The influence of terpenes on the release of volatile organic compounds and active ingredients to cannabis vaping aerosols

Dabbing and vaping cannabis extracts have gained large popularity in the United States as alternatives to cannabis smoking, but diversity in both available products and consumption habits make it difficult to assess consumer exposure to psychoactive ingredients and potentially harmful components. This work studies the how relative ratios of the two primary components of cannabis extracts, Δ9-tetrahydrocannabinol (THC) and terpenes, affect dosage of these and exposure to harmful or potentially harmful components (HPHCs). THC contains a monoterpene moiety and has been previously shown to emit similar volatile degradation products to terpenes when vaporized. Herein, the major thermal degradation mechanisms for THC and β-myrcene are elucidated via analysis of their aerosol gas phase products using automated thermal desorption-gas chromatography-mass spectrometry with the aid of isotopic labelling and chemical mechanism modelling. Four abundant products – isoprene, 2-methyl-2-butene, 3-methylcrotonaldehyde, and 3-methyl-1-butene – are shown to derive from a common radical intermediate for both THC and β-myrcene and these products comprise 18–30% of the aerosol gas phase. The relative levels of these four products are highly correlated with applied power to the e-cigarette, which indicates formation of these products is temperature dependent. Vaping THC–β-myrcene mixtures with increasing % mass of β-myrcene is correlated with less degradation of the starting material and a product distribution suggestive of a lower aerosolization temperature. By contrast, dabbing THC–β-myrcene mixtures with increasing % mass of β-myrcene is associated with higher levels of HPHCs, and isotopic labelling showed this is due to increased reactivity of β-myrcene relative to THC.


Table of Contents
Synthesis of -myrcene-d 6 . To a solution of hexadeutero isopropyl triphenylphoshine iodide salt (420 mg, 1.0 mmol, 1.1 eq) in THF (9 mL, 0.1 M) at 0 ⁰C was added n-butyllithium (1.6 M, 620 µL, 1.0 mmol, 1.1 eq). This solution was allowed to stir at 0 ⁰C for 30 min before a solution of 4methylenehex-5-enal (100 mg, 0.90 mmol, 1.0 eq) in THF (0.50 mL) was added dropwise. The ice bath was removed and the reaction was permitted to stir at room temperature for 2 hours before being quenched with saturated aqueous ammonium chloride and extracted with pentane. The combined organic fractions were dried over anhydrous magnesium sulfate, concentrated under reduced pressure, and purified via flash chromatography (100% pentane) to provide the title compound in 54% yield in a 6:1 ratio with pentane. As expected, NMR analysis shows a spectrum identical to that of myrcene except for the absence of six proton signals associated with the geminal dimethyl olefin, and confirming the presence of 7-(methyl-d 3 )-3-methyleneocta-1,6diene- 8 Figure S1. EIMS spectra of -myrcene and -myrcene-d 6 Synthetic cannabis oil. THC (Cayman Chemical, Ann Arbor, MI) was acquired as a 10 mg/mL solution in acetonitrile, which was concentrated in vacuo. Pure THC was assessed for purity by HPLC-UV and NMR. THC was used alone in vaping or dabbing experiments, or mixed with myrcene (Sigma Aldrich) or -myrcene-d 6 for studies using synthetic cannabis oil. THC and myrcene mixtures were homogenized in scintillation vials using a rotary evaporator slowly spinning at atmospheric pressure with the vial partially submerged in a 50 ⁰C water bath for 1 -2 hours. THC content was assessed by HPLC-UV on 5-point standard addition calibration curves by first creating analyte stock solutions. of the mixes at 1 -1.3 mg/mL in 1:1 CH 3 CN:H 2 O. 400 L of 1.0 mg/mL (-)- 9 -THC in methanol certified reference material standard soln. (Cerilliant Corporation, Round Rock, TX) were added to a 2 mL vol. flask, and the methanol was evaporated under a gentle stream of Ar, then brought up to volume in 1:1 CH 3 CN:H 2 O for a final conc. of 200 g/mL (THC spike soln.). 50 L of analyte stock soln. and 100, 150, 200, 300, or 400 L of THC spike soln. were added and to 2 mL. vol. flasks and brought up to volume in 1:1 CH 3 CN:H 2 O, and immediately analyzed by HPLC-UV monitoring at 254 nm.
Cartridge vaping experiments. Pure THC, THC with 7.2 % myrcene, THC with 14 % myrcene, and pure CBN were added to CCELL TH2 oil vape atomizer (CCELL) and warmed in a 40 ⁰C oven for 3 -4 hours oven to allow the oil to saturate the internal wick, and then used the following day in vaping experiments. The atomizers were connected to an iStick PICO (eLeaf) battery that was set to the wattage required for each experiment. The aerosol collection apparatus ( Figure S2  Given the variability of sorbent material packing in each ATD sorbent tube, each tube was calibrated on a 5-point calibration curve (CSM puff depth [V] vs. flowmeter flowrate [L/min]) in order to determine the puff depth setting on the CSM to match, as closely as possible, the CORESTA recommended setting for e-cigarette puffing: 50 mL puff volume in 3 s. 5 Knowledge of the exact puff volume facilitated air blank VOC correction. After calibration, VOC emissions from a single puff from the vaporizer were collected on the ATD sorbent tube, and the atomizer was massed before and after each puff. Air blanks were collected in triplicate in the exact same manner on the days experiments were performed and used to account for background levels of target VOCs in the samples. Benzene and toluene were the only target VOCs (Table 1) detectable. Air levels of benzene (4.30.2 ng/L) and toluene (2.00.4 ng/L) were taken as the mass of analyte collected on the sorbent tube vs. the total sampled air volume, including the calibration draws. Background contributions of benzene and toluene were subtracted from measured benzene and toluene levels in ATD sorbent tubes for vaping samples by accounting for the total sampled air volume for each (including calibration draws). ATD-GC-MS methodology. Sorbent tubes were stored at -20 C for not more than seven days before analysis. ATD sorbent tubes were thermally desorbed with a TurboMatrix 650 automated thermal desorber (ATD) unit. 20 ng fluorobenzene, 18.6 ng toluene-d 8 7 Where selected HPHCs were quantified, an ionizaton cross section is calculated to provide a more accurate result. When total the yield of total VOCs (VOC T ) were calculated, the ionization cross section of all components of the chromatogram was assumed to be equal to that of a chosen internal standard, fluorobenzene.

THC delivery analysis.
In GP samples generated from THC--myrcene-d 6 mixes, the coeluting deuterated and nondeuterated compounds prevented these from being estimated using the above non-target analysis method, which requires integration on the total ion chromatogram. To overcome this, response factors for HPHCs of interest were determined from previously collected quantitative ATD-GC-MS chromatograms. The mass of each HPHC in the sample (m HPHC, sample , ng) per mg particulate matter collected (m PM ) was determined using equation 1: where A HPHC is the area of HPHC's ion of interest in the selected ion chromatogram (SIC), A FB is the m/z = 96 SIC area of the fluorobenzene internal standard, RF FB fluorobenzene's response factor for m/z = 96 calculated from a blank run (A m/z=96 /m FB ), RF HPHC is the response factor of the HPHC's ion of interest calculated from an injection of pure standards, m FB is the mass of fluorobenzene added (20 ng) to each sample, and m HPHC,blank is the mass of HPHC present in the laboratory air blank. The response factor for a specific ion of interest of an HPHC was used for the equivalent ion in a deuterium isotopologue. For example, the RF for isoprene's m/z=67 amu ion was assumed to be equal to isoprene-d 5 's m/z=71 amu ion, because these both occur after loss of a methyl hydrogen.
Chemical mechanism modeling. A gas-phase oxidation mechanism for β-myrcene was derived using the SAPRC8-9 mechanism generation system, MechGen10, and product formation was predicted using a SAPRC box model. MechGen uses experimentally derived rate constants and branching ratios if data are available and otherwise uses estimated rate constants and branching ratios based on group additivity and other estimation methods. MechGen has been used previously in the development of the SAPRC-18 mechanism11 and in development of a detailed SAPRC furans mechanism for atmospheric modeling.12 In this work, MechGen was used to derive a β-myrcene oxidation mechanism under vaping conditions (significantly higher VOC levels and temperature than atmospheric conditions); the MechGen-derived mechanism was then implemented into a SAPRC box model to simulate vaping of a β-myrcene (300 ppm) and THC (700 ppm) mixture at 643 K and 1 atm with 5 ppb of NO. The SAPRC simulation duration was 10 minutes with a time step of 0.1 min, and the OH level was controlled between 2×10-8 and 5×10-7 ppm throughout the simulations. The SAPRC modeling was used to investigate observed ratios of product formation as a function of temperature and NO level.
To further investigate product formation mechanisms, a second gas-phase chemical mechanism generator, GECKO-A, was used to derive a β-myrcene oxidation mechanism under vaping conditions. GECKO-A is a nearly explicit chemical mechanism generator that relies on experimental data, structure-activity relationships, and a predefined protocol to generate detailed oxidation reaction schemes for organic compounds under atmospheric conditions (Aumont et al.  elutes immediately before 3MCA on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +6 amu mass shift on the molecular ion and a +6 amu mass shift on the isobutenyl cation. Figure S4. The EIMS spectra for 2-methyl-2-butene (2M2B) and its deuterium isotopologue 1,1,1-trideutero-2- (1,1,1-trideuteromethyl)-but-2-ene (2M2B-d 6 ) that are formed when -myrcene-d 6 is subjected to dabbing. 2M2B-d 6 elutes immediately before 2M2B on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +6 amu mass shift on the molecular ion and a +3 amu mass shift on its base peak.
S7 Figure S5. The EIMS spectra for isoprene and 1,1-dideutero-2-(1,1,1-trideuteromethyl)-1,3-butadiene (isoprene-d 5 ) that are formed when -myrcene-d 6 is subjected to dabbing. Isoprene-d 5 elutes immediately before isoprene on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +6 amu mass shift on the molecular ion and a +2 amu mass shift on the butadienyl cation. The presence of other ions such as m/z = 72, 56, and 57 suggest that another isoprene-d 5 isotopomer may be present, but the relatively higher abundance of m/z = 73, 71, 55, and 42 suggest that the proposed structure is the most abundant isotopomer. Figure S6. The EIMS spectra for isopentene and its deuterium isotopologue 4,4,4-trideutero-3-(1,1,1trideuteromethyl)-but-1-ene (isopentene-d 6 ) that are formed when -myrcene-d 6 is subjected to dabbing. Isopentened 6 elutes immediately before isopentene on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +6 amu mass shift on the molecular ion and a +3 amu mass shift on its base peak.
S8 Figure S7. The EIMS spectra for acetone and its deuterium isotopologue 1,1,1,3,3,3-hexadeutero-2-propanone (acetone-d 6 ) that are formed when -myrcene-d 6 is subjected to dabbing. Acetone-d 6 elutes immediately before acetone on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +6 amu mass shift on the molecular ion and a +3 amu mass shift on its base peak. Figure S8. The EIMS spectra for methacrolein (MACR) and its deuterium isotopologue 3,3-dideutero-2-(1,1,1trideuteromethyl)-prop-2-enal (MACR-d 5 ) that are formed when -myrcene-d 6 is subjected to dabbing. MACR-d 5 elutes immediately before MACR on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +5 amu mass shift on the molecular ion and a +5 amu mass shift on its base peak.
S9 Figure S9. The EIMS spectra for methyl vinyl ketone (MVK) and its deuterium isotopologue 1,1,1-trideuterobut-3en-2-one (MVK-d 3 ) that are formed when -myrcene-d 6 is subjected to dabbing. MVK-d 3 elutes immediately before MVK on the GC-MS chromatogram, and the structure was proposed primarily on the observation of a +3 amu mass shift on the molecular ion, an identical base peak which results from loss of the methyl group, and a +3 amu mass shift on the acetyl radical.     Table S2. All GP products from dabbing THC tentatively identified by GCMS presenting a match quality of >70 % with the NIST/Wiley mass spectral library. Figure S12. Proposed mechanism for the conversion of -myrcene to psi-limonene. psi-Limonene formation may occur as an intramolecular ene reaction of -myrcene or via a radical mechanism.
1a and 1b product distribution as a function of applied power In order to determine the influence of applied electrical power on the product distribution of the four products deriving from radical 1 (3MCA and 2M2B from resonance structure 1a, and isoprene and 3M1B from resonance structure 1b), relative ratios of integrations of the molecular ion of each were graphed as a function of power. The increase in isoprene:3M1B ratio (1b oxidation and reduction products) with respect to power and the decrease in 3MCA:2M2B ratio (1a oxidation and reduction products) is mirrored by a decreasing 3MCA:isoprene ratio with respect to power. The static 2M2B:3M1B ratio signals that the decreasing 1a:1b ratio with power is largely governed by a decreasing 3MCA:isoprene ratio.