The Gas-phase Ozonolysis of 1-Penten-3-ol, (Z)-2-Penten-1-ol and 1-Penten-3-one: Kinetics, Products and Secondary Organic Aerosol Formation

Transcription

1 Z. Phys. Chem. 224 (2010) DOI zpch by Oldenbourg Wissenschaftsverlag, München The Gas-phase Ozonolysis of 1-Penten-3-ol, (Z)-2-Penten-1-ol and 1-Penten-3-one: Kinetics, Products and Secondary Organic Aerosol Formation By M. A. O'Dwyer 1, T. J. Carey 1, R. M. Healy 1, J. C. Wenger 1, *, B. Picquet-Varrault 2, and J. F. Doussin 2 1 Department of Chemistry and Environmental Research Institute, University College Cork, Cork, Ireland. 2 Laboratoire Interuniversitaire des Systèmes Atmosphériques (LISA), UMR-CNRS 7583, Universités de Paris 12 et Paris 7, CNRS, Créteil, France Dedicated to Prof. Dr. Reinhard Zellner on the occasion of his 65 th birthday (Received December 23, 2009; accepted March 25, 2010) Ozonolysis. 1-Penten-3-ol. (Z)-2-Penten-1-ol. 1-Penten-3-one The gas-phase ozonolysis of the biogenic unsaturated compounds 1-penten-3-ol, (Z)-2-penten- 1-ol and 1-penten-3-one has been investigated in two atmospheric simulation chambers. The following rate coefficients (in units of 10 K17 cm 3 molecule K1 s K1 ) were determined at atmospheric pressure and 293±2 K using an absolute rate method: 1-penten-3-ol, (1.64±0.15); (Z)-2- penten-1-ol, (11.5±0.66); 1-penten-3-one, (1.17±0.15). Reaction products were identified by in situ FTIR spectroscopy and gas chromatography mass spectrometry (GC-MS). The major products and their average molar yields in the presence of a radical scavenger at relative humidity < 1% were: formaldehyde (0.49±0.02), 2-hydroxybutanal (0.46±0.03) and propanal (0.15±0.02) from 1-penten-3-ol; propanal (0.39±0.03) and glycolaldehyde (0.43±0.04) from (Z)-2-penten-1-ol; formaldehyde (0.37±0.02) and 2-oxobutanal (0.49±0.03) from 1-penten-3- one. The formation of secondary organic aerosol was also observed with yields ranging from for the unsaturated alcohols. Significantly lower yields of around 0.03 were measured for 1-penten-3-one. The results of this work are used to determine atmospheric lifetimes and reaction mechanisms for the gas-phase ozonolysis of 1-penten-3-ol, (Z)-2-penten-1-ol and 1- penten-3-one. The broader atmospheric implications of this work are also discussed. * Corresponding author. j.wenger@ucc.ie

2 1060 M. A. O'Dwyer et al. 1. Introduction The gas-phase reactions of ozone with unsaturated volatile organic compounds (VOCs) are of major importance in atmospheric chemistry [1, 2]. In addition to being an atmospheric sink for both ozone and unsaturated VOCs, ozonolysis reactions are also significant sources of oxygenated hydrocarbons, radicals and secondary organic aerosol (SOA) in the atmosphere [2]. In order to determine the atmospheric impact of a specific ozonolysis reaction, detailed information on the kinetics, products and SOA formation is required. The oxygenated pentene derivatives, 1-penten-3-ol, (Z)-2-penten-1-ol and 1- penten-3-one, are part of a family of compounds known as the green leaf volatiles which are emitted by plants following leaf damage caused by cutting, drying etc. [3]. In particular, it has been shown that these oxygenated pentene derivatives are emitted as a result of freeze-thaw cycles [4] and total mixing ratios of up to 5 ppbv were observed in air following a freezing spell in the Austrian Alps [5]. The subsequent atmospheric degradation of 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one can be initiated via gas-phase reaction with hydroxyl (OH) radicals, nitrate (NO 3 ) radicals and ozone. A number of laboratory studies have been performed to date. Orlando et al. [6] determined rate coefficients and products for the reaction of OH with 1-penten-3-ol and (Z)-2-penten-1-ol, whilst Jiménez et al. [7] recently reported temperature-dependent kinetic data for the photooxidation of 1-penten-3-ol and 1-penten-3-one. The rate coefficients for reaction of the three compounds with NO 3 have been determined by Pfrang et al. [8, 9], whilst Grosjean et al. investigated the kinetics and products of the ozonolysis reactions [10 13]. Apart from the reaction of 1-penten-3-ol with OH, each of the atmospherically relevant reactions has only been studied once. Further experimental work is therefore required in order to verify kinetic and mechanistic parameters and also to ascertain the SOA formation potential of these biogenic compounds. In this work the gas-phase ozonolysis of 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one has been performed in two atmospheric simulation chambers. Absolute rate coefficients have been determined and the reaction products identified by in situ FTIR spectroscopy and gas chromatography mass spectrometry (GC-MS). The formation of secondary organic aerosol during these reactions was also investigated. 2. Experimental Experiments were performed in the 3910 L atmospheric simulation chamber at University College Cork (UCC) and in the 977 L reactor at the Laboratoire Interuniversitaire des Systèmes Atmosphériques (LISA) in Paris.

3 The Gas-phase Ozonolysis L chamber at UCC The chamber consists of a cylinder (4 m long, 1.2 m diameter) made of chemically inert FEP Teflon foil sealed at the ends by two aluminium endplates coated with FEP foil. A detailed description is provided elsewhere [14, 15]. The chamber design has been modified to include fans, located at both ends of the cylinder, in order to promote rapid mixing of reactants. The chamber was operated at 293±2 K using purified dry air (relative humidity < 0.4%) at mbar above atmospheric pressure. The chamber was equipped with a multiple reflection optical arrangement coupled to an FTIR spectrometer (BioRad Excalibur) for chemical analysis by in situ FTIR spectroscopy using a path length of (229.6±0.6) m. Infrared spectra were obtained throughout the reactions at a resolution of 1 cm K1 using a narrow-band mercury cadmium telluride (MCT) detector and derived from the co-addition of 300 scans collected over 5 min. The oxygenated alkenes were added to the chamber using an inlet system in which known amounts of the substances were heated in a small flow of purified air. Ozone was produced by flowing oxygen through an ozone generator (Yanco Ozone Services GE60. FM5000) and into the chamber. Between experiments the chamber was cleaned by flushing with purified air at a flow rate of 150 L min K1 for a minimum of four hours and usually overnight. The humidity and temperature in the chamber was measured using a dew point hygrometer and temperature meter (DM70 Vaisala). Kinetics experiments were performed by measuring the rate of ozone decay in the presence of known excess concentrations of the oxygenated alkene [16, 17]. Ozone was added to the chamber and after a period of minutes, the alkene was rapidly added. The initial reactant concentrations were varied as follows; [O 3 ] = ( )!10 13 molecule cm K3 and [oxygenated alkene] = ( )!10 14 molecule cm K3. The estimated uncertainty in the concentration of the oxygenated alkenes is 1%. In all experiments, the initial alkene concentration was in excess by a factor of 10 or more to ensure that pseudo-first-order conditions prevailed. The concentration of ozone was measured at 10 second intervals throughout the reactions using an ozone analyser (Thermo Electron Corporation, model 49i) connected to the chamber via a Teflon tube. The ozone analyser works on the principle of UV absorption at 254 nm and although some of the reaction products (e.g. carbonyls, hydroperoxides) can absorb light at this wavelength, they were not generated in sufficient amounts to cause any interference with the ozone measurements. The loss of ozone to the walls of the chamber was measured periodically by performing control experiments where no alkene was present in the chamber. The loss of ozone was found to be negligible over the time scale of the kinetic experiments (typically of the order of minutes and seldom exceeding 15 minutes), confirming that reaction with the oxygenated alkenes was the only important removal process for ozone. Dilution of the reactant mixture due to sampling was also negligible. Product studies were performed with initial mixing ratios of ppmv for ozone and the oxygenated alkenes in the presence of carbon monoxide (7000

4 1062 M. A. O'Dwyer et al ppmv) which acts as a scavenger for any OH radicals produced during the reactions. The reactants and products were quantified using calibrated reference infrared spectra obtained by introducing known quantities of pure materials into the chamber. The formation and evolution of SOA was monitored using a scanning mobility particle sizer (TSI model 3936). During these experiments, O- (pentafluorobenzyl)-hydroxyl amine (PFBHA) derivatisation [18] was employed to identify the carbonyl compounds formed. At the end of each experiment the contents of the chamber were bubbled for 30 minutes at a flow rate of 0.5 L min K1 through two 10 ml glass impingers connected in series and filled with a solution of PFBHA dissolved in acetonitrile (10 mg L K1 ). A column filled with KI crystals was inserted between the chamber and the impingers to prevent ozone from entering the solutions. The impingers were immersed in a water-ice bath during sampling to increase the efficiency of the trapping process. A blank sample was also taken before the addition of O 3 to the chamber to determine background levels. Samples were kept in the dark for approximately 48 hours at room temperature to ensure complete conversion of the carbonyls to their oxime derivatives. One microlitre of the solution was analysed by GC-MS (Varian CP GC coupled to Saturn 2000 MS) using a CP-Sil8 fused silica capillary column (Varian, 30 m, 0.25 mm i.d., 0.5μm film thickness). The injector temperature was held at 200 C throughout the analysis. The oven temperature was programmed to maintain a temperature of 60 C for 1 min, then increased to 100 C at a rate of 5 C min K1, then to 280 C at a rate of 10 C min K1, and finally to 310 C at 30 C min K1 and held at this temperature for 5 min. The MS was operated in electron impact (EI) or chemical ionisation (CI) mode, using methane as the CI reagent gas, over the mass range 40 to L chamber at LISA Product studies were also carried out at 295±2 K and atmospheric pressure in the 977 L reaction chamber at the LISA in Paris. The chamber is a pyrex tube, 6 m in length, equipped with a FTIR spectrometer (Bomem DA8-ME) coupled to a multiple reflection White cell for in-situ analysis of gaseous species. Details concerning the chamber and the experimental procedures are provided elsewhere [19, 20]. Before each experiment the chamber was evacuated to a pressure of 1!10 K2 mbar to ensure the walls of the chamber were free of any residual compounds from previous experiments. The chamber was then filled to atmospheric pressure with a gaseous mixture comprising 80% N 2 and 20% O 2 and the mixing fans were started. The relative humidity was < 1%. The oxygenated alkenes were added to the chamber by flowing N 2 through known amounts of the reactant in a 0.55 litre bulb and into the chamber. Initial mixing ratios were in the range ppmv. Carbon monoxide ( ppmv), used as a radical scavenger, was flowed into the chamber at a rate of 2 L min K1. After a mixing period of around one hour, ozone ( ppmv), generated by passing oxygen through a silent discharge ozoniser (Kaufmann Umwelttechnik, GmbH),

5 The Gas-phase Ozonolysis 1063 was then flowed into the chamber. Chemical analysis was performed throughout the experiments using in-situ FTIR spectroscopy using a pathlength of 156 m. Once the chamber was filled with the N 2.O 2 mixture a background FTIR spectrum was recorded by co-adding 600 interferograms. Infrared spectra in the cm K1 range, with a resolution of 0.5 cm K1, were obtained for the duration of the experiment by co-adding 200 scans collected over approximately six minutes. The reactants and products were quantified using calibrated reference infrared spectra obtained by introducing known quantities of pure materials into the chamber. 2.3 Materials The following compounds, with stated purities in brackets, were obtained from Sigma Aldrich Chemical Company; 1-penten-3-ol (99%), (Z)-2-penten-1-ol (95%), 1-penten-3-one (99%), formaldehyde (99%), ethanal (99%), n-propanal (97%), n-butanal (99%), formic acid (99%), glycolaldehyde (99%), PFBHA ( 98%) and acetonitrile (99.9%). Carbon monoxide (99.99%) and methane (99.995%) were obtained from BOC gases and Air Liquide respectively. 3. Results and discussion 3.1 Kinetics studies Reactions of ozone with the oxygenated pentene derivatives were performed under pseudo first-order conditions; K d dt ln[o 3]=k'[oxygenated alkene] where k# is the pseudo-first-order rate coefficient, given by k# = k[o 3 ]. A plot of ln[o 3 ] versus time was obtained for each experiment and used to obtain the slope k#. The plots exhibited a strong linear decay with time over a broad range of concentrations during all experiments. Five or six experiments were performed on each oxygenated alkene. Plots of k# versus starting alkene concentration are shown in Figure 1, yielding straight lines with gradient k, the bimolecular rate coefficient for ozonolysis. A summary of the rate coefficients obtained in this work is presented in Table 1. The indicated errors associated with k are twice the standard deviation arising from the least squares fit of the second-order data. The available literature data are also listed in Table 1. Although the rate coefficient obtained for 1-penten-3-ol is in very good agreement with that reported by Grosjean et al. [11] the values for (Z)-2-Penten-1-ol and 1-penten-3-one differ by factors of around 1.5 and 2.0 respectively. Grosjean et al. [10, 11] also obtained the rate coefficients by measuring the decay of ozone by UV photometry in the presence of excess oxygenated alkene. However, a smaller number of experiments were performed on each compound (typically 3) and there was little

6 1064 M. A. O'Dwyer et al. Fig. 1. Plots of the pseudo-first-order rate coefficients (k#) versus initial alkene concentration for the reaction of ozone with 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one at 293±2 K. Table 1. Rate coefficients for the reaction of ozone with 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one. Compound k (10 17 cm 3 molecule s 1 ) T (K) Reference 1-penten-3-ol 1.64± ±2 This work 1.79± ±1 [11] (Z)-2-penten-1-ol 11.5± ±2 This work 16.9± ±1 [11] 1-penten-3-one 1.17± ±2 This work 0.60± ±1 [10] or no variation in the starting concentration of the oxygenated alkene. As a result, the rate coefficients reported here could possibly be more reliable. Nevertheless, further experiments using a different technique, e.g., the relative rate method might help to resolve the discrepancies. Reaction of ozone with the oxygenated alkenes proceeds via electrophilic addition of ozone to the carbon-carbon double bond. Thus reactivity will be affected by the nature, number and position of the substituents on the >C=C< bond. Compounds with internal carbon-carbon double bonds have a higher reactivity than those with terminal >C=C< bonds because the electron-donating alkyl substituents promote the electrophilic addition of ozone. Thus, (Z)-2-penten-1- ol, has a considerably higher reactivity than both 1-penten-3-ol and 1-penten-3- one. The parent alkenes, (Z)-2-pentene (k(o 3 ) = 20.9!10 K17 cm 3 molecule K1 s K1 [1]) and 1-pentene (k(o 3 ) = 1.1!10 K17 cm 3 molecule s K1 [1]) also exhibit the same reactivity pattern. There are two other observations of note.

7 The Gas-phase Ozonolysis 1065 Firstly, 1-penten-3-ol and 1-penten-3-one exhibit a similar reactivity to their parent alkene, indicating that the functional group does not significantly affect the rate coefficient. Secondly, 1-penten-3-ol is slightly more reactive than its parent alkene, while (Z)-2-penten-1-ol is slightly less reactive. As indicated by Grosjean et al. [11], these differences in reactivity may reflect a specific influence of the OH group, such as electronic and steric effects during the addition step, stability of the primary ozonide, or stability of the hydroxyalkyl-substituted Criegee intermediate. Theoretical calculations based on transition state theory, such as those performed by Sato et al. [21], may help to provide further information on the factors affecting the reactivity of the oxygenated alkenes towards ozone. 3.2 Product studies FTIR spectroscopy was used to identify and quantify products of the reaction of O 3 with 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one. FTIR spectra obtained for the ozonolysis of (Z)-2-penten-1-ol are shown in Figure 2. The data obtained from the spectral analysis were used to produce concentration-time profiles and product yield plots by subtraction of calibrated reference spectra. Wall loss rates were measured for reactants and all commercially available products, but were found to be negligible over the timescale of the experiments. Typical product yield plots are shown in Figures 3 5 and a summary of the products identified and quantified by FTIR spectroscopy is shown in Table 2. As shown in Figure 2, subtraction of the absorption features of reactants and known products results in a residual spectrum. For (Z)-2-penten-1-ol, the residual spectrum only contains weak absorption features, indicating that the majority of reaction products have been identified. In contrast, the residual spectra for the 1-penten- 3-ol and 1-penten-3-one reactions exhibit strong absorption bands in the carbonyl region, cm K1, Figure 6. Since 2-hydroxybutanal and 2-oxobutanal are major products of the reaction of ozone with 1-penten-3-ol and 1-penten-3- one respectively [12, 13], then it is highly probable that their absorption bands contribute to the residual spectra. Unfortunately, neither of these compounds is commercially available and calibrated reference spectra could not be obtained. However, the position of the main carbonyl peak is very similar to that observed in butanal, Figure 6, and an estimate of the yields of these products was obtained by careful subtraction of this carbonyl absorption feature from the residual spectra. The identification of 2-hydroxybutanal and 2-oxobutanal as major products of the reaction of ozone with 1-penten-3-ol and 1-penten-3-one respectively was confirmed by PFBHA derivatisation followed by GC-MS analysis. Carbonyl compounds were distinguished by their oxime derivatives, which showed the characteristic peak at m.z = 181 [18] due to the C 6 F 5 CH 2 + ion and were identified by their molecular ion. Using chemical ionisation (CI), the oxime derivatives also produce a strong protonated molecular ion (M+1) and hydroxycarbonyls exhibit a noticeable peak at (M-17) due to the loss of OH. Additional peaks that

8 1066 M. A. O'Dwyer et al. Fig. 2. FTIR spectra in the range cm K1 obtained from the reaction of ozone with (Z)-2-penten-1-ol in the 977 L chamber: 1. (Z)-2-penten-1-ol before ozone addition; min after the addition of ozone; 3. Propanal reference spectrum; 4. Glycolaldehyde reference spectrum; 5) Residual spectrum after subtraction of (Z)-2-penten-1-ol, propanal and glycolaldehyde. Note that the CO and CO 2 ( cm K1 ) region has been removed for reasons of clarity. Fig. 3. Typical product yield plot obtained for the ozonolysis of 1-penten-3-ol in the 977 L chamber.

9 The Gas-phase Ozonolysis 1067 Fig. 4. Typical product yield plot obtained for the ozonolysis of (Z)-2-penten-1-ol in the 3910 L chamber. Fig. 5. Typical product yield plot obtained for the ozonolysis of 1-penten-3-one in the 977 L chamber. were used to identify the carbonyl compounds include m.z = M-30 and M+181, resulting from loss of NO from the neutral molecule, or from the coupling of a C 6 F 5 CH 2 + ion with the neutral molecule. It should be noted that some of the derivatives exist as geometric isomers consisting of Z and E forms due to restricted rotation of the carbon-nitrogen double bond. In addition, dicarbonyl com-

10 1068 M. A. O'Dwyer et al. Fig. 6. FTIR spectra in the range cm K1 (obtained in the 977 L chamber): 1. Residual spectrum for reaction of 1-penten-3-ol with ozone; 2. Residual spectrum for reaction of 1- penten-3-one with ozone; 3. Butanal reference spectrum. pounds such as 2-oxobutanal can exist in both the single and double derivative forms. However, the long derivatisation times employed in this study ensured complete conversion of the dicarbonyls to their double derivatives [22]. A list of the carbonyl products detected using this method and the associated diagnostic ions is provided in Table 3. In general, there is very good agreement between the product yields obtained in both chambers. The slight variation in the yields observed for formaldehyde may be due to small differences in the reaction conditions during experiments performed in the chambers. For example, variations in the amount of scavenger or reactants present may influence the extent to which secondary reactions can occur. Another possibility is that the slightly different amounts of water vapour present in the chambers could affect the reaction pathways of the Criegee intermediates and thus alter product distributions. The results obtained in this work are also compared to those reported by Grosjean and Grosjean [12, 13], who performed the reactions at a relative humidity of ca. 55% and used the 2,4-dinitrophenylhydrazine (DNPH) derivatisation method coupled with liquid chromatography, for the identification of carbonyl compounds. Although there is a reasonable level of agreement between the yields measured in the two studies, it seems probable that the variations in product distributions may be due, at least in part, to the large difference in water vapour present during the experiments.

11 The Gas-phase Ozonolysis 1069 Table 2. Products resulting from the gas-phase ozonolysis of 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one. Compound Product Average Product Yield a Previous studies [12,13] UCC LISA 1-penten-3-ol Formaldehyde 0.42± ± ± Hydroxybutanal 0.44±0.03 b 0.47±0.03 b 0.30±0.05 Propanal 0.15± ± ±0.01 Formic acid c c - (Z)-2-penten-1-ol Propanal 0.38± ± ±0.02 Glycolaldehyde 0.44± ± ±0.04 Formaldehyde ±0.02 Ethanal d ±0.01 Methylglyoxal ± penten-3-one Formaldehyde 0.41± ± ± Oxobutanal 0.51±0.03 b 0.46±0.03 b 0.44±0.03 Formic acid c b - a Average yields are based on 3 4 experiments on each reaction system at UCC and 2 3 experiments at LISA. Errors are twice the standard deviation arising from the least squares fit of the data and represent precision only. b Estimated yield based on subtraction of butanal carbonyl peak in FTIR spectra. c Secondary product, not quantified. d Identified by GC-MS but not quantified. 3.3 Reaction mechanisms The ozonolysis of unsaturated hydrocarbons proceeds via electrophilic addition of the O 3 molecule across the >C=C< bond forming an ozonide which then cleaves to yield two primary carbonyls and the corresponding Criegee intermediates [1, 2]. The mechanisms for the reaction of ozone with 1-penten-3-ol, (Z)-2- penten-1-ol and 1-penten-3-one are presented in Figures 7 9. Assuming that the primary carbonyls are only produced from direct cleavage of the ozonide, the sum of their respective yields should be unity. The average of the product yields obtained in this work give total primary carbonyl yields of (0.96±0.05) for 1- penten-3-ol, (0.82±0.07) for (Z)-2-penten-1-ol and (0.86±0.05) for 1-penten-3- one. The small deviations from a yield of unity are likely to be a result of a combination of factors including: (i) uncertainties in quantification of the products, especially 2-hydroxybutanal and 2-oxobutanal; (ii) secondary reactions of the aldehydes; (iii) possible incorporation of the aldehydes into the aerosol phase, as outlined below. The relative yields of the primary carbonyls can also be used to determine the branching ratio between the two decomposition channels of the ozonide. Inspection of the data in Table 2 indicates that there is no significant preference for the formation of any particular primary carbonyl, with branching ratios of around 50:50. This observation is in line with results obtained for the corresponding alkenes [1] and

12 1070 M. A. O'Dwyer et al. Table 3. Carbonyl products identified using PFBHA derivatisation and GC-MS analysis. Compound Product m/z (EI) m/z (CI) 1-penten-3-ol Formaldehyde 225 (M); 406 (M + 181); 226(M + 1); 406 (M + 181) 195 (M-30); (M-30); 181 Propanal 253 (M); 434 (M + 181) 254 (M + 1); 434 (M + 181) 223 (M-30); (M-30); Hydroxy- 283 (M); 253 (M-30) 284 (M + 1); 464 (M + 181) butanal (M-30); 266 (M-17); 181 (Z)-2-penten- Ethanal 239 (M); 420 (M + 181) 1-ol 209 (M-30); 181 Propanal 253 (M); 434 (M + 181) 254 (M + 1); 434 (M + 181) 223 (M-30); (M-30); 181 Glycolaldehyde 255 (M); 436 (M + 181) 256 (M + 1); 436 (M + 181) 225 (M-30); (M-30); 238 (M-17); penten-3- Formaldehyde 225 (M); 406 (M + 181) 226(M + 1); 406 (M + 181) one 195 (M-30); (M-30); Oxobutanal a 476 (M); 279 (M-197) 477 (M + 1); 279 (M-197) 446 (M-30); (M-30); 181 a 2-Oxobutanal was only detected as a double derivative suggests that the presence of the oxygenated functional group, which is not directly linked to the >C=C< bond, does not greatly affect the reactivity of the ozonide. However, it is interesting to note that a 75:25 branching ratio is observed in the reaction of ozone with methyl vinyl ether [23], with the formation of the longer carbonyl and shorter Criegee intermediate preferred. In this case, the methoxy substituent is linked directly to the >C=C< bond and may therefore exert a stronger influence on the reactivity of the ozonide. The Criegee intermediates contain excess energy (denoted by *) and can undergo collisional stabilisation and follow a variety of decomposition pathways including; (i) O-atom channel; (ii) ester or hot acid channel and (iii) hydroperoxide channel. Although, in general, there is evidence for the existence of all three decomposition pathways, the hydroperoxide channel appears to be the principal decomposition route for the majority of Criegee intermediates, since the two former channels typically only account for < 5% of the reaction products [1, 2]. The branching ratio for the hydroperoxide channel can often be estimated from the yield of OH radicals produced during the reaction, which varies with the structure of the alkene, but is generally in the range 30 70% [1]. The simplest Criegee intermediate, (CH 2 OO)*, is a product of the ozonolysis of both 1-penten-3-ol and 1-penten-3-one. The major products arising from decomposition of (CH 2 OO)* are CO, OH, H 2 O, CO 2,H 2 and the stabilised Criegee species [1, 2]. The determination of product yields for H 2 O and CO 2 was not attempted due to interference from background levels of these compounds in the reaction chambers and in the housing for the optical components. The yield of

13 The Gas-phase Ozonolysis 1071 Fig. 7. Mechanism for the gas-phase reaction of ozone with 1-penten-3-ol. CO could also not be determined because it was used as the OH scavenger. The stabilised Criegee, CH 2 OO, is believed to undergo a variety of reactions including; (i) reaction with water to form HCHO and H 2 O 2 ; (ii) reaction with HCHO to produce formic acid, HCOOH; (iii) reaction with HCOOH to produce hydroperoxymethyl formate, HC(O)OCH 2 OOH; and (iv) reaction with water to produce hydroxymethyl hydroperoxide, HOCH 2 OOH [1, 2]. Whilst formic acid was positively identified as a reaction product using FTIR spectroscopy, its concentrationtime profile is indicative of a secondary product, suggesting that it may originate from the reaction of the stabilised Criegee with formaldehyde; CH 2 OO + HCHO / HCOOH + HCHO Note that formaldehyde is effectively a catalyst during this reaction. Hydroxymethyl hydroperoxide and hydroperoxymethyl formate may also be present in the reaction chambers and could thus be responsible for some of the absorption features in the residual infrared spectra. However, this cannot be confirmed due to the lack of reference spectra for these compounds. The ozonolysis of 1-penten-3-ol produces the Criegee intermediates (CH 2 OO)* and (CH 3 CH 2 CHOHCHOO)*. Assuming that, like other substituted Criegee intermediates, the major fate of (CH 3 CH 2 CHOHCHOO)* is decomposi-

14 1072 M. A. O'Dwyer et al. Fig. 8. Mechanism for the gas-phase reaction of ozone with (Z)-2-penten-1-ol. tion via the hydroperoxide channel, the expected organic products are propanoic acid and 2-oxobutanal, Figure 10. However, neither of these products was detected by FTIR spectroscopy or GC-MS. Furthermore, the three decomposition reactions (hydroperoxide channel, O-atom channel and ester.hot acid channel) do not account for the formation of propanal, which was identified by GC-MS and FTIR spectroscopy with a molar yield of This strongly suggests that (CH 3 CH 2 CHOHCHOO)* is involved in at least one other reaction pathway. The formation of propanal was also observed by Grosjean and Grosjean [12], who proposed that it may be produced from direct unimolecular decomposition of the excited Criegee, as shown in Figure 10. The ozonolysis of (Z)-2-penten-1-ol produces the Criegee intermediates (CH 2 OHCHOO)* and (CH 3 CH 2 CHOO)*. The latter species can produce methylglyoxal, formic acid, CO and OH radicals via the hydroperoxide channel. Methylglyoxal was not observed as a product in this work, although, Grosjean and Grosjean [12] did detect a small amount of methylglyoxal during the ozonolysis of (Z)-2-penten-1-ol and attributed this to decomposition of the hydroperoxide originating from (CH 3 CH 2 CHOO)*. For the other Criegee intermediate, (CH 2 OHCHOO)*, the hydroperoxide channel results in the formation of glyoxal and formic acid, Figure 11. However, glyoxal was not detected as a product, both in this work and by Grosjean and Grosjean [12], indicating that the hydrop-

15 The Gas-phase Ozonolysis 1073 Fig. 9. Mechanism for the gas-phase reaction of ozone with 1-penten-3-one. eroxide channel is not an important pathway for this species. Direct unimolecular decomposition of (CH 2 OHCHOO)* and (CH 3 CH 2 CHOO)* may possibly account for the observed formation of formaldehyde and ethanal respectively [12]. The ozonolysis of 1-penten-3-one produces the Criegee intermediates (CH 2 OO)* and (CH 3 CH 2 C(O)CHOO)*. It is of interest to note that the hydroperoxide channel is not possible for (CH 3 CH 2 C(O)CHOO)* since there is no hydrogen atom on the β-carbon. Thus it appears that unimolecular decomposition may dominate. This pathway, shown in Figure 12, results in the formation of the propionyl radical, CH 3 CH 2 CO, which rapidly reacts with oxygen to produce the corresponding peroxy propionyl radical. Under the experimental conditions employed in this study, this latter species can undergo reaction with other peroxy radicals to ultimately produce ethanal and CO 2 or react with HO 2 to yield a hydroperoxide or propanoic acid. Ethanal was not detected as a product in the experiments performed in this work and was also not detected by Grosjean and Grosjean [13]. It is possible that the hydroperoxide species, CH 3 CH 2 C(O)OOH, may be present in the reaction chambers and could thus be responsible for some of the absorption features in the residual spectra shown in Figure 6. However, this cannot be confirmed due to the lack of a reference spectrum. The results obtained in this work indicate that the hydroperoxide channel, which is usually the major reaction pathway for alkyl substituted Criegee inter-

16 1074 M. A. O'Dwyer et al. Fig. 10. Proposed reaction pathways for the (CH 3 CH 2 CHOHCHOO)* Criegee intermediate produced from the reaction of ozone with 1-penten-3-ol. mediates, is not as important for oxygenated Criegee intermediates. These species appear to undergo direct unimolecular decomposition more readily to yield so-called secondary carbonyls, as described by Grosjean and Grosjean [12]. It is also possible that the collisional stabilisation channel may be significant for these oxygenated Criegee intermediates [1, 2]. Although no information on the yields and possible reactions of the stabilized Criegee species can be obtained from the experiments performed in this study, it seems likely that they could react with HCHO, HCOOH and H 2 O, in an analogous manner to CH 2 OO, to produce acids and hydroperoxides. Another possibility worth considering is that the hydroperoxide channel is still important for the oxygenated Criegee intermediates, but that the resulting

17 The Gas-phase Ozonolysis 1075 Fig. 11. Proposed reaction pathways for the (CH 2 OHCHOO)* Criegee intermediate produced from the reaction of ozone with (Z)-2-penten-1-ol. hydroxylated hydroperoxides undergo different reaction pathways compared to their alkyl counterparts. For example, the hydroxylated hydroperoxides shown in Figures 10 and 11 may possibly undergo keto-enol tautomerism to produce CH 3 CH 2 C(O)CH 2 OOH and HC(O)CH 2 OOH respectively. However, such reactions remain speculative until further information becomes available on the reactivity of oxygenated Criegee intermediates. 3.4 Secondary organic aerosol formation The formation of secondary organic aerosol during the ozonolysis of the oxygenated pentene derivatives was observed in the 3910 L simulation chamber. The experiments were performed in the absence of seed aerosol and under dry conditions, which favoured the formation of new particles by nucleation. The temporal evolution of the aerosol size and number distributions were similar in all experi-

18 1076 M. A. O'Dwyer et al. Fig. 12. Proposed reaction pathway for the (CH 3 CH 2 C(O)CHOO)* Criegee intermediate produced from the reaction of ozone with 1-penten-3-one. Fig. 13. Number and size distribution of particles formed during the ozonolysis of 1-penten-3- ol in the 3910 L reaction chamber. The numbers in the legend refer to the time (in sec) elapsed since ozone was added to the chamber. ments. Typical data for 1-penten-3-ol are shown in Figure 13. A burst of particles with diameters in the range nm was observed immediately after the introduction of ozone. The subsequent growth of these particles via coagulation and condensation of gas-phase species was rapid and caused a drop in particle number. Towards the end of the experiments mean diameters in the range nm were measured.

19 The Gas-phase Ozonolysis 1077 Table 4. Yields of secondary organic aerosol produced during the ozonolysis of 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one. Compound Initial Mass Mass of Aerosol Yield b Conc. Reacted Aerosol (μg m 3 ) (μg m 3 ) Formed a (μg m 3 ) 1-Penten-3-ol (Z)-2-penten-1-ol Penten-3-one a Assuming a density of 1.0 g cm K3. b Calculated from (mass of aerosol formed).(mass of compound reacted) The yield of SOA produced in each experiment was determined from the ratio of the aerosol mass formed to the amount of hydrocarbon reacted at the point where the maximum particle concentration was observed. The aerosol mass was obtained from the measured volume concentration using a density of 1gcm K3 and was not corrected for possible wall loss. The calculated yields are listed in Table 4. Yields of were obtained for the alcohols, which compares with yields of ca and 0.09 reported for the ozonolysis of (Z)-3- hexen-1-ol and (Z)-3-hexenyl acetate respectively, under similar reaction conditions [24, 25]. It should be noted that in the experiments performed here and elsewhere [24 26], the initial mixing ratios of reactants are considerably higher than those observed in the ambient atmosphere. This is a general feature of simulation chamber experiments and arises from the requirement to generate sufficient aerosol mass for reliable measurements to be made. The extrapolation of SOA yields obtained from chamber experiments to realistic atmospheric conditions is subject to a high degree of uncertainty and remains a topical issue [26]. The large difference in aerosol yields observed for 1-penten-3-ol and 1- penten-3-one indicates that there are significant differences in the chemistry of the species produced during the ozonolysis reactions. As outlined above, the major difference appears to be the absence of the hydroperoxide channel for (CH 3 CH 2 C(O)CHOO)*. Thus it may be speculated that the hydroperoxide channel plays an important role in aerosol formation. However, recent work by Sadezky et al. [27] suggests that the stabilized Criegee intermediates may in fact combine or react with the parent alkene to produce oligomeric structures which can condense to form aerosol. At present, the reactions influencing SOA formation during the ozonolysis of unsaturated hydrocarbons are the subject of much uncertainty and speculation [26]. It seems that water plays an important role by acting as a scavenger for Criegee intermediates. Variations in the relative humidity can thus affect product distributions and SOA yields. Further work over a

20 1078 M. A. O'Dwyer et al. Table 5. Atmospheric lifetimes for 1-penten-3-ol, (Z)-2-penten-1-ol and 1-penten-3-one calculated using; 12 hour daytime average [OH] = 1.6!10 6 molecule cm K3, 24 hour average [O 3 ]= 7!10 11 molecule cm K3, 12 hour night-time average [NO 3 ]=5!10 8 molecule cm K3 [27]. Compound τ OH τ O3 τ NO3 1-penten-3-ol 2.6 h a 23.5 h c 40 h d (Z)-2-penten-1-ol 1.6 h a 3.7 h c 3.6 h d 1-penten-3-one 7.4 h b 29.2 h c 5910 h e a Using the rate coefficient determined by Orlando et al. [6]. b Using the rate coefficient determined by Jiménez et al. [7]. c Using the rate coefficient determined in this work. d Using the rate coefficient determined by Pfrang et al. [9]. e Using the rate coefficient determined by Pfrang et al. [8]. range of experimental conditions is clearly required to determine the key species and mechanisms responsible for aerosol production and growth. 3.5 Atmospheric implications The atmospheric lifetime of a compound with respect to reaction with a species X can be calculated by using the general equation (τ x )=1.k x [X], where k x is the bimolecular rate coefficient and [X] is the tropospheric concentration of the reactive species. The kinetic data obtained in this work have been used to determine the atmospheric lifetimes of the oxygenated pentene derivatives with respect to reaction with ozone. The data, along with the calculated lifetimes for reaction with the other important oxidising species, OH and NO 3 radicals, is presented in Table 5. The results indicate that reaction with OH radicals is the dominant atmospheric loss process for each compound. For (Z)-2-penten-1-ol, reaction with O 3 and NO 3 are also significant, while ozonolysis is only of minor importance for 1-penten-3-ol and 1-penten-3-one. Jiménez et al. [7] have also shown that direct photolysis by sunlight is not an important atmospheric degradation pathway for these compounds. Whether initiated by OH [6] or ozone [12, 13], atmospheric degradation of the oxygenated pentene derivatives results in the formation of aldehydes, which can react with OH and NO 3 radicals, and also undergo photolysis by sunlight, with lifetimes typically less than 1 day [28]. The high reactivity of the oxygenated pentene derivatives and their oxidation products implies that large emissions of these biogenic species will contribute to the formation of ozone and SOA in the atmospheric boundary layer and therefore impact on human health and climate. In addition, the observed formation of SOA from the oxygenated pentene derivatives may have interesting implications for the chemistry of the troposphere, since it gives further strength to the belief that smaller molecular weight biogenic compounds can give rise to aerosols [26], suggesting that the chemistry of the rural troposphere is more complicated than first thought.

21 The Gas-phase Ozonolysis 1079 Acknowledgement This work was funded by the European Commission (project EUROCHAMP, contract number RII3-CT ) and the European Science Foundation (INTROP programme). References 1. J. G. Calvert, R. Atkinson, J. A. Kerr, S. Madronich, G. K. Moortgat, T. J. Wallington, G. Yarwood, The Mechanisms of Atmospheric Oxidation of the Alkenes. Oxford University Press (2000). 2. D. Johnson, and G. Marston, Chem. Soc. Rev. 37 (2008) J. Kesselmeier, and M. Staudt, J. Atmos. Chem. 33 (1999) A. J. Fisher, H. D. Grimes, and R. Fall, Phytochemistry. 62 (2003) R. Fall, T. Karl, A. Jordon, and W. Lindinger, Atmos. Environ. 35 (2001) J. J. Orlando, G. S. Tyndall, and N. Ceazan, J. Phys. Chem. A. 105 (2001) E. Jiménez, B. Lanza, M. Antiñolo, and J. Albaladejo, Environ. Sci. Technol. 43 (2009) C. Pfrang, C. Tooze, A. Nalty, C. E. Canosa-Mas, and R. P. Wayne, Atmos. Environ. 40 (2006) C. Pfrang, M. T. Baeza Romero, B. Cabanas, C. E. Canosa-Mas, F. Villanueva, and R. P. Wayne, Atmos. Environ. 41 (2007) E. Grosjean, D. Grosjean, and E. Williams, Int. J. Chem. Kin. 25 (1993) E. Grosjean, and D. Grosjean, Int. J. Chem. Kin. 26 (1994) D. Grosjean, and E. Grosjean, J. Geophys. Res. 100 (1995) E. Grosjean, D. Grosjean, and J. H. Seinfeld, Int. J. Chem. Kin. 28 (1996) L. Thüner, P. Bardini, G. J. Rea, and J. C. Wenger, J. Phys. Chem. A. 108 (2004) G. M. Clifford, L. P. Thüner, J. C. Wenger, and D. E. Shallcross, J. Photochem. Photobiol. A: Chemistry. 176 (2005) J. Treacy, M. Curley, J. Wenger, and H. Sidebottom, J. Chem. Soc. Faraday Trans. 93 (1997) M. R. McGillen, T. J. Carey, A. T. Archibald, J. C. Wenger, D. E. Shallcross, and C. J. Percival, Phys. Chem. Chem. Phys. 10 (2008) J. Z. Yu, R. C. Flagan, and J. H. Seinfeld, Environ. Sci. Technol. 32 (1998) J. F. Doussin, D. Ritz, R. Durand-Jolibois, A. Monod, and P. Carlier, Analysis. 25 (1997) N. Carrasco, J.-F. Doussin, M. P. O'Connor, J. C. Wenger, B. Picquet-Varrault, R. Durand-Jolibois, and P. Carlier, J. Atmos. Chem. 56 (2007) K. Sato, B. Klotz, T. Taketsugu, and T. Takayanagi, Phys. Chem. Chem. Phys. 6 (2004) B. Temime, R. M. Healy, and J. C. Wenger, Environ. Sci. Technol. 41 (2007) B. Klotz, I. Barnes, and T. Imamura, Phys. Chem. Chem. Phys. 6 (2004) J. F. Hamilton, A. C. Lewis, T. J. Carey, J. C. Wenger, E. Borrás i Garcia, and A. Muñoz, Atmos. Chem. Phys. 9 (2009) J. F. Hamilton, A. C. Lewis, T. J. Carey, and J. C. Wenger, Anal. Chem. 80 (2008) M. Hallquist, J. C. Wenger, U. Baltensperger, Y. Rudich, D. Simpson, M. Claeys, J. Dommen, N. M. Donahue, C. George, A. H. Goldstein, J. F. Hamilton, H. Herrmann, T. Hoffmann, Y. Iinuma, M. Jang, M. Jenkin, J. L. Jimenez, A. Kiendler- Scharr, W. Maenhaut, G. McFiggans, Th. F. Mentel, A. Monod, A. S. H. Prévôt, J.

22 1080 M. A. O'Dwyer et al. H. Seinfeld, J. D. Surratt, R. Szmigielski, and J. Wildt, Atmos. Chem. Phys. 9 (2009) A. Sadezky, P. Chaimbault, A. Mellouki, A. Römpp, R. Winterhalter, G. Le Bras, and G. K. Moortgat, Atmos. Chem. Phys. 6 (2006) R. Atkinson, Atmos. Environ. 34 (2000) 2063.