Chemical Analysis of Exhaled Vape Emissions: Unraveling the Complexities of Humectant Fragmentation in a Human Trial Study

Electronic cigarette smoking (or vaping) is on the rise, presenting questions about the effects of secondhand exposure. The chemical composition of vape emissions was examined in the exhaled breath of eight human volunteers with the high chemical specificity of complementary online and offline techniques. Our study is the first to take multiple exhaled puff measurements from human participants and compare volatile organic compound (VOC) concentrations between two commonly used methods, proton-transfer-reaction time-of-flight mass spectrometry (PTR-ToF-MS) and gas chromatography (GC). Five flavor profile groups were selected for this study, but flavor compounds were not observed as the main contributors to the PTR-ToF-MS signal. Instead, the PTR-ToF-MS mass spectra were overwhelmed by e-liquid thermal decomposition and fragmentation products, which masked other observations regarding flavorings and other potentially toxic species associated with secondhand vape exposure. Compared to the PTR-ToF-MS, GC measurements reported significantly different VOC concentrations, usually below those from PTR-ToF-MS. Consequently, PTR-ToF-MS mass spectra should be interpreted with caution when reporting quantitative results in vaping studies, such as doses of inhaled VOCs. Nevertheless, the online PTR-ToF-MS analysis can provide valuable qualitative information by comparing relative VOCs in back-to-back trials. For example, by comparing the mass spectra of exhaled air with those of direct puffs, we can conclude that harmful VOCs present in the vape emissions are largely absorbed by the participants, including large fractions of nicotine.


S3
Table S1.Participants, vape devices, e-liquid, and inhaled puff information.Vapes are classified as either closed or open.Participant preferences and operating inputs (power, resistance, and voltage settings) are reported for open vapes.Flavors denoted with (*) were conducted instead of watermelon.Flavors denoted with (**) are alternatives for vanilla.The reported inhaled puff volumes are reported as the average amongst all puffs during one visit with uncertainty values reported to one standard deviation.Stated nicotine concentrations are those reported by the manufacturer, whereas actual nicotine concentrations are those determined by GC-MS measurements.E-liquid nicotine measurements were conducted using the method described in Pagano et al. (2015) for thirty three of the forty e-liquids included in this study. 1Gas chromatograph (6850 Series II, Network GC System, Agilent Technologies) coupled to a quadrupole mass spectrometer in electron ionization mode (5975B VL MSD, Agilent Technologies) was used to analyze the nicotine standards and e-liquid samples.Nicotine values typically deviated from those reported by the manufacturers.Appendix A. Proton-transfer-reaction time-of-flight mass spectrometer (PTR-ToF-MS) information.
Appendix A.1.General Description of Ionization in PTR-ToF-MS PTR-ToF-MS is a soft chemical ionization instrument.Ionization occurs through proton transfer reactions with volatile organic compounds (VOCs) and reagent hydronium ions (H3O + ) to form [M+H] + ions.A hallow-cathode plasma ion source produces H3O + reagent ions that enter a reaction chamber (drift tube) for exposure and collision with analyte VOC molecules.Proton transfer reaction, and subsequent ion detection by the mass spectrometer, will occur during collision only if the proton affinity (PA) of the analyte compound is greater than that of water (691 kJ mol -1 ). 2 Common constituents of air (N2, O2, Ar, CO2, and CO) have PAs less than 691 kJ mol -1 and, thus are not detected by PTR-ToF-MS.Most VOCs have PAs greater than water and can be detected, with the exception of alkanes smaller than C5. 3 Though energy upon collision of VOCs with H3O + is mild enough to leave most molecules intact, fragmentation often occurs for weak chemical bonds and/or large exothermicity of proton-transfer-reactions.

Appendix A.2. PTR-ToF-MS Experimental Settings and Tedlar ® Bags
Prior to each participant visit, mass-to-charge calibration was performed using four isotopic ions commonly observed in room air: m/z 21.0221 (H3 18 O + ), 33.9935 ( 18 O 16 O + ), 39.0327 (H2 18 O)H3 16 O + ), and 59.0491 [C3H6O + H] + .The PTR-ToF-MS drift tube settings for temperature, pressure, and voltage were 60 °C (Tdrift), 2.29 mbar (pdrift), and 600 V (Udrift), respectively, resulting in a ratio of the electric field (E) to the number density of the drift tube buffer gas molecules (N) of E/N ~132 Td (where Td = 10 17 V cm 2 ).Analyte sampling was conducted at a flow rate of 0.1 L min -1 through a PEEK tubing inlet (1.59 mm O.D.) heated to 70 °C.The data were acquired with a 1 s time resolution.Depending on the species measured, the limit of detection was in the range of several parts per billion by volume (ppbv) or better.PTR-ToF-MS calibration was conducted using a standard solution of VOCs (acetone, formic acid, acetic acid, acetaldehyde, and formaldehyde) at known concentrations (5, 10, 30, 50, 100, 150, 200, 250 ppbv).Through analysis of the calibration curves (Figure S1), it was determined that small acids were approximately 30% underestimated by the PTR-ToF-MS, ketones were approximately 10% overestimated, and aldehydes had varied results, acetaldehyde was correctly determined by PTR-ToF-MS but formaldehyde was 64% underestimated.These correction values were not applied to participant data due to much larger fragmentation pattern discrepancies discussed in the manuscript.
During each participant visit, a Tedlar ® bag was connected to the PTR-ToF-MS inlet for participant sampling.The most common material of Tedlar ® bags is polyvinylfluoride (PVF) (C2H3F)n. 4PTR-ToF-MS measurements of clean Tedlar ® bags show large peaks at m/z 88.0757 (N,N-dimethylacetamide, C4H10NO + ) and m/z 95.0491 (phenol, C6H7O + ).5][6][7][8] These compounds are quantitatively excluded from our results, as the background levels remain constant throughout sampling.Tedlar ® bags were cleaned 3x prior to each participant sampling through purging gas-phase pollutants by evacuating the bags and refilling them with clean air from a purge air generator (filtered and scrubbed of VOCs, CO2, and water vapor).This cleaning did not fully remove "sticky", residual compounds that were adsorbed on bag walls (i.e., nicotine).Removing residual compounds would require heating the bags to 60 °C and evacuating/filling them with clean air multiple times, which was not feasible within the turnaround time between all participants. 4Nicotine (m/z 163.1223.C10H15N2 + ) was present in the background prior to sampling but background levels did not affect the increases observed from new vapes into the bag. .Through analysis of these calibration curves, it was determined that small acids were approximately 30% underestimated by the PTR-ToF-MS, ketones were approximately 10% overestimated, and aldehydes had varied results, acetaldehyde was correctly determined by PTR-ToF-MS but formaldehyde was 64% underestimated.These trends were not factored into our reported concentrations, as our uncertainty in this human trial study far exceeds these errors.

Appendix A.3. Programmable Pump for Simulated Puffing (Direct Injection Measurements)
A customized programable syringe pump was manufactured to accommodate a Tomopal 50 mL glass syringe, facilitating the delivery of 20 mL of e-cigarette aerosol into a Tedlar® bag.Utilizing computer numerical control (CNC) machining, a syringe pump chassis was created from a 15.24 x 30.48 x 5.08 cm white ultra-high molecular weight polyethylene sheet.This chassis included two linear motion carbon steel shafts measuring 9.52 mm in diameter and 228.60 mm in length on each side, equipped with bushings to reduce friction during operation.Positioned between the linear motion shafts, a stepper motor connected to a threaded shaft measuring 300 mm in length was utilized to regulate the syringe pump's movement.The stepper motor was interfaced with a Raspberry Pi 3 model B+, which controlled the stepper motor through the general-purpose input/output (GPIO) pins using the Python programming language.The 20 mL aspiration and extraction processes of the syringe were represented by the step angle of the stepper motor, whereas the flow rate was determined by the delay value, which represents the delay between each step of the stepper motor.

Appendix A.4. PTR-ToF-MS Data Analysis
A peak list was generated from ions observed in the mass spectra of participant samples.Ions were identified by comparison with the publicly available library (www.tinyurl.com/PTRLibrary). 9This library contains 1000 trace gases detected by PTR-MS and displays the fragmentation patterns of compounds after proton transfer reactions.Once a peak list was compiled, a ratio of each ion's intensity in spectrum A (exhaled puff) to spectrum B (background breath) was taken.Ions in the peak list with ratio > 1 were focused on for further analysis.
After injection of each sample (baseline breath, exhaled vape puff, or direct vape injection) into the Tedlar ® bag, the signal plateaued and stabilized.For each VOC, an average of the signal was taken over 5 minutes after stabilization.This average signal was converted to mixing ratio of the corresponding gas.The PTR-MS Viewer software uses Equation S1 to convert average signal intensity into mixing ratio (C i ppb) in the unit of ppbv.

Equation S1
Other terms in Equation S1 are defined as: • k = ion-molecule collision reaction rate constant (= 2.0 x 10 -9 cm 3 s -1 ).While this rate constant can range from 1.5 to ~5 x 10 -9 cm 3 s -1 , a single value for the rate constant was used for all VOC. 3,9This variable contributes the largest to the uncertainty in the determination of VOC estimated concentrations, which is expected to be ± 50%.Concentration values (ppbv) determined by the PTR-Viewer Software were converted into mass concentrations (µg VOC puff -1 ) using Equations S2-6.∆ppbvBaseline is the concentration of baseline VOC subtracted from the clean air inside the bag (Equation S2).∆ppbvExhaled is the concentration of VOC in the exhaled puff subtracted from the background breath concentration (Equation S3).∆ppbvDirect Injection is the concentration of VOC in the direct injection subtracted from the exhaled puff VOC concentration, as the Tedlar ® bag was not cleaned after exhaled puff and the PTR-ToF-MS signal is additive (Equation S4).For the baseline breath, exhaled puff (after e-cigarette inhalation), and direct injection of e-cigarette aerosol, ∆ppbv for each VOC was converted into µg VOC in the bag using Equation S5 and the following: concentration of air molecules under ambient conditions (2.46 x 10 19 molecules cm -3 ), the molecular weight of each VOC, and the volume of clean air inside the Tedlar ® bag (80 L).

Equation S5
In the case of the direct injection (Equation S6), there was no participant interaction with these measurements and thus needed to be corrected using the inhaled volume of each participant's puff (from topography measurements, Table S1) and the volume of direct injection (20 mL).For the baseline breath and exhaled puff (after e-cigarette inhalation), pptv for each VOC was converted into µg VOC puff -1 using Equation S7 and the following: concentration of air molecules under ambient conditions (2.46 x 10 19 molecules cm -3 ), the molecular weight of each VOC, and the volume of the WAS canister (2000 cm 3 ).Since the participants often did not exhale 2000 cm 3 worth of breath into the WAS can, a pressure correction was made by applying the ratio of the pressure inside of the can to 760 Torr.2009) reported that, ethanol fragments into H3O + at higher E/N conditions, such as the ones used in this study (~132 Td).Consequently, the sensitivity at m/z 47 is reduced, consistent with our observation.(c) The PTRMS cannot distinguish between isomers at m/z 59.0491 (C3H7O + ), 57.070 (C4H9 + ), and 71.041 (C4H6O + ).(d) Analysis of the headspace of the pure standard showed that 2-propanol heavily fragments into its [M+H-H2O] + ion with an additional high intensity ion at m/z 41.040 (C3H5 + ), accounting for 65% of the m/z 43 (C3H7 + ) peak intensity.(e) Both MTBE and 1-butanol do not show a [M+H] + parent peak at m/z 89 and 75 respectively; but instead, both heavily fragment into a major ion at m/z 57.070 (C4H9 + ).Note that both compounds exhibit a small contribution to m/z 41.040 (C3H5 + ) as well accounting for about 26% and 15% of the m/z 57 peak for MTBE and butanol respectively, which overlaps with 2-propanol fragment.
(f) Note that a few compounds co-elute on our GC-FID column, which is the detector used for quantification.If two co-eluting species were both present at high concentration, such as in the VOC mixture, their respective quantification could not be conducted.This was the case for ethanol and acetonitrile (co-eluted at 8.963 min), as well as MTBE and MVK (co-eluted at 11.298 min).(g) 1-Butanol was not detected by GC.  b) Monoterpenes are known to fragment into m/z 81 but due to the complexity of the samples, we excluded this ion from quantification. 27,28        while acetone is presented exclusively for the GC measurements, the PTRMS signal is for the sum of acetone and propanal (m/z 59, C3H7O + ).In panel (f), due to the overwhelming presence of the humectant, the PTR-ToF-MS signal is given for the sum of GLY and toluene, while the GC measurements are reported for toluene exclusively.In panel (i), while ethyl acetate is presented exclusively for the GC measurements, the contribution of butanoic acid couldn't be excluded for the PTR-ToF-MS and the signal is for the sum of ethyl acetate and butanoic acid.Note that the slopes are included for reference only to demonstrate how the two measurements deviate from one another, not to suggest that there is correlation between the two datasets.

Figure S1 .
Figure S1.PTR-ToF-MS calibration was conducted using standard solutions of VOCs: a) acetaldehyde, b) formaldehyde, c) formic acid, d) acetic acid, and e) acetone at known concentrations(5, 10, 30, 50, 100, 150, 200, 250 ppbv).Through analysis of these calibration curves, it was determined that small acids were approximately 30% underestimated by the PTR-ToF-MS, ketones were approximately 10% overestimated, and aldehydes had varied results, acetaldehyde was correctly determined by PTR-ToF-MS but formaldehyde was 64% underestimated.These trends were not factored into our reported concentrations, as our uncertainty in this human trial study far exceeds these errors.

Figure S2 .
Figure S2.Expanded view of the PTR-ToF-MS spectra showing the separation of multiple peaks observed at nominal m/z 43, 57, and 93.Heated glycerol (bottom panel) is compared with exhaled puff (top) and direct injection (middle) measurements from participant 5's first visit.

Figure S4 .Figure S5 .
Figure S4.Exhaled puff PTR-ToF-MS values of micrograms per inhaled puff volume obtained from participant puff topography data.The puff volume reported is the volume each participant inhaled from their vape device prior to exhalation into the PTR-ToF-MS sampling unit.Twelve compounds were selected for this comparison.Red markers correspond to closed vapes and blue markers correspond to open vapes used in the human trial.

Figure S6 .
Figure S6.PTR-ToF-MS mass spectra for Participant 1's five visits using Puff Xtra Limited and Puffs Plus closed vapes in flavors of (a) mint, (b) watermelon skittles, (c) apple, (d) banana vanilla, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.The peak indicated with ( †) at m/z 88 corresponds to DMAC, a known impurity of Tedlar ® bags, which was incompletely removed by subtraction.Each spectrum corresponds to an average spectrum taken over 5 min.

Figure S7 .
Figure S7.PTR-ToF-MS mass spectra for Participant 2's five visits using a Smok® POZZ X open vape with e-liquid in flavors of (a) mint, (b) tobacco, (c) apple, (d) vanilla custard, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.Each spectrum corresponds to an average spectrum taken over 5 min.It is important to note that this participant did not smoke watermelon (one of the 5 flavor profiles in this study).Tobacco was initially one of the five profiles selected for this study, but due to feedback from participants 2 and 3, it was discontinued and switched out for watermelon.Mass spectra (a) and (b) had very low signal and the dominant peaks were acetone and isoprene (common in breath without vape).

Figure S8 .
Figure S8.PTR-ToF-MS mass spectra for Participant 3's five visits using a Smok® POZZ X open vape with e-liquid in flavors of (a) mint, (b) tobacco, (c) apple, (d) vanilla custard, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.The peak indicated with ( †) at m/z 88 corresponds to DMAC, a known impurity of Tedlar ® bags, which was incompletely removed by subtraction.Each spectrum corresponds to an average spectrum taken over 5 min.It is important to note that this participant did not smoke watermelon (one of the 5 flavor profiles in this study).Tobacco was initially one of the five profiles selected for this study, but due to feedback from participants 2 and 3, it was discontinued and switched out for watermelon.

Figure S9 .
Figure S9.PTR-ToF-MS mass spectra for Participant 4's five visits using Lucid Air and Blu closed vapes in flavors of (a) mint, (b) watermelon, (c) apple, (d) vanilla, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.The peak indicated with ( †) at m/z 88 corresponds to DMAC, a known impurity of Tedlar ® bags, which was incompletely removed by subtraction.Each spectrum corresponds to an average spectrum taken over 5 min.

Figure S10 .
Figure S10.PTR-ToF-MS mass spectra for Participant 5's five visits using a SMOK® Morph 2 open vape with e-liquid flavors of (a) mint, (b) watermelon, (c) apple, (d) vanilla custard, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.Each spectrum corresponds to an average spectrum taken over 5 min.

Figure S11 .
Figure S11.PTR-ToF-MS mass spectra for Participant 6's five visits using a SMOK® Alike open vape with e-liquid flavors of (a) mint, (b) watermelon, (c) apple, (d) vanilla, and (e) mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.The peak indicated with ( †) at m/z 88 corresponds to DMAC, a known impurity of Tedlar ® bags, which was incompletely removed by subtraction.Each spectrum corresponds to an average spectrum taken over 5 min.

Figure S12 .
Figure S12.PTR-ToF-MS mass spectra for Participant 7's five visits using Elfbar BC5000 and Hyde Rebel Pro closed vapes with e-liquid flavors of (a) lemon mint, (b) watermelon bubble gum, (c) apple, (d) vanilla, and (e) mango peach.The presented spectra are of the exhaled puff measurements with background breath subtracted.Each spectrum corresponds to an average spectrum taken over 5 min.

Figure S13 .
Figure S13.PTR-ToF-MS mass spectra for Participant 8's five visits using Flum Float closed vapes with e-liquid flavors of (a) mint, (b) watermelon, (c) apple, (d) tobacco cream, and (e) strawberry mango.The presented spectra are of the exhaled puff measurements with background breath subtracted.Each spectrum corresponds to an average spectrum taken over 5 min.Tobacco cream is Flum Float's equivalent product to vanilla that other participants smoked.

Figure S14 .
Figure S14.Scattered data for each PTR/GC exhaled breath measurement, in calculated micrograms per puff.Black dashed lines are the linear fit for participant data.Diagonal 1:1 lines (solid grey) illustrate the lack of correlation between measurements.All signal intensities reported for the PTR-ToF-MS correspond to the [M+H] + ion of each of the reported species.In panel (b) while acetone is presented exclusively for the GC measurements, the PTRMS signal is for the sum of acetone and propanal (m/z 59, C3H7O + ).In panel (f), due to the overwhelming presence of the humectant, the PTR-ToF-MS signal is given for the sum of GLY and toluene, while the GC measurements are reported for toluene exclusively.In panel (i), while ethyl acetate is presented exclusively for the GC measurements, the contribution of butanoic acid couldn't be excluded for the PTR-ToF-MS and the signal is for the sum of ethyl acetate and butanoic acid.Note that the slopes are included for reference only to demonstrate how the two measurements deviate from one another, not to suggest that there is correlation between the two datasets.

Figure S15 .
Figure S15.Exhaled puff GC values of micrograms per puff with respect to participant puff topography data.Twelve compounds were selected for this comparison.Red markers correspond to closed vapes and blue markers correspond to open vapes used in the human trial.

Table S2 .
VOC limit of detection and measurement precision and accuracy for the GC platform.

Table S3 .
Measured concentrations of a 13-component ambient air quality gas standard (Cylinder CC302254, 2000 psi, AiR Environmental Inc.).The concentrations in the rightmost column were recertified by the manufacturer in February 2024.

Table S4 .
Ions observed in the PTR-ToF-MS mass spectra (FiguresS5-12) from participant exhaled vape puff measurements.Structural isomers cannot be distinguished using the PTR-ToF-MS, hence, multiple compound assignments are given for certain peaks.