Secondary organic aerosol formation from acetylene ( C 2 H 2 ) : seed e ff ect on SOA yields due to organic photochemistry in the aerosol aqueous phase

Secondary organic aerosol formation from acetylene (C2H2): seed effect on SOA yields due to organic photochemistry in the aerosol aqueous phase R. Volkamer, P. J. Ziemann, and M. J. Molina Dept. of Chemistry and Biochemistry and CIRES, University of Colorado at Boulder, CO, USA Dept. of Chemistry and Biochemistry, U.C. San Diego, La Jolla, CA, USA Air Pollution Research Center, U.C. Riverside, 92521, Riverside, CA, USA Received: 1 July 2008 – Accepted: 5 July 2008 – Published: 5 August 2008 Correspondence to: R. Volkamer (rainer.volkamer@colorado.edu) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Secondary Organic Aerosol (SOA) formation is linked with debates about air quality, visibility, public health, climate, and the oxidative capacity of the atmosphere.Recent experimental findings indicate SOA contributes about 30-60% of the organic component of the aerosols (OA) in urban air, and >70% of OA in rural air and the free troposphere (Zhang et al., 2007).The contribution of SOA to aerosol optical depth (AOD) is estimated to be larger than that of primary organic aerosol (POA) or black carbon (BC) (Tsigaridis et al., 2006).This contribution is likely to increase as (1) current models significantly underestimate SOA production in large and very different compartments of the atmosphere (Heald et al., 2005;de Gouw et al., 2005;Volkamer et al., 2006), (2) current models assume the mass absorption cross-section (MAC) of OA to be zero, while laboratory and field evidence consistently suggests that OA absorbs light (Kirchstetter et al., 2004;Sun et al., 2007;Barnard et al., 2008), and (3) air quality regulations should result into a reduced contribution of sulphate in the future, while SOA precursor emissions are very poorly characterized (Volkamer et al., 2006) and could continue to increase disproportionately compared to sulfate (Tsigaridis and Kanakidou, 2007;Heald et al., 2008).Notably different models agree with respect to the importance of biogenic VOCs for SOA formation on regional and global Published by Copernicus Publications on behalf of the European Geosciences Union.
SOA is defined as the organic aerosol mass formed in the atmosphere by gas-to-particle partitioning of semivolatile products of the atmospheric oxidation of organic vapors.In the widely established view, partitioning theory (Pankow, 1994a) is employed to describe the partitioning of semivolatile organic vapors only between the gas-phase and the organic fraction of aerosol mass (including the aerosol water attracted into that phase from the hygroscopic properties of SOA) (Seinfeld and Pankow, 2003).This view uses as the key input the vapour pressure of gas-phase oxidation products, which decreases with the number of carbon-atoms of the precursor.Early work suggested the existence of a threshold in the number of carbon atoms, n=7, for a precursor to form SOA (toluene forms SOA, benzene does not) (Pandis et al., 1992).However, more recent experimental studies find SOA formation from benzene (n=6) (Martin-Reviejo and Wirtz, 2005) and isoprene (n=5) (Claeys et al., 2004).Some SOA models have also considered partitioning of SOA species to an aerosol water phase (Pun et al., 2003;Griffin et al., 2005).These models only consider the dissolution of the gas-phase reaction products, do not treat aqueous chemistry explicitly, and the mass fraction of SOA that partitions in this way is typically a few percent of the total SOA; the predominant SOA source in current models is from semivolatile vapors.
SOA formation from acetylene (C 2 H 2 , i.e., n=2), the lightest non-methane hydrocarbon (NMHC), questions volatility of the gas-phase oxidation products as the only controlling parameter of SOA formation.C 2 H 2 oxidation products are at least 6 orders of magnitude too volatile to explain partitioning to aerosols based on physical processes alone (i.e., condensation, absorption).These differences are too large to be explained by organic aerosol activity coefficients, which are thought to modify effective vapor pressures by less than one order of magnitude (Bowman and Melton, 2004).However, C 2 H 2 efficiently forms glyoxal (CHOCHO) upon oxidation by OH-radicals (Yeung et al., 2005).In Mexico City, C 2 H 2 oxidation accounts for about 8% to the over-all CHOCHO source (Volkamer et al., 2007).An increasing body of experimental evidence (Karpel vel Leitner and Dore, 1997;Schweitzer et al., 1998;Blando and Turpin, 2000;Sorooshian et al., 2006;Loeffler et al., 2006;Carlton et al., 2007;Altieri et al., 2008a, b) suggests the processing of CHOCHO and CHOCHO precursors in cloud and fog droplets is an important source for SOA (Ervens et al., 2003;Lim et al., 2005;Warneck, 2005;Ervens et al., 2008).CHOCHO uptake to aerosols forms an alternative pathway for SOA formation in the absence of clouds.Compared to clouds, aerosols contain 3-5 orders of magnitude less liquid water, and aerosol water has therefore been easily rejected as "too small" a volume to dissolve a significant amount of gas-phase organics.The experimental evidence for CHO-CHO uptake to particles is more scarce, and far less conclusive (Jang et al., 2002;Kroll et al., 2005;Liggio et al., 2005a;Corrigan et al., 2008).Conflicting evidence exists about the magnitude and mechanism of CHOCHO uptake, as well as the role of acid-catalysis to control CHOCHO reactivity in aerosols.Recent modeling studies deem CHOCHO and methylglyoxal responsible for 5-11 Tg yr −1 SOA formation globally, suggesting this SOA source could be comparable to the sum of SOA formed from monoterpenes, sesquiterpenes, isoprene, and aromatics (Fu et al., 2008;Myriokefalitakis et al., 2008).More than 85% of this SOA is deemed to form in clouds, mostly at altitudes below 1.5 km (Fu et al., 2008), where aerosols are mostly aqueous.
Current state-of-the-art SOA models use empirical fits to simulation chamber data to calculate SOA yields and partitioning coefficients (Odum et al., 1997;Robinson et al., 2007).Conditions in these chambers approximate the polluted atmosphere to a considerable extent, i.e., with respect to VOC/NO x ratios, NO x concentrations, temperatures, the pool of condensable species produced, OH-radical concentrations, light conditions, and oxidation time-scales.However, most SOA observed in chambers is attributed to secondand higher generation oxidation products (Hurley et al., 2001;Kroll and Seinfeld, 2005), while faster SOA formation is observed in the atmosphere (Volkamer et al., 2006;Riddle et al., 2008).Little attention has thus far been paid to the possible effect of the chemical composition of seed aerosols.In particular, most chamber studies of SOA yields to date have either used no seed at all (i.e., homogeneous nucleation of SOA), or added ammonium sulfate (AS) seed to provide a surface area for condensation (Odum et al., 1997;Martin-Reviejo and Wirtz, 2005;Ng et al., 2007;Kroll et al., 2007).Experiments with water containing seed are less common, but have been used to ascertain the effect of ionic strength on the partitioning of semivolatile products (Cocker et al., 2001a, b).More recently it was demonstrated that hydrophobic organic aerosol seed does not participate in the absorption processes that underlie the current concept of vaporpressure driven partitioning of gas-phase oxidation products to the organic aerosol phase (Song et al., 2007), further increasing the discrepancy between SOA measurements and models.There is currently no systematic study of the effect of chemical composition of seed aerosols on SOA yields.
Here we study directly the SOA formation from C 2 H 2 and CHOCHO on different inorganic, organic and mixed inorganic/organic seed aerosols, in a simulation chamber under both photochemical and dark conditions.The relative humidity and chemical seed composition studied resemble an important subset of particles found in the atmosphere.

Experimental
Experiments were conducted in a 7 m 3 indoor smog chamber equipped with 64 blacklights, and made of Teflon, with a surface/volume (S/V) ratio of 3 m −1 .The experimental setup is shown in Fig. 1.The chamber was filled with dry, clean air (Aadco pure air generator, <5 ppbv of hydrocarbons, 0.1% RH), followed by vapour addition of ultrapure water (Milli Q, 18 M ) to the desired relative humidity (RH).For all experiments seed aerosols were added prior to the organic by nebulizing dilute salt or salt/organic solutions, and passing the aerosols via a 210 Po neutralizer connected to the chamber.All chemicals were purchased from Aldrich, unless otherwise noted.The solutions were made up of either pure compounds (0.1 M) or mixtures of equal mass (about 0.1 M for each solute) of ammonium sulfate (AS, purity 99+%), ammonium bisulfate (ABS, Alfa Aesar, 99.9%), Suwannee River fulvic acid (FA, IHSS), Humic acid sodium salt (HA, tech.), and the amino acids (AA) Glycine (>98.5%), and Tyrosine (>98.5%).Succinic acid (SucA) seeds were prepared from vaporizing pure compound (99+%) in a stream of clean air connected to the chamber.C 2 H 2 (Airgass, grade 2.6) was further purified by passing a small flow via an acetone-slush bath held at a temperature T =−85 • C. Concentrations of C 2 H 2 , CH 3 ONO (Taylor et al., 1980), NO (Matheson) were determined from adding known volumes of these gases to the chamber.Typical initial concentrations were [C 2 H 2 ]=20 ppm, [CH 3 ONO]=5 ppm, [NO]=5 ppm.CH 3 ONO/NO/UV-light was used to generate OH-radicals.For CHOCHO, the reversible uptake to chamber walls, mixing fans, or sampling lines, has been identified as an uncertainty if experiments are conducted over extended experiment times (Kroll et al., 2005).This uncertainty was minimized here by homogeneously producing CHOCHO inside the chamber from the photo oxidation of C 2 H 2 with OH-radicals, thus eliminating the need to actively stir or mix the chamber volume, and systematically reducing experiment time.For the dark experiments authentic CHOCHO samples were prepared from flowing He through heated mixtures of glyoxal trimer dihydrate (GTD) with P 2 O 5 .About 1% CHOCHO/He mixtures were stable over several days if stored in the dark.The mixing ratio of CHOCHO in the gas mixture was periodically checked from measuring the distinct absorption features of CHOCHO in the UV and visible spectral range, and comparison to res- olution adjusted spectra of the absorption cross section of CHOCHO (Volkamer et al., 2005b).Known volumes taken from this bulb were added to the chamber.To convince ourselves that the CHOCHO concentration is reasonably constrained in our photochemical experiments, CHOCHO was measured inside the chamber by means of solid-phase microextraction (SPME) with on-fiber derivatization using o-(2, 3, 4, 5, 6-pentafluorobenzyl) hydroxylamine (PFBHA, 98+%) (Baker et al., 2005) in selected experiments.After addition of seed aerosols and chemicals to the chamber the chamber was operated as a static reactor.Temperature, RH, NO, NO y , aerosol size distributions, and in some experiments CO were periodically monitored over the course of an experiment.The Thermo Environmental Instruments NO-NO 2 -NO x analyzer sampling flow was 0.6 l min −1 .Size distributions were measured by a scanning mobility particle sizer (SMPS) system (long DMA and TSI 3010 particle counter).The DMA flow of 3 l min −1 was taken from the chamber; 2.5 l min −1 were filtered and used as sheath flow; 0.5 l min −1 was unfiltered aerosol sampling flow.The RH was measured in the DMA sampling line about mid-way between the chamber and the SMPS.The RH of sheath air was the same as that of the chamber.Diffusion losses in the sampling lines are <1% for diameters >50 nm.The TSI 3010 particle counter flow is 1 l min −1 (0.5 l min −1 aerosol and 0.5 l min −1 filtered make up flow).Number distributions are sampled over a range of 20 to 600 nm mobility diameter about every 120 s.Number and volume concentrations have been compared with a TSI 3936L72 SMPS system, and were found to agree within ±10% (Matsunaga and Ziemann, 2009).Aerosol wall loss was quantified over a period of about an hour prior to each experiment in terms of a volume loss rate τ (τ = ln (V t /V t=o )/ t=1.5 10 −5 s −1 in Fig. 2), and aerosol wall loss was corrected in the further analysis.
Flushing the chamber and filling it with humidified clean air removed NO y to below the detection limit (about 1 ppb; i.e., dilution factor >10 4 ).The dilution ratio was found smaller for CHOCHO, which is consistent with the release of CHOCHO from the walls, as had been observed previously (Kroll et al., 2005).This CHOCHO memory effect was quantified at 3% for the chamber cleaning procedure used between experiments.CHOCHO measurements compared reasonably well (within 30%) with simulations of the C 2 H 2 /CH 3 ONO/NO/light system based on a subset of the Master Chemical Mechanism (Bloss et al., 2005)  The experimental conditions were varied to study photochemical and dark reaction systems, and in terms of CHO-CHO concentration, OH radical concentration, RH, seed volume, seed chemical composition, liquid water content (LWC), and the mass of the organic portion of the seed (M om ), and are listed for individual experiments in Table 1.LWC of seed aerosols was calculated using an additive approach that sums the amount of water attracted from inorganic and organic components of the seed.The Aerosol Inorganic Model (Wexler and Clegg, 2002) was initiated at representative temperature and RH conditions to calculate LWC for inorganic seed components, accounting for metastable aerosols.For organic components the hygroscopic properties were taken from the literature: SucA (Ansari and Pandis, 2000;Peng et al., 2001), FA (Peng et al., 2001), AA (Chan et al., 2005); hygroscopic properties of HA were approximated as those of Suwannee River fulvic acid.M om was calculated from subtracting the LWC from the aerosol volume, and converting the volume portions of individual seed components based on their densities d (d AS =1.77 g cm −3 ; d ABS =1.79 g cm −3 ; d SucA =1.56 g cm −3 ; d FA =1.47 g cm −3 ; d HA =1.72 g cm −3 ; d Glycine =1.46 g cm −3 ; d Tyrosine =1.46 g cm −3 ).The density of 40% w/w glyoxal aqueous solution d Gly40% =1.27 g cm −3 corresponds to a dry equivalent density of glyoxal oligomers of 2.13 g cm −3 .In our calculation of SOA mass formed we converted the observed volume using a density d SOA =2 g cm −3 (a conservatively rounded number for the density of glyoxal oligomers in their dissolved equilibrium distribution).The actual density of oligomers is currently not known.It should be noted, however, that the value of density cancels out with relative comparisons between different experiments; any uncertainty in the density of oligomers thus does not affect the conclusions that are reached from relative arguments.Use of a different density value for oligomers will translate into changes of the absolute SOA yields and H eff values.
The SOA yield, Y SOA , was calculated as the mass of SOA formed divided by the mass of C 2 H 2 reacted.The uncertainty in the SOA yields reflects the combined error from: (1) uncertainty in the wall loss corrected volume change, (2) uncertainty in the amount of C 2 H 2 reacted as predicted by our box model, (3) uncertainty due to CHOCHO wall effects (described above); it does not include uncertainty in the density of CHOCHO-SOA.The uncertainty in (1) and (3) is systematically minimized here from using a photochemical CHO-CHO source (which overcomes the need for mixing), and facilitates reduced experiment times, especially in experiments conducted at higher OH concentrations.

Results
Significant SOA mass is formed from C 2 H 2 during all experiments where a seed was present.Oxidation of C 2 H 2 in the  The oxidation products from C 2 H 2 are very volatile.The products formed in our reaction system are predicted by the model as summarized in Table 2.For room temperature the vapour pressure for all products is well above 1 kPa (CRC Handbook, 2003), i.e., about 6 orders of magnitude too volatile to explain partitioning to aerosols as the result of physical processes (condensation, sorption into an organic liquid).Also shown are the respective Henry's law coefficients, or lower limits in the case of CHOCHO (Kroll et al., 2005;Sander et al., 2006).As is evident from Table 2, CHO-CHO is the only product that partitions to the aerosol aqueous phase in appreciable amounts.Notably, our conservative estimate of H eff,CHOCHO results in a lower limit for the amount of CHOCHO that is expected to partition to aerosols that accounts for more than 99% of the source for aerosol growth.Higher values for H eff have been reported in the presence of sulphate and from field evidence (>10 9 M atm −1 , Volkamer et al., 2007;Ip et al., 2009), and are in general agreement with the results presented here.The change in the size distribution of aqueous seed aerosols is therefore attributed to CHOCHO uptake to the seed aerosols.A, C, E) and liquid water content, LWC (B, D, F).No obvious correlation is found between Y SOA and M om , but Y SOA correlates well with LWC for all seed-types studied.For pure AS (solid line), AS+FA (dashed line), and pure ABS (dotted line) sufficient experiments were conducted to demonstrate that the increase in Y SOA with LWC is linear.R=0.76), and ABS (Y SOA =0.01+0.008×LWC,R=0.99) seeds.However, no obvious correlation was observed between Y SOA and M om .For example, for AS+FA seed among the highest Y SOA values were observed in experiments with a particularly low M om , while among the lowest Y SOA values were observed at moderately high values of M om (Fig. 3a).For some seed types Y SOA even anti-correlates with M om (e.g., ABS+FA in Fig. 3c).We take this as strong evidence that the aqueous aerosol phase is a chemical reactor for SOA formation from C 2 H 2 .

Dependence of SOA yields on seed chemical composition
No new particle formation was observed from the photochemical oxidation of C 2 H 2 .However, Y SOA from C 2 H 2 was non-zero on all seeds studied here.The actual value of Y SOA was found a strong function of the seed chemical composition, as is evident from the considerable scatter of the plotted data points in Fig. 3: Y SOA varied by a factor >20 (i.e., 1.1%≤Y SOA ≤25.2%) over the range of experimental conditions investigated.For comparable LWC, e.g., LWC=10 µg m −3 , Y SOA varied by more than a factor of 5, and increased in the following sequence: AS+FA+AA (3.2%), AS+FA+AA+SA (3.8%), SucA (4.7%), AS (5%), AS+FA+SA (5.8%), ABS (8.8%), ABS+HA (7.2-25.2%),ABS+FA (9.6-21%), AS+FA (17.2%),FA (17.3%).For this comparison the observed linear relation with LWC was used for AS, ABS, FA and AS+FA; for AS+FA+SA available data points were interpolated; for SucA, AS+FA+AA, AS+FA+AA+SA, ABS+FA, and ABS+HA linear interpolation through zero was applied (to derive a lower limit in the latter two cases); linear extrapolation between available data points was applied for ABS+FA, and this data was scaled for ABS+HA (to derive upper limits).
Interestingly, AS seed -by far the most widely used seed in chamber studies of SOA yields to date -showed among the lowest Y SOA from C 2 H 2 of any seed investigated.For aqueous AS Y SOA compares to that observed for SucA.However, addition of water soluble FA had a pronounced effect on Y SOA from AS.At comparable LWC, Y SOA is found to be 3.75 times larger for mixed AS+FA than for AS seed (Fig. 3b).Y SOA from ABS is slightly higher than for AS (Fig. 3d).However, the addition of FA (and HA) had less of an effect on Y SOA for ABS than for AS.For example, at LWC=5 µg m −3 , the Y SOA for AS+FA is about 9.6%, i.e., is larger than for ABS+FA (Y SOA =7.2%), despite the fact that Y SOA is higher for ABS than AS.This suggests that the relevant Y SOA for mixed ABS+FA (and likely ABS+HA) may be closer to the lower limit number of the range extrapolated above.
The lowest yields are observed for seeds containing amino acids (AA) like tyrosine and glycine.Over the timescales accessible by our setup, the presence of AA reduces Y SOA by a factor of 5.3 for AS+FA seed.It is important to note that this conclusion can only be drawn as the LWC for the AA studied here is well known.AA containing seed showed the largest LWC of all experiments conducted, and therefore yielded the largest Y SOA values observed (Fig. 3f).The apparent effect of AA to reduce Y SOA becomes visible only after LWC has been normalized; only then Y SOA is actually comparable to the other seeds studied.The apparent reduction in Y SOA becomes more pronounced over longer timescales.Furthermore, mild acidification with SA leaves Y SOA essentially unaffected for AS+FA+AA seed, but strongly reduces Y SOA for AS+FA seed.Experiments containing AA and SA yielded Y SOA values on the bottom scale of those observed.Similarly low yields were observed on SucA seed.Additional evidence on the role of acid comes from a comparison of ABS+FA and AS+FA seed.ABS behaves more acidic at lower RH, while AS is neutral at any RH.The LWC dependence of Y SOA is primarily determined from varying RH in our work, and hence ABS+FA seed at LWC=3 µg m −3 is more acidic than at LWC=5 µg m −3 .Interestingly, a factor of 3.5 lower Y SOA is observed for ABS+FA and ABS+HA at LWC=3 µg m −3 , while the difference is smaller (factor 1.5) for the less acidic conditions (Fig. 3d).We do not find evidence that would suggest higher Y SOA under acidic conditions.For comparable LWC and over the timescales accessible in our setup Y SOA values from AA, SA or SucA containing seed were lower than Y SOA from AS seed.Most experiments showed type A growth in the presence of light, followed by type B growth in the dark (see Table 1).This is illustrated in Fig. 4 for experiments that all used the same AS+FA seed solution.In some experiments the seed was acidified mildly by addition of AA (tyrosine and glycine), or SA to the AS+FA seed solution (pH∼4 for the AS+FA+SA seed).Panel b shows only experiments that were conducted at different RH, and used the same AS+FA seed, while in panel c experiments with addition of acids are shown.The data from panels a was normalized in panels b and c according to Eq. (1):

3.3
where V t is the observed volume at time t, V seed is the seed volume before the experiment, [Gly] t is the CHOCHO mixing ratio in units of ppb (normalized to 100 ppb here), and %LWC is the percent contribution of LWC to total seed volume (normalized to 50% here).(V norm,t −1) hence represents the volume change in units of percent, if 100 ppb CHOCHO were present in the gas-phase at 75% RH (for AS+FA seed in Fig. 3b; 50% RH for AS, not shown).For experiments #19, #20, and #27 the conditions varied by about a factor of 2 in seed volume, a factor of 2.5 in gas-phase CHOCHO (70-171 ppb) and span a large RH range (16%≤ RH≤86%), resulting in a factor of more than 6 difference in the observed volume change (see Table 1 and Fig. 4a).Interestingly, Eq. ( 1) fully accounts for these significant differences, and the normalized time profiles (Fig. 4b) are indistinguishable within error bars.Similarly, AS, ABS, FA, SucA, mixed ABS+FA and ABS+HA seed showed type A growth, but  the absolute volume change depends on the seed type (see Sect. 3.2 above).Normalization according to Eq. ( 1) accounts for differences in experimental conditions in the case of AS, ABS, FA and AS+FA seed.No dependence was observed on the light level used.Experiments conducted with a factor of 6 reduced light levels gave results that are essentially identical after normalization according to Eq. ( 1).This demonstrates that for these seeds the photochemical volume growth scales directly proportional to the gas-phase CHO-CHO concentration, and the liquid water of the seed, and further suggests that the availability of UV photons does not limit the rate of CHOCHO uptake.After addition of chemicals to the gas-phase, an immediate shift in aerosol volume is observed when C 2 H 2 is photochemically oxidized (twice here), followed by type B growth in the dark.Type B growth as parameterized in Fig. 3 for AS+FA can explain that observed for ABS+HA seed (solid line), see text.The rapid volume growth during periods when the light is turned ON is proportional to the amount of CHOCHO produced, and inversely proportional to LWC of the seed aerosols.
A reference experiment #32 was conducted in the dark chamber, with known amounts of CHOCHO added directly to the chamber.A considerably slower growth rate, and smaller relative volume growth is observed (type B).Notably,experiments #19,#20 (and #12,#17, show a significant volume increase after the lamps were turned OFF.The normalized volume growth rate between dark periods of these experiments agrees quantitatively with that observed in the dark CHOCHO experiment #32, and confirms directly that CHOCHO is a building block for the observed SOA formation.We conclude that type A+B growth is happening for AS+FA seed at high and low RH.Similarly, ABS, FA, mixed ABS+FA and ABS+HA seed showed type A+B uptake behaviour, while AS and SucA seed only showed type A uptake, but did not show continued growth in the dark chamber.Repeated type A volume growth is observed if the lamps were turned ON consecutively, followed by type B uptake during dark periods in between, as illustrated for ABS+HA seed in Fig. 5. Notably, no volume growth is observed if the seed is exposed to light in the absence of reactants in the gas phase (shown for two trials in Fig. 5).The type B growth rate shown in Fig. 5 does not represent a fit to the data, but was calculated using the type B growth rate derived from Fig. 4b for AS+FA seed.Interestingly this rate does explain type B growth rates observed for ABS+HA seed as well, despite different CHO-CHO and seed volume levels, reflecting that the normalization according to Eq. ( 1) applies also to type B growth.While type A growth is significantly lower for ABS+HA compared to AS+FA, type B growth rates seem to be controlled largely by gas-phase CHOCHO and LWC of the seed, and appear to be comparable within the measurement precision for both seed types.
Most notably, seeds containing AA like tyrosine and glycine, and/or SA did not show type A growth, despite the fact that the same photochemical system was used.The presence of these acids reduced the amount, and the rate of volume growth as illustrated in Fig. 4c.The normalized growth rate for these acidified experiments agrees reasonably well with type B growth observed for the dark periods of experiments #20 and #32 (type B growth from panel b is reproduced in panel c).We do not find an enhanced rate of volume growth for seeds that were mildly acidified with SA compared to the non-acidified seed.Seeds containing AA show a slightly faster initial volume growth, but this initial growth seems to level off over extended experiment times, and is below the type B growth rate after about 30 min experiment time.It should be noted however, that the smaller relative volume change associated with type B growth poses an experimental challenge that is not compensated easily by extended measurement times.The uncertainties from the correction of aerosol volume loss to the chamber walls and (reversible) uptake of gas-phase CHOCHO to the chamber walls then become dominant error sources.The reduction in uptake rates and relative volume changes for seeds that contained AA or SA however is beyond experimental doubt, and this observation is taken as evidence that the presence of acids reduces the amount of available water, or scavenges any excited states produced from organic photochemistry in the aerosol aqueous phase.The pronounced acceleration of the rate of volume change.For neutral seeds, the pronounced acceleration of the rate of volume change suggests that additional CHOCHO uptake is triggered as the result of organic photochemistry in the aerosol aqueous phase.The measured Y SOA number establishes as a result of competing uptake and scavenging rates of the intermediates of SOA formation in the aerosol water.

Dependence of effective CHOCHO solubility on relative humidity
Given that photochemical volume growth scales proportional to the gas-phase CHOCHO concentration, and LWC (Fig. 4b and Eq.1), the effective solubility of CHOCHO, H eff , was calculated from type A volume growth, and values are listed in Table 1.In experiments where also type B growth was observed, H eff was calculated based on the largest volume growth observed.In experiments where type B growth did not level off, H eff is listed as a lower limit number in Table 1.We calculate these H eff values primarily as a means of   CHOCHO uptake to aerosols; if calculations were based on the overall seed volume the actual H eff value could be underestimated by as much as a factor of 30 (for FA seed, Fig. 6b).This finding presents a deviation from the traditional definition of Molarity as the molar concentration of a solute per volume of solution (water+seed).
The effect of adding FA to AS seed is illustrated in Fig. 7a.The factor of 3.5 higher H eff for AS+FA seed is reflective of the higher Y SOA (Fig. 3b).Interestingly, the two experiments #12 and #20 (AS+FA seed at RH>80%) showed type A+B growth that did not level off over the timescales investigated; the respective H eff values in Table 1 represent lower limit values.For both experiments these lower limit values are higher than the values observed at lower RH (Fig. 7a).Both experiments were conducted above the deliquescence RH (DRH) of AS, and the observed H eff value is found to be approximately the sum of H eff values observed for pure AS and FA seed.For experiments below the DRH of AS, H eff values of mixed AS+FA seed were found essentially identical to those of pure FA seed.The higher H eff value for AS+FA seed at RH>DRH deserves further investigation.

Dependence of effective CHOCHO solubility on OH radical concentrations
It is of interest to explore whether our results depend on the conditions under which C 2 H 2 was oxidized.The OH-radical concentrations predicted by our model over the course of individual experiments varied by about 10% from the average values listed in Table 1; OH was systematically varied by more than a factor of 15 between different experiments.Over the range of OH-radical concentrations probed the H eff values for any given seed type did not depend on the OH concentration, as is shown in Fig. 7b.This is in line with a product distribution that did not vary much between experiments (see Table 2), and the conclusions in Sect.3, that SOA growth is essentially attributed to CHOCHO partitioning to the aerosol phase; in particular, the rate of any gas-phase polymerization reactions would be expected to depend on the OH radical concentration, which is not observed.Notably, studies of the OH radical initiated oxidation mechanism of C 2 H 2 show no change in the CHOCHO formation in the presence and absence of NO x , and suggest that the CHOCHO formation mechanism involves initial OH attack to the triple bond followed by a sequence of rapid O 2 addition/isomerization reactions that recycle the OH-radical without need for RO 2 →RO conversion (Hatakeyama et al., 1986).These simulation chamber experiments are consistent with OH-cycling experiments using laser-flash photolysis (Siese and Zetzsch, 1995), flow tube Chemical Ionization Mass Spectroscopy and computational studies (Yeung et al., 2005) that all show consistently that hydroxyl peroxy radical intermediates involved with the formation of products from C 2 H 2 are very short lived.Contrary to conventional RO 2 radicals, for which their fate could depend on the experimental conditions (reaction with NO vs. HO 2 /RO 2 ), the fate for these RO 2 is exclusively determined by rapid isomerization reactions.Yeung et al. (2005) infer the lifetime with respect to isomerization to be on the order of 700 ns.Assuming a rate constant of 9×10 −12 cm 3 molec −1 s −1 for the RO 2 +NO reaction, and [NO]=5 ppm, the fate of these RO 2 is determined to more than 99.9% by the isomerization reaction.The insensitivity of the OH-radical initiated oxidation mechanism, i.e., to form CHOCHO, is in general agreement with Fig. 7b, and support the robustness of our results also in the absence of NO x .
The lower range of OH radical concentrations in our experiments are comparable to those observed in the atmosphere (Shirley et al., 2006); typical OH concentrations were about 10-50 times higher than the global annual mean OH concentration (Prinn et al., 2001;Heard and Pilling, 2003).The lack of any OH-radical dependence suggests that our results can be extrapolated to atmospheric conditions.

Discussion and atmospheric relevance
There are currently three laboratory studies that investigate the uptake of CHOCHO to aerosols in the dark (Kroll et al., 2005;Liggio et al., 2005a;Corrigan et al., 2008), and one study reporting field evidence in support of the atmospheric relevance of SOA formation from CHOCHO (Volkamer et al., 2007).Our work is unique for three reasons: (1) the widely used concept of SOA yields is linked directly to a particularly volatile CHOCHO precursor NMHC; SOA formation from C 2 H 2 demonstrates directly that the current theory of SOA formation as it is widely used in atmospheric models is incomplete; (2) SOA yields from C 2 H 2 show a strong dependence on the chemical composition of seed aerosols, and raise questions whether a similar dependence of SOA yields exists also for other SOA precursors; and (3) we demonstrate that organic photochemistry in the aerosol aqueous phase is needed to accelerate SOA formation from CHOCHO, and dominates over dark processes.We also observe evidence for continued chemical processing of CHOCHO in the dark, but the amounts of SOA produced, and rate of SOA formation by dark processes are much smaller than by the photochemical reactions.

Comparison with literature values
Ours is the only study of CHOCHO uptake to aerosols conducted in the presence of gas-phase oxidants.The products formed in the presence of UV photons and OH-radicals are likely different from those formed under dark conditions in the absence of oxidants.Kroll et al. (2005) derived H eff based on the total seed volume, and a density of 1 g cm −3 for SOA from CHOCHO; both assumptions lead to an underestimate of the actual H eff value.The re-evaluated H eff value comparable to this work is a factor of 4.6 larger, i.e., H eff,AS,Kroll =1.2×10 8 M atm −1 at 50% RH.We find a 40% higher value of H eff for AS in our photochemical system, but the rate of CHOCHO uptake is enhanced by 2-3 orders of magnitude in our photochemical system.Notably the CHOCHO uptake kinetics for AS followed type A growth over the range of relative humidity conditions studied; no type B growth was observed in the following dark period of experiments.This is in contrast to the continued growth observed over several hours in dark systems (Kroll et al., 2005;Liggio et al., 2005a;Galloway et al., 2008) and UV-light irradiated systems in the absence of OH radical sources (Galloway et al., 2008), and surprising, as type B growth rates as those observed by other authors should have been observable in our system.Notably, the amounts of SOA formed after several hours in their experiments remain 40% below the SOA formed within 90 s in our study.While our results appear to confirm the conclusion for reversible CHOCHO uptake for AS (Kroll et al., 2005) over an extended RH range, the 40% higher uptake in our photochemical system is only marginally within error bars.The difference in amounts and the rate of CHOCHO uptake may reflect either of the following: (1) CHOCHO is irreversibly taken up by photochemical reactions; (2) photochemistry speeds up the reaction rates, but does not alter the reaction products; (3) photochemistry speeds up the reaction rates, and forms essentially different products than dark reactions; or (4) a combination of the above.The question of whether the oxidative CHOCHO uptake to AS is indeed fully reversible, i.e., can indeed be described by Henry's Law, remains to be answered.Kroll et al. (2005) had suggested that the uptake of CHO-CHO to AS is controlled by ionic strength.This appears to be somewhat unlikely based on the anti correlation between ionic strength and the missing CHOCHO sink observed in Mexico City (Volkamer et al., 2007).Based on the results from this work, any effect of ionic strengths is much smaller than the effect of organic photochemistry to enhance CHO-CHO uptake to aerosols.
There is one study available for discussion of SucA.Corrigan et al. ( 2008) did not observe any CHOCHO uptake in the dark under conditions where SucA is believed to be dry and solid (up to 54% RH).Our study was conducted at slightly higher RH (61% RH), and SucA containing about 36% v/v LWC showed type A uptake kinetics in the photochemical system.Consistent with findings by Corrigan et al. (2008) we did not observe type B growth in the dark, when the aqueous seed was exposed to about 90 ppbv CHO-CHO.It appears unlikely that aerosol water associated with dicarboxylic acids catalyzes any irreversible reaction with CHOCHO under dark conditions.Interestingly the uptake kinetics, and amounts of CHOCHO related growth for AS and SucA are comparable.In analogy to AS, we conclude that any irreversible reaction of CHOCHO with SucA in our system must be driven by photochemical reactions.Alternatively, the reaction rate to form oligomers from CHOCHO in the associated water phase could be strongly enhanced by photochemistry, yet follow a reversible mechanism.It remains to be elucidated whether the photochemically enhanced uptake of CHOCHO to AS and SucA follows a reversible or irreversible reaction mechanism.
The reaction of amines to form covalent N-C bonds has been exploited by analytical techniques for the detection of carbonyls including CHOCHO for a long time (Koshy et al., 1975;Yu et al., 1995b).Little is known however about the extent to which gas-phase reactions of amines play a role in the formation of particles, or affect the atmospheric chemistry of CHOCHO.More information is available on amino acids, which have recently been shown to catalyze the formation of oligomers from acetaldehyde via aldol condensation (pathway 1), or a Mannich Pathway (pathway 2) (Noziere and Cordova, 2008).While similar reactions may also trigger aldol condensation of α-dicarbonyls, the rate limiting step for both pathways consists in the formation of enamine, which may proceed also in the case of CHOCHO, as the additional aldehyde group in water undergoes rapid gem-diol chemistry.
For glycine, the most abundant AA in atmospheric particles, the rate of enamine formation is lower by about a factor of four than for arginine (Noziere and Cordova, 2008).Under the molar concentrations of glycine (and tyrosine) used in this study (about 2-4 M), the AA concentration is sufficiently high that any enamine that may form from CHOCHO would likely react via a Mannich pathway.Glycine concentrations used in this study are about two orders of magnitude higher than those found in atmospheric particles (up to 0.02 M) (Zhang and Anastasio, 2003;Matsumoto and Uematsu, 2005;Noziere and Cordova, 2008); however, our concentrations are low compared to those employed in Corrigan et al. (2008), where essentially pure solid AA were used.Under such particular experimental conditions Corrigan et al. (2008) argue that the presence of particle phase nucleophiles, sulfate and acidity all appear to increase CHOCHO uptake.However, growth factors observed on sodium sulfate are larger than growth factors on any AA.Arginine and glutamic acid experiments are the only AA systems that showed enhanced uptake over sodium chloride, which was chosen as a reference seed that only forms glyoxal oligomers in the associated water phase (no reactions with sodium chloride are expected).The observed growth factors on sodium chloride were larger than for glycine and other AA studied, including one of the two glutamic acid experiments.Overall, the enhancements attributed to AA catalysis over salt solutions were, if observable, rather small (Corrigan et al., 2008).In our study addition of glycine and tyrosine to AS+FA seed did reduce the rate of CHOCHO uptake by 2-3 orders of magnitude compared to experiments in the absence of AA in the photochemical system.We observe growth rates in the dark to be comparable in the presence and absence of AA, and do not find evidence that the most abundant AA in atmospheric aerosols, i.e., glycine, enhances the reactivity of CHOCHO in aerosols.
Similarly, acid catalyzed heterogeneous reactions of CHO-CHO are found second to photochemical reactions to enhance the CHOCHO uptake to particles.For the mildly acidified seeds investigated here (see Table 1) the photochemical rate of CHOCHO uptake is reduced by several orders of magnitude compared to identical seeds that were not acidified with SA.Differences in experimental conditions may explain results different from those observed by (Jang et al., 2002); those experiments were conducted with more acidic particles in a photochemical reaction system.An acid effect on CHOCHO uptake was also observed by Liggio et al. (2005a) and Corrigan et al. (2008), while -consistent with our results -no acid effect was observed by Kroll et al. (2005).Notably, in the apparent correlation of the growth rate with pK a (Corrigan et al., 2008) the strongest acid studied (oxalic acid) was an outlier to any apparent acid trend.Oxalic acid and SucA showed negligible CHOCHO uptake in the dark, despite rather high RH.This is consistent with the absence of type B growth on SucA observed here, and our finding that addition of SA had no effect on the rate of CHOCHO uptake in the dark.The apparent correlation of growth rate with pK a could hence equally be interpreted as LWC controlled CHOCHO uptake, and competing acid effects, as concluded in this work.LWC is not well constrained in Corrigan et al. (2008), and such interpretation seems likely also in light of the growth rate data for glycine and glutamic acid, which appears to be internally consistent only if regarded in context to the presumably higher LWC at higher RH.The reduced uptake of CHOCHO to acidified aqueous solutions observed in this work and Corrigan et al. (2008) appears to be analogous to the observation of reduced methylglyoxal uptake on more acidic solutions (55-85% w/w SA) (Zhao et al., 2006), which suggested a competition of acid for liquid water, and recognized the relevance of liquid water rather than acid as catalyst for SOA formation.Our study directly confirms such competition exists also for CHOCHO uptake due to organic photochemistry in aerosol water.At the pH found in atmospheric particles throughout the troposphere, acidity is also an inhibiting factor in SOA formation.In terms of the CHO-CHO reactivity enhancements, we conclude that sulfate is second to organic photochemical reactions; the presence of nucleophiles and SA reduces this reactivity.
The dependence of SOA yields on seed chemical composition demonstrates that the chemical reactions that form SOA depend on the chemical environment provided by the seed.Other than the direct role of providing reactive functional groups, indirect roles of the seed could include catalytic roles of aerosol water, seed chromophores facilitating photolytic processes, and particle pH influencing the reaction rate of CHOCHO processing in aerosol water.
4.2 Do SOA yields depend on the seed chemical composition also for other VOCs ?
Most studies of SOA yields to date used either no seed at all (e.g., Hurley et al., 2001 and references therein), or added AS seed in an attempt to overcome barriers in the Gibbs free energy towards homogeneous nucleation of semivolatile vapors, or ascertain the effect of ionic strength on the partitioning of semivolatile products (Odum et al., 1997;Cocker et al., 2001;Kroll et al., 2007).AS seed has been shown to accelerate the rate of SOA formation, but does not affect the SOA yield compared to experiments conducted in the absence of seed (Kroll et al., 2007).Notably, our observation of rapid photochemical SOA formation on AS could contribute to this acceleration, though AS is among those seeds that forms least SOA in this work.Cocker et al. (2001a, b) did not observe any measurable effect of RH on SOA yields from a-pinene, m-xylene and 1, 3, 5-trimethylbenzene, and attributed this to the existence of two separate SOA and aqueous salt phases in a single particle.This interpretation of phase separation is supported also by thermodynamic model considerations (Chang and Pankow, 2006).Consequently, experiments using seeds composed of inorganic aqueous salts, SOA, or mixtures of both will not be particularly sensitive to any aqueous phase organic photochemistry, as the amount of liquid water associated with SOA is much smaller than that associated with aqueous salts.Notably, the OHradical initiated oxidation of 1, 3, 5-trimethylbenzene does not form CHOCHO as oxidation product, and CHOCHO yields from a-pinene are very small.Also, CHOCHO yields from m-xylene are low compared to other alkylbenzenes (see Table 1 in Volkamer et al., 2007), reducing the sensitivity of all three systems further towards effects as those observed in this study.Regardless, CHOCHO and glyoxylic acid were identified in Cocker et al. (2001b) among the aerosol phase products.There is no currently known mechanism to produce glyoxylic acid via a gas-phase oxidation mechanism, while it is a known product from aqueous phase oxidation of CHO-CHO and other precursors (Ervens et al., 2003).The observation of glyoxylic acid among the constituents in SOA gives direct evidence that the molecular formation routes for SOA formation appear to have been affected by the presence of small amounts of water (either from multiphase chemistry at the interface between aqueous salt phases, or in water associated with SOA); CHOCHO and glyoxylic acid accounted for about 0.1% each of the extracted SOA mass from m-xylene under dry conditions.
To the best of our knowledge ours is the first study that systematically investigated the dependence of Y SOA on seed chemical composition.Unless SOA generated in chambers is equally water soluble as SOA generated in the atmosphere, our findings point to a possible bias in current simulation chamber data: a reduced interface for organic photochemistry in externally mixed SOA and water phases (within the same particle), compared to a greatly enhanced interface in bulk aqueous aerosols containing water soluble organic carbon (WSOC).WSOC is a major component of the seeds studied here, as well as the SOA produced in the atmosphere (Sullivan et al., 2006;Weber et al., 2007).If a similar effect of seed chemical composition was confirmed also for other SOA precursor VOCs, this would significantly increase the predicted amount of SOA formed on local, regional and global scales.4.3 Can CHOCHO uptake be described by Henry's Law ?Whether or not the partitioning of CHOCHO to aerosols can be described by Henry's Law is directly tied to the question whether the uptake mechanism is reversible or irreversible.The relevance of this question is illustrated from discussion in Volkamer et al. (2007): reversible CHOCHO uptake as described by Kroll et al. (2005) can explain several 10 ng m −3 of SOA mass for the typical urban CHOCHO concentrations found in Mexico City (Volkamer et al., 2005a); irreversible CHOCHO uptake (Liggio et al., 2005a) predict several 10 µg m −3 SOA.The missing CHOCHO sink to form SOA is closer to the upper end of this range, and can explain a significant fraction of the missing SOA source (Volkamer et al., 2006(Volkamer et al., , 2007)).Notably, this conclusion was reached without making any assumption about the reaction mechanism, reflecting that the uncertainty related to SOA formation from CHOCHO on aerosols spans several orders of magnitude.
For AS seed, Kroll et al. (2005) reproduced experiments by Liggio et al. (2005a) over extended experiment times, and observed that initial CHOCHO uptake slowed and reached a plateau after several hours.They concluded that CHO-CHO uptake to aqueous AS is reversible.This implies that the chemical bonds formed when CHOCHO reacts with sulfate in the dark (Liggio et al., 2005c), or forms oligomers in aerosol water (Kroll et al., 2005;Hastings et al., 2005;Liggio et al., 2005c) are not stable and appear to be broken already at ambient temperatures.Corrigan et al. (2008) observed continued growth upon exposing CHOCHO to (mostly dry) aspartic acid, sodium sulfate, arginine, glutamic acid aerosols (among others) in a dark chamber, and attribute this growth to reactions of CHOCHO with sulfate, nucleophiles (AA), and oligomer formation in thin water layers associated with mostly dry aerosols.The apparent growth rate appears to decrease with time, and it is not clear at present if it converges towards a plateau.It should be noted that the results by Corrigan et al. (2008) on dry seed (0<LWC<4% w/w showed most growth) and in the dark are somewhat difficult to compare with our results on mostly aqueous aerosols, where AA and/or SA largely reduce the photochemistry enhanced aerosol growth rate (see also Sect. 4.1.).Our findings that photochemical aerosol growth directly scales with (1) the CHOCHO concentration, and (2) the LWC for any given seed type are compatible with Henry's Law behavior, but allow no definite conclusion whether such growth is reversible or irreversible (see Sect. 4.1.).About half our experiments showed continued growth in the dark over accessible time scales; however, about an equal number showed no such growth.While an increasing number of studies now provide evidence that CHOCHO partitions to aerosols in amounts larger than expected, it remains to be demonstrated convincingly whether such CHOCHO uptake proceeds by a reversible or irreversible mechanism.Notably, we choose to present our results in terms of H eff values primarily for means of a quantitative comparison with other studies, and for discussion purposes, and do not mean to imply that photochemical CHOCHO uptake is necessarily reversible.

Atmospheric implications
It is of interest to place the results from the present work into perspective in terms of our understanding of the mechanism of SOA formation in the atmosphere.Traditional SOA models struggle with predicting the fast timing and the large amount of SOA produced in the atmosphere (Heald et al., 2005;de Gouw et al., 2005;Johnson et al., 2006;Volkamer et al., 2006;Riddle et al., 2008).Recent experimental work has further demonstrated that only a portion of the organic aerosol seems to participate in the absorption processes for semivolatile oxidation products (Song et al., 2007).This evidence further increases the already substantial mismatch between SOA predictions and observations.Traditional SOA models contain a high bias with respect to the predicted amount of SOA from semivolatile oxidation products in the atmosphere.Most SOA models represent SOA formation as the result of the gas-phase oxidation of VOCs to form semivolatile products that partition by absorption (dissolution) into an organic particle phase (Odum et al., 1997;Seinfeld and Pankow, 2003;Robinson et al., 2007).These models follow (Pankow, 1994a), according to which only the organic mass fraction of the aerosol (M om ) contributes to absorbing semivolatile gases; the partitioning coefficient for an absorptive process is expressed as a function of physical and thermodynamic properties of the semivolatile compound: where R is the ideal gas constant, T is temperature, M om (see above), MW is the mean molecular weight of the absorbing aerosol phase, ξ i is the activity coefficient of compound i in this phase, and p o i is the vapor pressure of compound i as a pure liquid.M om is defined to include any liquid water that is attracted due to hygroscopic properties of organic material (see Eq. ( 4) in Seinfeld and Pankow, 2003).According to Eq. (2) aerosol components other than M om , such as crustal material, inorganic aerosol, and liquid water attracted by inorganic aerosol do not contribute to SOA formation.In this view, for any given SOA precursor NMHC there is one set of Y SOA and M om data pairs; i.e., Y SOA is treated as a chemically inert quantity that should not depend on the chemical seed composition.The seed role is limited to accelerating SOA formation by providing a surface for condensation.
In this work the Y SOA values for any given seed do not correlate with M om , but instead show an excellent correlation with LWC (see Fig. 3,and Sect. 3.1.).Complementary evidence for the relevance of LWC for SOA formation from recent field observations (Hennigan et al., 2009) is consistent with our results.Moreover, at any given M om the Y SOA is found to vary as a strong function of the chemical seed composition (see Sect. 3.2.),providing direct evidence that SOA formation from C 2 H 2 is controlled by heterogeneous chemistry.Internal mixtures of inorganic and water soluble organic aerosol (fulvic and humic type matter) increase Y SOA over pure inorganic aerosol.This highlights an important function of inorganic aerosol, namely to attract LWC, and thus increase the reactor volume where organic photochemistry causes the Y SOA enhancements over pure salt solutions.Such SOA largely escapes the parameterization of SOA yields derived in simulation chamber experiments of other SOA precursors (see Sect. 4.4.4.).
Vapor pressure as the only parameter to control SOA formation clearly is insufficient to explain SOA formation from C 2 H 2 , and the variability of SOA yields at fixed M om observed in this study.SOA models require a conceptual expansion to also reflect SOA formation via the photochemical and dark solubility -reactivity scale in aqueous aerosols.The implementation of such SOA formation in atmospheric models requires knowledge of partitioning coefficients (e.g., effective Henry's law constants, reactive uptake coefficients), the oxidation mechanism and chemical kinetics of reactions that foster and inhibit SOA formation, ideally under conditions that resemble those found in the atmosphere.

Aerosol vs. cloud water as a chemical reactor
The chemical environment in aerosol water is fundamentally different from that in cloud or fog water, because (1) the liquid water content of aerosols is 3-5 orders of magnitude smaller than in clouds/fog droplets, and (2) the smaller diameter of aerosols results in 2-3 orders of magnitude higher surface to volume ratio for aerosols compared to cloud/fog droplets.The different chemical environment in aqueous aerosols compared to cloud/fog droplets is illustrated in Fig. 8. Based on measurements of the effective Henry's law constants of dilute solutions of CHO-CHO in water H eff =4.2×10 5 M atm −1 (Ip et al., 2009), and seawater H eff =3.6×10 5 M atm −1 (Zhou and Mopper, 1990), CHOCHO should partition in appreciable amounts only to clouds/fog, but not to aqueous aerosols.Even the higher CHOCHO solubility in aqueous AS particles, H eff =2.6×10 7 M atm −1 at 50% RH (Kroll et al., 2005), does not significantly affect the gas-phase CHOCHO budget in the absence of clouds.However, the presence of dissolved organic matter can photochemically enhance the CHOCHO uptake.Using our measured effective solubility of CHOCHO in aqueous fulvic acid particles, H eff =6.1×10 8 M atm −1 , and assuming 10 µg m −3 of aerosol liquid water, about 13% of the CHOCHO will be in the particle phase.For Mexico City, where organic aerosols comprise the largest fraction of fine particulate mass (Salcedo et al., 2006), our laboratory results predict up to 24% of the CHOCHO will partition to aerosols.Clearly, aerosols can affect the gas-phase CHOCHO budget to a measurable extent despite their small LWC.
The aerosol aqueous phase as a chemical reactor is distinctly different from cloud water.Notably, the source of oxidants is different.H 2 O 2 does not partition in appreciable amounts to aerosols; however, OH-radical uptake from the gas-phase provides a significant source of oxidants in particles (Ervens et al., 2003).Also, compared to the microto milli-molar dilute solutions characteristic of cloud or fog water (Carlton et al., 2007), reaction rates of CHOCHO are likely to be accelerated in 0.1-10 M solutions characteristic of aqueous aerosols.The nature of oligomer formation is non-linear (Barsanti and Pankow, 2005), and requires experiments to be conducted under essentially different conditions that resemble aerosols.

Anthropogenic vs. biogenic SOA sources
Global simulations indicate that biogenic NMHC oxidation is the major global source of SOA and this source has been enhanced by humans due to the anthropogenic increases in gas-phase oxidant and primary particle levels (Kanakidou et al., 2005;Fuzzi et al., 2006).This view is supported by 14 C measurements, which attribute a modern carbon age to most OA found in the atmosphere.However, field observations show no correlation between enhancements of OA mass and biogenic precursors (de Gouw et al., 2005;Sullivan et al., 2006;Weber et al., 2007), but instead show an excellent correlation with anthropogenic pollution factors (de Gouw et al., 2005;Volkamer et al., 2006;Sullivan et al., 2006;Weber et al., 2007).Such conflicting evidence from 14 C data and correlative analysis of time series data is difficult to resolve within the purely physical (vapor pressure driven) framework of SOA formation that underlies most SOA models; such models necessarily maintain the 14 C signature of precursors also in the 14 C signature of the semivolatile oxidation products.It should be noted, however, that a primarily modern 14 C age of OA is not necessarily indicative of biogenic NMHC precursors.Rather our interpretation of the information content of 14 C measurements depends on the framework that we choose to think about SOA formation.If, for example, the framework for SOA formation was expanded to include heterogeneous (photo) chemistry, other factors could increase or decrease particulate phase reactivity of SOA building blocks.While the possibility for such bias in the interpretation of 14 C age could in principle be caused by a biogenic or anthropogenic trigger of SOA formation, only an anthropogenic trigger to enhance SOA formation from biogenic NMHC is compatible with the available field evidence (de Gouw et al., 2005;Volkamer et al., 2006;Sullivan et al., 2006;Weber et al., 2007;Riddle et al., 2008).Further, there is no experimental evidence as yet that would demonstrate convincingly that equal amounts of OA are formed from biogenic NHMC precursors in the presence and absence of anthropogenic pollution factors.
The global source of α-dicarbonyls is estimated to be at least 185 Tg yr −1 (Fu et al., 2008).The CHOCHO source is estimated at 45-83 Tg yr −1 , and largely due to secondary sources (17% from C 2 H 2 , 70% from biogenic NMHCs) (Myriokefalitakis et al., 2008;Fu et al., 2008); an additional oceanic source (Wittrock et al., 2006) is currently largely uncertain.Aerosol uptake to form SOA could determine 75-95% of the atmospheric fate of CHOCHO in polluted air, and form about an equal amount of SOA as the sum of semivolatile oxidation products from traditional anthropogenic and biogenic SOA precursor NMHC in Mexico City (Volkamer et al., 2007).CHOCHO, and other α-dicarbonyls, Cl d Kroll et al. (2005)  aerosol liquid water.The LWC of clouds (blue line) and fog (orange line) is 3-5 orders of magnitude larger than that of aerosols (pink area).H eff values on the order of 3.5×10 4 M atm −1 (blue dot), 3.5×10 5 M atm −1 (orange dot) and 2×10 9 M atm −1 (black dot) characterize equal parts in the gas-and dissolved phase for clouds, fog and aerosols, respectively.Dotted lines indicate H eff -LWC pairs that favor the gas-phase; dashed lines indicate H eff -LWC pairs where the dissolved phase is the dominant reservoir (the partitioning ratio, defined as dissolved/gas-phase, is indicated next to each line).The vertical lines represent H eff values as measured for CHOCHO in (a) bulk water (Ip et al., 2009), (b) metastable AS aerosols (Kroll et al., 2005), (c) metastable AS+FA aerosols (this work), and (d) aqueous ambient aerosols in Mexico City (Volkamer et al., 2007).Aerosols are distinctly different aqueous/multiphase chemical reactors than clouds/fog.
are deemed responsible for 5-11 Tg yr −1 SOA production on global scales, which is comparable to the sum of SOA formed from monoterpenes, sesquiterpenes, isoprene, and aromatics (Fu et al., 2008;Myriokefalitakis et al., 2008); these estimates underestimate the role of aerosol seeds, and will further increase based on the results of this study.WSOC accounts for a major portion of OA in urban and rural areas (Sullivan et al., 2006;Weber et al., 2007).Our observation of fast SOA formation from WSOC photochemistry of CHO-CHO is consistent with field evidence of fast SOA formation due to first generation oxidation products, and helps explain part of the missing SOA source in Mexico City (Volkamer et al., 2006).However, our results fall short by a factor of 7-14 to explain the CHOCHO sink to form SOA in Mexico City (Volkamer et al., 2007).
Given the large mismatch between predicted and observed SOA on local and global scales, little trust should be placed in the dominance of biogenics on global scales based on these models alone.Clearly, the source of α-dicarbonyls, and the uncertainty associated with their atmospheric fate, www.atmos-chem-phys.net/9/1907/2009/Atmos.Chem.Phys., 9,2009 is large enough to affect the interpretation of 14 C measurements on scales from local to global.Anthropogenic precursors could be responsible for about one third of the global SOA production (Volkamer et al., 2006).Our results support the expanded definition of SOA formation, according to which SOA sources include gas-to-particle partitioning in the organic-and aqueous phase, organic phase processing, aqueous phase processing, and multiphase processing (i.e., surface and bulk) (Volkamer and Kanakidou, 2007).
4.4.4On the issue of double-counting SOA from CHO-CHO Most traditional SOA precursor hydrocarbons are also precursors for CHOCHO, i.e., about 80% of CHOCHO in Mexico City forms from aromatic precursors (Volkamer et al., 2007).The parameterization of SOA yields from traditional SOA precursor hydrocarbons that underlie current SOA models could already include SOA formed from CHO-CHO.If SOA from CHOCHO was treated as an explicit additional pathway in those models, would CHOCHO SOA be double counted?It appears unlikely for the following reasons: (1) the SOA source from CHOCHO closely compares to (and slightly exceeds) that of all other currently studied SOA precursor NMHC (Volkamer et al., 2007;Fu et al., 2008); all SOA would need to be CHOCHO related.This is obviously not the case, as most aerosol products that have been identified to date at the molecular level are unrelated to CHOCHO (Yu et al., 1995a;Forstner et al., 1997;Cocker et al., 2001).(2) Most chamber experiments were conducted under dry experimental conditions, and on AS seed.Such conditions essentially preclude SOA formation via the aerosol aqueous phase.AS showed among the lowest Y SOA values in this study.In the presence of relative humidity there is consistent experimental and theoretical evidence (Cocker et al., 2001a, b;Chang and Pankow, 2006) that strongly suggests that WSOC photochemistry is greatly reduced in current chamber data due to phase separation effects (see also Sect.4.2.).(3) The aerosol mass spectra of SOA generated in the presence of aqueous AS seed in chambers (Bahreini et al., 2005) are different from OOA spectra observed in ambient air (Zhang et al., 2005), and SOA generated from CHOCHO uptake to AS (Liggio et al., 2005a).Notably, CHOCHO uptake to AS accounts only for 2-3% of the missing CHOCHO sink to form SOA in ambient air (Volkamer et al., 2007); WSOC photochemistry is likely to result in different aerosol mass spectra than CHOCHO uptake to AS.Further, the differences in the mass spectra of (Zhang et al., 2005;Bahreini et al., 2005) appear to be largest for masses that have a high contribution from oxygen containing products (e.g., m/z 28, which corresponds to CO). ( 4) The oxygen-to-carbon ratio (O/C) for the di-hydrated CHOCHO monomer is O/C=2, and this ratio approaches O/C=1 for longer chain oligomers; O/C is likely to be high also for the presently unidentified products.As noted in (Volkamer et al., 2007) SOA from CHO-CHO will increase predicted O/C ratios (see also Dzepina et al., 2009).While the high O/C ratios of CHOCHO related SOA compare favorably to the O/C=0.9observed in ambient aerosol, the range of observed values in chamber experiments from toluene, 1, 3, 5-trimethylbenzene, isoprene, α-pinene is much lower, i.e., 0.3<O/C<0.4(Aiken et al., 2008).It is these low O/C ratios that are currently parameterized in atmospheric SOA models.Finally, (5) SOA generated in chambers is more volatile than ambient SOA (Huffman, et al., 2009).Thereby, the more oxidized fraction of the aerosol is least volatile.Typical values for the heat of vaporization H of SOA generated from toluene in chambers range between 11< H<31 kJ mol −1 (Takekawa et al., 2003;Offenberg et al., 2006).The H of SOA generated from single component CHOCHO aerosol is H=13.3 kJ mol −1 (Offenberg et al., 2006).While the volatility of CHOCHO related SOA is difficult to separate from the convoluted volatility of ambient or chamber generated organic aerosol, the CHO-CHO related SOA is likely to lower the effective volatility of ambient aerosol.Conversely, increasing the CHOCHO related portion of SOA in chamber experiments (see Sect. 3.2. and 4.3.)and ambient aerosol (Volkamer et al., 2007) will likely bring the predicted and observed volatility of SOA in closer agreement (see also Dzepina et al., 2009).
Overall, it appears likely that the predominant fraction of CHOCHO related SOA escapes the current parameterizations of SOA yields from chamber experiments (e.g., Odum et al., 1997;Ng et al., 2007) that underlie current estimates of SOA formation from atmospheric models (Kleeman and Cass, 2001;Chung and Seinfeld, 2002;Kanakidou et al., 2005;Fuzzi et al., 2006;Henze et al., 2008).

Conclusions
The formation of Secondary Organic Aerosol (SOA) from C 2 H 2 , the lightest Non Methane HydroCarbon (NMHC), is surprising, because all currently known oxidation products are at least six orders of magnitude too volatile to condense or partition into the aerosol organic phase to any appreciable extent.SOA formation from C 2 H 2 highlights the importance of heterogeneous chemistry in SOA formation.Our results point out the particular role of organic photochemistry in aerosol water as a SOA source from CHOCHO, and possibly other α-dicarbonyls.Our laboratory data is consistent with recent field observations (Hennigan et al., 2009), suggesting that organic photochemistry in LWC is very likely relevant also for other SOA precursor VOCs.This is to our knowledge the first systematic study of the effect of seed chemical composition and photochemistry on SOA yields, and the rate of SOA formation.The seed particles investigated here are representative for a large subset of aqueous atmospheric particles.The SOA yields for mixed inorganic/water soluble organic carbon (WSOC) seed aerosols are about four times larger compared to pure inorganic aerosol, but are comparable for the pure water soluble organic seed aerosols.We conclude that the inorganic aerosol component primarily determines the reactor size, i.e. the liquid water content (LWC), where WSOC photochemical reactions cause the increased SOA yields.Any ionic strength effects of concentrated salt solutions appear to be secondary to providing a sizable photochemical reactor.Further, ours is the only study to date that investigates CHOCHO uptake to aerosols in the presence of OH-radicals and UV light.The rate of SOA formation is about a factor of 500 higher under photochemical conditions compared to dark conditions, i.e., fast enough to compete with rapid gas-phase losses.
The implications for our understanding of atmospheric processes are significant: (1) atmospheric SOA models currently widely represent SOA yields from experiments conducted in the absence of seed, or in the presence of pure ammonium sulfate seed.Here we find the highest SOA yields for internally mixed inorganic/WSOC seed particles due to organic photochemistry in the bulk LWC of the particles.Most SOA precursors for which their chemical identity is known to date are major sources for CHOCHO, and other α-dicarbonyls.Hence, any low bias in the currently available SOA yield data might explain a significant portion of the missing SOA source in current atmospheric models.(2) The effective Henry's Law constants H eff for CHOCHO are sufficiently high to compensate for the 3-5 orders of magnitude smaller amount of liquid water associated with atmospheric particles compared to clouds/fog.We demonstrate that WSOC photochemistry of CHOCHO -and possibly other α-dicarbonyls -is a SOA source already in the absence of clouds/fog.Most notably, the SOA formation associated with these gases does not stop once a cloud droplet has evaporated, as inferred by current SOA models that represent cloud processing, but rather proceeds also in the absence of clouds, possibly even at an enhanced rate as aerosols are more concentrated solutions compared to clouds.(3) The atmospheric relevance of α-dicarbonyls as a source for SOA depends on the competition between the rate of uptake to particles compared to other gas-phase loss rates.Our results demonstrate that organic photochemistry driven by OHradical, and possibly light induced reactions is needed to enhance the rate of CHOCHO uptake to be competitive with rapid gas-phase losses.(4) The rate of CHOCHO uptake to particles in our photochemical system is fostered by most seeds studied but reduced by the presence of atmospherically relevant amino acids like glycine, and sulfuric acid.This finding is difficult to reconcile with the idea of acid catalysis to enhance the rate of CHOCHO uptake, and points to a more complex role of particle pH, i.e., identifies a strong pH dependence of the rate of photochemical reactions in the aqueous aerosol phase.(5) Our results explain up to 430 ng m −3 of the missing SOA source in Mexico City as CHOCHO related (Volkamer et al., 2006), and account for up to 7-14% of the missing CHOCHO sink in Mexico City (Volkamer et al., 2007).While our results reduce the existing imbalance in current gas-phase CHOCHO budgets most of the imbalance in gas-phase CHOCHO cannot be explained as yet based on the currently available laboratory data.
There is an urgent need for further laboratory studies to determine the reaction mechanism, kinetics, and products of this organic photochemistry.Atmospheric SOA models need to be expanded to explicitly represent the aerosol aqueous phase chemistry in form of dark and photochemical reactions in the presence and absence of clouds.

3. 1 Fig. 3 .
Fig. 3. SOA yield (Y SOA ) as a function of organic seed mass, M om (A, C, E) and liquid water content, LWC (B, D, F).No obvious correlation is found between Y SOA and M om , but Y SOA correlates well with LWC for all seed-types studied.For pure AS (solid line), AS+FA (dashed line), and pure ABS (dotted line) sufficient experiments were conducted to demonstrate that the increase in Y SOA with LWC is linear.
photochemical, type A

Fig. 5 .
Fig.5.Time series of the aerosol volume for type A uptake kinetics.Data from experiment #20 (ABS+HA) is shown here.Grey areas indicate times with lamps ON.No effect is observed if the seed is exposed to light in the absence of reagents in the chamber.After addition of chemicals to the gas-phase, an immediate shift in aerosol volume is observed when C 2 H 2 is photochemically oxidized (twice here), followed by type B growth in the dark.Type B growth as parameterized in Fig.3for AS+FA can explain that observed for ABS+HA seed (solid line), see text.The rapid volume growth during periods when the light is turned ON is proportional to the amount of CHOCHO produced, and inversely proportional to LWC of the seed aerosols.

Fig. 6 .
Fig.6.RH dependence of the effective Henry's law constants, H eff , of CHOCHO for different seed aerosols: (A) Ammoniumsulfate mixed with fulvic acid (AS+FA), (B) Fulvic acid (FA), (C) Ammoniumsulfate (AS), (D) Ammoniumbisulfate (ABS).Open symbols represent values of H eff derived under the assumption that CHO-CHO partitions to the overall seed volume (case 1).Solid symbols represent values of H eff derived under the assumption that CHO-CHO partitions only to the portion of the seed volume that corresponds to aerosol liquid water (case 2).The apparent RH dependence of H eff in case 1 is fully accounted in case 2, presenting direct evidence that CHOCHO uptake proceeds through the aerosol aqueous phase.

Fig. 7 .
Fig. 7. Water soluble organic carbon (WSOC) photochemistry enhancement of CHOCHO partitioning to AS+FA and pure FA aerosols compared to pure salt solutions (AS).
4.4.1 Conceptual expansion of SOA models: the photochemical solubility-reactivity scale

Fig. 8 .
Fig.8.Partitioning of water soluble molecules to cloud/fog vs. aerosol liquid water.The LWC of clouds (blue line) and fog (orange line) is 3-5 orders of magnitude larger than that of aerosols (pink area).H eff values on the order of 3.5×10 4 M atm −1 (blue dot), 3.5×10 5 M atm −1 (orange dot) and 2×10 9 M atm −1 (black dot) characterize equal parts in the gas-and dissolved phase for clouds, fog and aerosols, respectively.Dotted lines indicate H eff -LWC pairs that favor the gas-phase; dashed lines indicate H eff -LWC pairs where the dissolved phase is the dominant reservoir (the partitioning ratio, defined as dissolved/gas-phase, is indicated next to each line).The vertical lines represent H eff values as measured for CHOCHO in (a) bulk water(Ip et al., 2009), (b) metastable AS aerosols(Kroll et al., 2005), (c) metastable AS+FA aerosols (this work), and (d) aqueous ambient aerosols in Mexico City(Volkamer  et al., 2007).Aerosols are distinctly different aqueous/multiphase chemical reactors than clouds/fog.

Table 2 .
Products formed in the C 2 H 2 /CH 3 ONO/NO x /light system, effective Henry's law coefficients.Values in the right column have been normalized to CHOCHO.
remained constant in all experiments presented in this work.The aerosol volume growth corresponding to this shift in the aerosol size distribution was used to quantify SOA formation.
The timing of SOA formation: photochemical vs. dark systems ). Open symbols represent values of H eff derived under the assumption that CHO-CHO partitions to the overall seed volume (case 1).Solid symbols represent values of H eff derived under the assumption that CHO-CHO partitions only to the portion of the seed volume that corresponds to aerosol liquid water (case 2).The apparent RH dependence of H eff in case 1 is fully accounted in case 2, presenting direct evidence that CHOCHO uptake proceeds through the aerosol aqueous phase.normalization to facilitate a quantitative discussion between different seeds and experimental conditions, and do not mean to imply any conclusion about the reaction mechanism.H eff values for AS, ABS, FA and AS+FA are plotted as a function of RH in Fig.6.H eff values were calculated in two ways:(1) based on the measured seed volume (black open dots) and (2) based on the portion of the seed that was LWC (solid red dots).Calculations of H eff that are based on the measured seed volume give systematically lower values; those H eff values further show an apparent dependence of RH.Notably, this apparent RH dependence is fully accounted after normalization of the LWC portion of the seed.Such calculation gives essentially constant and higher H eff values for all seeds shown, over the entire range of RH probed.This further corroborates that LWC is a crucial quantity for determining the Volkamer et al.: WSOC photochemistry of CHOCHO V seed seed volume before the experiment V norm,t total aerosol volume after normalization, as defined in Eq. (1) WSOC water soluble organic carbon Y SOA SOA yield, Y SOA = M SOA / C 2 H 2 ζ i activity coefficient of compound i t observed aerosol volume at any given time t www.atmos-chem-phys.net/9/1907/2009/Atmos.Chem.Phys., 9, 1907-1928, 2009 R.