The trend held for all of the major components of the particle phase as well, including PG, VG, and nicotine. Clearly, PG is easier to aerosolize than VG. This is due to the differences in chemical structure, and correspondingly, viscosity, vapor pressure, and boiling point. VG has one more OH group than PG, which results in stronger hydrogen bond intermolecular forces in the e-liquidsolution. The order of magnitude higher viscosity of VG at room temperature, requires more energy for vaporizing the solution. Moreover, the coil temperatures in Table 3.2 already surpass the boiling point of PG but are below the boiling point of VG . This is consistent with the high aerosol production when pure PG was used, and the high PG fraction in the gas phase The difference in total aerosol mass when vaping pure PG versus pure VG e-liquids at 375 °F suggests that PG was lost from the e-liquid at 8 times the rate of VG. This observation was corroborated in the mixed e-liquid using the gaseous CIMS and particle filter GC-MS data for PG and VG. The combined analytical uncertainties from CIMS and CG-MS are larger than for the pure gravimetric analysis, but also suggested a significant acceleration of PG loss compared to VG by a factor of ~9. Figure 3.7 shows representative carbonyl compounds and nicotine emitted from vaping e-liquid with different PG:VG ratios. Generally, grow trays carbonyl and nicotine concentrations decrease as the VG percentage increases in the mixture; although the hydroxycarbonyls do not decrease as dramatically as the simple carbonyls. This is likely due to the lower total aerosol and particle mass production overall as VG content increases in the e-liquid .
A notable exception is acrolein , which is the only compound whose formation increased with increasing VG, even as total aerosol mass decreased. Thus, for the 100% VG e-liquid, acrolein was one of the most concentrated carbonyls inhaled, and its relative production exceeded that of formaldehyde. The enhancement of acrolein between 30:70 and 0:100Mechanistic differences were more apparent when carbonyl formation was normalized by the total aerosol mass . The normalized trends reverse the absolute trends for some compounds, such as hydroxyacetone . Although hydroxyacetone can be generated from both PG and VG , the increase in relative aerosol fraction of hydroxyacetone with higher VG percentage in the e-liquid suggests that it is more efficiently formed from VG through the dehydration mechanism. This is consistent with the previous discussion that the heat-induced pathway is much more favorable for VG at the vaping temperatures we tested, and supports the exponential temperature dependence of hydroxyacetone . Likewise, formaldehyde emissions were inversely proportional to VG content , but increased slightly when normalized by aerosol mass . Formaldehyde can originate from both PG and VG and from either thermal or radical pathways. The data suggest that it is formed at similar efficiencies from both precursors, perhaps slightly favoring VG, which is consistent with more pathways available from VG . The relative production trends clearly showed that PG decomposition was responsible for all of the propionaldehyde and most of the acetaldehyde , while VG decomposition was responsible for all of the dihydroxyacetone and nearly all of the acrolein .
The VG source of acrolein is well-studied, and is leveraged in the conversionof biomass to fuels. Although acrolein can be formed by PG, it is a secondary product of a minor compound that is formed by the primary alkyl radical intermediate instead of secondary , which limits the importance of the PG source. The isomeric lactaldehyde and 1- hydroxypropanal likely had opposite trends that overlapped since they are solely formed by PG and VG, respectively . Given the higher formation of the sum of lactaldehyde and 1-hydroxypropanal with increased VG percentage , it appears that the formation of 1-hydroxypropanal from VG dominates over lactaldehyde formation from PG. These data support the exponential temperature trends of the lactaldehyde/1-hydroxypropanal pair, given that 1-hydroxypropanal is formed via heat-induced dehydration from VG. So far, most of the data are consistent with PG/VG mechanisms from the literature as shown in Scheme 3.1. However, notable deviations may exist for acetaldehyde and acetone. Acetaldehyde is thought to be a coproduct of formaldehyde in the VG dehydration,80 which would elevate it to be a major VG product, yet it appeared to be formed almost exclusively from PG . From PG, there was a suggested acetaldehyde source via radical reaction, instead of heat-induced dehydration. Thus, there was no reason to expect such a large abundance, or an exponential temperature curve . These observations, together with the fact that propionaldehyde . Acetone is known to be formed by PG; however, the data suggest that it can be formed by both PG and VG at roughly equal efficiencies. The temperature results also suggest a radical mechanism is dominant for acetone formation. Combined with the relative production trends, it would suggest that a radical formation mechanism from VG is missing from Scheme 3.1.
We are not aware of any proposed mechanism in the literature stemming from VG, especially one that is radical-initiated. The nicotine percentage in the particle phase at the same vaping temperature fluctuated with different PG:VG ratios . The nicotine concentration range observed in the particle phase is comparable to that in the original e-liquid . These results are consistent with those of Baassrir et al. and the trials organized by the Cooperation Centre for Scientific Research Relative to Tobacco .A 3-mg/mL concentration of nicotine in the e-liquid translated to 1.2 – 3.4 mg/mL nicotine in the particle phase, with the lowest nicotine percentage for the 50:50 mixture and increasing in both directions . More research is needed to understand the robustness of, and underlying reasons for, this trend and whether it is conserved with different nicotine content in the e-liquid. Approximately 0.3 mg/mL nicotine was observed in the total aerosol compared to 3 mg/mL used in eliquid.Figure 3.9 shows that the mass of the particles and representative carbonyl compounds generally increased with puff duration, as expected. Given the simultaneous increase in both particle mass and carbonyl mass with puff duration at the same flow rate, which would increase the puff volume, the carbonyl mass yield as normalized by aerosol mass would more or less be invariable. Both linear and non-linear fits would have yielded acceptable correlation coefficients within the studied range of only three data points. As puff durations in realistic use cases are unlikely to exceed this range, we did not test further. The relative increase between carbonyl compounds were roughly the same, within uncertainty. These results agree with Son et al., who found that increases in puff duration will increase the formation of carbonyl compounds, OH radicals, and nicotine.Most of the aerosol mass ended up in the gas phase , i.e., not captured on the hydrophilic PTFE filter, regardless of the temperature or PG:VG ratio tested . It is challenging to understand how the carbon mass from the e-liquid loss was distributed in the gas phase because there is noconventional analytical technique to quantify PG and VG in the gas phase due to the semivolatile nature of these compounds. The results from CIMS , pruning cannabis demonstrated that the majority of the gas phase was PG and VG instead of unknown compounds that are not well-measured by targeted techniques. This is consistent with findings that CO and CO2 are not abundant e-cigarette emissions. The CIMS spectra also showed that PG and VG were orders of magnitude larger in concentration than hydroxycarbonyls, a result that is consistent with the carbonyl-DNPH analysis. For the 30:70 PG:VG condition at 375 °F coil temperature, the sum of PG and VG obtained by CIMS in the gas phase accounted for the missing mass that was not captured by the particle filter within uncertainty . As discussed in Section 3.2.4, CIMS may overestimate semivolatile distribution in the gas phase due to evaporation during the sample dilution . However, it is clear from the CIMS spectra that the gas phase was dominated by mainly PG . GC-MS analysis of filters also showed that PG and VG were dominant components in the particle phase, with nicotine making up much less than 1%. In the 30:70 sample , approximately three-fourths of the particulate fraction was VG and three-fourths of the gaseous fraction was PG. The particle-phase composition roughly mirrored the e-liquid composition . The particle phase content of nicotine at 2.2 mg/mL was also similar to the eliquid composition. Our particle-phase results are consistent with other accounts that PG, VG, water and nicotine are the main components of e-cigarette droplets, and that nicotine is a small fraction of the total aerosol and only found in the particle phase . Thermal degradation products of nicotine have been reported in other works, but were not found in the present study even though the GC-MS method we used can detect nicotine products.It can be assumed that regular e-cigarette users intake a median of 200 puffs per day. This translates to an exposure dose of approximately 4.5 g PG/day and 0.8 g VG/day through inhalation of e-cigarette aerosols produced from vaping 30:70 PG:VG e-liquid at a coil temperature of 375 °F for a duration of 3 seconds. Although the PG exposure is fairly high compared to other aerosol components, animal and human studies demonstrate that PG has low toxicity even at relatively high doses. Mild sensory and respiratory irritation effects may result at concentrations of > 871 mg/m3 for particle plus gas phase PG, which translates to ~ 17.5 g/day exposure assuming 20 m3 air intake per day for a 70 kg adult.
VG has similarly weak irritation effects, which is supported by the German occupational exposure limit of 200 mg glycerin/m3 to protect against sensory irritation effects in the workplace. In contrast, the thermal degradation products, such as carbonyls, are a concern for potential risk of acute and chronic adverse human health effects despite their low absolute concentration in our study . Carbonyls may be further enhanced in flavored e-liquid, and may approach or exceed unhealthy doses for toxicological exposure with or without additional flavors. Formaldehyde, acetaldehyde, and acrolein are classified as known or probable human carcinogens. The more abundant hydroxycarbonyls in e-cigarette aerosols, such as hydroxyacetone, do not have available toxicology data. Carbonyls are also found in combustible cigarettes, so it is informative to discuss carbonyl exposure risk compared to combustible cigarettes and normalized to nicotine, as e-cigarette users have been reported to self-titrate for nicotine intake. At a VG content of 100% in the e-liquid, exposure to VG products such as hydroxyacetone, 1-hydroxypropanal, and acrolein become increasing important. At 100% VG, the acrolein/nicotine ratio range increased by a factor of 20 compared to the 30:70 e-liquid at the same coil temperature range of 315 – 510 °F , which exceeds the acrolein/nicotine ratio in combustible cigarettes under some temperature conditions. A Chronic Reference Exposure Levels value of 0.35 μg/m3 was set by the California Office of Health Hazard Assessment for acrolein. If this is multiplied by 20 m3 inhaled volume of per day for a 70 kg adult, then a threshold of 7 μg/day may be considered safe for chronic exposure. However, at 100% VG, the acrolein e-cigarette exposure that is equivalent to replacing only 1 cigarette/day exceeds chREL threshold at all tested temperatures. Given the lower aerosol and nicotine production at high VG ratios in the e-liquid , users may increase temperatures, puff duration, or puff frequencies to achieve higher aerosolization rates, which will significantly increase carbonyl exposure.Although e-liquids with 100% VG can be readily found commercially, they also may be formed during the dynamic vaping process. Our data suggest, because the total e-liquid mass loss from PG was 8 times that of VG , the e-liquid will be more enriched in VG as vaping continues. This will shift the e-cigarette aerosol composition toward VG and its degradation products, particularly acrolein, as VG enrichment occurs. Likewise, total aerosol mass and total nicotine will decrease during the lifespan of the e-liquid. We can build a simple model to predict the e-liquid mass remaining when 100% enrichment occurs. The model assumes that, as the PG and VG ratio changes during vaping, the total amount of e-liquid lost also changes in accordance with the total aerosol data =16.92e. Thus, for an 8:1 aerosolization ratio for PG:VG, and for a 30:70 ratio PG:VG mixture, it can be estimated that approximately 30-40% of e-liquid mass will be consumed by the time the e-liquid reaches 100% VG .