Aqueous OH oxidation of ambient organic aerosol and cloud water organics: Formation of highly oxidized products

Authors


Abstract

[1] Aqueous chemistry can play a vital role in secondary organic aerosol (SOA) formation and aging. A novel analytical approach that allows for simultaneous photo-oxidation and atomization of reacting bulk solutions coupled to an aerosol mass spectrometer (AMS) investigates aqueous OH oxidation of ambient biogenic SOA, cloud water from a biogenic environment, glyoxal, and mixtures of glyoxal with α-pinene SOA components. This is the first study of aqueous oxidative aging of ambient SOA and cloud water organics. Starting with an AMS-based observational framework, we show that aqueous oxidation of biogenic SOA in the presence of glyoxal can better represent observed atmospheric aging than when glyoxal is absent. Oxidation of glyoxal alongside semi-volatile SOA components leads to the production of highly oxidized SOA.

1. Introduction

[2] Secondary organic aerosol (SOA) modeled to form from traditional gas-phase oxidation mechanisms cannot match observed SOA loadings [De Gouw and Jimenez, 2009; Heald et al., 2010] and degree of oxidation [Ng et al., 2010]. Aqueous chemistry has been proposed to explain these discrepancies [Chen et al., 2007; Carlton et al., 2008; Fu et al., 2008; Dzepina et al., 2009; Ervens and Volkamer, 2010]. In particular, water-soluble molecules such as glyoxal and methylglyoxal, formed from both biogenic and anthropogenic volatile organic compounds, are precursors of SOA because they strongly partition into condensed-phase water within aerosols, fogs, and clouds [Galloway et al., 2009; Ip et al., 2009; Volkamer et al., 2009], and can lead to formation of highly oxygenated species [Lim et al., 2010] through aqueous-phase oxidation [Herrmann et al., 2010, and references therein].

[3] An observational framework for the characterization of ambient oxygenated organic aerosols (OOA) has been created by Ng et al. [2010] from analysis of 43 Northern Hemispheric AMS field datasets. The ambient measurements cluster within a distinct space defined by the mass fraction of AMS spectral intensity at m/z 44 and 43 (f44, f43), as indicated by the grey shaded region in Figure 1a. Ambient semi-volatile OOA (SV-OOA or less oxidized OOA) and lab-SOA generated from gas-phase oxidation concentrate in the lower half of the space while ambient low-volatility OOA (LV-OOA or more oxidized OOA) mainly resides in the upper part, suggesting that lab-SOA is more similar to SV-OOA but less oxidized than LV-OOA. Although this may be due to higher aerosol loadings and limited residence time, oxidative aging will transform SV-OOA to LV-OOA [Jimenez et al., 2009] and lead to movement within the f44 vs. f43 space towards the upper vertex of the grey shaded triangle.

Figure 1.

Field measurements: (a) evolution pathways of aqueous OH oxidation of ambient SOA and cloud water collected at Whistler during summer 2010 on the f44 vs. f43 space. The time interval between each data point is 1 min. The grey shaded triangular region arises from analysis of AMS field datasets [Ng et al., 2010]. Ambient SV-OOA and LV-OOA mainly reside in the lower and upper part of the triangular region, respectively. Oxidation trajectories connect the SV-OOA and LV-OOA regions, as shown by the grey arrows. Organic AMS spectra of: (b) the water-soluble fraction of fresh biogenic SOA and (c) a photo-oxidized sample.

[4] We report the first studies of aqueous OH oxidation of the water-extractable fraction of fresh biogenic SOA, aged biogenic SOA, and the organic component of cloud water samples collected at Whistler, British Columbia. A novel analytical approach that allows for online mass spectrometric characterization of aerosolized organics from a photo-oxidation reactor was employed. The SOA and cloud water were collected within a coniferous forest during a warm summer period that resulted in high levels of organics and very low levels of sulfate. The goal of this study is to investigate whether the nature of the organic materials that form through aqueous oxidation is similar to that observed globally. Laboratory studies are used to interpret the field measurements.

2. Methods

[5] Aqueous-phase OH oxidation was performed using an ozone-free Hg lamp (UVP, 254 nm) inserted into a sealed, dark glass bottle at room temperature. OH was produced by photolysis of H2O2 (Sigma-Aldrich, 30 wt% in H2O). The solution was atomized continuously by ultrapure compressed air (BOC, Grade 0.1) throughout the OH exposure period using a TSI atomizer (Model 3076). The particles passed through a diffusion dryer and were analyzed by the Aerodyne Time-of-Flight AMS (CToF-AMS) with proper dilution (Figure 2).

Figure 2.

Experimental photooxidation set-up.

[6] Aqueous-phase OH oxidation of glyoxal (Sigma-Aldrich, 40 wt% in H2O) with and without cis-pinonic acid (Sigma-Aldrich, 99.8%) was conducted for 5 hours in a 1 L reaction vessel. The initial total molar concentrations of these two compounds were 500 μM, and the glyoxal to cis-pinonic acid (Gly:PA) molar ratios were: 0:1, 1:1, 3:1, 4.7:1, 9:1 and 1:0. The initial concentration of H2O2 was 13.3 mM. From the observed first-order decay of glyoxal and using the second-order rate constant of aqueous OH oxidation of glyoxal, the estimated steady state OH concentration was 10−12 to 10−13 M in the pure glyoxal oxidation experiment, similar to cloud water values. Note that there was no observable decay of glyoxal in control experiments with H2O2 absent (data not shown).

[7] Laboratory generated SOA was produced by the dark ozonolysis of α-pinene in a flow tube reactor [George and Abbatt, 2010]. The reactor flow passed through ozone and VOC denuders and the chemical composition of α-pinene SOA was directly determined by AMS. The α-pinene SOA was also collected on 47mm Teflon filters (2.0 μm pores) for 6 hours and then extracted by 18 MΩ water. The aqueous extracts underwent OH oxidation for 30 minutes while being atomized in a 100 ml reaction vessel with initial H2O2 concentration of 25 mM. The glyoxal to α-pinene SOA mass ratios (Gly:SOA) were 0:1, 1:1 and 2:1.

[8] Aerosol filter samples and cloud water were collected at Whistler, British Columbia, during summer 2010 (WACS 2010 campaign). The sampling site was located at a coniferous forest mountain site, allowing for the collection of fresh biogenic SOA. During a biogenic episode (i.e., a warm (Tmax ≥ 23°C) and dry period), over 85% of the total aerosol mass was organic. The fresh SOA sample was collected on July 10, several days after the onset of this event. The aged SOA sample was collected after the biogenic episode (July 19), (Tmax < 20°C) and the particles consisted of ∼30–50 wt% inorganics and had a higher m/z 44-to-43 ratio than fresh SOA. The ambient aerosol passed through a cyclone with PM1 cut size (UGR, Model 463) and was collected on the 47 mm Teflon filter for 24 hours. The cloud water sample was collected immediately after the biogenic episode (July 12) by an automated cloud water collector [Hutchings et al., 2009]. The aqueous extracts of the ambient SOA filters and cloud water were oxidized for 10 minutes on site using a 100 ml atomizer with initial H2O2 concentration of 70 mM.

3. Results and Discussions

[9] The initial coordinates of our ambient samples (fresh SOA, aged SOA, and cloud water) fall within the grey region of the f44 vs. f43 space (Figure 1a). The cloud water organic mass spectrum is similar to that of aged biogenic SOA (data not shown), and is consistent with the cloud droplets having formed from activated SOA particles. During aqueous OH oxidation, f44 values of the ambient samples increase continuously and their trajectories connect the less oxidized (SV-OOA) and more oxidized (LV-OOA) regions, as shown by the direction of the grey arrow. The organic mass spectra of fresh biogenic SOA and photo-oxidized aqueous extracts are shown in Figures 1b and 1c, respectively.

[10] By studying laboratory substances we demonstrate that OH oxidation of mixtures of glyoxal and biogenic SOA components matches this observed behavior of ambient samples. The oxidation pathway of glyoxal starts at the plot origin (Figure 3a) because it only has AMS major fragments at m/z 29, 30, 47 and 58 [Chhabra et al., 2010]. Oxidation of glyoxal leads to a significant increase of f44 but f43 remains at a low level. The increase in m/z 44 arises from the major condensed-phase oxidation products, glyoxylic and oxalic acids, under fog- and cloud-relevant glyoxal concentrations (e.g., <1 mM) [Lim et al., 2010]. The individual mass spectra of both glyoxylic and oxalic acid lie on the evolution pathway of glyoxal oxidation on the f44 vs. f43 space (Figure 3a).

Figure 3.

Laboratory Results: Evolution pathways of aqueous OH oxidation of (a) glyoxal, its oxidation products, and its mixtures with cis-pinonic acid, and (b) lab α-pinene SOA and its mixture with glyoxal. The time intervals between each data point in Figures 3a and 3b are 14 and 2 min respectively. The open red circle in Figure 3b reflects lab α-pinene SOA collected in situ. The dashed line bounded region is from Ng et al. [2010]. The yellow shaded area represents the boundary of the results from the oxidation of glyoxal and cis-pinonic acid. A simple reaction kinetic model (see text) that qualitatively reproduces the shape of the oxidation trajectory of cis-pinonic acid is shown in Figure 3a (red crosses), i.e., a species giving f43 is formed as an intermediate from oxidation of cis-pinonic acid, which is then oxidized to form a species that gives f44.

[11] Although the reaction between glyoxal and OH radical leads to formation of highly oxygenated organics (O:C ratios of glyoxylic and oxalic acid are 1.5 and 2, respectively) with an f44 value close to a very oxidized ambient aerosol, aqueous oxidation of glyoxal alone cannot explain the chemical characteristics and aging of ambient OOA, unless there is a large degree of oligomerization under particularly high concentrations [Lim et al., 2010]. In particular, the m/z 43 fragment (thought to be predominantly C2H3O+ for ambient OOA [Ng et al., 2010]) is largely absent at 10−4 M concentrations.

[12] Cis-pinonic acid, a major condensed-phase product from α-pinene ozonolysis, was also chosen for laboratory study because monoterpenes are the major SOA-forming species emitted in coniferous forests, as at Whistler. Its evolution pathway on the f44 vs. f43 space is notably different than that of pure glyoxal (Figure 3a). While f44 increases throughout the OH exposure, f43 increases initially but then decreases. This arises because m/z 43 (C2H3O+) arises from intermediates along the oxidation pathway, such as carbonyls, that are then converted to species that yield m/z 44, e.g., organic acids. A simple kinetic model where an organic precursor is oxidized by OH to f43 that is then oxidized in part to f44 qualitatively describes this behavior (Figure 3a). Further extending the OH exposure increases the f44 value only slightly, i.e., it is very hard to form products as oxidized as ambient LV-OOA from oxidation of such species, and the final mixture has a composition that is not observed in the field. Thus, oxidation of cis-pinonic acid also does not well represent the atmospheric aging of SOA. Similar results arise in the OH heterogeneous oxidation of squalane aerosol which requires unrealistically high OH exposures and produces aerosol outside the ambient regime within the f44 vs. f43 space [Kroll et al., 2009; Ng et al., 2010].

[13] Given that both highly-soluble small oxygenates such as glyoxal and less oxidized intermediates such as cis-pinonic acid are expected to form from oxidation of some common SOA precursors such as isoprene and α-pinene, we stress that simultaneous oxidation of these species may conceptually better represent ambient behavior in a biogenic-rich environment. Indeed, the evolution pathways of glyoxal/cis-pinonic acid mixtures fall within the yellow shaded area bounded by pure glyoxal and cis-pinonic acid (Figure 3a), and the oxidation trajectories are strikingly similar to the behavior of ambient organics (Figure 1a). Furthermore, the results indicate that addition of glyoxal can enhance the formation of highly oxidized organics. For all mixtures, oxidation to longer times than presented here only causes the evolution pathways to move along the right edge of the yellow shaded region instead of exceeding the boundary (data not shown).

[14] To more realistically match ambient behavior, the water-extractable fraction of α-pinene SOA formed from dark ozonolysis was also oxidized. The AMS mass spectrum of α-pinene SOA generated in the flow reactor is similar to that obtained previously [George and Abbatt, 2010]. The aqueous extracts of α-pinene SOA from the filter sample have lower f43 and slightly higher f44 than the values for particles measured in situ from the flow reactor (Figure 3b), likely due to evaporative loss of semi-volatile SOA components during aerosol collection and filter extraction. Aqueous OH oxidation of α-pinene SOA shows the same two-step kinetic behavior described earlier and the trajectory moves along the right boundary of ambient region, in a manner similar to that demonstrated by cis-pinonic acid. Also, as with the cis-pinonic acid experiment, addition of glyoxal to α-pinene SOA leads to the shifting of oxidation pathways from right to left in the f44 vs. f43 space.

[15] Because of the importance of m/z 29 to the AMS spectra of species such as glyoxal and methylglyoxal, monitoring changes in the organic m/z 29 fraction (f29) is useful for investigating aerosol aging when glyoxal is present. In particular, the two-step kinetic behavior of f29 can be observed from the aqueous OH oxidation of α-pinene SOA on an f44 vs. f29 plot (Figure 4a). Similar to the evolution of m/z 43, this suggests that m/z 29 likely corresponds to intermediates, such as aldehydes, producing the CHO+ fragment. The strong linear negative correlation between f44 and f29 observed with pure glyoxal implies the direct conversion of glyoxal to organic acids, and the coordinates of glyoxylic and oxalic acids lie along the oxidation trajectory (Figure 4a). For glyoxal mixed with the α-pinene SOA components, the correlations between f44 and f29 become less linear and deeper slopes are observed with decreasing initial glyoxal concentrations. This is because at lower glyoxal concentrations conversion of glyoxal to organic acids becomes less significant in affecting the aerosol composition.

Figure 4.

Evolution pathways of aqueous OH oxidation of glyoxal and its mixtures with α-pinene SOA, ambient SOA and cloud water (a) on the f44 vs. f29 space, and (b) on the f43-f44-f29 ternary diagram. The f29, f43 and f44 values on the ternary diagram are normalized to make their total equal to 1, and are denoted with a “primed” symbol. The dashed lines extended from the oxidation pathways in 4b emphasize the different trajectories. The coordinates of α-pinene SOA (red square) [Bahreini et al., 2005], glyoxal uptaken by H2SO4 particles (blue square) [Liggio et al., 2005], wood burning aerosols (green square) [Lanz et al., 2007] and levoglucosan (BBOA marker, orange square) [Schneider et al., 2006] are shown. The coordinates of standardized ambient hydrocarbon-like organic aerosol (HOA, green circle), LV-OOA (yellow circle), SV-OOA (blue circle) and BBOA (purple circle) from Ng et al. [2011] are also displayed.

[16] In comparison between our laboratory and field data, the oxidation pathways of fresh biogenic SOA, aged biogenic SOA, and cloud water are similar to those of α-pinene SOA-glyoxal mixtures on the f44 vs. f43 space (Figures 1a and 3b). In contrast, within the f44 vs. f29 space (Figure 4a), while the fresh biogenic SOA (July 10) evolved in a manner similar to α-pinene SOA oxidation, the oxidation paths of aged SOA (July 19) and cloud water (July 12) fell between the cases of α-pinene SOA and its mixture with glyoxal (1:1) instead. To detangle the interrelationships between m/z 29, 43 and 44 as they evolved during atmospheric oxidation, an f29-f43-f44 ternary diagram is presented to better describe our lab-SOA and field data. When plotted in this manner (Figure 4b), fresh biogenic SOA behaves in a comparable manner to α-pinene SOA, but now the aged SOA and cloud water are more easily seen evolving along trajectories closer to those of α-pinene SOA-glyoxal (1:1) mixtures.

[17] Note that all trajectories point towards the f44 vertex of the ternary diagram. Since the oxidation of fresh biogenic SOA cannot reproduce the initial f29 value of either aged SOA or cloud water organics, additional sources of water-soluble organics that yield intense m/z 29 fragments likely exist. Uptake of glyoxal by aerosols and cloud droplets in such biogenic-rich environments is a reasonable but not unique explanation, given that other species can contribute to m/z 29. For example, biomass-burning organic aerosols (BBOA) can be another major source of m/z 29 to AMS organics [Lanz et al., 2007; Ng et al., 2011] (see Figure 4b) but we note that there was no significant biomass burning observed within our sampling period at Whistler.

4. Summary

[18] This is the first study to investigate the effects of glyoxal on the aqueous oxidation of biogenic SOA. By comparing laboratory and field observations, aqueous oxidation of glyoxal, cis-pinonic acid or α-pinene lab SOA alone cannot accurately represent the aging and characteristics of the ambient organic aerosols or cloud water organics. However, aerosol that matches the composition of ambient organics is formed when mixtures of glyoxal and other organics are oxidized. Specifically, aqueous oxidation of SOA materials in the presence of glyoxal readily leads to the formation of heavily-aged, LV-OOA-like components. It is possible that other soluble species, co-oxidized with ‘traditional’ SOA material may have the same effect. Furthermore, the good agreement between the oxidative pathway of glyoxal and the coordinates of glyoxalic and oxalic acid in Figures 3a and 4a is consistent with current aqueous chemistry models [Lim et al., 2010].

[19] Oxidative aging may occur in the aqueous or gas phases, or heterogeneously. Depending on the hygroscopicity of SOA and inorganics present, particles can absorb water under increasingly humid conditions and can activate to form cloud droplets in supersaturated air. With significant global emissions [Fu et al., 2008] and its high effective Henry's law constant of glyoxal [Ip et al., 2009; Volkamer et al., 2009], glyoxal and biogenic SOA materials may co-exist and, depending on the degree of phase separation into aqueous and organic components, be well mixed in cloud water and possibly in sub-micron aerosol droplets. Condensed-phase OH radicals can be formed photolytically in aqueous solution or via uptake from the gas phase (up to 10−13 M in cloud water). The atmospheric significance of aqueous-phase oxidative aging will then depend on the water content and inorganic fractions of atmospheric droplets, solubility of pre-existing organics and OH radical production rate. Even if phase separation in cloud droplets or aerosol particles occurs into aqueous- and organic-rich fractions, such that highly soluble species such as glyoxal are subject to aqueous-phase oxidation and the organic phase to perhaps heterogeneous oxidation, the final composition of the oxidized aerosol ensemble may resemble that observed here, i.e., the aging of all organic aerosol components does not necessarily proceed in the same phase.

Acknowledgments

[20] This work was supported by NSERC and Environment Canada.

[21] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.

Ancillary