Product study of oleic acid ozonolysis as function of humidity

Abstract

The heterogeneous reaction of ozone with oleic acid (OA) aerosol particles was studied as function of humidity and reaction time in an aerosol flow reactor using an off-line gas chromatography mass spectrometry (GC–MS) technique. We report quantitative yields of the major C9 ozonolysis products in both gas and condensed phases and the effect of relative humidity on the product distribution. The measurements were carried out with OA aerosol particles at room temperature. The results indicate that the product yields are increasing with increasing relative humidity during the reaction. Nonanal (NN) was detected as the major gas-phase product (55.6 ± 2.3%), with 94.5 ± 2.4% of the NN yield in the gas, and 5.5 ± 2.7% in the particulate phase, whereas nonanoic, oxononanoic and azelaic acids were detected exclusively in the particulate phase. Using UV-spectrometry, we observed that peroxides make up the largest fraction of products, about half of the product aerosol mass, and their concentration decreased with increasing humidity.

Introduction

Organic species from a variety of anthropogenic and biogenic sources constitute a substantial fraction of the atmospheric aerosol (Ramanathan et al., 2001). The ozonolysis of particle phase organic molecules can lead to changes in the physicochemical properties of atmospheric aerosol (Rudich, 2003). Fatty acids are ubiquitously present in marine (Mochida et al., 2002), rural (Simoneit and Mazurek, 1982) and remote continental locations (Simoneit et al., 1988). Unsaturated fatty acids are a significant fraction of fatty acids, and in some polluted rural areas, they even exceed the concentrations of saturated fatty acids (Kawamura and Gagosian, 1987). Oleic acid (9-octadecenoic acid, OA) is found in the atmosphere at concentrations up to 33 ng m⁻³ in urban aerosol (Ketseridis et al., 1976). It has a low vapour pressure, a symmetric structure and a double bond that is susceptible to attack by ozone. The reaction of OA aerosol or of OA bulk samples with ozone has developed to a popular model for oxidative organic aerosol processing in the atmosphere (Zahardis and Petrucci, 2007).

Fig. 1 summarizes the OA–ozone reaction mechanism. Step 1 involves formation of an unstable primary ozonide that rapidly decomposes, which leads to branching into two separate routes (Criegee, 1975). The first one gives nonanal (NN) and a Criegee Intermediate with a carboxylic acid end (CI 1). Similarly, the second route gives 9-oxononanoic acid (ON) and another CI 2. In the gas phase, CIs are very short-lived and are known to rapidly rearrange to the corresponding acids (Horie and Moortgat, 1991), which for OA then leads to azelaic (AA) and nonanoic (NA) acids. In the condensed phase, CIs are longer lived due to more rapid thermalization. This opens the path to secondary chemistry as shown in Step 2 of Fig. 1. So far, the following pathways involving the CIs have been suggested:

• Reaction with water and formation of a hydroxyhydroperoxide (HHP, pathway 1, Step 2), that in turn decomposes heterogeneously to form either a carbonyl and/or a carboxylic acid, as is known from gas-phase studies (Hasson et al., 2001). Liquid phase studies also show that the CI can react with protic solvents including water. Moreover the efficiency of the reaction between CI and water or another protic species is higher than that of the CI–aldehyde reaction (Chen et al., 2008b).

• Reaction with the carbonyl group of NN or ON and formation of three secondary ozonides.

• Reaction with a carboxylic group of ON, NA and AA and formation of a variety of α-acyloxyalkyl hydroperoxides (AAHP). The CIs primarily react with the first generation products, namely ON, that leads preferentially to the formation of the corresponding AAHPs (Ziemann, 2005).

• Reaction with another CI to form a variety of diperoxides. However at 25 °C they rapidly decompose by loss of O₂ to form a pair of corresponding aldehydes (Fliszar and Chylinska, 1968). Thus reaction with CI 1 forms NN, while reaction with CI 2 forms the carboxylic group of ON.

Step 3 in Fig. 1 describes the route to high-molecular weight compounds. The reactivity of AAHP with the CI leads to formation of peroxidic polymers. Hung and co-workers have observed the presence of ester groups by IR after processing of OA droplets with ozone (Hung et al., 2005). They assigned ON and AA as polymerisation propagators by reaction with CI. Zahardis and co-workers also observed polymer formation in the OA–ozone reaction and also suggested a similar polymerisation mechanism, namely the addition of CI to the carboxyl group of OA forming AAHP products (Zahardis et al., 2006b).

In spite of a large number of previous OA ozonolysis studies conducted in both coated-wall flow tube and in aerosol flow reactors (Hearn and Smith, 2005, Hung et al., 2005, Katrib et al., 2004, Moise and Rudich, 2002, Rudich, 2003, Smith et al., 2002), neither the gas–condensed phase partitioning, nor the absolute yields of the direct products (NN, NA, etc.), nor the effect of humidity have been determined so far. Water vapour could influence the product distribution either by reaction with the CIs or by influencing the structure of the aerosol.

We used derivatization GC–MS to determine the product yields of the four primary reaction products NN, AA, ON, and NA, in both the gas and the particulate phases, from OA ozonolysis in an aerosol flow reactor. The ozone concentrations were 500 ppb, and the reaction time was varied from 60 to 360 s corresponding to atmospherically relevant ozone exposures in the range 0.3 × 10⁻⁴–1.8 × 10⁻⁴ atm s. Thereby, we assume that higher concentrations over shorter time periods are comparable to lower concentrations over longer time scales. Lee and Chan (2007) found no significant changes in the product chemistry, when the ozone concentration was increased to 10 ppm. In our study, we were in addition interested in the effects of relative humidity. We used a UV photometric method to quantify the total yield of peroxides, which make up the majority of other products (Docherty et al., 2005).

Thus this study was focused on (1) the product yields of the OA–ozone reaction at different ambient conditions, (2) the understanding of how relative humidity affects overall product yields and product distribution, (3) the partitioning of products between the gas and particulate phases and (4) the mechanism of this reaction in presence of gas-phase water.