 18.52 mg m -3 . The molecular weight of OMwall (MWOM) was estimated theoretically using the group contribution method (Barton, 1991;Jang et al., 1997;Jang and Kamens, 1998) (2010) To determine the coefficient for , , the backward elimination approach was performed by analyzing the R 2 , adjusted R 2 , and p-value. The processed statistical outcomes are shown in Table S4.

Section S5. Calculation of major atmospheric processes' characteristic time
"or" and "in" denote organic phase and inorganic aqueous phase. K or,i and K in,i are the partitioning coefficients of gas/or and gas/in, respectively. OMT and Min are particulate concentrations (µg m -3 ) of organic aerosol and inorganic aerosol, respectively. kor and kin are the reaction rate constants of SVOC in or and in phase, respectively.
s to h a) λ is the mean free path of air (=6.8 μm) and Dp and Np are assumed to be 100 nm, and 5× 10 4 cm -3 , respectively (Bowman et al., 1997).  The names of the structure in toluene products are from MCM products (Master Chemical Mechanism (MCM) version 3.3.1 (Jenkin et al., 2012)).

S12
c) The reactivity scale of SVOC was sourced from the UNIPAR SOA model. The rank of the reactivity scale is as follows: VF (Very Fast)>F (Fast)>M (Medium)>S (Slow)>P (no reaction).