The ESI-MS spectrum of the products of the bulk-phase reaction of glyoxal and methylamine are shown in Figure 1. Nucleophilic attack by the amine group on the glyoxal aldehyde group produces an imine in a drying aqueous droplet (Figure 2). A protonated, hydrated imine could be responsible for the peak detected at m/z 90 by ESI-MS (Figure 1). However, peaks assigned to 2nd generation imine reaction products are larger: a peak at m/z 97 is consistent with 1,3-dimethylimidazole, and a prominent mz = 18 series (Δm/z 125, 143, and 161) is likely due to protonated imine dimers incorporating varying amounts of water. A series of oligomeric groups with Δm/z of 12, 13 and 14, peaking at m/z 253, and extending well beyond m/z 800 as a continuous elevated baseline, are also detected by ESI-MS. None of these groups or patterns were observed in the absence of methylamine [Hastings et al., 2005], and the magnitude of the baseline signal was ten times lower than in Figure 1. These patterns, as will be shown below, are consistent with the formation of imine oligomers.
 In order to probe these reactions under more atmospherically relevant conditions, aqueous aerosol drying was simulated in chamber experiments. In 10 min at 90% RH, droplets containing 8.8 mM glyoxal and 17 mM methylamine lost >99.99% of their volume due to water loss, and the resulting aerosol particles contained significant amounts of oligomerized products, as seen by AMS (Figure 3). Decomposition during AMS aerosol evaporation adds to strong EI fragmentation at 70 eV (their combined effect we will refer to as “fragmentation” below), while unfragmented parent molecules (as H+ adducts) are detected by ESI-MS. While ESI-MS spectra are typically simpler than AMS spectra, in this case the complex oligomeric pattern detected with ESI-MS (Figure 1), resolves in the HR-ToF-AMS into superimposed Δm/z 13 and Δm/z 71 patterns (Figure 3), indicating that many oligomers decompose into the same “building-block” fragments. These patterns, analyzed at high resolution for exact mass information, suggest that oligomers are rapidly formed by repeated additions of the imine (CH3N=CHCHO, m/z 71). Potential product structures consistent with exact mass and NMR data (Figure S1 of the auxiliary material) are shown in Figure 2. A peak consistent with the protonated imine is observed by AMS at m/z 72. The ESI-MS peaks assigned to imine dimers are again visible by AMS at m/z 125 and 143, but peaks assigned to hydrated species (m/z 90, 161) are negligible due to water loss during fragmentation. The prominent peak at m/z 97 is consistent with 1,3-dimethylimidazole. See Figures S2–S4 for control runs.
 The proposed imine dimer structures with aldehyde groups have enol tautomers that can undergo aldol condensation with glyoxal or imine molecules, as shown in Figure 2. The m/z 214, 285, 356 “a” series could result from the repeated addition of imine molecules to a m/z 143 dimer. The addition of glyoxal by aldol condensation, followed by imine addition, could produce the m/z 201, 272, 343 “b” series. Most of these peaks are visible in the ESI mass spectrum (Figure 1). However, except for m/z 214, these peaks are not prominent in the AMS spectrum (Figure 3) likely because of H2CO loss (Δm/z -30) during fragmentation. Loss of H2CO can produce the corresponding fragment series m/z 171, 242, 313, 384, 455, 546, 597 (“b”) and 184, 255, 326, 297, 468, 539, 610 (“a”) shown in the left box in Figure 2. These fragment peaks dominate the AMS spectrum, but with the exception of m/z 171, are not prominently observed by ESI-MS, as expected. A third Δm/z 71 series is observed at 229, 300, 371, 442, and 513, likely due to hydrolysis and loss of a methylamine unit (Δm/z −13) from the series starting at m/z 171. The base of all of these fragment ion series appears to be the dimer fragment observed in Figure 3 at m/z 113 (C5H9N2O), formed either by H2CO loss from a m/z 143 imine dimer or by oligomer fragmentation.
3.1. Concentration Dependence
 The concentration dependence of glyoxal + methylamine reactions was studied by independently varying the concentrations of each reactant in aqueous droplets, evaporating the droplets at RH > 70%, and analyzing the resulting aerosol by Q-AMS. In the first set of experiments, glyoxal concentrations were varied from 0 to 8.8 mM while [methylamine] = 16.8 mM. Fifteen different product ions were analyzed (Figure S5) and were found to vary as a function of [glyoxal]1.0 ±0.2, indicating that the reaction is 1st order with respect to glyoxal, consistent with the reaction of glyoxal with ammonium salts under similar conditions (pH < 6 and [NH4+] < 2 M) [Nozière et al., 2009]. In a second experiment, [glyoxal] = 8.8 mM while methylamine concentrations were varied between 0.35 and 9.3 mM. Four product ions (that were not produced by glyoxal in the absence of methylamine) varied proportionally to [methylamine]0.8 ±0.3. The total reaction order is thus 1.8 ± 0.4 (2nd order overall), the same as that determined for glyoxal + amino acid reactions by a similar analysis and by 1H-NMR kinetics [De Haan et al., 2009]. The formation of oligomers and imidazole derivatives (that contain two or more molecules of each reactant) in reactions that are 1st order in each reactant suggests that imine formation is the rate-limiting step. The imine appears to be an intermediate in the production of oligomers, which is likely the reason it is not detected in longer, bulk-phase experiments.
3.2. Reactant Volatilization
 The glyoxal + methylamine experiments just described also provide information about volatilization of the reactants. In experiments where droplet glyoxal concentrations are very low relative to methylamine, if the excess methylamine evaporates, very little material will be left behind as aerosol when water evaporates from the droplet. This is exactly what is observed in three SMPS experiments (Figure S6). When the concentration of glyoxal is low, minimal aerosol is produced; furthermore, Q-AMS imidazole product peaks at m/z 97, 42, and 15 disappear. When glyoxal is added in, it reacts with an equivalent amount of methylamine and aerosol volumes are seen to increase sharply, demonstrating that glyoxal traps methylamine in the aerosol phase via the reactions described above.
 The behavior of glyoxal is fundamentally different. If all excess glyoxal partitioned to the gas phase upon droplet evaporation, very little aerosol material would be observed when methylamine concentrations are low. Instead, aerosol volumes remain high when methylamine concentrations are low. When methylamine is added in, glyoxal peaks decline and imidazole and imine peaks reappear in the aerosol mass spectrum (Figures S2, S7, and 3), but aerosol volumes increase by only 48% (Figure S6, circles). This nearly matches the theoretical volume increase of 47% if the same fraction of glyoxal remained in the aerosol phase but reacted with methylamine instead of itself (assuming that both reactions produce oligomers of the same density). Comparing these SMPS size distributions with that of NaCl aerosol generated under the same conditions indicates that about half of the glyoxal volatilizes from drying droplets initially containing 8.8 mM glyoxal in 1 h at RH > 70%, whether or not methylamine is present. This is consistent with recent observations of glyoxal volatilization from aqueous-phase aerosol at RH > 55% [Galloway et al., 2008].
 HR-ToF-AMS elemental analysis [Aiken et al., 2007, 2008] of aerosol produced from an aqueous solution containing 8.8 mM glyoxal and 16.8 mM methylamine reveals that the aerosol had a significantly lower N/C atomic ratio (0.32) than the starting mixture (0.49, calculated). Since molecules containing equal numbers of glyoxal and methylamine subunits (imine oligomers) are expected to have N/C ratios nearly identical (0.33) to those measured, the observed ratio is consistent with the formation of such oligomer products and the evaporation of excess methylamine. Thus, these experiments demonstrate that glyoxal can trap volatile amine compounds in the aerosol phase by oligomer-forming reactions that occur during cloud droplet evaporation. Aqueous-phase reactions between glyoxal and amines provide an explanation for the consistent co-detection of oligomers, amines and oxalic acid in Riverside, CA [Denkenberger et al., 2007]. While a fraction of aqueous-phase glyoxal may be oxidized to oxalic acid during the ∼15 min. lifetime of a cloud or fog droplet, the remaining glyoxal may oligomerize with amines upon droplet evaporation, producing aerosol containing oligomers, amines, and oxalic acid.
 While we have not measured reaction rates in this study, we have previously showed that glyoxal + amino acid reactions are fast enough to occur during and after cloud droplet evaporation under atmospheric conditions [De Haan et al., 2009]. The atmospheric significance of glyoxal + methylamine reactions will therefore likely depend on the relative concentrations of glyoxal and amines in clouds and aerosol. Alkyl amines are produced by marine organisms, animal husbandry, biomass and fossil fuel burning, meat cooking, and industrial activities [Murphy et al., 2007]. High (0.01–6 μg/m3) levels of amine compounds have been measured in aerosol and fog in California's Central Valley [Sorooshian et al., 2008; Zhang and Anastasio, 2001, 2003] and in winter aerosol in Logan, UT [Silva et al., 2008]. Methylamine concentrations averaged 9 μM in Central Valley fog [Zhang and Anastasio, 2003]. Glyoxal concentrations in rural fog are typically lower [Matsumoto et al., 2005; Munger et al., 1995]. Functional group analysis of filtered particles collected at a range of sites indicates that amine groups typically account for 2–13% of organic matter and appear to correlate with fossil fuel burning [Liu et al., 2009; Russell et al., 2009]. This is roughly consistent with aerosol atomic N/C ratios detected by HR-ToF-AMS of ∼0.02 in urban areas [Aiken et al., 2008]. It is likely that the highest coincident levels of glyoxal and alkyl amines will be in polluted regions, in biomass burning plumes, and over biologically active agricultural and marine regions.
 If we assume that amine concentrations typically exceed those of glyoxal in clouds and aerosol, then we can use recent estimates of SOA formation from glyoxal [Fu et al., 2008] to calculate an upper limit for the contribution of amines to global SOA budgets via reactions with glyoxal, using methylamine as a proxy since it is a dominant amine in cloudwater. 2.6 Tg C yr−1 SOA is formed globally by glyoxal if its uptake is irreversible [Fu et al., 2008], which equals 7.6 Tg SOA yr−1 if glyoxal reacts with itself. Using our observation that the addition of equimolar quantities of methylamine to glyoxal solutions increased aerosol volumes by 48%, and assuming no change in aerosol density, a total of 11.2 Tg SOA yr−1 would be formed when glyoxal reacts with amines instead of itself, of which the increase of 3.6 Tg SOA yr−1 can be attributed to amines. This corresponds to an additional SOA source from methylamine of 1.4 Tg C yr−1 and 1.6 Tg N yr−1, or about 3% of total WSOC [Fu et al., 2009], a level that is consistent with aerosol amine measurements [Liu et al., 2009; Russell et al., 2009].