The sulfate anion radical (SO4• –) is known to be formed in the autoxidation chain of sulfur dioxide, and from minor reactions when sulfate or bisulfate ions are activated by OH radicals, NO3 radicals, or iron. Here we report a new source of SO4• –, from the irradiation of the liquid water of sulfate-containing organic aerosol particles under natural sunlight and laboratory ultraviolet radiation. Irradiation of aqueous sulfate mixed with a variety of atmospherically relevant organic compounds degrades the organics well within the typical lifetime of aerosols in the atmosphere. Products of the SO4• – + organic reaction include surface-active organosulfates and small organic acids, alongside other products. Scavenging and deoxygenated experiments indicate that SO4• – radicals, instead of OH, drive the reaction. Ion substitution experiments confirm that sulfate ions are necessary for organic reactivity, while the cation is nearly irrelevant. The reaction proceeds at pH 1-6, implicating both bisulfate and sulfate in the formation of photoinduced SO4• –. Certain aromatic species may further accelerate the reaction through synergy. This new reaction may impact our understanding of atmospheric sulfur reactions, aerosol properties, and organic aerosol lifetimes when inserted into aqueous chemistry model mechanisms.
Materials and Methods
Materials. Organic reagents used in this study are shown in Figure S11. Erythritol (99%), EDTA (>99%) and magnesium sulfate hexahydrate (≥99%) were purchased from Fisher Scientific. D2O (99.8 atom% D), CDCl3 (99.8 atom% D), methanol (MeOH) (≥99%), acetonitrile (MeCN) (≥99%), cyclohexane (≥99%), pinonic acid (98%), ammonium sulfate (≥99%, reagent grade and molecular biology grade), ammonium carbonate (≥99%), sodium sulfate (≥99%), 2,4-dinitrophenyl hydrazine (DNPH) (97%) and 50 wt.% hydrogen peroxide (H2O2) in water were obtained from Sigma Aldrich. 4-nitrophenol (p-NP) (99%) and tert-butanol (>99%) were purchased from Alfa Aesar. DNPH was recrystallized prior to use, and its formic acid derivative (FADNPH) was synthesized via a previously reported procedure (65). 2-methylbut-3-ene-1,2-diol and 2,3-dihydroxy-2-methylbutane-1,4-dinitrate were each synthesized according to previous reports (66). All other purchased chemicals were used without further purification. Unless otherwise indicated, experiments were performed using reagent grade sulfate salts. Ultrapure H2O was obtained from a Milli-Q purification system (Millipore Sigma, 18 MΩ).
Characterization Methods. High Pressure Liquid Chromatography-High Resolution Mass Spectrometry (HPLC-HRMS) – HPLC-HRMS was used to quantify p-NP, pinonic acid, THB, and DHDN in bulk aqueous photochemical experiments, representative chromatograms shown in Fig. S12-S13. Analyses were performed on an Agilent 1100 HPLC coupled to a linear-trap-quadrupole (LTQ-XL) Orbitrap mass spectrometer (Thermo Corp., Waltham MA) operating at a mass resolving power of 60,000 m/Dm at m/z 400. Separation of DHDN and polyols was performed isocratically on a Shodex Asahipak NH2P-40 2D column (2 × 150 mm, 4 μm, 100 Å) at flow rate of 0.3 mL/min, column temperature of 40 °C, and eluent mixture 90:10 MeCN and water with 0.05% ammonium formate. Analysis of p-NP and organosulfate products was performed with an Agilent Poroshell EC-C18 column (2.1 × 100 mm, 2.7 μm, 120 Å) at flow rate of 0.27 mL/min, column temperature of 30 °C, and eluent mixture 40:60 MeCN and water with 0.1% ammonium formate. For solutions containing ammonium sulfate, a 100 μL aliquot of reaction sample was mixed with 900 μL MeOH. This precipitated out the ammonium sulfate, which was filtered off, and the filtrate was analyzed without further purification. For solutions in pure water, a 100 μL aliquot of reaction sample was diluted with 900 μL of MeOH and used without further purification.
Direct Infusion High Resolution Mass Spectrometry (HRMS) – Aerosol particles collected from chamber experiments were extracted with 0.5 mL LC-MS grade acetonitrile (Fisher Optima) and 2 minutes sonication in a bath sonicator. The extract was then directly introduced into the Orbitrap mass spectrometer (above) at the same tuning specifications but with a mass resolving power of 30,000 m/Dm at m/z 400 to improve sensitivity. External mass calibration is performed using commercial ESI calibration mix (Pierce Negative Mode Calibration solution). Calibrant analytes achieve a mass accuracy of < 2 ppm after recalibration. The sample mass spectra with signal to noise ratio (S/N) > 3 were processed by subtracting the mass spectra of the control filter extracts, deconvoluted with a quadratic fit model and deisotoped using Decon2LS tools (freeware from PNNL), mass corrected with the external calibration curve, and assigned to molecular formulas using a custom Matlab protocol based on heuristic mass filtering rules (67) and Kendrick Mass (KM) defect analysis (68, 69) with KM base of CH2.
1H Nuclear Magnetic Resonance (NMR) – Due to the large HOD/H2O and NH4 signals, 50 mM of 1,2-DHI was mixed with 3.7M AS in D2O to increase the analytical signal. Reactions for NMR analyses were performed by directly irradiating the sample within a 5 mm quartz NMR tube rated for 500 MHz. 1H spectra were collected on a 400 MHz Bruker instrument (400 MHz Bruker Avance III HD Nanobay Spectrometer) with water suppression, using an autosampler and analyzed using TOPSPIN. Water suppression was run using the standard WATERSUP parameters. Cyclohexane was used as an internal standard. Cyclohexane in CDCl3 (0.8 vol.%) was used as an internal standard; the internal standard mixture was loaded into a glass capillary, flame sealed, and dropped into an NMR tube containing the reaction mixture. The solution was irradiated for 1, 24, and 48 h. Aqueous 1,2-DHI first-order loss rates were extrapolated down to 1 mM by kinetic modeling.
Gas chromatography mass spectrometry (GC MS) – Analyses were performed on an Agilent 6890N gas chromatograph coupled to an Agilent 5973N quadrupole mass spectrometer using a silylation procedure previously reported for aqueous alcohols (70). Calibration was performed with pure sample in ammonium sulfate solutions to confirm the linearity of this method using our system.
UV-Vis spectroscopy – UV-vis spectroscopy was used to characterize the absorption of reagents, and for kinetic analyses of some reactions of p-NP (in addition to HPLC-HRMS). Spectra were obtained on a Shimadzu UV-1800 UV Vis spectrometer. Samples were directly inserted into the instrument between irradiation without any alterations. Spectra were collected every 15 minutes for 1 h. It is noted that UV-Vis determination of kinetics for p-NP taken at the ~ 320 nm peak are nearly identical to determinations using HPLC-HRMS (Table S1).
Inductively Coupled Plasma Mass Spectrometry (ICP-MS) – Ammonium sulfate solutions (0.15 M) in ultrapure water were analyzed for trace metals using a Thermo Scientific iCAP RQ ICP-MS operating in KED mode (He atmosphere). The analytical matrix is 2% v/v Omnitrace HNO3 and 0.5% v/v Omnitrace HCl. The instrument was calibrated using a 43-element calibration solution (Inorganic Ventures IV-ICPMS-71A). A 6-element internal standard (IV-ICPMS-71D) was added to samples prior to quantitation. Quality control checks were performed using NIST-1643f and Inorganic Ventures IV-STOCK-50 reference solutions.
Chemical Ionization Mass Spectrometry (CIMS) – 1,2-DHI, formic acid, and hydroxyacetone were quantified using a custom-built triple-quadrupole chemical ionization mass spectrometer (CIMS) using CF3O- as the reagent ion. Details of the instrument and the humidity-dependent calibration methods have been described previously (71, 72). Authentic standards were used to calibrate the CIMS; formic acid and hydroxyacetone were purchased from Sigma Aldrich, and 1,2-DHI was synthesized as previously described (23). Formic acid was detected as its F-transfer ion at m/z 65. Hydroxyacetone and 1,2-DHI were detected as clusters with CF3O- at m/z 159 and m/z 187. Calibrated signals in CIMS have a quantification uncertainty of 20-30%.
Bulk aqueous experiments. Photochemical reactions of bulk solutions (Figs. 1, 2, 4, 5, and 6) were performed in a photochemical enclosure equipped with a UV-B broadband fluorescent light with peak wavelength emission at 310 nm (Fig. S1). For all non-NMR studies prior to irradiation, a 5 mL solution of the organic compound of interest at the desired organic concentration with 3.7M ammonium sulfate in milli-Q water was prepared. For pH-dependent studies, H2SO4 was added to the solution until pH 1-2 was achieved. All other determinations were performed at pH 5-6. The pH of solutions was measured with a micro pH electrode (LE422) that was calibrated with commercial pH standards. The aqueous solution was transferred to a 3.5 mL capped quartz cuvette with stopper (Thor Labs). This was placed in the photochemistry chamber to irradiate for the desired reaction time. Aliquots of approximately 100 μL volume were taken for the chemical analysis procedure specific to the target organic analyte (GC-MS for erythritol, NMR for 1,2-DHI, HPLC-HRMS for others; see instrument-specific subsections above) using Hamilton gas-tight syringes. Reactions performed using natural sunlight as the irradiation source were prepared in the same way as those using artificial lighting. One control sample (organic in water) and one treatment sample (organic + 3.7M AS in water) in quartz vials were placed on the roof of Meyer Hall at the UC Davis main campus lying on a reflective aluminum sheet (Anomet Inc.) and exposed to sunlight. Samples were collected over several hours and analyzed. Deoxygenated experiments for the p-NP reaction were performed using water sparged with ultra-high purity nitrogen (N2) for 2 hours at room temperature. Solutions were then prepared in quartz vials, irradiated inside of a glove bag under pure N2 atmosphere, capped, and taken out of the glove bag for UV-Vis analyses at various time intervals. Analytical uncertainties represent 1 standard deviation of data across repeated experiments for the majority of experiments. For the 1,2-DHI experiment, uncertainties also include errors resulting from data extrapolation using a model.
Chamber Experiments. The aerosol experiments shown in Fig. 3 were conducted using a 10 m3 Teflon atmospheric chamber, temperature controlled to 20 °C. Prior to experiments, the chamber was cleaned by humidifying to 80% RH, injecting with excess H2O2 and irradiated with UV lights for 12 hours to remove any deposited organic materials before being flushed with dry, purified air for at least 24 hours. For each experiment, the chamber was humidified to ~80% RH prior to all injections using ultrapure water (18 MΩ, Millipure Milli-Q) at 35 °C circulated through a Nafion membrane humidifier while purified air flowed through the humidifier and into the chamber. Temperature and RH were monitored continuously by a membrane probe (Vaisala Inc.) calibrated with saturated salt solutions in the RH range of 11 – 95 %. Seed particles were generated by atomizing 20mM ammonium sulfate through a wet wall denuder. 1,2-DHI was injected in the gas phase by flowing ultrapure nitrogen gas through a clean glass bulb with a few drops of the neat standard until desired concentration is achieved, as monitored by the CIMS. A scanning mobility particle sizer (SMPS), comprised of an electrostatic classifier (TSI 3080) and a condensation particle counter (TSI 3772), was used to determine the particle size distribution and number concentration. Approximately 1500 μg/m3 total particle mass (including water) was used for experiments. Particles were sampled shortly after lights were turned on in the chamber via a 20 LPM flow through a hydrophilic PTFE membrane filter with 0.2 μm pores (Omnipore, Millipore Sigma). Approximately 6 hours of collection time was used to obtain sufficient organic signal for the extraction to avoid sample concentration.