Aqueous SOA formation from the photo-oxidation of vanillin: Direct photosensitized reactions and nitrate-mediated reactions

Abstract. Vanillin (VL), a phenolic aromatic carbonyl abundant in biomass burning emissions, forms triplet excited states (3VL*) under simulated sunlight leading to aqueous secondary organic aerosol (aqSOA) formation. This direct photosensitized oxidation of VL was compared with nitrate-mediated VL photo-oxidation under atmospherically relevant cloud and fog conditions, through examining the VL decay kinetics, product compositions, and light absorbance changes. The majority of the most abundant products from both VL photo-oxidation pathways were potential Brown carbon (BrC) chromophores. In addition, both pathways generated oligomers, functionalized monomers, and oxygenated ring-opening products, but nitrate promoted functionalization and nitration, which can be ascribed to its photolysis products (•OH, •NO2, and N(III), NO2- or HONO). Moreover, a potential imidazole derivative observed from nitrate-mediated VL photo-oxidation suggested that ammonium may be involved in the reactions. The effects of secondary oxidants from 3VL*, pH, the presence of volatile organic compounds (VOCs) and inorganic anions, and reactants concentration and molar ratios on VL photo-oxidation were also explored. Our findings show that the secondary oxidants (1O2, O2•-/•HO2, •OH) from the reactions of 3VL* and O2 play an essential role in VL photo-oxidation. Enhanced oligomer formation was noted at pH < 4 and in the presence of VOCs and inorganic anions, probably due to additional generation of radicals (•HO2 and CO3•-). Also, functionalization was dominant at low VL concentration, whereas oligomerization was favored at high VL concentration. Furthermore, guaiacol oxidation by photosensitized reactions of VL was observed to be more efficient relative to nitrate-mediated photo-oxidation. Lastly, potential VL photo-oxidation pathways under different reaction conditions were proposed. This study indicates that the direct photosensitized oxidation of VL, which nitrate photolysis products can further enhance, may be an important aqSOA source in areas influenced by biomass burning emissions.



S3
consisted of A (water) and B (acetonitrile). The gradient elution was performed at a flow rate of 0.2 mL/min: 0-1 min, 10% eluent B; 1-25 min, linear increase to 90% eluent B; 25-29.9 min, hold 90% eluent B; 29.9-30 min, decrease to 10% eluent B; 30-35 min, re-equilibrate at 10% eluent B for 5 min. Standard solutions of VL and GUA ranging from 10 to 130 µM were analyzed along with samples and blanks using the channels with UV absorption at 300 and 274 nm, respectively.
The calibration curves for VL and GUA standard solutions are shown in Figure S2.
Text S4. IC analyses of small organic acids.
The small organic acids were analyzed using an ion chromatography system (IC, Dionex ICS-1100, Sunnyvale, CA) equipped with a Dionex AS-DV autosampler (Sunnyvale, CA). The separation was achieved using an IonPac TM AS15 column (4×250 mm) with an AG15 guard column (4×50 mm). The isocratic gradient was applied at a flow rate of 1.2 mL/min with 38 mM sodium hydroxide (NaOH) as the eluent. The total run time was set at 20 min. The standard solutions (1-50 μM) of formic, succinic, and oxalic acid were analyzed three times along with the samples and water blank. Formic, succinic, and oxalic acid had retention times of 3.6 min, 8.3 min, and 11.9 min, respectively.
The characterization of reaction products was performed using a UHPLC system (ExionLC TM AD, ABSciex, Concord, Canada) coupled to a quadrupole time-of-flight mass spectrometer (qToF-MS) (TripleTOF 6600+, ABSciex). The settings (e.g., column, mobile phase, gradient, oven temperature) in the UHPLC system were the same as those used in UHPLC-PDA (Text S3). The mass spectrometer was equipped with an electrospray ionization (ESI) source and operated in the positive ion mode (the negative ion mode signals were too low for our analyses) at S4 a resolving power (full width at half-maximum (fwhm) at m/z 300) of 30000 in MS and 30000 in MS/MS (high-resolution mode). Information-dependent acquisition (IDA) scanning was adapted for product identification. The acquisition using IDA consisted of a ToF-MS scan and informationdependent trigger events. The ToF-MS scan had an accumulation time of 250 ms and covered a mass range of m/z 30-700 with a declustering potential (DP) of 40 and collision energy (CE) of 10 eV. The accumulation time for the IDA experiment was 100 ms, and the MS/MS scan range was set from m/z 30-700 in high-resolution mode. The IDA criteria were as follows: 5 most intense ions (number of IDA experiments) with an intensity threshold above 50 cps, isotope exclusion was switched off, and dynamic background subtraction was switched on. The automated calibration device system (CDS) was set to perform an external calibration every four samples. The ESI source conditions were as follows: temperature, 500 °C; curtain gas (CUR), 25 psi; ion source gas 1 at 50 psi; ion source gas 2 at 50 psi; and ion-spray voltage floating (ISVF) at 4.5 kV.
All parameters in the liquid chromatography system and mass spectrometer were controlled using Analyst TF Software 1.8 (ABSciex). The high-resolution LC-MS data were processed with PeakView and Analyst in the SCIEX OS software 1.5 (ABSciex). Peaks from the blank sample were subtracted from the sample signals. In addition to a minimum signal-to-noise ratio of 30, a peak was determined as a product if the difference in peak area between the samples before and after irradiation is ≥10 times. The formula assignments were carried out using the MIDAS molecular formula calculator (http://magnet.fsu.edu/~midas/) with the following constraints: C ≤35, H ≤70, N ≤5, O ≤20, Na ≤1, and the mass error was initially set as 10 ppm. were constrained by a mass error mostly <5 ppm, which is a requirement for product identification using positive ion mode (Roemmelt et al., 2015). The double bond equivalent (DBE) values and carbon oxidation state (OSc) of the neutral formulas were calculated using the following equations (Koch and Dittmar, 2006) DBE = C − H/2 + N/2 + 1 (Eq. S1) where C, H, O, and N correspond to the number of carbon, hydrogen, oxygen, and nitrogen atoms in the neutral formula, respectively. Based on the identified products, the average oxygen to carbon  (Table S1) and DBE values, examples of structures for products identified from VL (and GUA) photooxidation experiments were proposed (Table S3).
In this work, 2-nitrobenzaldehyde (2NB), a chemical actinometer, was used to determine the photon flux in the aqueous photoreactor. We first measured the relative intensity of light passing through the empty reactor, then the reactor containing 50 μM 2NB using a high-sensitivity spectrophotometer (Brolight Technology Co. Ltd, Hangzhou, China) equipped with an optical fiber (Brolight). Then, the average relative intensity absorbed by 2NB solution as a function of wavelength was calculated. Briefly, the photolysis of 50 μM 2NB in the reactor was monitored by determining its concentration every 5 min for a total of 35 min, during which 2NB was almost completely decayed. The concentration of 2NB was measured using UHPLC-PDA, and the settings (e.g., column, mobile phase, gradient, oven temperature) were the same as those for VL decay analysis (Text S3). The channel with UV absorption at 254 nm was used for the quantification of 2NB. The concentration of 2NB in the reactor followed exponential decay, and its decay rate constant was determined using the following equation: where NA is Avogadro's number, λ ′ is the actinic flux (photons cm −2 s −1 nm −1 ), ∆λ is the wavelength interval between actinic flux data points (nm), and 2NB,λ and 2NB,λ are the base-10 molar absorptivity (M −1 cm −1 ) and quantum yield (molecule photon −1 ) for 2NB, respectively.
Values of 2 , (in water) at each wavelength under 298 K and a wavelength-independent 2NB value of 0.41 were adapted from Galbavy et al. (2010). Finally, λ ′ was estimated through Eq. S5.
The estimated photon flux in the aqueous reactor is shown in Figure S2.   Table S2. Reaction conditions, initial VL (and GUA) decay rates, normalized abundance of products, and average carbon oxidation state (<OSc>) in each experiment. Except where noted, the reaction systems consisted of VL (0.1 mM); GUA (0.1 mM), AN (1 mM); sodium nitrate (SN) (1 mM); VOC (IPA) (1 mM) or inorganic anions (NaBC) (1 mM) under air-saturated conditions after 6 h of simulated sunlight irradiation. Analyses were performed using UHPLC-qToF-MS equipped with an ESI source and operated in the positive ion mode.
a Irradiation time for VL* (0.01 mM, A14) was 3 h. b The data fitting was performed in the initial linear region. c Ratio of the normalized abundance of the 50 most abundant products to that of total products, except for direct GUA photodegradation (A17), GUA+VL (A18), and GUA+AN (A19) whose ratios are based on the absolute signal area of products. d The normalized abundance of products was calculated using Eq. 2. e <OSc> of the 50 most abundant products.       The normalized abundance of products was calculated from the ratio of the peak area of the product to that of VL (Eq. 2). The 50 most abundant products contributed more than half of the total normalized abundance of products, and they were identified as monomers (blue) and dimers (green). Grey peaks denote peaks with low abundance or unassigned formula. Examples of high-intensity peaks were labeled with the corresponding neutral formulas. The color bar denotes the normalized abundance of products. The grey dashed lines indicate the carbon oxidation state values (e.g., OSc = -1, 0, and 1).    The normalized abundance of products was calculated from the ratio of the peak area of the product to that of VL (Eq. 2). The 50 most abundant products contributed more than half of the total normalized abundance of products, and they were identified as monomers (blue), dimers (green), trimers (red), and tetramers (orange). Grey peaks denote peaks with low abundance or unassigned formula. Examples of highintensity peaks were labeled with the corresponding neutral formulas. Note the different scales on the y-axes.  x x+2 ) (brown dashed line). Data points within the shaded area are potential BrC chromophores. Light grey circles show the classification of the data points as monomers, dimers, trimers, or tetramers. Figure S13. Reconstructed mass spectra of assigned peaks from (a) 0.01 mM VL* (A14), (b) 0.01 mM VL + 1 mM AN (A16), and (c) 0.01 mM VL + 0.01 mM AN (A15) at pH 4 under air-saturated conditions after 6 h of simulated sunlight irradiation. The normalized abundance of products was calculated from the ratio of the peak area of the product to that of VL (Eq. 2). The 50 most abundant products contributed more than half of the total normalized abundance of products, and they were identified as monomers (blue) and dimers (green). Grey peaks denote peaks with low abundance or unassigned formula. Examples of high-intensity peaks were labeled with the corresponding neutral formulas. (d) Contributions of different m/z ranges to the normalized abundance of products from experiments with low [VL] = 0.01 mM (A14-A16) and high [VL] = 0.1 mM (A5 and A7) at pH 4 under air-saturated conditions after 6 h of simulated sunlight irradiation.