Technical note: Preparation and purification of atmospherically relevant α-hydroxynitrate esters of monoterpenes

Organic nitrate esters are key products of terpene oxidation in the atmosphere. We report here the preparation and purification of nine nitrate esters derived from (+)-3carene, limonene, α-pinene, β-pinene and perillic alcohol. The availability of these compounds will enable detailed investigations into the structure–reactivity relationships of aerosol formation and processing and will allow individual investigations into aqueous-phase reactions of organic nitrate esters.


Introduction
Biogenic volatile organic compound (BVOC) emissions account for ∼ 88 % of non-methane VOC emissions. Of the total BVOC estimated by the Model of Emission of Gases and Aerosols from Nature version 2.1 (MEGAN2.1), isoprene is estimated to comprise half, and methanol, ethanol, acetaldehyde, acetone, α-pinene, β-pinene, limonene, ethene and propene together encompass another 30 %. Of the terpenoids, α-pinene alone is estimated to generate ∼ 66 Tg yr −1 (Guenther et al., 2012). These monoterpenes can be oxidized by nitrate radicals that are projected to account for more than half of the monoterpene-derived secondary organic aerosol (SOA) in the US (Pye et al., 2010). Nitrate oxidation pathways have been shown to be important particularly during nighttime. A large portion (30 %-40 %) of monoterpene emissions occur at night (Pye et al., 2010). These emissions can then react with NO 3 radicals, formed from the oxidation of NO 2 emissions by O 3 (Pye et al., 2010).
The full role of organic nitrates (ONs) is complicated with many different sources and sinks (Perring et al., 2013). Fully deconvoluting the atmospheric processing of terpene- derived ON is difficult, particularly due to partitioning into the aerosol phase in which hydrolysis and other reactivity can occur (Bleier and Elrod, 2013;Rindelaub et al., 2014Rindelaub et al., , 2015Romonosky et al., 2015;Thomas et al., 2016). Hydrolysis reactions of nitrate esters of isoprene have been studied directly (Jacobs et al., 2014) and the hydrolysis of ON has been studied in bulk (Baker and Easty, 1950). These and other studies have shown that the hydrolysis of ON is dependent on structure . For example, primary and secondary ON are thought to be relatively stable (Hu et al., 2011). In contrast, tertiary nitrates have been shown to hydrolyze on the order of hours (Boyd et al., 2015;Liu et al., 2012) to minutes . To the best of our knowledge there is only one study of the hydrolysis of an isolated terpene-derived hydroxynitrate (2 in Fig. 1) (Rindelaub et al., 2016a).
Furthermore, fully understanding the atmospheric processes of organic molecules is restricted by the ability to identify these species (Nozière et al., 2015). Part of this challenge is, of course, related to the lack of available standards. While one certainly cannot synthesize all of the atmospherically relevant ON, having access to representative Published by Copernicus Publications on behalf of the European Geosciences Union. 4242 E. A. McKnight et al.: Preparation and purification of atmospherically relevant nitrate esters compounds from monoterpenes would enable key studies. With these molecules in hand, the atmospheric chemistry community could directly study the ON reactivity, such as hydrolysis, and deconvolute the structure-reactivity relationships. Additionally, novel method development would be enabled and validated (Rindelaub et al., 2016b). For example, Nozière and co-authors called attention to "the lack of NMR spectra libraries for atmospheric markers" as a barrier to utilizing nuclear magnetic resonance (NMR) spectroscopy in atmospheric science. While the preparation of α-hydroxynitrates of terpenes has been alluded to in a handful of reports, we have only been able to identify NMR data for two species (Fig. 1). Two main methods appear to have been attempted, treating an epoxide with either fuming nitric acid (Rollins et al., 2010) or bismuth nitrate (Rindelaub et al., 2016b;Romonosky et al., 2015). The utility of the former is limited by the extreme hazards involved with mixing fuming nitric acid with organic materials (Parker, 1995;Univ. of California, Berkeley, 2019) and only provides characterization data for a single compound. On the surface the latter appears to be a usable method but on closer inspection lacks spectral data, perhaps due to intractable and inseparable mixtures (Romonosky et al., 2015), and has been the subject of a retracted study (Pöschl et al., 2011).

Perillic alcohol epoxide (15)
Perillic alcohol (5 mL, 31.46 mmol) was combined with dichloromethane (62 mL) in a round bottom flask with stir bar and cooled to 0 • C under inert atmosphere. m-CPBA (5.97 g, 34.61 mmol) was added in five portions over the course of 10 min. After 30 min the reaction was warmed back up to room temperature and allowed to stir for an hour at room temperature. The reaction mixture was filtered through celite, washed with DCM (3 × 30 mL) and then transferred to a separatory funnel where it was washed with sodium bicarbonate (50 mL) and brine (50 mL). The organic layer was dried (Na 2 SO 4 ) and concentrated to yield a crude oil (5.1 g). Purification by column chromatography (80 g spherical SiO 2 ; 0 %-100 % EtOAc / Hex over 15 CVs) yielded the epoxide (15)

β-Pinene oxide
Potassium peroxide monosulfate (4.750 g, 31.20 mmol) was dissolved in deionized water (60 mL). Sodium bicarbonate (4.019 g, 47.84 mmol) was placed into a 125 mL Erlenmeyer flask and acetone (40 mL) was added, followed by β-pinene (1.59 mL, 9.974 mmol). The solution of potassium peroxide monosulfate mixture was slowly added into the βpinene mixture over the course of 3 min via syringe while the mixture was stirring at 800 rpm. The reaction mixture was stirred for exactly 30 min while stirring at 800 rpm at room temperature. The reaction mixture was transferred to a separatory funnel and extracted with dichloromethane (2 × 40 mL), dried (MgSO 4 ) and concentrated to produce a clear oil (1500 g

General nitration method
A round bottom flask was charged with a solution of epoxide (2 mmol) in toluene, dioxane or dichloromethane (10 mL, 0.2 M). Bismuth nitrate (1.164 g, 2.4 mmol, 1.2 equiv.) was added to the reaction mixture. The reaction mixture was stirred for 30-60 min. When TLC indicated complete consumption of starting material, the reaction was filtered through a 1 in. (2.5 cm) celite pad and washed with DCM (2 × 15 mL). The filtrate was transferred to a separatory funnel and washed with sodium bicarbonate (3 × 15 mL). The organic layer was dried with sodium sulfate, filtered and concentrated to yield a crude liquid. Desired products were isolated via column chromatography.

Nitration of α-pinene oxide
Nitration of α-pinene oxide was carried out according to the general method. Crude oil from a 10 mmol reaction was purified by column chromatography (40 g SiO 2 , 0 %-40 % EtOAc / Hex over 30 CVs) yielded the following isomerization products and nitrate esters.

Synthesis
Previous reports have described the opening of epoxides and aziridines using bismuth nitrate in acetonitrile (Das et al., 2007). Others report optimal conditions in dichloromethane (Rindelaub et al., 2016b), or 1,4-dioxane with undesired side reactions in acetonitrile (Pinto et al., 2007). Thus, we began our investigations into the nitration of trans-carene oxide (3) using bismuth nitrate in various solvents (Fig. 2). In our hands, dioxane clearly outperformed dichloromethane, yielding 53 % of the desired product after 45 min (entry 3). A diol was isolated as a major (16 %) byproduct. Attempts to mitigate hydrolysis by the use of base or molecular sieves were ineffective. Acetonitrile produced a complex intractable mixture with a high amount of diol and caronaldehyde (entry 2). Other solvents, such as tetrahydrofuran (THF) and nitromethane, also produced complex mixtures with no trace of desired nitrate. Methanol, interestingly, only produced an undesired methyl ether. It is unclear if this product was generated through methanolysis of the nitrate ester or direct substitution of the epoxide. The regiochemistry of the nitration was easily elucidated with gHSQC NMR data. A doublet of doublets at 3.69 ppm correlated with a carbon shift at 71.1 ppm and was thus assigned to the alcohol methyne carbon. A tetrasubstituted carbon, with no correlations in the gHSQC at 95.6 ppm was consistent with the tertiary nitrate ester. Finally, the IR of compound (4) displayed the expected strong absorbances at 1616, 1291 and 868 cm −1 .
Next, we aimed to investigate the impact of relative stereochemistry on the nitration. We hypothesized that the secondary nitrate would predominate due to classical stereoelectronic effects. (+)-3-Carene was treated with Nbromosuccinimide and calcium carbonate to produce a bromohydrin which subsequently cyclized under basic conditions to produce cis-carene oxide (Cocker and Grayson, 1969). Interestingly, we never observed the desired nitrate ester. Instead a 1,2-hydrogen shift generated cis-4-caranone (7) in 53 % yield.
The preparation of 8,9-limonene oxide (9) began with a Diels-Alder reaction to form 1-(4-methyl-3-cyclohexene) ethenone (8) in 88 % yield. Corey-Chaykovsky addition of a sulfur ylide cleanly produced 8,9-limonene oxide (9). Nitration of (9) was attempted in a variety of solvents (DCM, benzene, dioxane, acetonitrile, nitromethane). In acetonitrile, oxazoline (10a) (1 : 1 d.r.) was cleanly produced presumably via nucleophilic addition of acetonitrile and subsequent cyclization to the oxazoline. Pinto et al. (2007) observed a similar bismuth nitrate mediated epoxide opening with acetonitrile. In dioxane, aldehyde (11) (1 : 1 d.r.) was observed along with a product consistent with a tertiary nitrate ester (10) (1 : 1 d.r.) in a 0.25 : 1 mixture. Formation of 11-enol presumably proceeds via a facile intramolecular elimination of nitric acid. The IR of the crude mixture showed the expected strong absorbances at 1613, 1288 and 865 cm −1 . A set of signals in the 13 C NMR at 97.40 and 97.17 ppm, with no gHSQC correlations, is consistent with a tertiary nitrate. Attempts to purify this mixture on silica gel gave small amounts of (11) and unidentifiable decomposition products. Furthermore, the fast hydrolysis rates of tertiary nitrate esters hinder the ability to isolate these products (Boyd et al., 2015;Darer et al., 2011;Liu et al., 2012). No reaction was observed when non-polar solvents such as DCM and benzene were employed. We hypothesize that the steric hindrance at the β-carbon severely limits reactivity.
Commercially available 1,2-limonene oxide was resolved into pure samples of cisand trans-1,2-limonene oxide by treating with cyclic amine bases (Steiner et al., 2002). The nitration of the cis isomer was predicted to produce  the tertiary nitrate ester via stereoelectronic effects. The dominant conformation of cis-1,2-limonene oxide is expected to be a pseudo-half chair (Fig. 5). Nucleophilic substitution at the less substituted position would proceed through a lower-energy chair-like transition state, whereas attack at the more substituted carbon leads to a very unstable twist-boat. The nitration of cis-1,2-limonene oxide (cis-12) proceeded smoothly in both dioxane (53 %) and dichloromethane (63 %). The trans isomer was expected to produce the secondary nitrate ester. The nitration of trans-1,2-limonene oxide (trans-12) also proceeded smoothly in both dioxane (63 %) and dichloromethane (61 %). Contrary to previous reports (Romonosky et al., 2015), we did not observe the desired nitrate esters in acetonitrile. gHSQC NMR data were used to verify nitrate ester (13) was tertiary and (14) was secondary. For (13), a broad singlet at 4.12 ppm correlated with a carbon shift at 69.2 ppm and was thus assigned to the methyne adjacent to the alcohol. A tetrasubstituted carbon at 91.4 ppm was consistent with the tetrasubstituted nitrate ester. Similarly for (14), a broad triplet at 5.00 ppm correlated with a carbon shift at 84.45 ppm and was thus assigned to the methyne adjacent to the nitrate ester. A tetrasub-stituted carbon at 69.5 ppm was consistent with the tertiary alcohol.
Under acidic conditions, β-pinene oxide undergoes facile rearrangement to form perillic alcohol. Accordingly, we prepared the epoxide (15) by treating perillic alcohol with one equivalent of m-CPBA. The inseparable mixture of diastereomers was treated with bismuth nitrate under the standard reaction conditions. Nitrate esters (16) and (17) were easily separated by column chromatography and assigned to the secondary and tertiary nitrate esters, respectively. These isomers were formed from the cisand trans-epoxide diastereomers in an analogous fashion to the limonene isomers. For (16), a broad singlet at 5.21 ppm correlated with a carbon shift at 80.1 ppm and was thus assigned to the methyne adjacent to the nitrate ester. A tetrasubstituted carbon at 70.9 ppm was consistent with the tertiary alcohol. Similarly for (15), a broad singlet at 4.22 ppm correlated with a carbon shift at 65.6 ppm and was thus assigned to the methyne adjacent to the alcohol. A tetrasubstituted carbon at 92.8 ppm was consistent with the tertiary nitrate ester.
The facile rearrangements of α-pinene with both Brønsted and Lewis acids (Kaminska et al., 1992) can be thought to proceed via a nonclassical isobornyl cation (18) (Kong et al., 2010). We expected the reaction with bismuth nitrate to generate a complex mixture of products due to the many reactive sites. As shown in Fig. 7, three rearrangement products (campholenic aldehyde 19, diene 20 and transcarveol 21) and three nitrate esters (22,23,24) were observed. The structure of nitrate ester (23) was identified by a clear singlet at 4.64 ppm (s, 1 H). Two-dimensional-NMR data (gCOSY, gHSQC and gHMBC) along with comparison to 6-exo-hydroxyfenchol, the analogous diol, confirmed the structural assignment (Miyazawa and Miyamoto, 2004). Correspondingly, nitrate ester (24) displayed a distinct doublet of doublets of doublets at 5.31 ppm that correlated to the methyne adjacent to the nitrate ester. Again, 2-D-NMR data (gCOSY, gHSQC and gHMBC) were used in conjunction with the literature spectra for platydiol and its trans diastereomer to verify the assignment (Kuo et al., 1989).   Under all conditions campholenic aldehyde was the major product (20 %-28 % yield). First generation nitrate ester (25) was not found under all conditions. All six components were isolated in small amounts when the reaction was run in dichloromethane (Fig. 7, entry 1). Cooling the reaction to −78 • C completely shut down all reactivity (entry 2). Interestingly, adding one equivalent of tetrabutylammonium nitrate (TBAN) as an external nitrate source in addition to bismuth nitrate at −78 • C generated mostly nitrate (22) (entry 3). More polar solvents, like dioxane (entry 4), generated slightly higher amounts of nitrates at room temperature. Aromatic solvents, such as benzene and toluene, provided the best yield of (23). Interestingly, adding TBAN to the nitration in toluene dramatically increased the amounts of diene (20) and trans-carveol (21). Using acetonitrile as a solvent gen-erated a complex mixture that appeared to be mainly diols. Zirconium nitrate produced a similar distribution of products (Das et al., 2006), but other metal nitrate complexes (Y(NO 3 ) 3 q 6H 2 O, Co(NO 3 ) 2 q 6H 2 O) resulted in no observable reaction.
The nitration of β-pinene was similarly complicated by isomerization pathways through the corresponding nonclassical carbocation (Fig. 8). Nitrate (27) initially co-eluted with myrtenol (26) but was separable upon a second purification by column chromatography. The methyne proton of (27) at 5.51 ppm is correlated with a carbon at 85.3 ppm in the gHSQC NMR. These data are consistent with the secondary nitrate ester. Similarly, the methyne proton in nitrate (29) is a doublet at 4.91 ppm (J 4 = 1.6 Hz; W coupling). This signal correlates with a carbon at 89.5 ppm. Interestingly, nitrates (30) and (31) were not observed.

Spectral data
As shown in Table 1, all nitrate esters displayed the expected strong nitrate ester absorbances at ∼ 1600, 1300 and 900 cm −1 . Methyne protons next to the secondary nitrate esters appeared between 4.64 and 5.51 ppm. Carbon chemical shifts were found over a broader range than anticipated from 80.1 to 95.6 ppm. Finally, we observed masses consistent with methanolysis products (185.1 m/z M+;  223.1 m/z M + K) when nitrate esters were analyzed (GC-MS) as solutions in methanol. Further experimentation is necessary to evaluate the implications of methanolysis.

Stability and storage
We were surprised to find that five of the nitrate esters (4, 16, 23, 24 and 29) are solids. We have stored these compounds at 0 • C for up to 9 months with no noticeable decline in purity. The remaining nitrate esters are stable at 0 • C for 2-4 weeks. Interestingly, some of these compounds appear to deviate from the expected stability patterns (vide supra). For example, tertiary nitrate ester (4) is particularly stable stored as a solid at 0 • C. In contrast, when secondary nitrate ester (14) was stored as a neat oil at 0 • C, it decomposed to a complex mixture within a few weeks. Freezing the sam-ples as a solution in benzene is recommended for longer-term (∼ 6 months) storage.

Conclusions
We have clearly delineated successful methods required to synthesize and purify nine nitrate esters derived from monoterpenes. Seven of these compounds are undescribed in the literature and the remaining two had gaps in their characterization. Using our methods it is possible to cleanly isolate 50-100 mg even of the least prevalent isomers from the nitration reactions of α-pinene and β-pinene. Interestingly, we did not observe the formation of α-pinene oxide and β-pinene oxide products that retained their bicyclic ring structures. This is consistent with the solution-phase synthetic literature (Kaminska et al., 1992) but in contrast to many atmospheric reports (for example, see Rindelaub et al., 2016b;Duporte et al., 2016). We believe that the availability of these compounds will enable further study of the structure-reactivity relationships. For example, comparing the specific hydrolysis rates for tertiary versus secondary nitrate esters in (13) and (14) as well as (17) and (16) could help deconvolute the fates of each terpene in the atmosphere. We also believe that these compounds will assist in confirming the identities of organic nitrates that have previously been limited to detection by MS-based methods (Rindelaub et al., 2016b). A forthcoming report will describe our investigation into the behavior of these authentic compounds in mass spectrometers. In par- ticular, the detailed MS-MS data, impact of various ionization conditions (ESI, chemical ionization (CI), etc.) and sample preparation will be described. Finally, the availability of these compounds is important for further studies into the influence of terpene structure on the fate and roles of organic nitrates in SOA formation.
Data availability. All spectral data for new compounds are available in the Supplement to this article.