Photochemical Aging of Light-Absorbing ... - ACS Publications

Mar 18, 2013 - Samar G. Moussa,. ‡ and V. Faye McNeill*. Department of Chemical Engineering, Columbia University, New York, New York 10027, United ...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JPCA

Photochemical Aging of Light-Absorbing Secondary Organic Aerosol Material Neha Sareen,† Samar G. Moussa,‡ and V. Faye McNeill* Department of Chemical Engineering, Columbia University, New York, New York 10027, United States S Supporting Information *

ABSTRACT: Dark reactions of methylglyoxal with NH4+ in aqueous aerosols yield light-absorbing and surface-active products that can influence the physical properties of the particles. Little is known about how the product mixture and its optical properties will change due to photolysis as well as oxidative aging by O3 and OH in the atmosphere. Here, we report the results of kinetics and product studies of the photochemical aging of aerosols formed by atomizing aqueous solutions of methylglyoxal and ammonium sulfate. Experiments were performed using aerosol flow tube reactors coupled with an aerosol chemical ionization mass spectrometer (Aerosol−CIMS) for monitoring gas- and particle-phase compositions. Particles were also impacted onto quartz windows in order to assess changes in their UV−visible absorption upon oxidation. Photooxidation of the aerosols leads to the formation of small, volatile organic acids including formic acid, acetic acid, and glyoxylic acid. The atmospheric lifetime of these species during the daytime is predicted to be on the order of minutes, with photolysis being an important mechanism of degradation. The lifetime with respect to O3 oxidation was observed to be on the order of hours. O3 oxidation also leads to a net increase in light absorption by the particles due to the formation of additional carbonyl compounds. Our results are consistent with field observations of high brown carbon absorption in the early morning.

1. INTRODUCTION Water-soluble volatile organic compounds (WSVOCs) in the atmosphere may be absorbed by aqueous aerosol particles or cloud droplets. Once in the aqueous phase, these species can react to form lower-volatility secondary organic aerosol material (aqSOA).1,2 Water-soluble carbonyl compounds like glyoxal and methylglyoxal, which have both biogenic and anthropogenic sources, have been shown to be likely precursors for aqSOA.1−5 The products of methylglyoxal reacting in aerosol mimics containing ammonium salts have been shown to include lightabsorbing, surface-active, and high-molecular-weight oligomeric species.3,4 The reactive processing of methylglyoxal in aqueous aerosols is thus a potential chemical source of humic-like substances (HULIS) or aerosol brown carbon (BrC). Sareen et al. (2010) identified organic acids, (hemi)acetals, aldol condensation products, and possible nitrogen- and sulfurcontaining compounds as the products of methylglyoxal in aqueous solutions containing ammonium sulfate.3 Besides adding to tropospheric organic aerosol mass, these product species may affect aerosol optical properties and/or cloud droplet formation.3−6 Some could also potentially be used as tracer species for aqSOA formation chemistry. However, the formation of these species will compete simultaneously with their photochemical degradation in the atmosphere. Their lifetimes in the oxidizing environment of the atmosphere, as well as the chemical identities and properties of their oxidation © 2013 American Chemical Society

products, are largely unknown. Most of these compounds are expected to be highly vulnerable to oxidation by the hydroxyl radical (OH); carbonyls and unsaturated species such as aldol condensation products may also be oxidized by ozone (O3). Photooxidation may result in revolatilization of aerosol organic material,7−10 although the aqueous-phase OH oxidation of glyoxal and methylglyoxal has been shown to be a source of aqSOA via the formation of lower-volatility organic acid products.1,2,11 Additionally, oxidation may decrease (through the breaking of CC or C−N bonds) or increase (through the creation of carbonyls) UV light absorption by the oligomeric aqSOA products.3−5,12 In this study, we aim to quantify the products and kinetics of photochemical aging of secondary organic material formed via the aqueous reaction of methylglyoxal with ammonium sulfate. The effects of oxidation via OH and O3 were studied, as well as photolysis. Aerosol flow tube reactors coupled to an aerosol chemical ionization mass spectrometer (Aerosol−CIMS) for detection of gas- and particle-phase organics were used for this purpose. Particles were collected on quartz windows before and after reaction in order to characterize the effect of photochemical processing on the light-absorbing properties of these aerosols. Received: September 21, 2012 Revised: March 15, 2013 Published: March 18, 2013 2987

dx.doi.org/10.1021/jp309413j | J. Phys. Chem. A 2013, 117, 2987−2996

The Journal of Physical Chemistry A

Article

Figure 1. Schematic diagram for the (a) O3 and (b) OH reaction setups. The O3 reactor has an injector in the reactor. The OH reactor is slightly modified, with the OH generator lamp in the reactor.

OH was formed in situ in an aerosol flow cell (described in the following section) by reacting O(1D) with water, where O(1D) was generated from the photolysis of O3.

2. METHODS SECTION Experiments were performed using aerosol flow tube reactors coupled to a custom-built Aerosol−CIMS for product characterization and a scanning mobility particle sizer (SMPS) (TSI 3081, 3775,189 AIM v. 8.1) to monitor particle concentration. 2.1. Aerosol Generation. Aerosols were generated in a similar manner as that described in Sareen et al. (2010).3 Briefly, mixtures containing 1.62 M methylglyoxal (Sigma Aldrich or Acros Organics) and 3.1 M (NH4)2SO4 were prepared using Millipore water (18.2 MΩ cm) and allowed to react for >24 h. The mixtures were then diluted with Millipore water to attain a salt concentration of 0.2 M. These dilute solutions were then aerosolized with N2 using a constant output atomizer (TSI 3076), mixed with a dry N2 stream to maintain the relative humidity at 64−68% measured with a hygrometer (Vaisala), and then introduced into the reactor. The particles were fully deliquesced before entering the reactor as they were not passed through a dryer before equilibrating at the above relative humidities, allowing them to remain on the upper portion of the hysteresis curve. Also, no liquid−liquid phase separation was expected to occur because the O/C ratio of hydrated methylglyoxal and the oligomer products was expected to be greater than 0.7.13 Typical number concentrations were 1.2 × 105 cm−3, with a mean surface areaweighted radius of 140 ± 2.0 nm and a mean volume-weighted radius of 193 ± 23 nm. The surface tension, light absorption properties, and product characterization for this system has been described in detail by Sareen et al. (2010).3 2.2. Gas-Phase Reactants. O3 was generated by flowing 0.2−5 sccm of oxygen with a carrier gas of 200 sccm N2 through a photocell containing a Hg pen-ray lamp (Jelight) operating at 185 nm. This produced O3 concentrations ranging from 0.2 to 0.8 ppm. The concentration of O3 generated was monitored using a UV absorption cell, 10 cm long with 1/4” inlet and outlet ports for O3 flow. Light from another Hg penray lamp (Jelight) coated to primarily emit 254 nm passes through the cell, was absorbed by O3 and detected by a photodiode (Hamamatsu) on the other end.

O3 + hν → O(1D) + O2

(1)

O(1D) + H 2O → 2OH

(2)

The concentration of OH generated was 9.5 × 106 molecules cm−3, as determined by calibration experiments with SO2 following the procedure described in McNeill et al. (2008).9 2.3. Flow Reactors. The O3 reactor setup is shown in Figure 1a. O3 oxidation experiments were performed in a similar manner as described by McNeill et al. (2007).14 The humidified aerosol stream was introduced through a perpendicular side arm into a 7.5 cm ID, 55 cm long Pyrex flow tube. A 1/8” Teflon tube fitted inside of a 0.25” OD stainless steel moveable injector was used to introduce the O3/ N2 stream into the flow tube. The position of the injector was changed to vary the reaction times (0−23 s) and hence attain kinetic information. The O3 concentration in the reactor varied from 0.2 to 0.8 ppm. The relative humidity in the reactor was maintained at 62−67%. The photolysis and OH reaction studies occurred in a continuous flow photocell reactor shown in Figure 1b. The reactor consisted of a 25 cm long, 5 cm in diameter quartz photocell with a 1 cm inner sleeve. An ozone-free Hg pen-ray lamp (Jelight) was inserted into the inner sleeve. The reactor featured four different inlet ports placed at intervals of 8 cm that were used to vary the reaction times (0−10 s). The humidified aerosol stream (RH ≈ 62−67%) and the O3/N2 stream were introduced via separate ports to prevent the aerosols from being exposed to O3 before entering the reactor. In a typical experiment, the aerosol stream was introduced via a perpendicular port at the rear of the reactor, while the position of the O3/N2 stream was varied using the remaining ports on the reactor. Experiments were also performed in this reactor in order to study the effect of photolysis on these aerosols in the absence of oxidants. All reactions were studied here under NOx-free conditions. 2988

dx.doi.org/10.1021/jp309413j | J. Phys. Chem. A 2013, 117, 2987−2996

The Journal of Physical Chemistry A

Article

The average flux of the lamp was determined to be 1.27 × 1012 photons cm−2 s−1 (between 258 and 329 nm), based on O3 photolysis using the following equation (further calculation details are included in the Supporting Information) J o3 =

∑ Fλϕλσλ

2.6. Multiphase Reaction Kinetics. Briefly, in a typical oxidation kinetics experiment, the Aerosol−CIMS is used to monitor the concentrations of the aqSOA species and oxidation products of interest. We previously characterized the products of the aqueous reactions of methylglyoxal in the presence of NH4+ in detail using Aerosol−CIMS.3 Aerosols formed by atomizing bulk solutions containing the methylglyoxal−NH4+ reaction mixture were exposed to O3 or OH in their respective reactors. As mentioned earlier, for the O3 flow tube reactor, the position of the injector through which O3 was introduced was changed in order to achieve different reaction times. To change the reaction time in the OH reactor, O3 was introduced through one of three different ports positioned along the axis of the reactor. Peaks corresponding to species of interest were monitored for changes in signal with reaction time. The reactive uptake coefficient, γ, was calculated based on the measured rate constants for the reaction of the oxidants with individual aqSOA products, as described in the following paragraphs. The organic reactants of interest were distributed throughout the aqueous volume of the droplet. The oxidants (O3 and OH) were also water-soluble; therefore, the oxidation reactions likely occurred in solution. OH is expected to react with most of the organic species in the droplet volume, whereas O3 will react with aldol condensation products featuring a double bond. On the basis of the CIMS signal at 271.5 amu, the instrument sensitivity for organics of this type,3 and the particle volume concentration (398 μm3 cm−3), we estimate the concentration of aldol condensation products with the formula C6H8O4 within the particles to be 2 × 10−4 M. By assuming an aqueous-phase diffusion constant of 10−5 cm2 s−1 and a very fast aqueous-phase reaction (i.e., diffusion-limited, kII ≈ 109 M−1 s−1),20 we can estimate a lower bound for the diffusoreactive length for O3 oxidation, l = (D/kII[Org])1/2 ≥ 77 nm ([Org] is the in-particle concentration of C6H8O4).21 This is on the same order of magnitude as the volume-weighted average particle radius, suggesting that the reactions take place within a significant fraction of the particle volume and are not confined to the near-surface region. Therefore, the following expression applies

(3)

where JO3 is the photolysis rate constant for ozone (2.5 × 10−4 s−1), Fλ is the average flux of the lamp, ϕλ is the quantum yield, and σλ is the absorption cross section for ozone across the wavelength range above (1.150 × 1023 cm2); ϕλ and σλ are determined from the NASA JPL handbook15 and FinlaysonPitts and Pitts (1999).16 JO3 is the rate constant calculated from the OH calibration experiments with SO2. The SO2 signal is monitored with time as it is exposed to OH in the reactor. The calculated rate constant represents the photolysis rate of O3 to form OH. We note that the “ozone-free” lamp emits at 253.7 nm and above. The wavelengths at which O3 absorption is maximum in this range (258−329 nm) overlap the ∼280 nm absorption peak of the methylglyoxal/(NH4)2SO4 products studied here (∼280 nm).3,4 2.4. Aerosol−CIMS Detection of Products. A custombuilt CIMS coupled to a volatilization flow tube inlet (Aerosol− CIMS) was used to monitor the aerosol chemical composition during the photooxidation experiments. The instrument, its operations, and relevant ionization chemistry were described in detail by Sareen et al. (2010).3 The effluent from the flow tube reactors was sent through a volatilization flow tube (VFT) kept at 135 °C in order to vaporize the particle-phase organic material for detection using CIMS. Some experiments were performed without applying heat to the inlet in order to identify species that are volatile at room temperature. A quadrupole mass spectrometer (Extrel CMS) was used to detect the organics as the products of their reactions with the reagent ion, I−, or its clusters with water. I− is ideal for detection of organic acids in aerosols,9,14 and we have previously used this approach to detect the oligomeric products of the methylglyoxal−(NH4)2SO4 reaction in bulk samples.3 An orifice was used downstream of the VFT to control the flow through the reactors and into the CIMS. Total flow was maintained at 2.3 SLPM through the O3 reactor and 1.8 SLPM for the OH reactor. A fraction of this flow (0.3 SLPM) was sent to the SMPS, in order to characterize the particle population. 2.5. Particle Collection and UV−vis Spectrophotometry. The effect of oxidation on the UV−visible light absorption of these particles was also studied by collecting particles before and after reaction by impaction on quartz windows (25 × 2 mm, Edmund Optics) using a custom-made impactor. The windows were held in place using a custommade filter holder that was modified so that gases can escape while particles are collected on the windows. Prior to collection on the quartz windows, particles from the reactor were passed through a dryer, and the collection period was 90 min at a rate of 1.5 LPM.17 UV−visible spectra were measured immediately after particle collection. The impact windows were placed in a custom-made holder that held them in position. Spectra were collected using a single-beam spectrophotometer (HP 8453). A spectrum of the blank window was measured each time before particle collection, and all subsequent spectra were baseline corrected at 1000 nm with respect to the blank window. Three trials were conducted for each experiment.

d[Org]aq dt I

= −k I[Org]aq

(4)

II

Here, k = k HPO3 is the observed pseudo-first-order rate constant, which can be used to determine the reactive uptake coefficient using the following equation19 γ=

4RT R = k I[Org]aq ωPO3 3

(5)

where R is the universal gas constant, T is the temperature, ω is the mean speed of the oxidant molecules, and R is the volumeaveraged radius of the particles. The Henry’s Law solubility constant for O3, H, is ∼0.008 M atm−1.18 Using our experimental observations, we confirm a posteriori that the diffusoreactive length, l = 3.4 μm, is much greater than the size of the particles. In order to calculate the diffusoreactive length for OH oxidation, we again consider a very fast aqueous-phase reaction, but we take into account the scavenging of OH by other organics besides the reactant of interest by using the total organic concentration in our calculation. The total organic concentration in the particle is estimated to be ∼22 M, based on the initial concentration of methylglyoxal in the atomizer 2989

dx.doi.org/10.1021/jp309413j | J. Phys. Chem. A 2013, 117, 2987−2996

The Journal of Physical Chemistry A

Article

(2010),3 shows the different aqSOA product peaks detected with Aerosol−CIMS using the I− detection scheme and the observed change upon exposure to O3 or OH. In general, upon exposure to O3, a prompt decrease of signal is observed at some of the higher masses, accompanied by an increase of signal at lower masses, indicative of O3 reacting with the aqSOA products to form lower-molecular-weight species. Figure 2 shows an Aerosol−CIMS difference spectrum comparing the results for aerosols before and after exposure to

solution and the estimated concentration factor upon equilibration of ammonium sulfate aerosols at ∼65% RH. We therefore find that l < 1 nm, indicating that oxidation is confined to a reactive zone near the gas−particle interface, and Henry’s Law equilibrium does not apply. The volume-averaged oxidation rate may therefore be matched with the flux of oxidant to the surface, scaled by the proportion of the total organic content consisting of the species of interest d[Org] 3 nOH(g)ω [Org]nC,A =− γ R dt 4 CT

(6)

where nOH(g) is the bulk gas-phase concentration of OH, nC,A is the number of carbon atoms in species Org, and CT is the total molar concentration of organic carbon in the particle. The observed decay rate is then used to calculate γ according to γmeas =

I R 4kobsC T 3 nox(g)nC,A ω

(7)

3. RESULTS AND DISCUSSION The photolysis and multiphase reactions of O3 and OH with aqSOA formed by methylglyoxal and (NH4)2SO4 were observed to result in changes in aerosol light absorption and the formation of volatile organic acid products. 3.1. Ozonolysis Products for aqSOA from (NH4)2SO4 and Methylglyoxal. Table 1, adapted from Sareen et al.

Figure 2. Spectrum showing the difference between (NH4)2SO4 and methylglyoxal particles before and after exposure to 0.2 ppm of O3 for 15 min.

0.2 ppm of O3 for 40 s in the flow tube reactor. The peaks that lie above the x-axis increase in the presence of O3, whereas the peaks below the x-axis decrease. The (hemi)acetals and aldol condensation products at m/z 271.5, 273.5, and 289.5 amu decrease by ∼70−90% when exposed to O3. As the exposure time increases, the peaks at 271.5 and 289.5 amu decrease further. The peaks at 273.5 and 275.6 amu did not show a clear trend with exposure time over the range of conditions studied here. Upon exposure to O3, the peak at 217 amu also decreases by ∼30%. This peak is attributed to either singly hydrated methylglyoxal or oxalic acid. These compounds are not expected to react with O3. The decrease in signal at this mass could be due to a shift in the aqueous-phase equilibrium caused by the oxidation of other aqSOA products. Most of the compounds that undergo ozonolysis break down into smaller, more volatile organic acids. Two such possible pathways are shown in Scheme 1, in which two separate aldol condensation products detected at m/z 271.5 amu, with the formula C6H8O4, undergo ozonolysis to form acetic, formic, and pyruvic acids and CO2. Note that only the most likely primary ozonide decomposition pathways (i.e., those that will form the most stable Criegee intermediates) are shown. Figure 2 shows a significant increase in the formic acid peak at 173 amu when the aerosols are exposed to O3, consistent with our proposed reaction mechanisms. The observed increase in 187 and 200.9 amu represents an increase in acetic acid and glyoxylic acid, respectively, both of which are predicted to form under ozonolysis of the compounds with formulas C6H8O4 and C6H10O5 (detected at m/z 271.5 and 289.5 amu). A broad peak that encompasses masses from 213 to 215 amu corresponds to oxopropanedial (213 amu) and pyruvic acid (215 amu) products, respectively, from the reaction of O3 with the peak at 271.5 amu. Similar mechanisms are also observed for the ozonolysis of the (hemi)acetal and aldol condensation products observed at m/z 289.5 amu; the products include compounds like formic acid, acetic acid, methanol, and formaldehyde. There is also an increase in the peak at m/z 225 amu, which

Table 1. Proposed Peak Assignments for Aerosol−CIMS Mass Spectra with I− As the Reagent Ion for Solutions Initially Containing Methylglyoxal and (NH4)2SO4a,b

a Adapted from Sareen et al. (2010).3 bThe up and down arrows in the two columns on the right indicate an increase or decrease with increasing exposure times, respectively, in signal at that mass when exposed to O3 or OH. NC indicates that no discernible change was seen with increasing reaction times.

2990

dx.doi.org/10.1021/jp309413j | J. Phys. Chem. A 2013, 117, 2987−2996

The Journal of Physical Chemistry A

Article

Scheme 1. Ozonolysis of the Product at 271.5 amu, Leading to the Formation of (a) Formic Acid and CO2 and/or (b) Acetic Acid, Pyruvic Acid, and CO2

corresponds to an organosulfate compound.3 While this species is not believed to be a product of the O3 reaction with any known aqSOA product, the apparent increase in its abundance may be caused by an equilibrium shift due to the oxidation of other aqSOA products. Table 2 lists the various observed oxidation product species and their proposed structures. Control experiments were also conducted, whereby the oxidants were introduced to the reactor in the absence of aerosols. Formation of the smaller acids is not observed, leading to the conclusion that they are not generated via reactions with the flow tube walls; rather, the oxidation products are formed solely from the aerosol oxidation chemistry. 3.2. Kinetics of Ozonolysis. Figure 3 shows the signal decay for a sample peak traced at 271.5 amu plotted as a function of reaction time to attain kIobs values. The values for average kIobs, γ, and τ, the lifetime of the specific species in the atmosphere, are listed in Table 3. Details regarding the atmospheric lifetime calculations are discussed in section 4.

Ozone concentrations were varied between 0.2 and 0.8 ppm. No clear trend for the rate constant with ozone concentration over this concentration range was seen for most of the traced peaks. The peak at 271.5 amu showed a slight increase in kI with increasing ozone concentration, as seen in Figure 4. 3.3. Photolysis and OH Oxidation Products and Kinetics. In the OH reactor, we have three processes that can possibly affect the reaction chemistry, photolysis due to the lamp, reaction with O3 (which is used as a OH precursor), and reaction with OH. During the daytime, degradation via photolysis can be a major fate of carbonyl-containing, lightabsorbing organic compounds such as the aqSOA material studied here. Nizkorodov and co-workers recently showed that photolytic aging is more efficient in aqSOA as compared to dry SOA.22,23 Hence, prior to studying the effect of OH on these aqueous compounds, we conducted experiments to assess the impact of photolysis. The lamp used in the photolysis and OH experiments transmits at wavelengths > 290 nm, which are relevant for the troposphere. In order to study photolysis, we 2991

dx.doi.org/10.1021/jp309413j | J. Phys. Chem. A 2013, 117, 2987−2996

The Journal of Physical Chemistry A

Article

Table 2. Product Compounds Formed upon the Ozonolysis, Photolysis, and/or OH Oxidation of the SOA Products of (NH4)2SO4 and Methylglyoxala

Figure 4. Average kI values for the peak at 271.5 amu over varying ozone concentrations.

the aqueous phase as methylglyoxal does in air. The aldol condensation and (hemi)acetal product peaks at m/z 271.5 and 289.5 amu decrease by ∼40 and ∼30%, respectively. Table 4 lists the photolysis rate constants observed in our system for these peaks. Table 4. Kinetic Parameters for the Photolysis and OH Oxidation of the Aqueous Products of (NH4)2SO4 and Methylglyoxala m/z [amu] 271.5 289.5

a

The m/z represents the mass of the compound with the reagent ion, I−, as seen in the Aerosol−CIMS.

JSOA [s−1] −4

1.15 × 10 1.22 × 10−5

kIOH [s−1]

γOH

τphot [min]

τOH [h]

2(±1) × 10−2 1(±1) × 10−2

>1 >1

3 30