Reaction Kinetics of Green Leaf Volatiles with Sulfate, Hydroxyl, and Nitrate Radicals in Tropospheric Aqueous Phase

Green plants exposed to abiotic or biotic stress release C-5 and C-6 unsaturated oxygenated hydrocarbons called Green Leaf Volatiles (GLVs). GLVs partition into tropospheric waters and react to form secondary organic aerosol (SOA). We explored the kinetics of aqueous-phase reactions of 1-penten-3-ol (PENTOL), (Z)-2-hexen-1-ol (HEXOL), and (E)-2-hexen-1-al (HEXAL) with SO4•–, •OH, and NO3•. At 298 K, the rate constants for reactions of PENTOL, HEXOL, and HEXAL with SO4•– were, respectively, (9.4 ± 1.0) × 108 L mol–1 s–1, (2.5 ± 0.3) × 109 L mol–1 s–1, and (4.8 ± 0.2) × 108 L mol–1 s–1; with •OH – (6.3 ± 0.1) × 109 L mol–1 s–1, (6.7 ± 0.3) × 109 L mol–1 s–1, and (4.8 ± 0.3) × 109 L mol–1 s–1; and with NO3• – (1.5 ± 0.15) × 108 L mol–1 s–1, (8.4 ± 2.3) × 108 L mol–1 s–1, and (3.0 ± 0.7) × 107 L mol–1 s–1. The rate constants increased weakly with temperatures ranging from 278 to 318 K. The diffusional limitations of the rate constants appeared significant only for the GLV–•OH reactions. The aqueous-phase reactions appeared negligible in deliquescent aerosol and haze water but not in clouds and rains. The atmospheric lifetimes of GLVs decreased from many days to hours with increasing liquid water content and radicals’ concentration.


■ INTRODUCTION
The impact of Volatile Organic Compounds (VOCs) on the air quality and formation of ozone in the troposphere became recognized in the 1950s. 1 Their crucial role as precursors of Secondary Organic Aerosol (SOA) was noticed in 1960 2 and earned global attention in the 1990s. 3 The contribution of SOA to the Earth's solar radiative budget, climate change, and cloud formation, as well as its impact on human health through gas-and aqueous-phase processes, have been progressively investigated. 4−6 Still, however, the vast fraction of potentially important SOA sources and transformation processes remain unknown. 5,7,8 Hydrophilic aerosol particles serve as Cloud Condensation Nuclei (CCN) and promote cloud formation playing a significant role in cooling the Earth's climate. 4 Among biogenic VOCs (BVOCs), isoprene 9−15 and monoterpenes 16−20 have already gained ample attention, while green leaf volatiles (GLVs), also potentially precursors of SOA, 21 have been much less studied.
Researching novel atmospheric compounds requires understanding their kinetics to predict their fate in the atmosphere. 44,45 In this work, we explored for the first time the kinetics of aqueous-phase reactions of three GLVs1penten-3-ol (PENTOL), (Z)-2-hexen-1-ol (HEXOL), and (E)-2-hexen-1-al (HEXAL) (Scheme 1)with tropospheric radicals • OH, SO 4 •− , and NO 3 • . Our main goal was to determine the rate constants and evaluate the atmospheric significance of the reactions. The examined GLVs may be effective precursors of aqueous SOA formation like other GLV, 37 even though they are moderately water-soluble and intermediary volatile. Their physical properties were estimated using the EPI suite 2012 from EPA 46 (Table S1). The global annual emission of GLVs (hexenal, hexenol, and hexenyl acetate) is 10−50 Tg C/yr, 47 giving rise to 1−5 Tg C/yr SOA, i.e., at least one-third of that from isoprene. 48 The local concentrations of the several GLVs, including 1-penten-3-ol, in the vicinity of stressed plants reach a few ppb. 49,50 Heiden et al. 51 and Jardine et al. 25 observed the high emission of many GLVs, including 1-penten-3-ol and (Z)-2-hexenal, under stress from pathogen attack or ozone exposure. Common anthropogenic activities like harvesting the cereal and biofuel grasses or residential grass mowing cause significant GLVs emissions that influence the local air quality. 52−54 Novel agricultural, horticultural, and forestry practices based on the fumigation of plants with GLVs for better resistance against pathogens and abiotic stress will probably increase the GLV emissions. 55,56 Thus, GLVs chemistry can play an essential role in the atmosphere and requires thorough attention in atmospheric chemistry and air quality models. Our work increases the chemical-kinetic database for the aqueous-phase reactions that is indispensable for such modeling, as generally postulated in several major reviews. 4 Figure S1) was used to measure the rates of the aqueous-phase oxidation of GLVs by the relevant radicals. A detailed description of the setup is available elsewhere. 57,59 The LFP-LLPA method applied is like that used by Schone et al. and Otto et al. 60,61 A freshly prepared aqueous solution containing a GLV compound and radical precursors was transferred into the solution tank. The solution flowed down through the measurement cell (4 cm × 3.5 cm × 2 cm) thermostated with Water Thermostat (Julabo or S LAUDA). In the measurement cell, the radical precursors' photolysis occurred by excimer laser (COMPEX 201 series) pulses of microsecond width triggered at 4 Hz (DG535 Digital Delay Generator, Standford Research Systems). A continuous-wave (CW) laser measured the radicals' light absorption after passing the beam across the cell several times by a White mirror setup. The signal's final intensity was measured with a photodiode and recorded with an oscilloscope (Data SYS 944, Gould) and a computer for further data processing to obtain a second-order rate constant k second of the reaction. The temperature of measurements was constant and varied from 278 to 318 K. The GLVs studied do not undergo ion speciation, so all experiments were carried out at pH close to 7. Table S4 shows the LFP-LLPA configuration, while Table  S5 shows the initial concentrations of GLVs and radical precursors. Figure S2 and Table S2 present the recorded UV spectra and calculated molar absorption coefficients of the GLVs. For experiments with PENTOL and HEXOL, the 248 nm excimer laser and 407 nm CW laser (LSR 407 nm, Coherent) were used to generate the • OH and SO 4 •− radicals and follow the reactant concentrations, respectively. Due to the strong absorption of light by HEXAL at 248 nm (ε 248 nm = 1722.1 L mol −1 cm −1 ), a 308 nm excimer laser and 473 nm CW laser (LasNova Series 40 blue, LASOS) were used for exploring the kinetics of HEXAL reactions with • OH and SO 4 •− (ε 308 nm (HEXAL) = 51.8 L mol −1 cm −1 ). The 473 nm CW laser secured better light absorption at low concentrations of radicals. Figure S3 shows a typical absorbance time trace in the experiments following a laser flash photolysis. The NO 3 • kinetics for all the three GLVs was followed using 351 nm excimer laser and 635 nm continuous-wave laser (Radius, Coherent  Figure S3) was the average of eight separate recordings. The intensity was converted to the concentration of SO 4 •− radicals using the molar extinction coefficients (ε 407 nm = 1260 L mol −1 cm −1 and ε 473 nm = 1389 L mol −1 cm −1 ). 62 A pseudo-first-order rate constant k first for the reaction was calculated from the slope of the concentration vs time plot. The pseudo-first-order k first constants were plotted against the initial concentrations of the GLV to obtain the second-order rate constant k second for the reaction. 60,61 • OH Kinetics. Because of weak light absorption and spectra of • OH overlapping with those of the organic constituents, • OH radicals are difficult to detect directly. 62−64 Therefore, the competition kinetics method 65 The dithiocyanate radical-anion ((SCN) 2 •− ) strongly absorbs light in the visible region of the spectrum (400−550 nm). 62 The solution's absorbance was measured using a CW laser at 407 nm for PENTOL and HEXOL and at 473 nm for HEXAL. The k second for the reaction GLV + • OH was calculated using eq 1 from Schaefer and Herrmann 66 and eq 2 from Zhu et al. 67 presence of GLV at the concentration X in the reaction solution, k ref is the reference rate constant for reaction b. The GLV compound absorbs UV light at both wavelengths of excimer laser (248 nm for PENTOL and HEXOL, and 308 nm for HEXAL). Thus, GLV act as internal filters of the UV light and reduce the initial • OH concentrations measured in experiments. Therefore, A [(SCN) 2 − ] 0 in eq 1 was corrected for each GLV at all temperatures using the procedure of Schaefer and Herrmann. 66 Table S3 shows the changes in the initial • OH concentrations (<0.05% for PENTOL, 0.2−0.9% for HEXOL, and 1−3% for HEXAL).
NO 3 • Kinetics. The reaction of GLV with NO 3 • radicals started by the photolysis of solutions containing NaNO 3 , Na 2 S 2 O 8 , and a GLV in the measurement cell, using 351 nm excimer laser. The NO 3 • radicals were generated by reactions f and g. 59,60,68 Light absorbance by NO 3 • was measured using a red diode CW laser (635 nm) and converted to concentrations using the molar extinction coefficient ε 635 nm = 1120 M −1 cm −1 . 69 The intensity−time traces were processed using the same method as for SO 4 •− kinetics to get the k second (GLV + NO 3 • ).
The uncertaintiy of each k second determined in the present study was calculated as a product of the standard deviation and the Student's t-factor taken with the 95% confidence level. Each rate constant determined for a single GLV at a single temperature was backed by 40 measurements (8 replicates for a single GLV concentration). Diffusion Limitations of Rate Constants. The radical reactions with k second on the order of 10 9 L mol −1 s −1 or higher can be controlled by the diffusion of reactants, at least in part. Therefore, we analyzed the experimental rate constants (i.e., the constants obtained from the LFP-LLPA experiments, k obs ) for diffusion limitations using a simple resistance-in-series approach 70 to split them into the true rate constants (k reac ) and the rate constants for the diffusion of reactants (k diff ): where all k are the second-order rate constants (L mol −1 s −1 ), D is a diffusion coefficient of reactants A and B (m 2 s −1 ), r is the radius of reactant molecules A and B (m), and N is the Avogadro number (for details, see Section S6, SI). Kinetic Modeling of Reactions. We used the COmplex PAthway SImulator of biochemical systems (COPASI from Bioinformatics, 71 to evaluate the bias of the rate constants determined for reactions of GLV with nitrate radicals. We chose the evolutionary programming method (number of generations 200, population size 20) for parameter estimation and the deterministic ordinary differential equation solver (LSODA) 71−73 for simulating the reaction time courses.

Reactions of SO 4
•− Radical-Anions with PENTOL, HEXOL, and HEXAL. Previous studies 74 showed that SO 4 •− radical is a strong oxidizing agent and reacts with many organic compounds at the rates nearly controlled by the diffusion of reactants. The experimental rate constants (k obs ) for the aqueous-phase reactions of PENTOL, HEXOL, and HEXAL with SO 4 •− determined in this study at 278−318 K range from (4.2 ± 0.2) × 10 8 to (2.9 ± 0.6) × 10 9 L mol −1 s −1 (Table S7).
The Arrhenius plots ( Figure 1) and eqs 5−7 show the rate constants weakly increase with temperature.
Reactions of NO 3 • Radicals with PENTOL, HEXOL, and HEXAL. The rate constants for the aqueous-phase reactions of PENTOL, HEXOL, and HEXAL with NO 3 • radicals range from (2.0 ± 0.6) × 10 7 to (9.8 ± 3.9) × 10 8 L mol −1 s −1 (Figure 3, Table S9). We could not determine the rate constant for HEXOL at 318 K, so we got the fifth value at 293 K. Figure 3 and eqs 11−13 show the temperature variation of the experimental rate constants.
The contribution of diffusion to k obs was 1−2% for PENTOL, 7−8% for HEXOL, and 0.2−0.4% for HEXAL (Table S9, Figure S4c), so all those reactions are fully chemically controlled. Large error bars in the Arrhenius plots result from low light absorption values measured in the experiments but still fall within the 95% confidence interval. The magnitude of the rate constants for the aqueous-phase reactions of GLV with radicals was 10 9 for • OH, 10 8 for SO 4 •− , and 10 7 L mol −1 s −1 for NO 3 • . Only for HEXOL, the rate constant for NO 3 • was larger than for SO 4 •− . HEXOL appeared the fastest reacting compound of the three GLV studied. The possible explanation is that the HEXOL molecule has a CC double bond position available for radical addition and two allylic positions available for H-abstraction by the radicals, while PENTOL and HEXAL have one CC position and only one allylic position available. The difference in the rate constants between HEXOL and PENTOL or HEXAL is more prominent in the case of SO 4 •− and NO 3 • reactions, as they are more chemically controlled than the partially diffusion-controlled reactions with • OH. Table 1 shows the rate constants and activation energies for aqueous-phase reactions of several GLVs and structurally similar compounds with • OH. 37,57,62 The activation energies range from 6 to 17 kJ mol −1 , and the rate constants from 2 ×     (Table 1). 37 The rate constants and activation energies for the reactions of structurally similar oxygenated compounds with • OH radicals, such as methyl isobutyl ketone and isobutyraldehyde, are also similar to those for GLV (Table 1).

Bias of the Experimental Rate Constants for Reactions of GLV with NO 3
• . The experimental method used to determine the rate constants assumed that NO 3 • radicals were consumed only in the reaction with a GLV. However, NO 3 • can participate in other reactions, e.g., with − OH, H 2 O, HO 2 • , S 2 O 8 2− , and organic peroxides that form by the autoxidation of alkyl compounds produced by the reaction of GLV with SO 4 •− radicals. To assess the influence of "neglected" reactions, we constructed a chemical-kinetic Model_1 including those reactions and used it to evaluate the rate constants for GLV+ NO 3 • reactions (for details, see Section 8, SI). Table 2 compares the experimental rate constants, experimental uncertainties from the LFP-LLPA procedure, and model-derived rate constants. The relative algebraic difference between the constants (eq 14) estimates the bias of the experimental constants due to the "neglected" sinks of NO 3 The bias is the largest for HEXAL, which reacts with NO 3 • most slowly among the GLVs studied. The experimental rate constants are overestimated by 6−25%. The effect decreases with temperature, probably due to the relative acceleration of the HEXAL − NO 3 • reaction. For PENTOL and HEXOL, the bias is smaller, with the rate constants overestimated by less than 15%. In most cases, the intrinsic uncertainty of the experimental rate constants determined by the LFP-LLPA procedure is significantly larger than the bias due to the "neglected" NO 3 • sinks, including the reactions with peroxy intermediates. Few exceptions occurred for the slowest reacting HEXAL at 288 and 308 K. Probably, the dataprocessing unit of the LFP-LLPA method can be modified based on the present results to reduce the bias for the rate constants of reactions with NO 3 radicals.
Activation Parameters. The activation parameters are essential in understanding the chemical mechanisms of reactions. Table 3 presents the activation parameters for reactions of GLVs with atmospheric radicals (SO 4 •− , • OH, and NO 3 • ). The equations for activation enthalpies (ΔH ‡ ), activation entropies (ΔS ‡ ), Gibb's activation energy (ΔG ‡ )   (Figures 1−3) and follow eq 15 where k is the rate constant, E A is the activation energy, A is the pre-exponential factor, R is the gas constant, and T is the absolute temperature. The Arrhenius expressions for the aqueous-phase reactions of PENTOL, HEXOL, and HEXAL with SO 4 •− , • OH, and NO 3 • are provided for the first time (eqs 5−13). The ratio of E A to the average kinetic energy (RT) directly influences the reaction rate constant. The E A values lower than 18 kJ mol −1 explain the weak temperature dependence of the reactions rates. The low activation enthalpies ΔH ‡ (2 to 15 kJ mol −1 ) and negative activation entropies ΔS ‡ (−72 to −23 J mol −1 K −1 ) explain the decreasing randomness of molecules within the system. They indicate that the aqueous-phase reactions studied mainly proceed via the radical addition to the double bond or associative pathway and warrant further theoretical and experimental investigation. The activation parameters for the diffusion-corrected rate constants were calculated following a similar procedure, and the activation energies are still less than 20 kJ mol −1 (Table S10).

■ ATMOSPHERIC IMPLICATIONS
Estimating GLV fluxes removed from the atmosphere by gasphase reactions, aqueous-phase reactions, and other processes like deposition to land and aquatic ecosystems requires extensive modeling of individual scenarios beyond the scope and size of this paper. To estimate the proportion of the gas-phase and aqueous-phase fluxes, we scaled the GLV removal rates dividing them by the concentration of GLV. That descriptor compares the corresponding GLV fluxes independent of the GLV concentration. Besides, we evaluated the atmospheric significance of gas-phase and aqueous-phase reactions of GLV with radicals using the commonly accepted method of atmospheric lifetimes and relative rates of removal.
Scaled GLV Removal Rates. Table 4 shows the scaled removal rates due to gas-phase and aqueous-phase reactions of PENTOL, HEXOL, and HEXAL with • OH, NO 3 • , and SO 4 •− in dry air, urban clouds, remote clouds, and urban aerosol (SO 4 − radicals occur only in the aqueous phase). The calculation was based on equations S12−S13 and data in Tables S7−S9, S11, and S12.
Data in Table 4 show that only in urban and remote clouds of high liquid water contents, the fluxes of 1-penten-3-ol and (Z)-2-hexen-1-ol removed by aqueous-phase reactions are comparable to the fluxes by gas-phase reactions. (E)-2-Hexen-1-al was removed faster by the gas-phase reactions in all clouds. In urban aerosol, the gas-phase removal of all GLV studied dominated the aqueous-phase one by several orders of magnitude. So was the case with all clouds.
Atmospheric Lifetimes. The atmospheric lifetime (t) of a GLV removed by the gas-phase and aqueous-phase reactions with a radical X is the time after which the initial GLV concentration in the gas phase [GLV] 0,g drops to [GLV] 0,g /e (eq 16, Section 7 in SI). The GLV and X partition between the phases according to Henry's Law (Equation S22 and Figure  S10 in the SI).

Notes
The authors declare no competing financial interest.