1. IntroductionVolatile organic compounds (VOCs) are introduced indoors by outdoor ventilation, emissions from building materials, and the use of various cleaning products (Nazaroff and Weschler, 2004; Singer et al., 2006). In indoor environments, these VOCs can react with oxidants such as ozone (O3) and/or hydroxyl radicals (OH•) in the gas phase or on indoor surfaces and can transform into a variety of intermediate and stable oxygenated organics (e.g. peroxyl radicals, aldehydes, ketones, di- and tricarbonyls, and carboxylic acids). Peroxyl radicals may further react with NO or NO2 to generate organic nitrates (e.g alkyl nitrates, peroxyacyl nitrates (PANs), hydroxynitrates, and dinitrates) (Finlayson-Pitts and Pitts, 2000). Indoor concentrations of O3, NO, and NO2 in the US have been measured with average values of 50, 50, and 25 ppb, respectively (Nazaroff and Cass, 1986; Weschler and Shields, 1997; Weschler et al., 1994). Although, hydroxyl radical concentrations have not been measured indoors, they have been estimated to be in the range of 0.12–2 × 106 mol cm−3 (0.48–8 × 10−5 ppb) (Alvarez et al., 2013; Sarwar et al., 2002; Waring and Wells, 2015).
Given these measured oxidant concentrations indoors and the reactivity of specific VOCs (e.g. terpenes such as α-pinene, limonene, terpinolene), it is expected that oxidation products are formed and lead to potential indoor exposures. As an example, the bimolecular rate constant for terpinolene + O3 is 19.0 × 10−16 cm3 molecule−1 s−1 (0.169 ppb−1 h−1) (Atkinson and Arey, 2003; Nazaroff and Weschler, 2004). Assuming an indoor O3 concentration of 50 ppb, the pseudo-first order rate for terpinolene ozonolysis would be 8.45 hr−1 indicating terpinolene would likely be removed by reaction with O3 before removal by a typical air-exchange of 0.6 hr−1 (Wilson et al., 1996). Therefore, identifying reaction products from terpene ozonolysis that occurs indoors is critical to characterizing occupant exposures.
Limonene (1-methyl-4-(prop-1-en-2-yl)cyclohexene), is a prevalent terpene with a strong orange-like fragrance found in a number of household consumer products used indoors. The National Library of Medicine’s (NLM) Household Products Database (HHS/NIH, 2015) lists 166 consumer products that contain D-limonene as an ingredient. A significant fraction (59 of 166) of these products are used inside the home (e.g., in cleaning agents) which frequently use D-limonene as an odorant and for its antimicrobial properties. Recent work by Singer et al. determined the one hour concentration of limonene after the application of a full strength cleaning product to be 300–6000 μg/m3 (~80–1600 ppb) (Singer et al., 2006).
The ozonolysis of limonene has been extensively studied using a variety of analytical techniques. However, most of this research has focused on the characterization of secondary organic aerosols (SOAs) from the formation of gas-phase species (Donahue et al., 2014; Ebben et al., 2012; Jiang et al., 2012; Pan et al., 2009; Pathak et al., 2012b; Youssefi and Waring, 2014). This research has provided information about the particle size distribution, aerosol yields and chemical composition, but only limited information of the gas-phase yields from limonene ozonolysis has been determined. Questions still remain on the carbon mass balance of limonene oxidation. The answers may be related to undetected highly oxygenated products (e.g. tricarbonyls). Reaction models (e.g. Master Chemical Mechanism) propose the formation of tri-carbonyl species from limonene ozonolysis (Carslaw, 2013; Jenkin et al., 2015; Norgaard et al., 2013; Pathak et al., 2012a). Recently, the tricarbonyl (3-acetyl-6-oxoheptanal (3A6O)) from limonene ozonolysis was detected using the new derivatization agent, TBOX (Wells and Ham, 2014).
In this study, limonene ozonolysis with and without addition of nitric oxide (NO) and cyclohexane (OH• scavenger) was investigated using a Teflon® impinger to capture and characterize gas-phase reaction products. Identification and quantification of the reaction products (i.e., aldehydes, ketones, and di- and tri-carbonyls) was made using O-tert-butylhydroxylamine hydrochloride (TBOX) to derivatize the carbonyl products (Wells and Ham, 2014). This method provides the sensitivity, ease of use, and applicability needed for detection of carbonyl compounds at expected indoor air concentrations.
2. Experimental methods2.1. Chemicals and solventsAll compounds were used as received and had the following purities: from Sigma-Aldrich/Fluka (St. Louis, MO): O-tert-butyl-hydroxylamine hydrochloride (TBOX, 99%), limonene (97%), toluene (HPLC grade, 99+%), cyclohexane (HPLC grade, 99+%), cyclohexanone (98%), methylglyoxal (40 wt% in water), and glutaraldehyde (50 wt% in water). Methanol (HPLC grade, 99+%) was purchased from Fisher Scientific (Pittsburgh, PA). Water (DI H2O) was distilled, deionized to a resistivity of 18 MΩ cm, and filtered using a Milli-Q® filter system (Billerica, MA). Helium (UHP grade), the carrier gas, was supplied by Butler Gas (McKees Rocks, PA) and used as received. Experiments were carried out at (297 ± 3) K and 1 atm pressure. Compressed air from the National Institute for Occupational Safety and Health (NIOSH) facility was passed through anhydrous CaSO4 (Drierite, Xenia, OH) and molecular sieves (Drierite) to remove both moisture and organic contaminants. This treated dry air from the NIOSH facility flowed through a mass flow controller and into a humidifying chamber and was subsequently mixed with dry air to the pre-determined relative humidity (RH) of 50%. An 80 L Teflon® reaction chamber contained in a light-tight wooden box was filled through a heated syringe injection port facilitating the introduction of liquid reactants into the chamber. Background measurements of the NIOSH facility air showed concentrations of O3, NO, and NO2 at less than 1.0, 1.2, and 0.5 ppb, respectively. All reactant mixtures were generated by this system.
Ozone was produced by photolyzing air with a mercury pen lamp (Jelight, Irvine, CA) in a separate Teflon® chamber. Aliquots of this O3/air mixture were added to the Teflon® reaction chamber using a gas-tight syringe. O3 concentrations were measured using a Thermo Electron (Waltham, MA) UV photometric ozone analyzer Model 49C. Aliquots of NO were added to the reaction chamber from a 100 ppm tank (Butler Gas, McKees Rocks, PA) using a gas-tight syringe. NO and NO2 concentrations were measured using a Thermo Electron (Waltham, MA) NOx analyzer Model 49i.
2.2. Methods2.2.1. CalibrationExperiments to measure the gas-phase carbonyls (cyclohexanone, and glutaraldehyde, Table 1) used for calibration of the gas-phase reaction products formed from the reaction of limonene with O3 were conducted with a previously described apparatus (Ham et al., 2015; Wells and Ham, 2014). A brief description is provided here. Reactants were introduced and samples were withdrawn through a 6.4-mm Swagelok (Solon, OH) fitting attached to an 80 L Teflon®-film chamber. The chamber was filled with 50% relative humidity (RH) air (described above).
Calibration plots were made by analyzing triplicate measurements of standard solutions that were injected into an 80 L Teflon® chamber at 50% RH, ranging in concentration from 5 to 30 ppb (1.2–7.4 × 1011 mol cm−3). Samples were obtained by pulling 60 L of air from the chamber using a pump (URG 3000-02Q, Chapel Hill, NC) into 25 mL of DI H2O in a 60 mL Teflon® impinger (Savillex, Eden Prairie, MN). After collection, samples were decanted into 40 mL vials, then derivatized with 100 μL aqueous 250 mM TBOX, and placed in a heated water bath at 70 °C for 2 h. After removing the vial from the water bath and allowing to cool to room temperature, 0.5 mL of toluene was added to the vial. The vial was then shaken for 30 s and allowed to separate into a toluene layer and aqueous layer. Then 100 μL of the toluene layer was then removed with a pipette and placed in a 2 mL autosampler vial with a 100 μL glass insert (Resetk, Bellefonte, PA). Then 1 μL of the extract was analyzed by gas chromatography-mass spectrometry (GC/MS) (conditions described below).
2.2.2. Limonene + O<sub>3</sub> reactionsIn an 80 L volume of air at 50% RH, O3 (20–100 ppb; 0.5–2.5 × 1012 molecule cm−3) was added to 1.7 ppm limonene (4.25 × 1013 molecule cm−3), and allowed to react for 30 min. After the reaction, 60 L of sample was collected into 25 mL of deionized water using an impinger, TBOX derivatized, extracted, and analyzed (as described above). Additional experiments included the addition of cyclohexane (CH) to the reaction mixture to scavenge OH• formed from Criegee intermediates of limonene ozonolysis (Aschmann et al., 2002; Carslaw, 2013; Criegee, 1975; Forester and Wells, 2009). NO’s effect on limonene ozonolysis with and without OH• was also investigated. The results from each of these experiments are described below. Each experiment was done in duplicate.
All samples were analyzed using a Varian (Palo Alto, CA) 3800/Saturn 2000 GC/MS system operated in the electron impact (EI) mode. Compound separation was achieved by an Agilent (Santa Clara, CA) HP-5MS (0.25 mm I.D., 30 m long, 0.25 μm film thickness) column and the following GC oven parameters: 40 °C for 2 min, then 5 °C min−1 to 200 °C, then 25 °C min−1 to 280 °C and held for 5 min. One μL of each sample was injected in the splitless mode, and the GC injector was returned to split mode 1 min after sample injection, with the following injector temperature parameters: 130 °C for 2 min then 200 °C min−1 to 300 °C and held for 10 min. The Saturn 2000 ion trap mass spectrometer was tuned using perfluorotributylamine (FC-43). Full-scan EI ionization spectra were collected from m/z 40–650.
4. DiscussionAs stated earlier, the ozonolysis of limonene has been extensively studied using a variety of analytical techniques (Hakola et al., 1994; Larsen et al., 2001; Leungsakul et al., 2005; Wells and Ham, 2014). Ozone can react with limonene via addition to either the endocyclic or exocyclic carbon-carbon double bonds with calculated rate constants (AOPWIN v.1.92a) of 43 and 1.2 × 10−17 cm3 molecule −1 s−1, respectively (EPA, 2000). These numbers suggest that the endocyclic O3 addition is favored by about 35 to 1 over the exocyclic O3 addition.
When O3 adds to either of these double bonds, a primary ozonide is formed which subsequently decomposes to a Criegee intermediate (Aschmann et al., 2002; Atkinson et al., 1992; Criegee, 1975; Forester and Wells, 2011). Further unimolecular decay of the Criegee intermediate may then lead to formation of OH• which can further react with excess limonene or newly formed reaction products. The OH• yield from ozonolysis of limonene has been reported to be 64–80% (Aschmann et al., 2002; Forester and Wells, 2009; Herrmann et al., 2010). The OH• can add to double bonds (similarly to O3) and/or abstract available hydrogens leading to product formation. The calculated OH• rate constants for reaction with limonene from the endocyclic double bond and the exocyclic double bond obtained from AOPWIN are (in units of 10−12 cm3 molecule −1 s−1) 86.9 and 51.4, respectively. To determine which oxidation products were formed via ozonolysis alone and/or a combination of OH• and O3, the hydroxyl radical scavenger (cyclohexane) was added to the system. The effect of OH• on the formation of each product is described below.
4.1. Limonene + O<sub>3</sub> and OH• reaction products and yieldsPrevious investigations of limonene ozonolysis have reported two of the reaction products (LimaKet, IPOH) shown in the current investigation (Clausen et al., 2001; Forester and Wells, 2011; Hakola et al., 1994). Moreover, use of TBOX as the derivatization agent has afforded the observance of three additional oxidation products that have been predicted through models such as the MCM. A proposed reaction mechanism based on the data is described below.
LimaKet, 7H6O, and 2A5O are primary oxidation products formed only from the limonene + O3 reaction, and not from secondary OH• formation as observed in Fig. 3 and Table 3. When OH• are scavenged, no statistically significant effects (7, 10, and 15% change, respectively) based on molar yields were observed, indicating that OH• is not influential in their formation. However, the formations of IPOH and 3A6O were strongly dependent on the presence of OH•. When OH• was scavenged, the molar yields decreased by 68% and >95%, respectively. This data further emphasizes the importance of OH• chemistry and how it strongly influences indoor reactions.
The mechanisms for the formation of LimaKet, IPOH, and 3A6O from limonene ozonolysis have been previously discussed (Carslaw, 2013; Lee et al., 2006; Wells and Ham, 2014). A mechanism for their formation can be found in the Supplementary Information, Figure S4. The ozonolysis product, 7H6O, is likely formed via ozone addition to the endocyclic carbon-carbon double bond, formation of a primary ozonide, and then subsequent cleavage to form the radical LIMOOA, as seen in the MCM. Subsequent decomposition of LIMOOA leads to the dicarbonyl peroxyl radical (LIMALBO2) which reacts with an alkoxyl radical to form the stable oxidation product, 7H6O.
The ozonolysis product, 2A5O, can be formed through multiple mechanisms as detailed in the MCM v.3.3.1. One mechanism (Figure S4) may involve initial ozone addition to the exocyclic double bond to form LimaKet, as previously described (Hakola et al., 1994; Wells and Ham, 2014). Subsequent, ozone addition to the endocyclic double bond of LimaKet, followed by cleavage of the primary ozonide, leads to the radical CH3C(= O)CH2CH2CH(C(= O) CH3)CH2C·(OO•)H. The reaction then likely proceeds via hydrogen abstraction of the adjacent CH2 and decomposition to form CH3C(= O)CH2CH2CH(C(= O)CH3)CH2• and CH2·OOH. The larger radical adds O2 and stabilizes to form the tricarbonyl, 2A5O. Although this proposed mechanism does explain the formation of 2A5O through plausible steps, secondary addition of ozone to LimaKet seems unlikely due to the low concentration of ozone used in these experiments. Alternatively, it is possible that the reaction proceeds through initial O3 addition to the endocyclic bond of limonene to form an intermediate similar to IPOH. This intermediate then reacts with another O3 molecule to form the tricarbonyl. While the mechanism for 2A5O is not clear, the observation of this product highlights the formation of tricarbonyl species.
4.2. Nitric oxide (NO) effectWhen NO is added to the limonene + O3 reaction system, there are several pathways for NO to affect the reaction system. NO can react with O3 leading to NO2 with a rate constant of kO3+NO = 2 × 10−14 cm3 molecule−1 s−1 (Wallace et al., 1980). NO can also react with peroxyl radicals (RO2•) generated from OH• and/or O3 reactions to form stabilized carbonyls and/or organic nitrates (Finlayson-Pitts and Pitts, 2000). These reactions are typically fast with a rate constant of kRO2• +NO = 7 × 10−12 cm3 molecule−1 s−1 (Finlayson-Pitts and Pitts, 2000).
Addition of NO (20, 50, and 100 ppb) to the limonene + O3 (50 ppb) system had no statistical effect in the formation of LimaKet or IPOH, Fig. 3. However, when NO is held constant (20 ppb) and O3 was increased (30, 50, and 100 ppb), the molar yields of both LimaKet and IPOH decreased by 31.6% and 31.2%, respectively (Table 3). As O3 was increased, the ratio of RO2•/NO also increased which affected the rate of the RO2• + NO reaction leading to the formation of organic nitrates. The addition of NO had a more significant effect on 7H6O, 2A5O, and 3A6O. A decrease in their formation was observed with the addition of NO (20, 50, and 100 ppb) to the limonene + O3 (50 ppb) system, Fig. 3. Furthermore, when NO is held constant (20 ppb) and O3 was increased (30, 50, and 100 ppb), the molar yields of 7H6O, 2A5O, and 3A6O decreased by 61.9%, 63.3%, and 46.7%, respectively, Table 3. This suggests that peroxyl radical intermediates likely react with NO to form organic nitrates.
Interestingly, 7H6O and 2A5O (Fig. 3) increase when cyclohexane is added in the presence of NO. This may be explained by the side reaction of NO with cyclohexylperoxyl (CHRO2•) radicals. When OH• is scavenged by cyclohexane, CH• can be formed, which then reacts in a diffusion-controlled manner with O2 to form CHO2• (Finlayson-Pitts and Pitts, 2000; Neuenschwander et al., 2010). These CHO2• effectively out-compete RO2• that form 7H6O and 2A5O for reaction with NO, resulting in higher yields of these two products.