Interactive comment on “ Molecular corridors and parameterizations of volatility in the evolution of organic aerosols ” by Y

I enjoyed very much reading the interesting paper. However, I am missing a paragraph discussing and referring the readers to the problem that our knowledge of saturation pressures of low-volatility compounds are limited because of measurement challenges, e.g. Bilde et al. (2015). All estimation methods rely on training sets of well-established vapor pressures. Those are typically biased toward compounds with saturation vapor pressures in the range of 103 to 105 Pa. For partitioning we are however, mostly interested in compounds with saturation vapor pressures in the range of about 10−7 Pa to 1 Pa (O’Meara et al. 2014). Those with larger saturation pressure are entirely in the gas phase whereas those with lower saturation pressures will partition entirely into the aerosol. The authors state that the EPI Suite software is “...accepted as


Introduction
Organic aerosols (OA) consist of a myriad of chemical species and account for a substantial mass fraction (20-90 %) of the total submicron particles in the troposphere (Jimenez et al., 2009;Nizkorodov et al., 2011).They influence regional and global climate by affecting radiative budget of the atmosphere and serving as nuclei for cloud Figures droplets and ice crystals (Kanakidou et al., 2005).OA play a central role in air quality by causing haze formation in urban air (Huang et al., 2014;Fuzzi et al., 2015;Zhang et al., 2015) and inducing adverse health effects (Pöschl and Shiraiwa, 2015).OA are introduced into the atmosphere either by being directly emitted from fossil fuel combustion and biomass burning, or formed by multigenerational oxidation of gaseous precursors.
Secondary organic aerosols (SOA) pose a wide range of volatility, hygroscopicity and reactivity (Hallquist et al., 2009).SOA evolution is a complex process involving both chemical reaction and mass transport in the gas and particle phases (Kroll and Seinfeld, 2008;Ziemann and Atkinson, 2012;Shiraiwa et al., 2013), but most aerosol models do not resolve multiphase processes explicitly.
Several two-dimensional (2-D) frameworks have been proposed for efficient SOA representation in chemical transport models.These 2-D frameworks were built based mainly on OA properties including volatility or saturation mass concentration, number of carbon and oxygen atoms in a molecule, mean carbon oxidation state, and atomic O : C or H : C ratios (Donahue et al., 2006;Jimenez et al., 2009;Pankow and Barsanti, 2009;Heald et al., 2010;Donahue et al., 2011;Kroll et al., 2011;Cappa and Wilson, 2012;Zhang and Seinfeld, 2013).The volatility basis set (VBS) approach uses volatility and O : C ratio that can be constrained by chamber experiments (Donahue et al., 2006(Donahue et al., , 2011(Donahue et al., , 2012)).The total organic mass is classified into volatility bins and their distribution between gas and aerosol phases is calculated according to absorptive equilibrium partitioning assuming that gas-phase formation of semi-volatile organic compounds as a limiting step of SOA formation (Pankow, 1994).VBS has been extensively applied in chemical transport models, improving prediction of organic aerosol concentrations (Lane et al., 2008;Tsimpidi et al., 2010;Shrivastava et al., 2011;Ahmadov et al., 2012;Jathar et al., 2012;Murphy et al., 2012;Matsui et al., 2014;Morino et al., 2014).
Volatility is a consequence of molecular characteristics of molar mass and chemical composition and structure.Even though molar mass is an explicit parameter in computing absorptive SOA partitioning (Pankow, 1994;Pankow and Barsanti, 2009), the current VBS method does not account for the dependence of volatility on molar mass, Figures assuming that the products distributed in all volatility bins have the same value of molar mass, e.g., 150 g mol −1 for anthropogenic SOA and 180 g mol −1 for biogenic SOA (Murphy and Pandis, 2009;Hayes et al., 2015).Shiraiwa et al. (2014) have shown that SOA from a variety of biogenic and anthropogenic precursors can be represented well by molecular corridors with a tight inverse correlation between molar mass (M) and volatility (C o ) of SOA oxidation products.The slope of these corridors corresponds to the increase in molar mass required to decrease volatility by one order of magnitude.Molecular corridors can help to constrain chemical and physical properties as well as reaction rates and pathways with characteristic kinetic regimes of reaction-, diffusion-, or accommodation-limited multiphase chemical kinetics involved in SOA evolution (Shiraiwa et al., 2014).

Conclusions References
Tables Figures

Back Close
Full Although nitrogen-and sulfur-containing organic compounds are important components of atmospheric aerosols, their physical properties and chemical behaviors are still poorly understood (Nozière et al., 2015), and thus are untreated in air quality models so far.In this study, we apply the molecular corridor-based approach to analyze ambient and indoor observations of organic aerosols including nitrogen and sulfur containing organic compounds, to provide insights on the physical and chemical processes driving OA evolution in the atmosphere.

Characterization of 31 066 organic compounds in molecular corridors
More than thirty thousand organic compounds were analyzed to examine whether molecular corridors can constrain a variety of organic compounds.The used dataset is the National Cancer Institute (NCI) open database (http://cactus.nci.nih.gov/download/nci/), which contains 31 066 organic compounds and corresponding SMILES codes.A list of the compounds was also provided in Wei et al. (2012).Volatility of each compound was estimated by the Estimation Programs Interface (EPI) Suite software (version 4.1) developed by the U.S. Environmental Protection Agency, which is accepted as a good estimation method of volatility of organic compounds including nitrogen-and sulfur-containing compounds (http://www.epa.gov/sab/pdf/sab-07-011.pdf).
We classified these organic compounds into six classes based on chemical composition: CH, CHO, CHN, CHON, CHOS, and CHONS, with the number of compounds of 328, 8420, 2968, 13 628, 925, and 3367, respectively.These compounds cover a molar mass from 41 to 1779 g mol −1 .In total we consider 22 structural sub-classes including N-containing compounds of amine (primary, secondary, tertiary, and quaternary), amide, azo, azide, amino acid, imine, nitroso, nitro, alkyl nitrite, nitrile and organonitrate as well as S-containing compounds of organosulfate, sulfonate, sulfone, sulfoxide, sulfite, heterocyclic ring, thioate, and thiocarbamate.It indicates that the 2-D space of molar mass and volatility can constrain most of the organics, including compounds containing heteroatoms of N and S.About 1000 compounds with branched structures lie above the linear alkane line among the CHO class (Fig. 1b), as volatilities of branched compounds are higher than those of compounds with linear structures.The compounds with high O : C are organosulfate and sulfite (the mean O : C of ∼ 1), followed by compounds in sub-classes of sulfonate, sulfone, sulfoxide, organonitrate, nitro, alkyl nitrites, nitroso, or amino acid, which tend to occupy the space close to the sugar alcohol line.
For the impact of molar mass on volatility, Fig. 1c shows clearly that mean values of volatility decrease as more hydrogen atoms of an ammonia structure are replaced by alkyl or aryl groups, forming the primary-, secondary-, tertiary-and quaternary-amine.Figure 1d-f show that the molar mass of CHON, CHOS, and CHONS can be up to 1000 g mol −1 and the volatility can be as low as 10 −30 µg m −3 .
The molar mass of VOC is usually less than 160 g mol −1 ; 130-250 g mol −1 for IVOC, 180-330 g mol −1 for SVOC, 210-400 g mol −1 for LVOC, and larger than 260 g mol −1 for ELVOC.As the volatility decreases, the covered molar mass range becomes wider, indicating more compounds are encompassed with increasing complexity.Meanwhile, the volatility decreases as the average molar mass of every composition class increases.For example, the average molar mass of the CHO class in the VOC group is 130 g mol CHO and CHN compounds are populated in the IVOC and SVOC groups.More than 80 % of the CHON and CHOS compounds are located in the range covering from IVOC to LVOC, and about 10 % of them belong to the ELVOC group.90 % of the CHONS compounds cover the range from SVOC to ELVOC.Given that the molar mass is observationally constrained, it is suitable to use the molar mass to constrain the volatility of organic compounds.

Parameterization of volatility by elemental composition
Accurate prediction of volatility requires structural information of the organic compounds, which is often difficult to be obtained in field measurements.We predict volatility as a function of elemental composition that is often determined by soft-ionization high-resolution mass spectrometry.Donahue et al. (2011) has developed a parameterization to estimate volatility as log 10 C o = f (n C , n O ) for the CHO compounds.We broaden this formulation to log 10 to be applicable to the N and S-containing compounds:  (Donahue et al., 2011).The non-linear terms of nitrogen-oxygen and sulfuroxygen are not considered, as they impact the predicting results only slightly.Values of b coefficients were obtained by fitting with multi-linear least squares analysis to the thirty thousand compounds for each class (CH, CHO, CHN, CHON, CHOS, and CHONS).The best-fit parameters obtained in this analysis are presented in Table 1.Introduction

Conclusions References
Tables Figures

Back Close
Full The developed volatility estimation method was compared with the EPI Suite for the thirty thousand compounds as shown in Fig. 3 and Table S1 in the Supplement.The agreement between the two methods was assessed by statistical measures of correlation coefficient (R), mean absolute gross error (MAGE), and mean bias (MB).As shown in Fig. 3, our new method performs well with R above 0.8 for all composition classes.Some of variations at a low C o may be related to uncertainty in converting a super-cooled liquid value to a solid-phase value, or the structural differences in the compounds that cannot be captured by our expression.Given that compounds with a low C o , e.g.LVOC and ELVOC, are almost in the condensed phase under typical ambient conditions, relatively large estimation errors at a low C o will not exert a significant impact on the calculation of SOA partitioning.Our method works well with good statistical measures (R > 0.8, MAGE < 1.7 µg m −3 , MB = −0.06-0.08 µg m −3 ) for compounds with the molar mass less than 500 g mol −1 .
In the CHN class, our method performs well for primary, secondary, and tertiary amines, heterocyclic rings, imines and nitriles (Table S1).Our method overestimates the volatility of the quaternary amine located in LVOC and ELVOC ranges.For the CHON class, relatively large errors are found for quaternary amine and amino acid.Note that there are relatively few data for organonitrate.When nitrate functionality appears in the amine, they are assigned to the amine class.Nitroso and alkyl nitrite are treated similarly.For structural classes in the CHOS and CHONS, our method works well with R > 0.82.
Our method was also tested with a different set of organic compounds and a different volatility estimation method.We used 704 SOA oxidation products of CHO compounds (Shiraiwa et al., 2014) with volatility estimated using the EVAPORATION model (Compernolle et al., 2011).As shown in Fig. 4, our newly developed parameterization also agrees well with EVAPORATION predictions and shows better agreement than Donahue et al. (2011).The above comparisons and validations show that the new volatility estimation parameterization derived from a large dataset is sufficiently good to predict C o for various structural organic classes.Thus, this new method is suitable to be used Introduction

Conclusions References
Tables Figures

Back Close
Full to predict the volatility of ambient OA, e.g., the compounds with elemental composition measured by high resolution mass spectrometry techniques.

Application of molecular corridors to atmospheric aerosols
Applying the newly developed volatility estimation method to laboratory experiments and field campaigns, the observed organic compounds were mapped into the molecular corridor with an alternative representation displaying volatility as a function of molar mass, which appears more straightforward for direct comparisons to mass spectra (Shiraiwa et al., 2014).The used observation dataset is summarized in Table 2.In total 9053 organic compounds were collected from chamber experiments for new particle formation (Ehn et al., 2012;Schobesberger et al., 2013) and field measurements at a boreal forest (Ehn et al., 2010(Ehn et al., , 2012)), at a mountain site (Holzinger et al., 2010), in urban areas (Lin et al., 2012;Stone et al., 2012;Ma et al., 2014;O'Brien et al., 2014;Tao et al., 2014), in biomass burning plumes (Laskin et al., 2009), in radiation fog (Mazzoleni et al., 2010), in super-cooled cloud (Zhao et al., 2013), and in rain (Altieri et al., 2009a(Altieri et al., , b, 2012) ) as well as indoor measurements of OA originated from tobacco smoke (Sleiman et al., 2010a, b) and human skin lipids (Wisthaler and Weschler, 2010).The organic compounds were categorized into CH (55), CHO (3042), CHN (152), CHON (4074), CHOS (954), and CHONS (776) classes.These large data sets provide insights into the chemical and physical nature of OA from different sources and their evolution upon chemical transformation.servation events.Figure 5a shows abundant organic compounds found in new particle formation (NPF) experiments performed at the Cosmics Leaving Outdoor Droplets (CLOUD) chamber at CERN (Schobesberger et al., 2013) and the Jülich Plant Atmosphere Chamber (Mentel et al., 2009;Ehn et al., 2012).The values of O : C ratio are all above 0.3 and some organic compounds are remarkably highly oxidized (O : C up to 1.4).The compounds cover the mass range of 184-558 g mol −1 in the volatility range of IVOC-ELVOC, spreading in a large space of the molecular corridor.These oxidized organic compounds play an important role in formation and growth of OA particles in ambient conditions (Ehn et al., 2012(Ehn et al., , 2014;;Riipinen et al., 2012;Kulmala et al., 2013;Schobesberger et al., 2013;Riccobono et al., 2014;Wildt et al., 2014).

CH and CHO compounds
Figure 5b shows highly oxidized compounds observed at the boreal forest research station in Hyytiälä, Finland during NPF events (Ehn et al., 2010(Ehn et al., , 2012)).The average O : C ratio is as high as 1.1.The locations of organic compounds observed in the CLOUD experiments (Fig. 5a) and at Hyytiälä (Fig. 5b) in the molecular corridor are similar by occupying the space close to the sugar alcohol line (high O : C corridor; Shiraiwa et al., 2014), indicating that chemical properties of these organic compounds are similar and chamber experiments represent ambient observations well.Such lowvolatility and highly oxygenated compounds may be generated by autoxidation reactions (Crounse et al., 2013;Ehn et al., 2014;Mentel et al., 2015).
Figure 5c shows organic compounds observed in the remote area at the Mt.
Sonnblick, Austria (Holzinger et al., 2010).The observed compounds were less oxidized and lie close to the alkane line.Aqueous-phase processing of organic compounds leads to formation of highly oxidized compounds in fog (Fig. 5d; Mazzoleni et al., 2010), cloud (Fig. 5e; Zhao et al., 2013) and rain (Fig. 5f; Altieri et al., 2009aAltieri et al., , b, 2012) ) suggesting that atmospheric water is enriched in polar compounds compared to atmospheric particulate matter.Higher oxidized compounds tend to have lower molar mass than less oxidized compounds, indicating that fragmentation is an important pathway in aqueous-phase oxidation (Sun et al., 2010;McNeill et al., 2012;Carlton and Trupin, 2013;Daumit et al., 2014;Erven, 2015).In addition, aqueous processing can Introduction

Conclusions References
Tables Figures

Back Close
Full produce high molar-mass and high O : C compounds through oligomerization (Altieri et al., 2008;Lim et al., 2010;Ervens et al., 2011), as seen in super-cooled cloud water collected at a remote site on the Mt.Werner in the US in Fig. 5e. Figure 5g shows the oxidation products from reactions of ozone and human skin lipids (Wisthaler and Weschler, 2010).The majority of products are VOC and IVOC with O : C < 0.7 and molar mass < 200 g mol −1 , mainly occupying the space close to the origin of the molecular corridor.Some products with higher molar mass (> 300 g mol −1 ) are mainly first-generation products of ozonolysis of skin lipids, including hydroxyl geranyl acetone, polyunsaturated aldehydes and fatty acids.These products can be further oxidized by ozone, generating fragmented secondary products with a relatively higher O : C ratio with carbonyl, carboxyl, or α-hydroxy ketone groups (Wisthaler and Weschler, 2010).
Cigarette smoke is another pollutant frequently encountered in indoor air and residual secondhand tobacco smoke absorbed to indoor surfaces can react with atmospheric species such as ozone (Destaillats et al., 2006;Sleiman et al., 2010a) and nitrous acid (HONO) (Sleiman et al., 2010b) to form thirdhand smoke hazards.As shown in Fig. 5h, most of the nicotine products have M < 300 g mol −1 and O : C < 0.5 with log 10 C o > 2 µg m −3 .As oxidant levels in indoor air are relatively low compared to outdoor air (Weschler, 2011), indoor OA tend to be less oxidized and have lower molar mass.

Nitrogen-and sulfur-containing compounds
Figure 6 shows N-containing compounds plotted in the molecular corridor.Figure 6a shows 23 compounds observed at Hyytiälä, Finland during NPF events (Ehn et al., 2010(Ehn et al., , 2012)).They are mainly amines with a small molar mass range (69-169 g mol −1 ) and intermediate volatility, covering the space close to the origin of the molecular corridor.Amines can stabilize sulfuric acid clusters efficiently and their role in nucleation may be significant (Loukonen et al., 2010;Smith et al., 2010;Wang et al., 2010;Erupe et al., 2011;Zhang et al., 2011;Kulmala et al., 2014).Introduction

Conclusions References
Tables Figures

Back Close
Full N-containing organic compounds are important components of biomass-burning organic aerosols (BBOA) (Lobert et al., 1990;Simoneit, 2002).Figure 6b presents these compounds including amine, urea, alkyl amide, alkyl nitrile, amino acid and Nheterocyclic alkaloid compounds (Laskin et al., 2009).These compounds spread separately in two parts in molecular corridors.Some compounds are assembled in the upper left space bounded by log 10 C o > 0 µg m −3 and M < 300 g mol −1 .A part of these compounds may be a consequence of oxidative fragmentation or thermal decomposition.Compounds clustered in the lower right space are CHON compounds with molar mass higher than 300 g mol −1 covering the range from LVOC to ELVOC.
More than two hundred N-containing compounds were identified at the Mt.Sonnblick (Fig. 6c; Holzinger et al., 2010).Less oxidized N-containing organic compounds could be formed through reactions transforming carbonyls into imines or reactions of NO with organic peroxy radicals (O'Brien et al., 2014).Highly oxidized organonitrates (O : C > 1) were suggested to be formed by nitrate radical chemistry (Holzinger et al., 2010).Many highly oxidized N-containing compounds with a range of O : C ratios from 1 to 2 were observed in fog (Fig. 6d; Mazzoleni et al., 2010), cloud (Fig. 6e; Zhao et al., 2013), rain in New Jersey (Fig. 6f; Altieri et al., 2009a, b) and Bermuda (Fig. 6 g; Altieri et al., 2012).Aqueous-phase processing can form a variety of highly-functionalized nitrated organic compounds including organonitrates, hydroxynitrates, carbonyl nitrates and dinitrates (Zhao et al., 2013).Reduced nitrogen compounds were also identified in atmospheric water, mainly occupying the space close to the alkane line.
Figure 6h shows N-containing compounds found in the secondhand and thirdhand tobacco smoke.Nitrosamines were found in third-hand smoke hazards when the residual nicotine reacts with HONO (Sleiman et al., 2010b).Similar to the trend of CHO compounds in the molecular corridors (Fig. 5h), N-containing compounds also mainly occupy the region close to the origin of the molecular corridor.Some high molar mass N-containing compounds (m/z 400-500) were also detected, but their elemental compositions and structures have not been identified (Sleiman et al., 2010a).Introduction

Conclusions References
Tables Figures

Back Close
Full Figure 7 shows S-containing compounds plotted in the molecular corridor.Organosulfate and nitroxy-organosulfate were frequently identified in fine aerosols especially in urban areas including Shanghai (Ma et al., 2014;Tao et al., 2014) and Guangzhou (Lin et al., 2012) in China, Taipei (Lin et al., 2012), Lahore in Pakistan (Stone et al., 2012), Bakersfield (O'Brien et al., 2014) and Los Angeles (Tao et al., 2014) in the US as shown in Fig. 7a.Most of them are LVOC or ELVOC with O : C ratio higher than 0.8 and molar mass in 200-400 g mol −1 .Some uncommon organosulfates with higher molar mass but a lower degree of oxidation were found in Shanghai and long-chain alkanes from vehicle emissions were suggested to be their precursors (Tao et al., 2014).Compared to the organosulfates identified in urban areas, oxygenated sulfur-containing compounds characterized at the Mt.Sonnblick have a relatively lower O : C ratio (∼ 0.5 in an average) and molar mass (mostly less than 250 g mol −1 ) and higher volatility (IVOC or SVOC) (Fig. 7b).Similar to highly oxidized N-containing compounds observed in fog, cloud and rain, S-containing highly oxidized compounds including functionalized (nitro-and nitrooxy-)organosulfates were formed in atmospheric water (Fig. 7c-f), which locate close to the sugar alcohol line in the range of LVOC and ELVOC.Compounds close to the alkane line are mostly reduced sulfur compounds, e.g., aromatic structures containing only one S and one O, which may be emitted from primary sources.
Figure 8 summarizes the molecular corridor for amine, organonitrate, organosulfate, nitroxy organosulfate and reduced sulfur compounds.The small markers are compounds identified in the studies included in Figs. 6 and 7.Among these species, nitroxyorganosulfates have the highest O : C ratio (> 0.9) and the lowest volatility falling into the ELVOC group with molar mass up to 400 g mol

Summary and conclusions
From the analysis of measured OA locating into the molecular corridor, we can conclude that the molecular corridor characterized by molar mass, volatility, and O : C ratio has successfully grasped the properties of organic compounds from different sources and formed in various atmospheric conditions.Figure 9 shows the trend of observed organic compounds with surrogate compounds with the mean values of molar mass, volatility, and O : C ratio derived from every observation event.The symbol size is scaled with the ratio of the number of compound in each class (e.g., CH, CHO, CHON, CHONS) to the total number of compounds in each observation.
OA in indoor environments have relatively lower molar mass and higher volatility, mainly occupying the space close to the origin of the molecular corridor.Outdoor OA are constrained to a corridor in the range of IVOC-LVOC with a molar mass of up to ∼ 400 g mol −1 .Atmospheric water of fog, cloud and rain droplets often contain many highly-oxygenated, high molar-mass, and low-volatility compounds, extending to a wide space with higher molar mass and lower volatility.Molecular corridors are a useful framework for analysis and interpretation of measurements by a high-resolution mass spectrometer to visualize distribution of organic compounds providing insights into the evolution of OA properties.
Explicit consideration of molar mass in an OA model would also be useful in inferring particle phase state (liquid vs. semisolid vs. amorphous solid), as the molar mass correlates with the glass transition temperature of organic compounds (Koop et al., 2011).The phase state has been shown to affect various gas-particle interactions including heterogeneous and multiphase chemistry, SOA formation and evolution as well as activation to cloud droplets and ice crystals (Pöschl and Shiraiwa, 2015, and references therein).Molecular corridors may serve as a basis for better treatment of SOA proper-Introduction

Conclusions References
Tables Figures

Back Close
Full ties and interpretation of model outputs in detailed SOA models (e.g., Shiraiwa et al., 2012;Cappa et al., 2013;Roldin et al., 2014;Zaveri et al., 2014) as well as for compact representation of OA formation and evolution in regional and global models of climate and air quality.Full  Full

The
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 1
Figure1shows that most of the organic compounds (small markers color-coded by atomic O : C ratio) fall into the molecular corridor with upper and lower boundaries rep- −1 , which increases to 445 g mol −1 in the ELVOC group.The symbol sizes are scaled with relative abundance of compounds in each composition class.Many of the CH compounds (∼ 69 %) are distributed in the VOC and IVOC groups.70 % of the Discussion Paper | Discussion Paper | Discussion Paper | 1) where C o (µg m −3 ) is the volatility or the saturation mass concentration of an organic compound; n 0 C is the reference carbon number; n C , n O , n N , and n S denote the numbers of carbon, oxygen, nitrogen, and sulfur atoms, respectively; b C , b O , b N , and b S denote the contribution of each atom to log 10 C o , respectively, and b CO is the carbon-oxygen nonideality Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 5
Figure 5 shows the CH and CHO compounds measured in different atmospheric conditions plotted in molecular corridors.The small markers show individual observed compounds color-coded by atomic O : C ratio.The larger symbols with error bars represent mean values of molar mass, volatility, and O : C ratio derived from every ob- Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | −1 .Organosulfates and organonitrates have an O : C ratio generally higher than 0.7, covering the range of IVOC to ELVOC with a broad molar mass range (100-600 g mol −1 ) to occupy the high O : C corridor.Reduced sulfur compounds have low O : C ratio (< 0.4) and are located close to the alkane line.Amine and N-heterocyclic alkaloid compounds found during new parti-Discussion Paper | Discussion Paper | Discussion Paper | cle formation and biomass burning have the lowest O : C ratio and molar mass and the highest volatility (in VOC and IVOC groups), following the low O : C corridor.

Figure 1 .Figure 2 .Figure 5 .Figure 6 .Figure 7 .Figure 9 .
Figure 1.Molecular corridors of molar mass vs. volatility for organic compounds in elemental composition classes of (a) CH, (b) CHO, (c) CHN, (d) CHON, (e) CHOS, and (f) CHONS.The data comprise 31 066 compounds from the NCI open database.The dotted lines represent linear alkanes C n H 2n+2 (purple with O : C = 0) and sugar alcohols C n H 2n+2 O n (red with O : C = 1).The small markers correspond to individual compounds identified in each structural subclass (see Sect. 2), color-coded by atomic O : C ratio.The larger symbols indicate the surrogate compounds with the mean values of molar mass, volatility, and O : C ratio computed for each of the structural sub-classes with error bars indicating standard deviations.Note that the data points above the linear alkane line in (b) represent molecules with branched structures.

Table 1 .
Composition classes and the n 0 C and b values for volatility parameterizations obtained by least-squares optimization using the NCI database.

Table 2 .
Summary of the number of organic compounds in the composition classes (CH, CHO, CHN, CHON, CHOS, and CHONS) measured in 11 observation events.