Radical Generation from the Gas-Phase Activation of Ionized Lipid Ozonides
Reaction products from the ozonolysis of unsaturated lipids at gas–liquid interfaces have the potential to significantly influence the chemical and physical properties of organic aerosols in the atmosphere. In this study, the gas-phase dissociation behavior of lipid secondary ozonides is investigated using ion-trap mass spectrometry. Secondary ozonides were formed by reaction between a thin film of unsaturated lipids (fatty acid methyl esters or phospholipids) with ozone before being transferred to the gas phase as [M + Na]+ ions by electrospray ionization. Activation of the ionized ozonides was performed by either energetic collisions with helium buffer-gas or laser photolysis, with both processes yielding similar product distributions. Products arising from the decomposition of the ozonides were characterized by their mass-to-charge ratio and subsequent ion-molecule reactions. Product assignments were rationalized as arising from initial homolysis of the ozonide oxygen–oxygen bond with subsequent decomposition of the nascent biradical intermediate. In addition to classic aldehyde and carbonyl oxide-type fragments, carbon-centered radicals were identified with a number of decomposition pathways that indicated facile unimolecular radical migration. These findings reveal that photoactivation of secondary ozonides formed by the reaction of aerosol-bound lipids with tropospheric ozone may initiate radical-mediated chemistry within the particle resulting in surface modification.
KeywordsOzonolysis Secondary ozonide Electrospray ionization Free radical Collision-induced dissociation Photo dissociation Aerosols
Large amounts (Tg year–1) of short-chain alkenes, in the form of volatile organic compounds, are emitted into the atmosphere every year from a variety of biogenic and anthropogenic sources [1, 2, 3]. Less volatile long-chain molecules such as fatty acids and other lipids are also emitted into the atmosphere through a variety of natural and anthropogenic processes and are well known to be components of atmospheric aerosols [2, 3, 4, 5, 6]. Lipids are important components of all cell membranes and thus form a significant part of the organic layer on the surface of sea water following the decomposition of marine organisms [7, 8, 9, 10, 11, 12]. Wave action on the ocean surface leads to the formation of sea salt based aerosols onto which organic material from the surface layer—including lipids derived from marine life—become bound. Once formed, marine aerosols are exposed to oxidizing species in the atmosphere. The oxidation of organic matter on aerosol particles will alter their hygroscopic properties and thus impact mechanisms of cloud condensation [9, 10, 13, 14].
In the gas-phase, however, the aldehyde and carbonyl oxide intermediates can become rapidly separated, allowing for comparatively little recombination. Furthermore, in the absence of collisional partners, relaxation of energetically excited carbonyl oxide intermediates is slow, thus allowing for unimolecular rearrangement or decomposition . The formation of stable secondary ozonides is also partly determined by the carbon chain length of the reacting alkene. For example, in larger molecules such as lipids, the excess energy can be redistributed amongst many internal degrees of freedom, making formation of a stable secondary ozonide more probable.
While it is accepted that tropospheric ozone reacts with organics adsorbed onto aerosol particles, and that such reactions can lead to the formation of stable secondary ozonides, the fate of these secondary ozonides in the atmosphere is not as well understood [9, 10, 17, 19]. It is possible that activation of secondary ozonides, either thermally or by photolysis, can facilitate their decomposition into a number of lower mass components. Several studies have investigated the thermal decomposition of secondary ozonides in both the gas-phase and solution-phase, and in all cases aldehydes and carboxylic acids are observed as major products [20, 21, 22, 23, 24, 25]. Studies by Hull et al.  and Khachatryan et al.  have provided evidence for a radical mechanism leading to the formation of aldehydes and carboxylic acids. This involves initial homolysis of the oxygen–oxygen bond to produce a biradical followed by rearrangement and dissociation pathways to give an aldehyde and carboxylic acid. The intermediacy of the biradical is also consistent with a variety of other products observed upon ozonide activation. While it is well known that ozonolysis of alkenes produces hydroxyl radicals [26, 27, 28, 29, 30], work by Pryor and co-workers investigating the thermal decomposition of the secondary ozonide of allylbenzene found evidence for the formation of carbon- and oxygen-centered radicals . UV-photolysis of secondary ozonides has also been reported with several studies observing spectroscopic signatures consistent with products of biradical decomposition [17, 24, 31, 32, 33]. While prior studies have found evidence for radical formation following secondary ozonide activation, the identity of the radicals themselves is most often not determined.
Mass spectrometry is a powerful tool for observing and studying the structure of reactive intermediates in the gas-phase. Using this approach, Harrison and Murphy  studied the fragments arising from collision-induced dissociation (CID) of phospholipid secondary ozonides. Along with the characteristic aldehyde and carboxylic acid fragments (or isomers thereof), ions with mass suggestive of radical ions were observed. These were assigned as carbon-centered radicals formed from the biradical by a β-scission mechanism; however, the structures of the radicals were not investigated further. In the present study, we have probed the intrinsic unimolecular decomposition of lipid secondary ozonides analogous to those likely to be formed on marine aerosols. This has been undertaken by electrospray ionization tandem mass spectrometry (MS/MS), where surface-formed secondary ozonides have been gently transferred to the gas-phase for subsequent activation. The molecular structure of radical products formed by activation has been studied using multi-stage tandem mass spectrometry (MSn) and ion-molecule reactions. These data are compared with decomposition products (including radicals) formed by photodissociation (PD) using wavelengths within the actinic window. These investigations support the possibility of unimolecular decomposition of secondary ozonides formed from aerosol lipids as being a source of radicals in the atmosphere.
All phospholipids standards were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Fatty acids were purchased from Nu-Chek Prep (Elysian, MN, USA) with the exception of partially deuterated oleic acid (11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,18-D 17-9Z-octadecenoic acid, 99%), which was purchased from Cayman Chemical (Michigan, MI, USA). Methanol, chloroform, and pentane (all HPLC grade) were purchased from Crown Scientific (Sydney, NSW, Australia). Sodium acetate was purchased from Ajax Chemicals (Sydney, Australia). Boron trifluoride was purchased from Sigma Aldrich (Castle Hill, NSW, Australia). All compounds were used without further purification.
Fatty Acid Methyl Ester Preparation
Approximately 1 mg of each fatty acid standard was dissolved in 1 mL of 10% boron trifluoride in methanol and stirred for 20 min at room temperature. Water (0.5 mL) and pentane (1 mL) were then sequentially added to the solution to separate aqueous and organic components. The organic layer containing the fatty acid methyl ester (FAME) component was then removed from the organic layer at a concentration of approximately 3 mM in pentane.
Ozonide preparation was modeled on the method reported by Harrison and Murphy . First, 20–50 nmol of lipid present as 0.2–3.0 mM solutions in either methanol, 2:1 methanol:chloroform (v/v) (phospholipids), or pentane (fatty acid methyl esters) was deposited into a 10 mL glass vial and the solvent removed under a stream of dry nitrogen. The vial containing the dried lipid film was placed in a Drechsel bottle connected to an ozone generator (HC-30 ozone generator; Ozone Solutions, Sioux Center, IA, USA). Oxygen (backing pressure 140 kPa and a flow of 100–150 mL min–1) was passed through the ozone generator and directed into the glass vial. The power output of the ozone generator was 30 (arbitrary units) producing an estimated 15% w/w ozone in oxygen. Thus produced ozone was passed over the lipid film for 2–5 min, after which time the ozone generator was turned off and the system flushed with oxygen for 10 min. During this procedure, excess ozone exiting the Drechsel bottle was destroyed by bubbling through an aqueous solution of sodium thiosulfate. Sodium iodide and Vitex (a starch-based indicator) were also present in the solution to provide a rapid visual indication if the sodium thiosulfate was exhausted . The vial was then removed and lipids and their oxidation products dissolved by addition of ~1.5–2.0 mL 2:1 methanol:chloroform (v/v) giving a final lipid concentration of 10–25 μM. Methanolic sodium acetate was then added to a final salt concentration of 50 μM. This solution was then analyzed by mass spectrometry.
Electrospray Ionization-Mass Spectrometry
Electrospray Ionization-Mass Spectrometry (ESI-MS) was performed on a Thermo Finnigan LTQ mass spectrometer (now Thermo Fisher Scientific, San Jose, CA, USA) operating Xcalibur 2.0 software. Samples were infused at a flow rate of 3–5 μL min–1 with a spray voltage of +4 kV. The automatic tune function was used to optimize the ion optics for the detection of secondary ozonides of each lipid class. The capillary temperature was typically 200 °C and typical tube lens voltages were +235 V for phospholipid ions and +65 V for fatty acid methyl esters. For CID, ions were isolated with an isolation window of 2–3 Th and a resonant excitation was applied using a normalized collision energy  between 20% and 30% for a period of 30 ms. In some instances, where isolation of individual isotopologues was difficult with a single isolation step, ions were isolated using a nested isolation procedure in which isolation was repeated three times using a sequence of isolation widths of 5, 2, and 3–5 Th.
For ion-molecule reactions involving background oxygen, the radical precursor was first generated by CID of the corresponding ozonide and then isolated in the ion trap without additional collisional activation (with CE = 0). The reaction time with background oxygen was controlled by adjusting the ion activation time to between 500 and 7000 ms at the relevant step of the MSn experiment. Oxygen concentrations within the instrument under normal operating conditions (pressure in the ion trap region of 2.5 mTorr) have previously been determined to be ca. 3 × 109 molecules cm−3 . Each spectrum represents an average of at least 50 scans.
Ozone-induced dissociation (OzID) was performed as previously described . Ozone was generated as described above and collected in a plastic syringe. Ozone was introduced by attaching the plastic syringe to a PEEK-sil tubing restrictor (100 mm L × 1/16′′ o.d. × 0.025 mm i.d.; SGE Analytical Science) connected to the helium supply line via a shut-off ball valve and T-junction. The helium flow rate was controlled using a metering valve. A backing pressure was applied to the syringe (25 μL min–1) using a syringe pump, thereby introducing ozone into the ion trap. Ozonolysis reaction time was controlled by adjusting the ion activation time. For phospholipids, the reaction time was 10 s, following which the ion appearing 48 Th above the mass-selected pseudo-molecular ion was isolated and fragmented via CID. For analysis of secondary ozonide decomposition products, the surface-synthesized ozonide was subjected to CID and the product ion resulting from the 46 Da neutral loss was re-isolated in an MS3 experiment and allowed to undergo OzID with a reaction time of 1 or 10 s for FAMEs and phospholipids, respectively.
Photodissociation (PD) using a Quanta-Ray INDI Nd:YAG pumped optical parametric oscillator (OPO) laser system (Spectra-Physics, Santa Clara, CA, USA) was performed as previously described [39, 40]. Briefly, a quartz viewport was fitted to the back plate of the LTQ chamber to allow transmission of a laser pulse into the vacuum region. Optical access to the ions within the quadrupole ion trap is afforded by the 2 mm orifice centered on the back ion lens. To ensure isolated ions were only activated with a single laser pulse, a mechanical shutter is placed at the exit of the aperture of the OPO and synchronized with the activation sequence of the mass spectrometer using a TTL pulse generated at the beginning of the appropriate MSn activation step. Ions were isolated with a nested isolation sequence of 5, 2, and 10 Da to ensure no fragmentation occurred during application of the isolation waveform. After isolation ions were irradiated using either a 260 or 300 nm laser pulse (~5 ns pulse width at ~1 mJ pulse–1) and spectra averaged over 200 scans.
Results and Discussion
Decomposition of Secondary Ozonides Formed from Fatty Acid Methyl Esters
The product ions observed at m/z 209 and 225 in Figure 1b correspond in mass to the aldehyde and carbonyl oxide, respectively, although in the latter case, rearrangement of the carbonyl oxide to other isomers is likely (cf. Scheme 1). Given that the structure of the m/z 225 ion is ambiguous, we refer to it here (and in similar instances) as the “Criegee ion.” These ions may form via either the biradical mechanism (Scheme 2b) or by cycloreversion of the secondary ozonide (Scheme 2h). Both of these fragments are well known decomposition products of lipid secondary ozonides [20, 34].
The even mass of the m/z 180 product ion observed following CID of the [M + O3 + Na]+ ion in Figure 1b is consistent with a radical species formed by loss of a neutral fragment with the composition, C10H19O3 • . This is likely formed from the initial biradical intermediate following activation of the secondary ozonide through a β-scission mechanism (Scheme 2c). An analogous radical ion has previously been observed following CID of the secondary ozonide formed from the monounsaturated phospholipid 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (PC 16:0/9Z-18:1) . If the nascent radical fragment undergoes further reaction within the ion–dipole complex prior to dissociation, analogous β-scission pathways could also account for a number of the other products observed, including (i) the neutral loss of octane to form m/z 253 (–114 Da) (Scheme 2d); (ii) loss of formic anhydride to form m/z 293 (–74 Da, Scheme 2e), and (iii) loss of dihydrogen to form m/z 365 (–2 Da, Scheme 2f). Analysis of the isomeric vaccenic acid methyl ester (FAME 11Z-18:1) secondary ozonide gave rise to an analogous suite of product ions (m/z 208, 281, 237, and 253) supporting the assignment of these species (Supporting Information, Figure S1).
Decomposition of Secondary Ozonides Formed from Phospholipids
The CID spectrum of the phospholipid [M + O3 + Na]+ ion at m/z 830 is shown in Figure 3b. In this spectrum, the m/z 771 product ion arises from loss of trimethylamine from the phosphocholine head group (–59 Da), a characteristic fragment arising from activation of sodiated PCs [48, 49]. The m/z 769 ion is assigned to the subsequent loss of dihydrogen from the [M + O3 + Na – N(CH3)3]+ ion and likely has an anhydride structure analogous to that assigned for decomposition of the FAME 9Z-18:1 ozonide (cf. Scheme 2f). The product ions at m/z 672 and 688 are assigned as the aldehyde and Criegee ion [18, 34, 38], whereas the m/z 613 and 629 ions correspondingly arise from further loss of trimethylamine from these species. Formic acid loss is also observed from the PC (16:0/9Z-18:1) ozonide resulting in a product ion at m/z 784. Further interrogation of the structure of this ion by OzID revealed an abundant loss of 96 Da (Supporting Information, Figure S5), thereby confirming formic acid loss occurs via a mechanism identical to that described for FAME 9Z-18:1 ozonide (Scheme 4).
The radical ion observed at m/z 559 in Figure 4b can arise from radical migration and carbon–carbon bond cleavage yielding an acetate-like radical analogous to the m/z 96 ion formed from FAME 9Z-18:1 (Figure 2b). Further evidence for this assignment is the observation of a similar reactivity towards molecular oxygen as that observed for m/z 643 (data not shown) and its prior observation as a CID product of the corresponding [M + O3 + H]+ ion . This radical cation can then undergo decomposition via radical migration to produce the fragment ions at m/z 304, 500, and 501 (Supporting Information, Figure S8).
PD of Ozonides
The most abundant PD product ions in the spectra in Figure 5 can be assigned as aldehyde (m/z 672), Criegee (m/z 688), and alkyl radical (m/z 643) ions. Additional product ions observed at lower abundance include m/z 784, 756, and 716, and are assigned to losses of formic acid, formic anhydride, and octane, respectively [24, 31]. These product channels are attributed to decomposition via the biradical intermediate (cf. Scheme 2). Furthermore, lower mass ions at m/z 585 and 559 are detected and assigned to decomposition of alkyl radicals following radical migration (Scheme 5). It is interesting to note that ions at m/z 756 and 716 are also observed in the corresponding CID spectrum (cf. Figure 3b) but at relatively low abundance. Another key comparison is that in the λ = 260 nm PD spectrum (Figure 5a), the abundance of odd-electron products relative to the combined abundance of the even-electron (e.g., aldehyde and Criegee ions) is higher than that observed from CID (Figure 3b). One possible explanation is an electronic excited state process that delivers selectivity for oxygen–oxygen scission (and thus radical-driven dissociation) over competing even-electron dissociation upon photoactivation. Further evidence for this is the absence of any product ions in the PD spectrum not associated with ozonide decomposition (e.g., loss of N(CH3)3). CID presumably occurs via electronic ground state pathways whereas UV photoexcitation accesses electronically excited states. Whether dissociation following UV photolysis occurs directly from excited electronic states or on the electronic ground state following state-crossing processes is not known but an intriguing question for further study.
Comparison of Ozonides Formed on Surfaces Versus the Gas Phase
The ability to observe gas-phase ozonolysis of ionized lipids within an ion trap provides a unique opportunity to directly compare surface-formed secondary ozonides (discussed above) with their gas-phase counterparts. It is generally accepted that secondary ozonides are not formed to a significant extent in the gas phase because of the lack of stabilization of the vibrationally excited carbonyl oxide. This ultimately results in unimolecular decomposition and/or rearrangement before any recombination with the partnering carbonyl compound can occur . Despite this, under certain conditions several studies have detected secondary ozonides from the ozonolysis of small alkenes in the gas phase [52, 53, 54]. However, the gas-phase secondary ozonide formation for larger lipid molecules has not been investigated.
Ion-trap mass spectrometry has provided direct evidence for the products arising from the unimolecular dissociation of lipid secondary ozonides of the type likely present on the surface of both marine and other aerosols. These findings complement recent evidence for the formation of secondary ozonides upon reaction of lipid-based aerosols with ozone [55, 56]. The unequivocal demonstration that carbon-centered radicals are formed upon both collisional- and photo-activation of these lipid ozonides indicates that these processes may need to be considered as pathways for surface modification of aerosols in the troposphere. Interestingly, some of the mechanisms proposed to account for radical-driven dissociation of lipid ozonides posit hydroxyl radical as a co-product. The release of hydroxyl radicals during decomposition of lipid secondary ozonides represents an important avenue for chain propagation of free radical oxidation. Furthermore, many other products of ozonolysis (such as aldehydes and ketones) can open additional pathways to photolysis and free radical production, and may initiate further reactions and surface modifications .
Finally, mechanisms for radical production from activation of lipid secondary ozonides all proceed from initial homolysis of the oxygen–oxygen bond and subsequent dissociation and rearrangement of the resulting biradical intermediate. Recent studies suggest that similar ozonolysis mechanisms in polymers proceed via surface-crossing from the singlet to the triplet biradical surfaces . Future identification of minimum energy crossing points for lipid biradicals may provide insight into the competition between even- and odd-electron processes in lipid ozonide decomposition.
S.R.E. was supported by an Australian Postgraduate Award and H.T.P. was funded by the Australian Research Council (DP0986628) and the University of Wollongong. S.J.B. acknowledges research funding from the ARC Center of Excellence for Free Radical Chemistry and Biotechnology (CE0561607). S.J.B, T.W.M., and A.J.T. acknowledge research funding from the ARC Discovery Program (DP120102922). T.W.M. and M. in het P. are grateful to the Australian Research Council for fellowship support (FT110100249 and FT0990846, respectively). Part of this research has been made possible with the support of the Dutch Province of Limburg. The authors acknowledge Dr. Gabriel DaSilva (University of Melbourne) for helpful discussions and thank Dr. Christopher Hansen and Dr. Benjamin Kirk for their assistance with PD experiments.
- 2.Guenther, A., Hewitt, C.N., Erickson, D., Fall, R., Geron, C., Graedel, T., Harley, P., Klinger, L., Lerdau, M., McKay, W.A., Pierce, T., Scholes, B., Steinbrecher, R., Tallamraju, R., Taylor, J., Zimmerman, P.: A global model of natural volatile organic compound emissions. J. Geophys. Res. 100, 8873–8892 (1995)CrossRefGoogle Scholar
- 3.Finlayson-Pitts, B.J., Pitts Jr., J.N.: Chemistry of the Upper and Lower Atmosphere. Academic Press, San Diego (2000)Google Scholar
- 36.Available at: https://www.thermofisher.com/content/dam/tfs/ATG/CMD/CMD%20Marketing%20Material/MassSpectrometryCollateral/Other/PSB104-Normalized-Collision-Energy-Technology-EN.pdf, Accessed: December/21 (2016)
- 37.Harman, D.G., Blanksby, S.J.: Investigation of the gas phase reactivity of the 1-adamantyl radical using a distonic radical anion approach. Org. Biomol. Chem. 5, 3495–3503 (2007)Google Scholar
- 39.Kirk, B.B., Trevitt, A.J., Blanksby, S.J., Tao, Y., Moore, B.N., Julian, R.R.: Ultraviolet action spectroscopy of iodine labeled peptides and proteins in the gas phase. J. Phys. Chem. A 117, 1228–1232 (2012)Google Scholar
- 44.Kuczkowski, R.L.: The structure and mechanism of formation of ozonides. Chem. Soc. Rev. 21, 79–83 (1992)Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.