Gas-Phase Water-Mediated Equilibrium Between Methylglyoxal

Gas-phase water-mediated equilibrium between SPECIAL FEATURE methylglyoxal and its geminal diol

Jessica L. Axsona,b, Kaito Takahashic, David O. De Haand, and Veronica Vaidaa,b,1

aDepartment of Chemistry and Biochemistry, and bCooperative Institute for Research in Environmental Sciences, Campus Box 215, University of Colorado, Boulder, CO 80309; cThe Institute of Atomic and Molecular Sciences, P.O. Box 23-166, Taipei, Taiwan 10617, Republic of China; and dDepartment of Chemistry and Biochemistry, University of San Diego, 5998 Alcala Park, San Diego, CA 92110

Edited by Barbara J. Finlayson-Pitts, University of California, Irvine, CA, and approved December 15, 2009 (received for review October 20, 2009)

In aqueous solution, aldehydes, and to a lesser extent ketones, mined computationally that the aldehydic C ¼ O is more favor- −1 hydrate to form geminal diols. We investigate the hydration of ably hydrated in solution (ΔG ¼ −1.4 kcal mol ) than the −1 methylglyoxal (MG) in the gas phase, a process not previously ketonic C ¼ O(ΔG ¼þ2.5 kcal mol ) (21). In solution, MG considered to occur in water-restricted environments. In this study, is present primarily as MGD (60% diol to 40% tetrol) with we spectroscopically identified methylglyoxal diol (MGD) and the aldehydic group forming a geminal diol (21, 23). MGD obtained the gas-phase partial pressures of MG and MGD. These has a lower vapor pressure than MG, which allows the molecule results, in conjunction with the relative humidity, were used to ob- to partition more easily into the particle phase, lending to the tain the equilibrium constant, KP, for the water-mediated hydration formation of SOA. of MG in the gas phase. The Gibbs free energy for this process, ΔG°, obtained as a result, suggests a larger than expected gas-phase diol O OH concentration. This may have significant implications for under- CH [R1] standing the role of organics in atmospheric chemistry. + HOH 3 OH 3CH O O hydration ∣ equilibrium constant ∣ water clusters CHEMISTRY tmospheric aerosols are a major topic of current atmospheric Although the hydration of small aldehydes in aqueous solu- Astudies, given the important, but not fully understood, role tions is known to be extensive, gas-phase hydration of carbonyls that aerosols play in the Earth’s radiative balance (1). Aerosols has not yet been considered in atmospheric models because it is affect global radiative forcing directly by absorbing or scattering commonly believed that there is not enough water present to radiation and indirectly by enhancing cloud albedo (1). Because make such reactions favorable. Gas-phase studies of glyoxylic of their changing chemical composition, aerosol optical and acid performed in our laboratory observed hydration of glyoxylic physiochemical properties vary, which greatly complicates the acid and allowed for the identification of the gem diol through IR quantification of their global radiative forcing. Efforts at model- spectroscopy, suggesting this chemistry can occur in water re- ing aerosol effects on climate, as assessed by the Intergovernmen- stricted environments (28). In a CCl4 matrix with restricted tal Panel on Climate Change (1), are primarily based on sulfate water present, similar results were obtained for the hydration aerosol studies, though there is growing evidence that organic of pyruvic acid. aerosols play an important role in climate change (2–4). Organic In this IR spectroscopic study, we identify and assign the vibra- molecules formed by the oxidation of biogenic and anthropogenic tional features of gas-phase MGD and characterize the water- mediated gas-phase equilibrium between MG and MGD. The organic emissions have been identified as important components K of atmospheric aerosols (2, 5). The formation pathways of these equilibrium constant, P, is calculated using spectroscopically de- secondary organic aerosols (SOA) remain highly speculative, termined concentrations of MG, MGD, and water. This study leading to uncertainties in predictions of atmospheric models shows gas-phase hydration to be significant even under relatively (6–8). A large SOA source is missing from models as illustrated dry environmental conditions. This affects the gas-phase/aqueous by simultaneous field measurements of volatile organic com- particle partitioning of MG and could provide insight into the pounds (VOCs) and aerosol particles (6, 7, 9, 10). discrepancy between measured and modeled amounts of SOA Organic acids, especially oxalic and pyruvic acid, are found in (29, 30). SOA, though the origin of these acids is not predicted correctly by Results and Discussion the gas-phase chemistry considered in models (11–14). Recent Spectral Identification of MG and MGD. Fundamental gas-phase studies suggest that these acids are made in the aqueous phase, spectra MG and MGD. The fundamental gas-phase IR spectrum particularly in cloud water, by the oxidation of aldehydes (e.g., of MG has been previously observed (31) and provides the basis glyoxal and methylglyoxal) with hydroxyl radicals and other aque- for determining the presence of MGD in this study (Fig. 1). In the ous radical species (14–17). Methylglyoxal (MG) is one of the lower energy region of the MG spectrum (Fig. 1A), from 1,050 to most abundant α-dicarbonyls present in the atmosphere and is −1 1; 500 cm , there are three distinguishable bands which are as- produced from VOCs of both biogenic and anthropogenic origin −1 signed to the CCC asymmetric stretch at 1; 240 cm (ν4), the CH (14, 18–20). −1 bend (ν6)at1; 373 cm , and the CH3 bends (ν7) in the 1,415 to Water-mediated aldehyde chemistry is expected to have impor- −1 1; 440 cm region (Table1). MG contains a ketonic and aldehydic tant consequences to the formation of SOA. For example, labora- −1 C ¼ O, which are present in the MG spectrum at 1; 723 cm (ν8) tory studies of MG hydration performed using theoretical (21) −1 and 1; 741 cm (ν9), respectively (Fig. 1A). The CH stretching and spectroscopic techniques (14, 22–24) suggest that in aqueous environments MG can become hydrated to form methylglyoxal diol (MGD) via reaction 1 and undergo further reactions to form Author contributions: J.L.A., D.O.D.H., and V.V. designed research; J.L.A. and K.T. performed oligomers (14, 21, 24). Like other aldehydes and ketones in research; J.L.A. and V.V. analyzed data; J.L.A. wrote the paper. aqueous solution, MG hydrates through proton addition to the The authors declare no conflict of interest. aldehyde carbonyl and reaction to form MGD (21–23, 25–27). This article is a PNAS Direct Submission. MG has both an aldehydic and ketonic C ¼ O and it was deter- 1To whom correspondence should be addressed: E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.0912121107 PNAS Early Edition ∣ 1of6 Downloaded by guest on October 2, 2021 MGD features under very low relative humidities. In the absence of literature assignments, theoretical frequency and intensity calculations aided in assigning MGD vibrational modes (Table 1). The formation of MGD can be observed by the coincidence of the distinct MGD vibrational modes that appear in the lower energy region from 1,050 to 1; 500 cm−1 (Fig. 1B, Inset), the decrease in the ν9 intensity relative to ν8 intensity, and the appear- −1 δ ance of OH stretching vibrations at 3; 505 cm (ν13) and −1 δ 3; 585 cm (ν14) (Fig. 1B) with the addition of water vapor. With water present, the CH stretching mode of MGD is observed at −1 δ 2; 895 cm (ν12). Because the lower energy region of the spectra from 1,050 to 1; 500 cm−1 becomes increasingly complex due to overlapping bands from the addition of water, the intensities of δ δ the ν8, ν9, ν13, and ν14 stretches were used to quantify and follow the formation of MGD.

C ¼ O fundamental and overtone stretching region. With the addition of trace amounts of water, the C ¼ O stretching region is altered by the decrease in intensity of ν9 relative to the ν8, pre- sumably due to the formation of MGD. To better understand these changes in the fundamental C ¼ O stretching region, the intensity and peak width (FWHM) of the fundamental and overtone C ¼ O stretches for MG and MGD were modeled by two Lorenzian curves. The fundamental MG C ¼ O stretching region shows ν8 and ν9 with roughly the same intensity (Fig. 2A). As MG is hydrated to form MGD, the aldehydic C ¼ O is hydrated to form the gem diol, which can be observed spectroscopically by the decrease in the relative intensity of ν9 in Fig. 2B, intensity ν8∶ν9 ¼ 1.0∶0.75. In addition to the MG and MGD fundamental C ¼ O stretches near 1; 730 cm−1, there is a third peak around −1 δ 1; 780 cm (ν10), which is attributed to the diol because of its intensity increase mirroring the increase in partial pressure of 1; 050 −1 3; 600 −1 water.TheMGandMGDC¼ O first overtone region near Fig. 1. Fundamental gas-phase spectra from cm to cm of −1 −1 (A) MG and (B) MG and MGD. MGD frequencies are labeled with δ. 3; 450 cm has the ketonic C ¼ Oovertoneat3; 443 cm −1 (2ν8) and the aldehydic C ¼ O overtone at 3; 458 cm (2ν9). vibration of MG has been previously assigned (31) and occurs at In both MG and MGD spectra, the intensity of the first overtone −1 C ¼ O stretches drops by at least a factor of 10 from the intensity 2; 835 cm (ν11). The MG CH stretch is very prominent and was ¼ ¼ used to compare the MG and MGD spectra and to identify MGD. of the fundamental C O stretches. In the MG C O first over- −1 C 2ν ∶2ν ¼ The peak at 3; 450 cm , which normally corresponds to the OH tone spectrum in Figure 2 , the relative intensity of 8 9 1 00∶0 23 D 1 00∶0 17 stretching region, we assign to the first overtone, 2ν8 and 2ν9, of the . . , whereas in Figure 2 it is . . . This relative in- ν D ν8 and ν9 C ¼ O stretches as seen in Fig. 1A. This peak has been tensity decrease seen for 9 in Figure 2 is consistent with the previously observed, but not assigned (31). decrease in the fundamental C ¼ O, reinforcing the suggestion The addition of water to the MG sample hydrates MG to form that the fundamental aldehydic C ¼ O intensity is decreasing be- MGD (Fig. 1B). The spectrum in Fig. 1B shows both MG and cause of hydration of the carbonyl to generate diol.

Table 1. Theoretical and observed experimental frequencies for MG and MGD with vibrational modes assignments Theoretical frequency, cm−1 Theoretical intensity , km mol−1 Experimental frequency, cm−1 Mode assignments δ 1,099 182 1,088 ν1 Hb C-O stretch δ 1,185 73 1,173 ν2 COH wag δ 1,180 18 1,199 ν3 CCC stretch 1,207 22 1,240 ν4 CCC stretch δ 1,275 48 1,292 ν5 Hb COH bend 1,335 2 1,373 ν6 CH bend 1,366–1,415 — 1,415–1,440 ν7 3 × CH3 bend 1,754 132 1,723 ν8 C ¼ O stretch 1,774 114 1,741 ν9 HC ¼ O stretch δ ——1,780 ν10 Diol 2,811 72 2,835 ν11 CH stretch δ 2,838 43 2,895 ν12 CH stretch 3,497 — 3,443 2ν8 C ¼ O overtone 3,530 — 3,458 2ν9 C ¼ O overtone δ 3,478 86 3,505 ν13 Hb OH stretch δ 3,623 55 3,585 ν14 Fr OH stretch

B3LYP∕6-31 þ Gðd; pÞ theoretical frequencies were scaled using wavenumber linear scaling method. δ, MGD vibration; Hb, hydrogen bonding; Fr, free OH.

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912121107 Axson et al. Downloaded by guest on October 2, 2021 SPECIAL FEATURE CHEMISTRY

Fig. 2. Fundamental C ¼ O stretching region of (A) MG and (B) MGD and the first overtone, 2ν8 and 2ν9,of(C)MGand(D) MGD showing a decrease in the relative intensity of ν9 in (B) when compared to (A) and a decrease in the 2ν8 and 2ν9 aldehydic peak in (D) when compared to (C).

OH stretching region of MGD. In the MGD spectrum (Fig. 3), there Determination of MG, MGD, and Water Partial Pressures. Based on are two OH vibrational stretching modes, with the δ denoting the spectroscopic identification of MGD, we quantified the −1 δ −1 MGD vibration, located at 3; 505 cm (ν13)and3; 585 cm amount of MG, MGD, and water present in each spectrum as δ −1 (ν14), in addition to the 3; 450 cm peak assigned to 2ν8 and outlined below. Spectra that were saturated were not used in δ 2ν9. The first OH vibrational stretch at ν13 is attributed to the the calculation of partial pressures, but were useful in making δ hydrogen-bonded OH of MGD and the ν14 is attributed to the MG and MGD mode assignments. δ free OH of MGD. Our theoretical frequency puts the ν13 at 3; 478 cm−1, which is slightly red shifted compared to the experimental value of 3; 505 cm−1. This discrepancy between the theoretically and experimentally derived frequencies is consistent δ δ with results from other studies (28, 32–35). The ν13 and ν14 could also have a contribution from tetrol OH stretching vibrations which all fall near to those of MGD, making it difficult to distin- guish or model them. In solution, the ratio of diol to tetrol is approximately 60∕40 (23). Although this ratio has not been measured for the gas phase, at much lower water concentration in our gas-phase experiments we expect MGD to be favored over the δ tetrol. The ν14 stretch is present in very small amounts even in δ our lowest relative humidity MG spectrum. The ν14 stretch in- creases in intensity in response to increasing the partial pressure of water in the experiments. The region from 3,100 to 3; 350 cm−1 contains water clusters and hydrated complexes of MG and MGD. The long red shift relative to computations and very broad appearance of these features are consistent with observations and predictions of – hydrogen-bonded water clusters (28, 32 34, 36, 37). The broad Fig. 3. The 3; 000 cm−1 to 3; 650 cm−1 region of (A) MGD and (B) MG, show- −1 −1 and overlapping features in this region make it difficult to qua- ing the appearance of water clusters (3; 100 cm to 3; 350 cm ), the 2ν8 and δ δ litatively and quantitatively identify these clusters. 2ν9 of C ¼ O, and the ν13 and ν14 of MGD.

Axson et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on October 2, 2021 Table 2. Partial pressure of water 1; 550 −1 1; 918 −1 3; 920 −1 N∕v −3 P Experiment Int abs cm Int abs cm Int abs cm Avg , molecule cm H2O, atm 1 0.003 0.003 0.002 1.3 × 1015 5.4 × 10−5 2 0.003 0.003 0.002 1.3 × 1015 5.4 × 10−5 3 0.003 0.003 0.002 1.3 × 1015 5.4 × 10−5 4 0.008 0.006 0.005 3.2 × 1015 1.3 × 10−4

P Partial pressures, H2O, of water determined from experimental spectra and HITRAN. Integrated absorbance (Int abs) for each experimental water line was taken. Water lines at 1,550; 1,918; and 3; 920 cm−1 had the following cross-sections: 2.88 × 1020, 2.93 × 1020,and2.61 × 1020 cm molecule−1 (55).

Partial pressure of water. Experimentally observed spectro- phase (21, 22). There have been few thermodynamic studies of scopic water lines were compared with high-resolution transmis- the hydration of MG and none performed in the gas phase. sion molecular absorption database (HITRAN) water lines and As shown in Table 5, under very dry conditions like those used line strengths to determine the partial pressure of water in this study (RH% < 5%), easily detectable quantities of MGD (Table 2), PH2O, and percent relative humidity, RH%, in each ex- are formed. Assuming that this reaction involves equilibrium R1, −1 periment. Three water lines at 1,550; 1,918; and 3; 920 cm were and that the tetrol formation is inefficient under conditions like isolated and analyzed. Each observed water line was integrated those in our investigation, the equilibrium constant KP can be and the H2O number density (N∕v) was determined via Eq. 1: derived from the partial pressures extracted from our spectra using Eq. 2. N∕v ¼ integrated absorbance : [1] L σ P H2O K ¼ MGD [2] P P P L σ −1 Mgly · H2O The path length of the cell, , was 71 cm and H2O (cm molecule ) P is the HITRAN line strength for each waterline. The H2O was The KP values for the four experiments (Table 5) fall within the determined using the average N∕v from the three water lines range of 154–161 (13.0) and reflect MGD production. The low and converted to %RH using the ideal gas law and the vapor pres- relative humidity conditions of the experiments did not allow for sure of water. the water concentration to be varied over a large range and there was little clustering observed by FTIR. As noted previously, there P σ Partial pressure of MG. The partial pressure of MG, MG,was was a discrepancy between the theoretical and literature MG, determined spectroscopically using the integrated absorbance which leads to ∼11% discrepancy in the KP values determined of the MG CH stretch at 2; 835 cm−1 (Table 3) and a theoretically using the same spectra, with the theoretical values being slightly σ 1 19 × 10−17 −1 K calculated MG cross section, MG,of . cm molecule higher than the literature P values. σ ΔG° K (Table 1). The theoretical and literature value for MG were com- The Gibbs free energy ( ) was calculated from P using pared: There is a discrepancy, with the literature value being Eq. 3 higher (31). In this work we chose to use the theoretically calcu- ΔG ¼ −RTð K Þ [3] σ ° ln P lated MG because of the agreement of the calculated and experimental frequencies and intensities (Table 1). For consistency, the Our experimentally determined ΔG° values, which range from σ K −2 98 −3 10 −1 0 051 theoretical MG was used to determine the P from these experi- . to . kcal mol ( . ) (Table 5) reflects production ments. The MG N∕v was determined for each experiment using of MGD in the gas phase. These values for MGD formation are 1 P Eq. , and the MG was then determined using the ideal gas law. significantly more favorable than theoretical predictions, even ones made in the aqueous phase (21, 40). P Partial pressure of MGD. The partial pressure of MGD, MGD, was determined spectroscopically using the MGD OH stretching Experimental. Sample preparation. MGðCH3COCHOÞ in aqueous vibration at 3; 585 cm−1 (Table 4). This peak was used because solution (40 wt%) (Sigma-Aldrich) was dried and distilled before of its presence in each of the spectra containing MGD and its use. A 20 mL sample of MG solution was attached to a vacuum response to water. It is assumed that this peak contains only line at around 15 m torr and gently heated and stirred at 313 K for MGD (although it may also contain a small tetrol contribution) ∼12–15 h. Under these conditions, the solution became increas- andthisfeaturewasintegratedtoobtainanintegratedabsor- ingly viscous. An equimolar amount of P2O5 was added to the bance. The integrated absorbance along with the theoretically sample, which was heated to 393 K and put under vacuum. σ 9 14 × 10−18 −1 determined cross-section, OH,of . cm molecule , Two condensers were employed in the distillation, each cooled whichisinagreementwithatypical alcohol cross-section with cold water, similar to Gurnick et al. (41). A liquid nitrogen (34, 38, 39), was used in Eq. 1 to determine MGD N∕v.The bath was used to trap the purified MG. The MG solution was P 1 MGD was then determined using Eq. and ideal gas law. The orange-brown in color in contrast to the pure MG, which is bright experimentally obtained partial pressures are shown in Table 5. yellow-green, and MGD, which is colorless. The MG was put under nitrogen and kept frozen so it would remain in its aldehydic Water-Mediated Gas-Phase Equilibrium Constant. Previous MG hy- form until the experiment. The MGD was prepared in situ, by dration studies have been primarily performed in the solution introducing small amounts of water to the sample in the spectro-

Table 3. Partial pressures of MG Table 4. Partial pressures of MGD −1 N∕v −3 P −1 N∕v −3 P Experiment Int abs, cm , molecule cm MG,atm Experiment Int abs, cm , molecule cm MGD, atm 1 26.05 3.08 × 1016 1.25 × 10−3 1 0.172 2.65 × 1014 1.08 × 10−5 2 25.93 3.07 × 1016 1.24 × 10−3 2 0.169 2.60 × 1014 1.06 × 10−5 3 25.66 3.04 × 1016 1.23 × 10−3 3 0.171 2.63 × 1014 1.07 × 10−5 4 33.15 3.92 × 1016 1.59 × 10−3 4 0.505 7.78 × 1014 3.16 × 10−5

P P Partial pressures of MG, MG, determined from experimental spectra. Partial pressures of MGD, MGD, determined from experimental spectra. Integrated absorbance (Int abs) for the MG CH stretch at 2; 835 cm−1 was Integrated absorbance (Int abs) for the MGD Fr OH stretch at 3; 585 cm−1 was σ 1 19 × 10−19 −1 σ 9 14 × 10−18 −1 taken. Theoretical calculated MG of . cm molecule is used. taken. Theoretically calculated OH of . cm molecule is used.

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912121107 Axson et al. Downloaded by guest on October 2, 2021 Table 5. Experimental equilibrium constant and gibbs free energy bation theory developed by Barone (51) using the B3LYP method SPECIAL FEATURE ° −1 is reported along with the intensity calculated by the double har- Experiment P ,atm P 2 ,atm P , atm KP ΔG , kcal mol MGD H O MG monic approximation. The optimized structures indicate that 1 1.08 × 10−5 1.25 × 10−3 5.40 × 10−5 159 -3.13 trans 1 06 × 10−5 1 24 × 10−3 5 40 × 10−5 -MG is the most stable structure of MG, with the 2 . . . 157 -3.12 cis 4 9 −1 3 1.07 × 10−5 1.23 × 10−3 5.40 × 10−5 161 -3.13 -MG structure being . kcal mol higher in energy. Calcula- 4 3.16 × 10−5 1.59 × 10−3 1.29 × 10−4 154 -2.98 tions also confirmed that MG would hydrate at the aldehydic group. The theoretical frequencies and intensities along with Experimentally determined gas-phase water-mediated equilibrium the experimental frequencies are presented in Table 1. constant, KP , values calculated using Eq. 2 for reaction R1 between MG and MGD and the experimentally determined Gibbs free energy, ΔG°, Conclusion K values calculated using Eq. 3. Temperature was 298 K. P values have an MG is a known product of VOC oxidation and is prevalent in the error of 13.0 and ΔG° values have an error of 0.05 kcal mol−1. atmosphere. In this work, we obtain and assign the gas-phase IR spectrum of MGD and find that the gas-phase hydration of MG is scopic cell, and allowing MG and water vapor to equilibrate over possible at low relative humidity (RH% < 5%). One of the con- – a period of about 10 30 min. sequences of this gas-phase water-mediated chemistry is a change in the electronic state of the molecule, eliminating the n → π Fourier transformation infrared spectra. The mid-IR absorption transition of the aldehyde carbonyl which is well known to under- spectra of MG and MGD were measured between 1,000 and – 8; 000 −1 0 5 −1 go near-UV photochemistry (52 54). Instead, the OH vibrational cm using FTIR spectroscopy at . cm resolution in chromophore of the diol may react through excitation of the OH a static cell. A Bruker IFSv 66 spectrometer equipped with a glo- vibrational overtone in the near IR to form new products by bar source, KBr beamsplitter, and mercury cadmium telluride or dehydration, decarboxylation, and decarbonylation, as suggested MCT detector was used for optimal mid-IR absorption spectro- – scopy. This setup has been described previously (42, 43). The recently for a number of alcohols and acids (32, 55 58). Using spectroscopic cell was both pressure- and temperature-controlled MG and MGD spectral features, we determine the gas-phase K K and was pumped down to approximately 15 m torr and then water-mediated equilibrium P. This P suggests that the gas- closed off to the pump. The MG sample was introduced to phase formation of the diol in the atmosphere is possible and the cell and water vapor was subsequently added and allowed could be expected to affect gas-particle partitioning of MG and to equilibrate. All experiments were performed at 298 K. its potential to form SOA. CHEMISTRY

Theoretical calculations. The structures and fundamental vibra- ACKNOWLEDGMENTS. V.V. and J.A. would like to thank B. Ervens and K. Plath tional mode frequencies of MG and MGD were calculated using for their insightful comments and R. Talukdar and J.Brukholder for sharing unpublished MG spectra. V.V., D.D.H., and J.A. thank the National Science the hybrid density functional theory method of B3LYP (44, 45) Foundation for support of this work. D.D.H. acknowledges a Cooperative In- with the 6 − 31 þ Gðd; pÞ basis set (46–50) using the Gaussian stitute for Research in Environmental Sciences Fellowship. J.A acknowledges 03 program (48). The anharmonic frequencies using the pertur- a NASA Earth and Space Science Fellowship.

1. Solomon S, et al. (2007) IPCC, 2007: Climate Change 2007: The Physical Science Basis. 17. Warneck P (2003) In-cloud chemistry opens pathway to the formation of oxalic acid in Contribution of Working Group I to the Fourth Assessment Report of the Intergovern- the marine atmosphere. Atmos Environ 37(17):2423–2427. mental Panel on Climate Change (Cambridge Univ Press, Cambridge, UK). 18. Fu TM, et al. (2008) Global budgets of atmospheric glyoxal and methylglyoxal, 2. Kanakidou M, et al. (2005) Organic aerosol and global climate modelling: A review. and implications for formation of secondary organic aerosols. J Geophys Res-Atmos Atmos Chem Phys 5:1053–1123. 113(D15):D15303. 3. De Gouw J, Jimenez JL (2009) Organic aerosols in the Earth’s atmosphere. Environ Sci 19. Jang MS, Kamens RM (2001) Characterization of secondary aerosol from the Technol 43(20):7614–7618. photooxidation of toluene in the presence of NOx and 1-propene. Environ Sci Technol 4. Rincon AG, Guzman MI, Hoffmann MR, Colussi AJ (2009) Optical absorptivity 35(18):3626–3639. versus molecular composition of model organic aerosol matter. J Phys Chem A 20. Tuazon EC, Macleod H, Atkinson R, Carter WPL (1986) Alpha-dicarbonyl yields from 113(39):10512–10520. the NOX-air photooxidations of a series of aromatic-hydrocarbons in air. Environ 5. Zhang Q, et al. (2007) Ubiquity and dominance of oxygenated species in organic aero- Sci Technol 20(4):383–387. sols in anthropogenically-influenced Northern Hemisphere midlatitudes. Geophys Res 21. Krizner HE, De Haan DO, Kua J (2009) Thermodynamics and kinetics of methylglyoxal Lett 34(13):L13801. dimer formation: A computational study. J Phys Chem A 113(25):6994–7001. 6. Heald CL, et al. (2005) A large organic aerosol source in the free troposphere missing 22. Betterton EA, Hoffmann MR (1987) Kinetics, mechanism, and thermodynamics from current models. Geophys Res Lett 32(18):L18809. of the reversable-reaction of methylglyoxal (CH3COCHO) with S(IV). J Phys Chem 7. Volkamer R, et al. (2006) Secondary organic aerosol formation from anthropogenic air 91(11):3011–3020. pollution: Rapid and higher than expected. Geophys Res Lett 33(17):L17811. 23. Nemet I, Vikic-Topic D, Varga-Defterdarovic L (2004) Spectroscopic studies of methyl- 8. Finlayson-Pitts BJ (2009) Reactions at surfaces in the atmosphere: Integration of glyoxal in water and dimethylsulfoxide. Bioorg Chem 32(6):560–570. experiments and theory as necessary (but not necessarily sufficient) for predicting 24. Loeffler KW, Koehler CA, Paul NM, De Haan DO (2006) Oligomer formation in the physical chemistry of aerosols. Phys Chem Chem Phys 11(36):7760–7779. evaporating aqueous glyoxal and methyl glyoxal solutions. Environ Sci Technol 9. Fast J, et al. (2009) Evaluating simulated primary anthropogenic and biomass burning 40(20):6318–6323. organic aerosols during MILAGRO: Implications for assessing treatments of secondary 25. Bell RP, Rand MH, Wynnejones KMA Kinetics of the hydartion of acetaldehyde. organic aerosols. Atmos Chem Phys 9(16):6191–6215. T Faraday Soc 52(8):1093–1102. 10. De Gouw JA, et al. (2005) Budget of organic carbon in a polluted atmosphere: Results 26. Buschmann HJ, Dutkiewicz E, Knoche W (1982) The reversable hydration of carbonyl- from the New England Air Quality Study in 2002. J Geophys Res-Atmos 110(D16): compounds in aqueous-solution 2. The kinetics of the keto gem-diol transition. Ber D16305. Bunsen Phys Chem 86(2):129–134. 11. Ervens B, Feingold G, Frost GJ, Kreidenweis SM (2004) A modeling study of aqueous 27. Wolfe S, Kim CK, Yang K, Weinberg N, Shi Z (1995) Hydration of the carbonyl group—A production of dicarboxylic acids: 1. Chemical pathways and speciated organic mass theoretical-study of the cooperative mechanism. J Am Chem Soc 117(15):4240–4260. production. J Geophys Res-Atmos 109(D15):D15205. 28. Plath KL, et al. Gas-phase vibrational spectra of glyoxylic acid and its gem diol mono- 12. Tervahattu H, Juhanoja J, Kupiainen K (2002) Identification of an organic coating on hydrate Implications for atmospheric chemistry. React Kinet Catal L 96(2):209–224. marine aerosol particles by TOF-SIMS. J Geophys Res-Atmos 107(D16):1–7. 29. Bao L, et al. (2009) Gas/particle partioning of low-molecular-weight diacarboxylic acids 13. Veres P, et al. (2008) Development of negative-ion proton-transfer chemical-ionization at the suburban site in Saitama, Japan. Atmos Environ, in press. mass spectrometry (NI-PT-CIMS) for the measurement of gas-phase organic acids in the 30. Matsunaga S, Mochida M, Kawamura K (2004) Variation on the atmospheric concen- atmosphere. Int J Mass Spectrom 274(1-3):48–55. trations of biogenic carbonyl compounds and their removal processes in the northern 14. Altieri KE, et al. (2008) Oligomers formed through in-cloud methylglyoxal reactions: forest at Moshiri, Hokkaido Island in Japan. J Geophys Res-Atmos 109(D4):D04302. Chemical composition, properties, and mechanisms investigated by ultra-high resolu- 31. Zhou SM, et al. (2008) Atmospheric Chemistry of Acetylacetone. Environ Sci Technol tion FT-ICR mass spectrometry. Atmos Environ 42(7):1476–1490. 42(21):7905–7910. 15. Ervens B, et al. (2008) Secondary organic aerosol yields from cloud-processing of 32. Havey DK, Feierabend KJ, Vaida V (2004) Vapor-phase vibrational spectrum of glycolic isoprene oxidation products. Geophys Res Lett 35(2):L02816. acid, CH2OHCOOH, in the region 2000–8500 cm−1. J Phys Chem A 108(42):9069–9073. 16. Carlton AG, et al. (2007) Atmospheric oxalic acid and SOA production from glyoxal: 33. Plath KL, Takahashi K, Skodje RT, Vaida V (2009) Fundamental and Overtone vibra- Results of aqueous photooxidation experiments. Atmos Environ 41(35):7588–7602. tional spectra of gas-phase pyruvic acid. J Phys Chem A 113(26):7294–7303.

Axson et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on October 2, 2021 34. Vaida V, Feierabend KJ, Rontu N, Takahashi K (2008) Sunlight-initiated photochemis- 46. Clark T, Chandrasekhar J, Spitznagel GW, Schleyer PV Efficient diffuse function- try: Excited vibrational states of atmospheric chromophores. Int J Photoenergy augmented basis-sets for anion calculations 3. The 3 − 21 þ G basis set for 1st-row 2008:1–13. elements, LI-F. J Comput Chem 4(3):294–301. 35. Vaida V (2009) Spectroscopy of photoreactive systems: Implications for atmospheric 47. Frisch MJ, Pople JA, Binkley JS (1984) Self-consistent molecular-orbital methods. 25. – chemistry. J Phys Chem A 113(1):5 18. Supplementary functions for Gaussian-basis sets. J Chem Phys 80(7):3265–3269. 36. Vaida V, Daniel JS, Kjaergaard HG, Goss LM, Tuck AF (2001) Atmospheric absorption of 48. Frisch MJ, et al. (2004) Gaussian03 (Gaussian Inc, Wallingford, CT), C.02 Ed. near infrared and visible solar radiation by the hydrogen bonded water dimer. Q J Roy 49. Krishnan R, Binkley JS, Seeger R, Pople JA (1980) Self-consistent molecular-orbital Meteor Soc 127(575):1627–1643. methods. 20. basis set for correlated wave-functions. J Chem Phys 72(1):650–654. 37. Vaida V, Kjaergaard HG, Feierabend KJ (2003) Hydrated complexes: Relevance to 50. McLean AD, Chandler GS (1980) Contracted Gaussian-basis sets for molecular calcula- atmospheric chemistry and climate. Int Rev Phys Chem 22(1):203–219. Z ¼ 11–18 – 38. Lange KR, Wells NP, Plegge KS, Phillips JA (2001) Integrated intensities of O-H stretch- tions. 1. Second row atoms, . J Chem Phys 72(10):5639 5648. ing bands: Fundamentals and overtones in vapor-phase alcohols and acids. J Phys 51. Barone V (2005) Anharmonic vibrational properties by a fully automated second-order Chem A 105(14):3481–3486. perturbative approach. J Chem Phys 122(1):014108. 39. Miller BJ, Kjaergaard HG, Hattori K, Ishiuchi S, Fujii M (2008) The most stable confor- 52. Chen YQ, Wang WJ, Zhu L (2000) Wavelength-dependent photolysis of methylglyoxal mer of benzyl alcohol. Chem Phys Lett 466(1-3):21–26. in the 290–440 nm region. J Phys Chem A 104(47):11126–11131. 40. Barsanti KC, Pankow JF Thermodynamics of the formation of atmospheric organic 53. Staffelbach TA, Orlando JJ, Tyndall GS, Calvert JG (1995) The UV-visable absorption- particulate matter by accretion reactions - 2 Dialdehydes, methylglyoxal, and spectrum and photolysis quantum yields of methylglyoxal. J Geophys Res-Atmos 100 diketones. Atmos Environ 39(35):6597–6607. (d7):14189–14198. 41. Gurnick M, Chaiken J, Benson T, McDonald JD (1981) Vibrational and rotational spec- 54. Meller R, Raber W, Crowley JN, Jenkin ME, Moortgat GK (1991) The UV-visable absorp- troscopy of the 1st electronically allowed transition of alpha-dicarbonyls. J Chem Phys tion-spectrum of methylglyoxal. J Photoch Photobio A 62(2):163–171. 74(1):99–105. 55. Takahashi K, Plath KL, Skodje RT, Vaida V (2008) Dynamics of vibrational overtone 42. Rontu N, Vaida V (2006) Vibrational spectroscopy of perfluoroproplonic acid in the excited pyruvic acid in the gas phase: Line broadening through hydrogen-atom region between 1000 and 11000 cm−1. J Mol Spectrosc 237(1):19–26. chattering. J Phys Chem A 112(32):7321–7331. 43. Dunn ME, Shields GC, Takahashi K, Skodje RT, Vaida V (2008) Experimental and 56. Donaldson DJ, Tuck AF, Vaida V (2003) Atmospheric photochemistry via vibrational theoretical study of the OH vibrational spectra and overtone chemistry of gas-phase – vinylacetic acid. J Phys Chem A 112(41):10226–10235. overtone absorption. Chem Rev 103(12):4717 4729. 44. Becke AD (1993) Density-functional thermochemistry 3. The role of exact exchange. 57. Mills MJ, et al. (2005) Photolysis of sulfuric acid vapor by visible light as a source of the J Chem Phys 98(7):5648–5652. polar stratospheric CN layer. J Geophys Res-Atmos 110(D8):D08201. 45. Lee CT, Yang WT, Parr RG (1988) Development of the Colle-Salvetti correlation-energy 58. Vaida V, Kjaergaard HG, Hintze PE, Donaldson DJ (2003) Photolysis of sulfuric acid formula into a functional of the electron-density. Phys Rev B 37(2):785–789. vapor by visible solar radiation. Science 299(5612):1566–1568.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.0912121107 Axson et al. Downloaded by guest on October 2, 2021