Aqueous-Phase OH Oxidation of Glyoxal: Application of a Novel

ARTICLE pubs.acs.org/JPCA

Aqueous-Phase OH Oxidation of Glyoxal: Application of a Novel Analytical Approach Employing Aerosol Mass Spectrometry and Complementary Off-Line Techniques Alex K. Y. Lee, R. Zhao, S. S. Gao, and J. P. D. Abbatt* Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, M5S 3H6, Canada

bS Supporting Information ABSTRACT: Aqueous-phase chemistry of glyoxal may play an important role in the formation of highly oxidized secondary organic aerosol (SOA) in the atmosphere. In this work, we use a novel design of photochemical reactor that allows for simultaneous photooxidation and atomization of a bulk solution to study the aqueous-phase OH oxidation of glyoxal. By employing both online aerosol mass spectrometry (AMS) and offline ion chromatography (IC) measurements, glyoxal and some major products including formic acid, glyoxylic acid, and oxalic acid in the reacting solution were simultaneously quantified. This is the first attempt to use AMS in kinetics studies of this type. The results illustrate the formation of highly oxidized products that likely coexist with traditional SOA materials, thus, potentially improving model predictions of organic aerosol mass loading and degree of oxidation. Formic acid is the major volatile species identified, but the atmospheric relevance of its formation chemistry needs to be further investigated. While successfully quantifying low molecular weight organic oxygenates and tentatively identifying a reaction product formed directly from glyoxal and hydrogen peroxide, comparison of the results to the offline total organic carbon (TOC) analysis clearly shows that the AMS is not able to quantitatively monitor all dissolved organics in the bulk solution. This is likely due to their high volatility or low stability in the evaporated solution droplets. This experimental approach simulates atmospheric aqueous phase processing by conducting oxidation in the bulk phase, followed by evaporation of water and volatile organics to form SOA.

’ INTRODUCTION Secondary organic aerosols (SOA) contribute a significant mass fraction of atmospheric particulate matter. In addition to a traditional SOA formation pathway (i.e., SOA formed via partitioning of semivolatile organics produced from gas-phase oxidation), there is growing evidence that water-soluble volatile organics can strongly partition into condensed-phase water, which can be found in aerosol, fog, and cloud droplets and further react in the aqueous phase to form SOA.1 8 Recent atmospheric models predict that uptake of small gas-phase dicarbonyls by aqueous droplets in the atmosphere can produce SOA mass comparable to that produced from the traditional pathways.9,10 Furthermore, models that account for both gasphase and aqueous-phase chemistry can well predict oxalate concentration in rural and remote locations.11 In the center of Atlanta, a significant increase of particulate water-soluble organic carbon was observed when the relative humidity was higher than 70% during the summer of 2007.7 This increase was greater than the prediction of Henry’s law partitioning alone, suggesting aqueous-phase reactive chemistry may be occurring. Glyoxal is a C2 dicarbonyl compound (C2H2O2) with significant production from gas-phase oxidation of isoprene9 and probably from marine environment.12 Glyoxal and its gasphase oxidation products are too volatile for it to be an effective r 2011 American Chemical Society

precursor of traditional SOA. However, due to its atmospheric abundance and unexpectedly high effective Henry’s law constant into inorganic and organic aerosol,3,5,8 glyoxal has been employed as a model precursor to investigate SOA formation through aqueous-phase chemistry.13,14 On the basis of recent laboratory observations, glyoxal can react in aqueous solution to form SOA materials including organic acids, oligomers, organosulfates, and nitrogen-containing organics via numerous mechanisms such as photochemical oxidation, acid-catalyzed reaction, and self-oligomerization.6,14 18 In particular, aqueous-phase oxidation of glyoxal and other similar water-soluble organics can generate low-volatility products with high oxygen-to-carbon (O/C) atomic ratios,14,19 21 and therefore may be a substantial source of highly oxidized ambient organic aerosol that is difficult to produce through traditional smog chamber experiments.22,23 The Aerodyne aerosol mass spectrometer (AMS) can measure nonrefractory components in submicrometer aerosol particles with high-time resolution.24 It has been widely employed to investigate the chemical characteristics of organic aerosols in field and laboratory SOA formation studies,23,25 but its application to Received: May 3, 2011 Revised: August 17, 2011 Published: August 19, 2011 10517

dx.doi.org/10.1021/jp204099g | J. Phys. Chem. A 2011, 115, 10517–10526

The Journal of Physical Chemistry A

ARTICLE

Figure 1. Schematic diagram of the experimental setup used for monitoring the chemical composition of reacting solution by AMS.

aqueous oxidative processing has been limited. Developing a reaction chamber that operates at extremely humid condition without significant droplet deposition but providing sufficient retention time for oxidation is a challenge. In the current study, we demonstrate the use of a novel approach that allows for simultaneous photo-oxidation and mass spectrometry measurement of aerosolized organics using AMS to investigate aqueous OH oxidation of glyoxal in a bulk solution. With the aerosol drying process prior to the AMS measurement, the experimental setting simulates aqueous oxidation in the bulk aerosol or cloud droplets followed by evaporation of water and volatile organics to form SOA. Although AMS is generally not treated as a speciesspecific instrument due to extensive fragmentation of organics with electron-impact ionization, we show that it can be used to quantify some of the major organic species present in the glyoxal aqueous oxidation system, with other species in the bulk solution measured by complementary off-line techniques. We also address the degree to which AMS is sensitive to the overall particulate organics that are formed in the reacting solution. In addition to the mechanistic information, the chemical characteristics of the oxidized materials observed here can be directly compared to those of AMS-measured organic aerosol components in field and other laboratory studies, thus, making possible a better evaluation of the atmospheric relevance of the aqueous oxidative processes23 and permitting a stronger connection between laboratory studies and field observations.

’ EXPERIMENTAL SECTION Aqueous Oxidation and AMS Characterization. Aqueous-

phase OH oxidation of glyoxal (Sigma-Aldrich, 40 wt % in H2O) was performed for 5 h using an ozone-free mercury lamp (UVP, 254 nm) inserted into a sealed, 1 L glass reaction vessel covered with aluminum foil at room temperature. OH radicals were produced continuously by photolysis of H2O2 (Sigma-Aldrich, 30 wt % in H2O). The aqueous solution was atomized continuously by ultrapure compressed air (BOC, grade 0.1) throughout the OH exposure period using a TSI atomizer (Model 3076). The aqueous particles then passed through a diffusion dryer and were subsequently analyzed by the Aerodyne time-of-flight AMS (CToF-AMS) with proper dilution throughout the oxidation period, as shown in Figure 1.26 The AMS can only measure

Figure 2. Calibration curves of (a) glyoxal, (b) glyoxylic acid, and (c) oxalic acid measured by AMS. (1) Concentration of target compound in the reacting solution (red circle), (2) m/z 44 mass-to-sulfate mass ratio (m/z 44/SO42 , blue diamond), and (3) organic mass-to-sulfate mass ratio (Org/SO42 , green triangle) against molecular ion mass-to-sulfate mass ratio (i.e., Org58/SO42 for glyoxal, Org74/SO42 for glyoxylic acid and Org90/SO42 for oxalic acid). All calibration curves have R2 values greater than 0.98.

particle-phase materials (i.e., SOA) but not the gas-phase products from the reaction. The working principle of AMS has been reviewed by Canagaratna et al.,24 and therefore, the detail is not described here. The initial molar concentrations of glyoxal and H2O2 in each experiment were about 3 and 13.3 mM, respectively. Control experiments without UV or H2O2 were also performed. AMS Calibration and Quantification of Target Organics. Ammonium sulfate ((NH4)2SO4) was added to the reacting solution as an internal standard with a concentration of 0.2 mM. Assuming that the inorganic sulfate did not significantly react with organics and oxidants in the reacting solution and was internally mixed with organics in the aerosolized solution droplets, changes in any sulfate-normalized organic fragment intensity/mass measured by the AMS can be used for quantitative 10518

dx.doi.org/10.1021/jp204099g |J. Phys. Chem. A 2011, 115, 10517–10526

The Journal of Physical Chemistry A analysis (see next paragraph for calibration procedure). Note that NH4+ and SO42 may react with glyoxal in aqueous solution to form organonitrates and organosulfates, respectively,16,17,27 but their yields are still highly uncertain and are expected to be much lower than other OH oxidation products formed in our experiment. Furthermore, it has been reported that the sulfur content of organosulfates almost completely fragments in the AMS to HxSOy+ series, with a pattern indistinguishable from inorganic sulfate, instead of organosulfur fragments (e.g., CH3SO2+).28 We, therefore, assume that organosulfate formation only introduces minimal uncertainty to the AMS-measured inorganic sulfate mass and, thus, the quantitative analysis. Glyoxal (Gly) and its potential oxidation products, including formic acid (FA), glyoxylic acid (GA), and oxalic acid (OA), are the target compounds to be quantified in the reacting solution. Glyoxal is selected as an example to demonstrate the calibration procedure. First, a series of known concentrations of glyoxal solution in the presence of 0.2 mM (NH4)2SO4 were atomized and subsequently analyzed by the AMS as described above. The background organic spectrum originating from water and glassware was measured by atomizing a 0.2 mM (NH4)2SO4 solution without any additional organics. A calibration curve was constructed by plotting the glyoxal concentrations against their corresponding ratios of background-subtracted molecular ion mass of glyoxal to sulfate mass (i.e., ([Org58/SO42 ]Gly = ([Org58/SO42 ]background), as ([Org58/SO42 ]measured shown in Figure 2a (red circle data). For quantification of glyoxal in the reacting solution, the measured Org58/SO42 was corrected by subtracting the signals arising from background organics and identified products including glyoxylic acid and oxalic acid (i.e., [Org58/SO42 ]Gly = [Org58/SO42 ]measured [Org58/SO42 ]GA [Org58/ [Org58/SO42 ]background 2 2 SO4 ]OA). [Org58/SO4 ]GA and [Org58/SO42 ]OA were determined from the calibration data set of glyoxylic acid and oxalic acid when their solution concentrations are known. Because formic acid is too volatile to be detected by the AMS, it does not affect the intensity of any molecular ions. Using the calibration curve and [Org58/SO42-]Gly of the reacting solution, the concentration profile of glyoxal throughout the oxidation period can be determined. The AMS-measured total organic mass (Org) and organic mass at m/z 44 (Org44, thought to be largely due to CO2+ from organic acids) contributed by glyoxal were quantified using the same normalization and calibration approach, and their corresponding calibration curves are shown in Figure 2a (green triangle data for [Org/SO42 ]Gly and blue diamond data for [Org44/SO42 ]Gly). The AMS data analysis program Squirrel (version 1.49B1) was used for data treatment. The sulfate mass used for normalization was determined using the standard sulfate fragments assignment and a default relative ionization efficiency (RIE) of 1.2. Note that glyoxal and other target organics can give non-negligible signals at m/z 30 and 46, but Squirrel recognizes these fragments predominately contributed by nitrate instead of organics. Because there was no inorganic nitrate added into the reacting solution, the nitrate contributions in the standard fragmentation table of Squirrel were removed. For Org determination, Squirrel summed the mass of all organic fragments determined by the standard organic fragments assignment and a default organic RIE of 1.4 was used. The same RIE was also used to calculate the mass of individual organic fragments like Org58 and Org44. The molecular ions of glyoxylic acid and oxalic acid are m/z 74 and 90, respectively, and the calibration curves of

ARTICLE

Figure 3. Organic AMS spectra of aerosol particles measured at (a) 0, (b) 2.5, and (c) 5 h OH oxidation. Organic AMS spectra of (a) glyoxal, (d) glyoxylic acid, and (e) oxalic acid particles.

these two organic acids are shown in Figure 2b,c. All calibration curves have high linearity with R2 values greater than 0.98. It is important to note that this calibration procedure accounts for the ionization efficiency of each species and, with normalization to sulfate mass, is independent of the collection efficiency of aerosol particles in the AMS. Organic AMS spectra of glyoxal, glyoxylic acid and oxalic acid are shown in Figure 3. All these spectra are similar to those reported in the literature except the intensity at m/z 18 and 28.29,30 In particular, Chhabra et al.29 showed that the ratio of Org28 to Org44 (or Org CO+/CO2+) is about 5 based on their high-resolution AMS data, and Takegawa et al.30 reported that Org CO+/CO2+ of glyoxylic acid and oxalic acid are approximately equal to 1.33 and 0.67, respectively. Apparently, the Org CO+/CO2+ ratio of the reacting solution should keep changing if 10519


The Journal of Physical Chemistry A glyoxylic acid and oxalic acid are the major oxidation products. However, due to the mass resolution limitation of the CToF-AMS and using air as a carrier gas, we were not able to differentiate organic CO+ fragment from the strong intensity of nitrogen fragments, and hence, Org28 was estimated to be equal to Org44 in our measurements (a default setting in Squirrel for unknown organics). Similarly, Org18 is another major fragment that is associated to Org44 in Squirrel. Nevertheless, because fragmentation assignments are the same for the Org/SO42 of the reacting solution and the organic solution used for calibration, quantitative comparison between the Org/SO42 directly measured by the AMS and that determined from speciation analysis is still valid. Other calibration curves should not be affected by this issue. Offline Analytical Methods. The major drawback to the AMS measurements is that volatile species formed in the reaction vessel such as formic acid cannot be detected as they escape to the gas-phase after aerosolization. Ion chromatography (IC) has been widely used to quantify water-soluble organic acids in aerosol filter samples. To provide complementary offline information to the AMS analysis, 10 mL of the reacting solution were drawn from the photochemical glass reactor every 20 30 min for quantification of water-soluble organic acids using IC. Because a rapid destruction of H2O2 is needed to avoid reaction between glyoxylic acid and H2O2 in the solution samples,31 a volume of 100 μL of peroxide catalase from bovine liver (Sigma-Aldrich) was immediately added to each sample to quench H2O2. The samples were analyzed within 10 min after collection to further minimize the effects caused by any interaction between organics and H2O2 awaiting IC analysis.31 The instrument was equipped with an IonPac ICE-AS6 column (Dionex), a conductivity detector (Altech 550), and an isocratic pump (Prekin Elmer Series 200). The detector was operated at 40 °C. To optimize the performance of peak separation, a mixture of 1.0 mM heptafluorobutyric acid (HFBA) and water (70:30) was used as an eluent and a flow rate of 1 mL/min was used. Sample injection volume was 20 μL. Overall, the instrument can detect these organic acids in concentrations of 10 6 M and resolve the peaks of a formic, glyoxylic, malonic, tartaric, and succinic acid standard mixture very well. The peak of oxalic acid overlaps with the solvent signal making the quantification of oxalic acid difficult when its concentration is low. The IC analysis has a measurement uncertainty of less than 10%. Total organic carbon (TOC) analyzer is a common technique to determine the TOC content in aerosol samples. A fully automated TOC analyzer (Shimadzu TOC-VCPN) was used to determine the TOC content from individual oxidation experiments. The instrument measured the total carbon and inorganic carbon of the samples, and TOC was calculated by taking the difference between the two. Total carbon was measured by converting all carbon-containing species to CO2 in a combustion tube filled with an oxidation catalyst and then detected by nondispersive infrared (NDIR) gas analyzer. Inorganic carbon was measured by reacting samples with HCl to produce CO2, which was then volatilized by sparging the samples with air and detected by the NDIR gas analyzer. Potassium hydrogen phthalate standards and sodium bicarbonate/sodium carbonate standards were used to calibrate total carbon and inorganic carbon measurement, respectively. All the above details can be found in the user manual of the TOC analyzer. For the TOC analysis, 10 mL samples were collected without quenching of H2O2. They were stored in the ice bath to slow down the reaction kinetics and

ARTICLE

Figure 4. Concentration profiles of glyoxal, formic, glyoxylic, and oxalic acid measured by AMS and IC analysis.

analyzed within 2 h. Note that TOC was measured from individual oxidation experiments at a later date than the AMS and IC speciation measurements. We have ensured that the same degree of oxidation had occurred in all experiments by scaling to the similar amount of oxalic acid formed, as measured by the AMS. The TOC analysis has a measurement uncertainty of less than 10%.

’ RESULTS AND DISCUSSION Product Identification and Concentration Dynamics. The AMS organic spectra of the evaporated droplets measured at 0, 2.5, and 5 h of the oxidation experiment are shown in Figure 3. Initially, glyoxal has major fragments at m/z 29, 47, and 58, which is consistent with those observed by Chhabra et al.29 Because glyoxal solution was prepared 12 h prior to the photo-oxidation, the glyoxal dimer fragment signal (e.g., m/z 105) was much lower than that in fresh prepared glyoxal solution, and we believe that most of glyoxal was in the mono- or dihydrated form. Recent laboratory work has shown that the presence of inorganic SO42 shifts the hydration equilibrium of glyoxal from the unhydrated carbonyl form to the hydrated form.27 When the oxidation took place, the signature of glyoxal started to deplete (i.e., m/z 29 and 58), whereas organic acid fragments formed dramatically (i.e., m/z 44). In particular, the AMS spectra of glyoxylic acid and oxalic acid are very similar to those of evaporated droplets measured at 2.5 and 5 h, respectively (Figure 3). This is an overall indication 10520



ARTICLE

Figure 5. Simplified reaction scheme of aqueous-phase OH oxidation of glyoxal. The reaction between glyoxal and H2O2 is proposed as a new channel for glyoxylic acid formation. The intermediate (2-hydroxy-2hydro-peroxyethanal, HHPE) and its AMS fragments at m/z 63 and 46 are shown in the gray shaded region. The yellow shaded area represents the oligomerization process, and dashed arrows represent tentative pathways.

that glyoxal was oxidized in the aqueous solution to produce these two organic acids as the major reaction products and they were able to retain in the particle-phase after droplet evaporation prior to the AMS analysis. Figure 4 shows the concentration profiles of glyoxal, formic acid, glyoxylic acid, and oxalic acid in the reacting solution measured by different techniques. The glyoxal and oxalic acid concentrations shown in Figure 4a were determined from the AMS measurement assuming that there are no significant mass spectral overlap between these species and unidentified organics at the molecular fragments of glyoxal and oxalic acid (i.e., m/z 58 and 90). Note that glyoxal cannot be detected by our IC system. Despite the initial oxalic acid concentrations being initially below the detection limit of the IC, AMS- and IC-measured oxalic acid concentrations match quite well to each other at the later stages of oxidation. Glyoxylic acid concentrations determined by the AMS were much higher than that measured by IC (Figure 4b) likely due to positive interferences of m/z 74 fragments from unidentified products (see later discussion). Although Tan et al.31 suggested significant loss of glyoxylic acid through a reaction with H2O2 to produce formic acid prior to IC analysis, a short waiting period used and sufficient addition of the H2O2quenching catalase in our experiments likely made the effects of this reaction negligible. In the control experiments, the oxidized samples without quenching of H2O2 result in similar glyoxylic acid and formic acid levels to those with addition of quenching catalase measured in another oxidation experiment (see Figure S1 in Supporting Information). This indicates that a quick IC analysis is a critical factor to make the results of glyoxylic acid concentrations reliable. The AMS and offline IC analysis provide complementary speciation data on the aqueous oxidation of glyoxal. In the following discussion, the speciated-AMS data of glyoxal and oxalic acid are used, whereas glyoxylic and formic acid refer to the IC measurements. First of all, a rapid decay of glyoxal coincided with an instantaneous formation of glyoxylic acid (Figure 4a,b), indicating that glyoxylic acid was a first-generation product of aqueous-phase OH-glyoxal oxidation.31 Assuming the first-order decay of glyoxal and using the second-order rate constant of aqueous-phase OH oxidation of glyoxal (1.1 109 M 1 s 1),31

Figure 6. (a) Comparison between the Org/SO42 ratios measured by the AMS and those determined from the total of glyoxal (Gly), glyoxylic acid (GA), and oxalic acid (OA). Formic acid is not included in the calculation because it is too volatile to be detected by the AMS. (b) The correlation between the AMS measured and speciated Org/SO42 ratios from three different data sets. The dashed line represents the 1:1 line.

the estimated concentration of OH radicals at the start of the experiment was on an order of 10 13 M, which is relevant to atmospheric cloud conditions.32 Second, formation of formic and oxalic acid was delayed relative to the rapid glyoxal decay suggesting that these two organic acids were more likely second- or multiplegeneration products. Substantial production of oxalic acid after most of the glyoxal has been consumed provides further evidence that oxalic acid was not directly produced from glyoxal. In particular, glyoxylic acid is a potential precursor of these two acids. While formic acid can be formed through the reaction between glyoxylic acid and H2O2,20,31 OH oxidation of glyoxylic acid can lead to formation of oxalic acid according to previous studies.14 The simplified reaction scheme of the above pathways is shown in Figure 5. Formic and oxalic acid can be further oxidized to produce carbon dioxide as a final product.14 Organic Mass Balance of Aerosolized Solution Droplets. Because AMS can only measure particle-phase materials, the contributions of glyoxal, glyoxylic acid, and oxalic acid to the AMS-measured total organic mass (Org) can provide an insight whether there were any other important unknown organics presented in the evaporated droplets. This can be achieved by comparing the Org/SO42 of aerosol particles determined by direct Squirrel output of AMS data analysis to that determined by the sum of Org/SO42 contributed by these identified species.The Org/SO42 contribution from each species to the evaporated droplets can be determined using the calibration curves shown in Figure 2 once their IC- or AMS-measured solution concentrations are known. Figure 6a,b shows that the two Org/SO42 ratios match very well with each other, suggesting that glyoxylic 10521



ARTICLE

Figure 8. Cumulative yield curves of identified products after 5 h OH oxidation: oxalic acid (blue area, AMS), glyoxylic acid (green area, IC), and formic acid (red area, IC). Yield is defined as mass of organic formed divided by mass of nonhydrated glyoxal originally in the reaction vessel.

Figure 7. (a) Comparison between the Org44/SO42 ratios measured by the AMS and those determined from the total of glyoxal, glyoxylic acid, and oxalic acid. Formic acid is not included in the calculation because it is too volatile to be detected by the AMS. (b) The correlation between the AMS measured and speciated Org44/SO42 ratios from three different data sets. The dashed line represents the 1:1 line.

and oxalic acid are the most abundant particle-phase products that are AMS measurable. Moreover, the intensity of the organic fragment at m/z 44 (CO2+) is often used as a measure of the organic acids.22,30 In particular, Takegawa et al.30 observed that the Org44 accounts for 15 and 34% of the Org of glyoxylic and oxalic acid, respectively. Figure 7a,b compares the Org44/SO42 of direct AMS measurements to that determined from the speciation data using our calibration (Figure 2). Continuous increase of Org44/SO42 was predominately due to the substantial production of oxalic acid. Similar to the comparison of Org/SO42 , glyoxylic and oxalic acid largely accounted for the entire measured Org44/SO42 , further confirming the existence of insignificant levels of other organic acids in the evaporated droplets. Tan et al.31 observed the formation of larger organic acids such as malonic, tartaric, and succinic acid formed in low concentrations (10 4 10 6 M) via aqueous oxidation of glyoxal. However, we did not observe detectable amounts of these proposed species in the bulk solution using IC analysis. Because the large acids are expected to be less volatile than glyoxal and the identified organic acids, the IC result is consistent with the comparison of organic mass balance discussed above. On the basis of ultrahigh resolution Fourier transform ion cyclotron resonance electrospray ionization-mass spectrometry (FTICR-MS), Lim et al.14 proposed that large organic acids are predominately produced through radical radical oligomerization. Their mass spectra contain series of peaks with a difference of 74 m/z bins. The major monomeric unit (Figure 5) can be generated through the hydrogen atom abstraction of glyoxal dihydrate by OH radicals followed by dehydration.14 In particular, tartaric acid

is the dimer of their self-reaction. Although we did not observe formation of tartaric acid and other organic acids, we observed substantially higher m/z 74 fragment intensity than should have arisen from the molecular ion of glyoxylic acid in our AMS measurements as mentioned in the previous section (Figure 4b). This difference is likely due to the presence of large acids and oligomers arising from this radical even though their contribution to the AMS-measured total organic mass seems to be very minor. Under the assumption that the oligomers, being relatively involatile, would be detected by the AMS, we estimate that they constituted no more than 10% of the total organic mass. As mentioned above, we cannot rule out the possibility of organonitrates and organosulfates formation but they are not likely the major particle-phase products. Yields of Major Oxidative Products. Yields of formic, glyoxylic, and oxalic acid, defined as the mass of an individual organic produced divided by the mass of nonhydrated glyoxal reacted, were determined. All these organic acids are not the absolute end products of the reaction, and thus, their yields are a function of reaction time. Figure 8 shows the cumulative yield of formic, glyoxylic, and oxalic acid. The yield of oxalic acid increased with OH exposure, ultimately reaching 57%. The maximum yield of glyoxylic acid was 40% at 100 min and then gradually dropped to 21%. The total yield of these two particlephase organic acids was about 78% at the end of our experiments, and it may further increase with prolonged OH exposure before reaching the potential maximum. Formic acid yield peaked (21%) at 120 min. Note that the yields reported here, especially for formic acid, only represent lower limits, as their evaporative loss throughout the course of the atomization process may be non-negligible. This problem is expected to be much less significant for glyoxylic and oxalic acid. It is important to note that the total initial yield of these organic acids was quite low (e.g., only about 20% at 20 min), highlighting considerabe conversion of glyoxal to other reaction intermediates that have not been identified yet at the early stage of oxidation (see next section). However, with the organic mass balance results of evaporated droplets, it can be speculated that those intermediates were formed inside the reaction vessel but did not contribute much to the AMS-measurable organic mass. Reaction between Glyoxal and H2O2. To investigate the possible reaction between glyoxal and H2O2 and photolysis of 10522



ARTICLE

Figure 10. Carbon concentration profiles determined by the sum of glyoxal and formic, glyoxylic, and oxalic acid (red circle) concentrations and that measured directly by the TOC analyzer (blue and orange cross) during aqueous oxidation experiments. Intensity profiles of Org63/ SO42- (green line) and Org46/Org44 (red line).

Figure 9. Normalized time series profiles of (a) Org58/SO42 and Org/SO42 , (b) Org63/SO42 and Org46/SO42 of UV and H2O2 control experiments obtained by AMS measurement. Increasing intensity of fragment m/z 63 is potentially due to the formation of HHPE by H2O2 oxidation of glyoxal.

glyoxal at the early stage of aqueous oxidation, two sets of control experiments were conducted to determine the individual effects of UV light and H2O2 on the aqueous glyoxal chemistry. Figure 9a shows the evolution of Org58/SO42 , where m/z 58 is the molecular ion of glyoxal, in these control experiments. While there was no measurable decay of Org58/SO42- by UV photolysis in the absence of H2O2 (H2O2 control), about 15 20% of glyoxal was consumed within 2 h when it was mixed with H2O2 without any light illumination (UV control). In the UV control experiment, the decay of glyoxal coincided with a rapid increase of normalized Org63/SO42 and Org46/SO42 followed by a slow approach to a stable value (Figure 9b). This observation indicates that a reversible reaction between glyoxal and H2O2 took place and equilibrium was established in the absence of UV light. A reduction of AMS-measured total organic mass by 10 15% was also observed. This points out that only a small portion of the products that contributed to organic fragments at m/z 63 and 46 were retained in the evaporated droplets for AMS detection, consistent with the conclusions made at the end of the last section. It is believed that the formation of organic fragments at m/z 64 and 46 arises from a known aqueous-phase reaction between glyoxal and H2O2.33 In particular, one of the carbonyl groups of glyoxal is nucleophilically attacked by H2O2 to form 2-hydroxy-2hydroperoxyethanal (HHPE). Fragmentation of this intermediate due to electron-impact ionization may lead to the m/z 63 and 46 organic fragments as shown in Figure 5. Figure 10 shows that there was fast formation of HHPE in the glyoxal OH-oxidation

experiments, as indicated by an enhancement of Org63/SO42 . The glyoxal OH oxidation products can produce both organic fragments at m/z 46 and 44 significantly in the AMS, and therefore, the Org46/Org44 ratio was used instead to visualize the contribution of HHPE to the m/z 46 organic fragment. Subsequent photochemical decomposition of HHPE may lead to formation of glyoxylic acid33 and other products and, thus, the decay of the related fragments. The profiles of Org63/SO42 and Org46/Org44 measured in the oxidation experiments were well correlated to each other as shown in Figure 10. No other organic fragments had the similar signal profiles, suggesting that HHPE was likely a single product resulting in the signal profiles of Org63/SO42 and Org46/Org44. Although only 15 20% of glyoxal was reacted to form HHPE in the UV control experiment (Figure 9a), depletion of HHPE via photolysis or oxidation may shift this reaction equilibrium to the right and, hence, more glyoxal may interact with H2O2 to form additional HHPE overall. It is, however, not possible to quantify the production of HHPE because of the unavailability of HHPE chemical standard. Carbon Balance of the Reacting Solution. Another important part of this study was to determine whether the coupled in situ AMS and offline IC analytical techniques were able to measure full carbon balance of the reacting solution during the course of the reaction. Figure 10 shows the result of carbon balance calculations (red circle), which were determined based on the solution concentrations of glyoxal and formic, glyoxylic, and oxalic acid. After 5 h of aqueous oxidative processing, all these identified species accounted for about 60% of the initial carbon concentration. Excluding formic acid, approximately 50% of the initial carbon remained in the particle phase at the end of oxidation. The carbon loss can be due to decomposition of organics to carbon dioxide and also the evaporative loss of volatile organic species produced such as formic acid. However, it is important to note that the carbon concentration dropped significantly reaching its lowest level (∼50%) at around 60 min before rising to its final value of about 60%. This abnormal kinetic behavior demonstrates that our speciated-AMS and IC analysis are incomplete in assessing the carbon content of the solution, especially within the first hour of oxidation. In addition to the observation from yield calculations, this provides direct evidence 10523


The Journal of Physical Chemistry A to show that the formation of HHPE was potentially important at the beginning of the oxidation period. Offline TOC analysis is an ideal approach to determine the carbon content of aqueous solution. Figure 10 shows that TOC of the reacting solution decreased gradually throughout the oxidative period consistent with a previous study.31 A large positive difference between the TOC and speciated carbon content strongly suggests that there was at least one intermediate that has not been identified and quantified in the bulk solution. Because glyoxal, glyoxylic acid, and oxalic acid account very nicely for the AMS-measured Org/SO42 (Figure 6), HHPE and some unknown species not being fully detected by the AMS are expected to be either too volatile or unstable to remain in the evaporated droplets for AMS detection. We believe that under measurement of formic acid and evaporative loss of other volatile species can explain part of this discrepancy. The speciated carbon content built up after reaching the minimum point (Figure 10), suggesting that glyoxylic acid and oxalic acid can be subsequently produced through photochemical processing of HHPE and larger organic molecules with two or more carbon atoms (Figure 5),34 which are expected to have lower volatility than glyoxal and its oxidation products. Therefore, stability of HHPE and other unknown large organics in the evaporated droplets likely play an important role in this discrepancy as well. This observation has potentially important implications for laboratory, but perhaps more importantly, field studies of aqueous systems containing glyoxal and its oxidative products that use the AMS as the primary analytical instrument. Comparisons with Literature Studies of Aqueous Glyoxal Oxidation. The studies of Turpin and co-workers have nicely illustrated the complexities that arise in the aqueous OH oxidation of glyoxal. In particular, major conclusions from their work include (1) substantial amount of formic acid may be produced rapidly by the OH oxidation of glyoxal,34 but Tan et al.31 suggested that this may be due to the ongoing reaction in the sample solution awaiting offline analysis; (2) glyoxylic acid is the major first-generation product and can be subsequently oxidized to oxalic acid;14,31 (3) formation of oligomers via radical radical reaction is possible and can be significantly enhanced at high initial glyoxal concentration (e.g., >10 3 M);14 (4) glyoxylic acid and oxalic acid can be formed through OH oxidation of large organic acids and oligomeric products that produced by aqueous oxidation of glyoxal;34 (5) acidity of the oxidizing solution has little influence on oxalic acid yield.31 Some of these observations are now compared to the results obtained from the current study. Whether formic acid is produced directly from glyoxal oxidation remains unclear. Although Carlton et al.34 observed significant formation of formic acid coinciding with complete consumption of glyoxal within an initial reaction time, suggesting that formic acid is a direct oxidative product from glyoxal, Tan et al.31 found that their observation is likely due to the reaction between glyoxylic acid and H2O2 in solution samples awaiting offline analysis. In their later study, Lim et al.14 further proposed a minor formic acid formation pathway that OH oxidation of glyoxal in the presence of oxygen produces peroxy radicals that then self-react to produce an alkoxy radical, which eventually decomposes to formic acid. On the contrary, our observation clearly shows the formation of formic acid within the OH exposure period, which is more likely a second-generation product possibly formed by H2O2 oxidation of glyoxylic acid or some unknown reaction pathways inside the reaction vessel (Figure 4a). Formation of formic acid from HHPE is also

ARTICLE

possible. Apparently, more research is needed to better understand the formation mechanism of formic acid through aqueousphase oxidation given the divergent experimental results. One addition to the mechanisms proposed in the literature to date is our evidence that glyoxal can react with H2O2 to produce HHPE, which we believe can further decompose to form glyoxylic acid or some other products in the presence of UV light. This reaction channel, however, has not been incorporated into aqueous photo-oxidation models.14,20,31,34 In those kinetic models, glyoxal converts to glyoxylic acid primarily through the OH radical reaction channel. Formation of glyoxylic and oxalic acid through OH oxidation of large multifunctional compounds or oligomers is also included.34 Due to the fact that H2O2 concentrations are substantially lower in ambient aqueous solution droplets, the atmospheric relevance of this HHPE chemistry may be limited. Nevertheless, high initial concentrations of H2O2 are frequently used as an OH radical precursor in laboratory investigations of aqueous photo-oxidation. Including this additional reaction channel may be necessary to improve the performance of models that interpret laboratory data and are used to make atmospheric predications. Other atmospherically relevant organics frequently used in laboratory studies of aqueous aerosol chemistry and in-cloud processing such as methylglyoxal may also react with H2O2 in the similar way. In addition to glyoxylic acid and oxalic acid, Turpin and coworkers have shown that aqueous OH oxidation of glyoxal can lead, at the expense of lower oxalic acid yields, to formation of higher order organic acids and oligomers with carbon numbers larger than C2 when the initial glyoxal concentration is at least 10 3 M.31,34 Similar observations were also reported for other organic precursors such as glycolaldeyhyde and methylglyoxal.20,21 While the predicted maximum mass yield of oxalic acid is 140% at cloud-relevant glyoxal concentrations (10 5 10 4 M), the production of oxalic acid becomes negligible and the maximum yield of oligomers can be over 80% at aerosol-relevant glyoxal concentrations (>1 M).14 Taking into account that our study did not observe oxalic acid concentrations peaking, the final oxalic acid yield (∼57%) reported here is in a good agreement with the model prediction of Lim et al..14 As indicated above, no detectable levels of large organic acids were observed in the current study. Because Tan et al.31 observed the rapid formation of large organic acids at the early stage of oxidation, this discrepancy is probably due to insufficient initial organic radicals concentrations for oligomer formation in our experiment. This is consistent with the fact that OH radical concentration (10 13 M) used in this study is about a factor of 10 lower than that observed in their previous works (10 12 M).31 In addition, Tan et al.31 ran their experiments in the absence of NH4+ ion, the overall effects of NH4+ ion on the oxidation mechanism as well as oligomerization are still highly uncertain and need to be further investigated.

’ ATMOSPHERIC IMPLICATIONS Atmospheric aerosol particles absorb water under increasingly humid conditions and can activate to form cloud droplets when the air is supersaturated. With high global glyoxal production9 and a large effective Henry’s law constant,3,5,8 a significant amount of gas-phase glyoxal can partition into atmospheric aqueous droplets and further react with OH radicals to form numerous gas-phase species and SOA materials. In this study, formic acid is the major volatile product identified, although the 10524



Figure 11. (a) O/C ratio profile of SOA materials calculated based on the concentration of each identified particle-phase products (blue circles) and that estimated by using f44 as a surrogate (red triangles). (b) The f44 vs f43 space developed based on AMS measurements. The worldwide ambient data sets mainly distributed within the dash line bounded triangular region.23 The oxidation pathway of aqueous glyoxal oxidation is shown in orange circles.

yields and effects of H2O2 on its formation are still highly uncertain. Recent laboratory studies also suggested that OH oxidation of other organic precursors can produce formic acid as well.20,21,35 In addition to the experimental evidence, it has been suggested that aerosol oxidation may be a significant source of formic acid based on the modeling observation,36 able to partially reconcile the discrepancy between gas-phase formic acid sources and sinks in the atmosphere. However, the current knowledge of formic acid formation via aqueous aerosol and cloud chemistry is rather limited, in part because laboratory experiments often utilize higher concentrations of oxidants than that present in the atmosphere. Much research has been performed on SOA formation driven by gas-phase photo-oxidation of biogenic and anthropogenic precursors followed by subsequent gas-particle partitioning of less volatile products. However, laboratory SOA generated through this traditional pathways are usually less oxidized than the ambient SOA,22,23 indicating that there may be some unrecognized SOA formation and aging processes in the atmosphere. Specifically, aqueous oxidation of glyoxal can be a source of highly oxidized SOA. Assuming glyoxal, glyoxylic acid and oxalic acid are the predominant species contributing to SOA formation in our oxidation experiment, the calculated O/C ratio was equal to 1 initially and kept increasing to about 1.8 at the end of the experiment, as shown in Figure 11a. Inclusion of these species with the traditional SOA products may thus lead to a better agreement with observed high O/C levels of ambient SOA. Furthermore, our results are in agreement with those of

ARTICLE

Chhabra et al.29 who reported that the organic mass fraction at m/z 44 (f44) is a poor surrogate for estimating the oxygen content of pure glyoxal. Figure 11a compares the O/C ratio profiles of particle-phase materials calculated based on the concentration of each identified products and that estimated by f44.22 It was found that the O/C ratio estimated by f44 is always much lower than that calculated using our speciated measurements, indicating that f44 may not be a good surrogate to predict the O/C ratio of SOA predominantly originating from aqueous oxidation of glyoxal and possibly other similar organics. Recently, Ng et al.23 have developed an observational framework for the characterization of ambient oxygenated organic aerosols (OOA) based on 43 northern hemispheric AMS data sets, as shown in Figure 11b. The ambient measurements cluster within a distinct space defined by the mass fraction of AMS organic spectral intensity at m/z 44 and 43 (f44, f43), as indicated by the dashed line bounded triangular region shown in Figure 11b. Note that m/z 43 and 44 of ambient OOA are usually dominated by fragments of C2H3O+ from less-oxidized organics and COO+ from organic acids, respectively. Therefore, while the ambient low-volatility OOA (LV-OOA or more oxidized OOA) component resides on the upper part of the triangular region, the ambient semivolatile OOA (SV-OOA or less oxidized OOA) component concentrates in the lower half.23 Taking advantage of worldwide usage of AMS in field measurements, the aqueous oxidation pathway of glyoxal is shown on the f44 vs f43 space to compare with the global AMS field data for evaluating the atmospheric relevance of aqueous OH-glyoxal oxidation in SOA formation. Although aqueous OH oxidation of glyoxal can produce highly oxidized organics with an f44 value close to a very oxidized ambient LV-OOA component, the evolution pathway is located far away from the ambient triangular region. Thus, aqueous oxidation of glyoxal alone cannot explain the chemical characteristics of ambient OOA, unless there is a large degree of oligomerization. Because traditional SOA produced from gas-phase oxidation usually has a lower degree of oxidation (i.e., giving higher f43 but lower f44 values in AMS measurements) than the ambient SOA,23 this observation highlights the potential importance of co-oxidation of glyoxal and traditional SOA in aqueous aerosols or cloud droplets and also physical mixing of glyoxal SOA (or other SOA formed through aqueous chemistry) and traditional SOA to explain the observed highly oxidized SOA materials in ambient environment, as discussed in more detail in another paper from our group.26

’ CONCLUSIONS This is the first study to demonstrate the use of a novel design of photochemical reactor that takes advantage of both off-line IC and TOC measurements, as well as online AMS measurements to study the aqueous oxidation of glyoxal. Three major oxidative products were identified and their yields were quantified: formic acid, glyoxylic acid and oxalic acid. The reaction between H2O2 and glyoxal is proposed as a new pathway to produce glyoxylic acid that may need to be incorporated into cloud chemistry models. In contrast to previous work, no formation of large organic acids were observed at the glyoxal concentrations used, possibly due to the lower OH radical concentration employed. Furthermore, the AMS is not able to detect all dissolved organics in the reacting solution, likely due to their high volatility or low stability in the evaporated solution droplets. Overall, aqueous oxidation of glyoxal can produce highly oxidized SOA materials 10525


The Journal of Physical Chemistry A that likely coexist with traditional SOA materials, thus, potentially better matching field observations of organic aerosol mass and degree of oxidation. Because it is challenging to develop a smog chamber or reaction flow tube that can operate in high relative humidity or water supersaturated conditions without significant droplet deposition, the simple experimental approach described in this paper can be useful as an alternative approach to investigate aqueous oxidation processes on SOA formation.

’ ASSOCIATED CONTENT

bS

Supporting Information. Figure S1 shows the concentration profile of glyoxylic acid and oxalic acid in both quenched and nonquenched experiment. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Tel./Fax: (416) 946 7358. E-mail: [email protected].

’ ACKNOWLEDGMENT This work was supported by NSERC.

ARTICLE

(21) Tan, Y.; Carlton, A. G.; Seitzinger, S. P.; Turpin, B. J. Atmos. Environ. 2010, 44, 5218–5226. (22) Aiken, A. C.; et al. Environ. Sci. Technol. 2008, 42, 4478–4485. (23) Ng, N. L.; et al. Atmos. Chem. Phys. 2010, 10, 4625–4641. (24) Canagaratna, M. R.; et al. Mass Spectrom. Rev. 2007, 26, 185–222. (25) Zhang, Q.; et al. Geophys. Res. Lett. 2007, 34, L13801. (26) Lee, A. K. Y.; Herckes, P.; Leaitch, W. R.; Macdonald, A. M.; Abbatt, J. P. D. Geophys. Res. Lett. 2011, 38, L11805. (27) Yu, G.; Bayer, A. R.; Galloway, M. M.; Korshavn, K. J.; Fry, C. G.; Keutsch, F. N. Environ. Sci. Technol. 2011, 45, 6336–6342. (28) Farmer, D. K.; Matsunaga, A.; Docherty, K. S.; Surratt, J. D.; Seinfeld, J. H.; Ziemann, P. J.; Jimenez, J. L. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 6670–6675. (29) Chhabra, P. S.; Flagan, R. C.; Seinfeld, J. H. Atmos. Chem. Phys. 2010, 10, 4111–4131. (30) Takegawa, N.; Miyakawa, T.; Kawamura, K.; Kondo, Y. Aerosol Sci. Technol. 2007, 41, 418–437. (31) Tan, Y.; Perri, M. J.; Seitzinger, S. P.; Turpin, B. J. Environ. Sci. Technol. 2009, 43, 8105–8112. (32) Jacob, D. J. J. Geophys. Res., [Atmos.] 1986, 91, 9807–9826. (33) Sun, Z. C.; Eli, W.; Xu, T. Y.; Zhang, Y. G. Ind. Eng. Chem. Res. 2006, 45, 1849–1852. (34) Carlton, A. G.; Turpin, B. J.; Altieri, K. E.; Seitzinger, S.; Reff, A.; Lim, H.; Ervens, B. Atmos. Environ. 2007, 41, 7588–7602. (35) Vlasenko, A.; George, I. J.; Abbatt, J. P. D. J. Phys. Chem. A 2008, 112, 1552–1560. (36) Paulot, F.; et al. Atmos. Chem. Phys. 2011, 11, 1989–2013.

’ REFERENCES (1) Blando, J. D.; Turpin, B. J. Atmos. Environ. 2000, 34, 1623–1632. (2) Gelencser, A.; Varga, Z. Atmos. Chem. Phys. 2005, 5, 2823–2831. (3) Kroll, J. H.; Ng, N. L.; Murphy, S. M.; Varutbangkul, V.; Flagan, R. C.; Seinfeld, J. H. J. Geophys. Res., [Atmos.] 2005, 110, D23207. (4) Ervens, B.; Carlton, A. G.; Turpin, B. J.; Altieri, K. E.; Kreidenweis, S. M.; Feingold, G. Geophys. Res. Lett. 2008, 35, L02816. (5) Ip, H. S. S.; Huang, X. H. H.; Yu, J. Z. Geophys. Res. Lett. 2009, 36, L01802. (6) De Haan, D. O.; Corrigan, A. L.; Tolbert, M. A.; Jimenez, J. L.; Wood, S. E.; Turley, J. J. Environ. Sci. Technol. 2009, 43, 8184–8190. (7) Hennigan, C. J.; Bergin, M. H.; Dibb, J. E.; Weber, R. J. Geophys. Res. Lett. 2008, 35, L18801. (8) Volkamer, R.; Ziemann, P. J.; Molina, M. J. Atmos. Chem. Phys. 2009, 9, 1907–1928. (9) Fu, T.; Jacob, D. J.; Wittrock, F.; Burrows, J. P.; Vrekoussis, M.; Henze, D. K. J. Geophys. Res., [Atmos.] 2008, 113, D15303. (10) Fu, T.; Jacob, D. J.; Heald, C. L. Atmos. Environ. 2009, 43, 1814–1822. (11) Myriokefalitakis, S.; Tsigaridis, K.; Mihalopoulos, N.; Sciare, J.; Nenes, A.; Kawamura, K.; Segers, A.; Kanakidou, M. Atmos. Chem. Phys. 2011, 11, 5761–5782. (12) Sinreich, R.; Coburn, S.; Dix, B.; Volkamer, R. Atmos. Chem. Phys. 2010, 10, 11359–11371. (13) Ervens, B.; Volkamer, R. Atmos. Chem. Phys. 2010, 10, 8219–8244. (14) Lim, Y. B.; Tan, Y.; Perri, M. J.; Seitzinger, S. P.; Turpin, B. J. Atmos. Chem. Phys. 2010, 10, 10521–10539. (15) De Haan, D. O.; Tolbert, M. A.; Jimenez, J. L. Geophys. Res. Lett. 2009, 36, L11819. (16) Liggio, J.; Li, S. M.; McLaren, R. Environ. Sci. Technol. 2005, 39, 1532–1541. (17) Galloway, M. M.; Chhabra, P. S.; Chan, A. W. H.; Surratt, J. D.; Flagan, R. C.; Seinfeld, J. H.; Keutsch, F. N. Atmos. Chem. Phys. 2009, 9, 3331–3345. (18) Shapiro, E. L.; Szprengiel, J.; Sareen, N.; Jen, C. N.; Giordano, M. R.; McNeill, V. F. Atmos. Chem. Phys. 2009, 9, 2289–2300. (19) Carlton, A. G.; Turpin, B. J.; Lim, H. J.; Altieri, K. E.; Seitzinger, S. Geophys. Res. Lett. 2006, 33, L06822. (20) Perri, M. J.; Seitzinger, S.; Turpin, B. J. Atmos. Environ. 2009, 43, 1487–1497. 10526


Aqueous-Phase OH Oxidation of Glyoxal: Application of a Novel

Recommend Documents