Carbonyl compounds are the most abundant toxic chemicals emitted from electronic cigarettes (e-cigarettes) [1
]. Among the carbonyls found in e-cigarette emissions (e-vapor), two of the major carbonyls (i.e., formaldehyde and acetaldehyde) are known human carcinogens [5
]. Glyoxal, which is known to cause allergic reactions, was also found in e-vapor [6
]. Several flavoring carbonyls in e-liquids (e.g., vanillin, cinnamaldehyde) have shown increased cell toxicity [7
]. It is also worth mentioning that 2,3-butanedione (diacetyl) and 2,3-pentanedione (acetylpropionyl) were found in e-vapors with ‘buttery’ flavored e-liquids [9
] which are known to cause bronchiolitis obliterans (aka ‘popcorn lung’) [10
]. However, factors affecting carbonyl emissions from e-cigarettes are still not fully understood. Most current studies report that higher e-cigarette power outputs significantly increase formaldehyde and acetaldehyde emissions [11
]. However, there is a 1000-fold difference across the literature in formaldehyde emission rate data from e-cigarettes at comparable e-cigarette device power outputs [12
], suggesting that carbonyl emissions are not dependent on e-cigarette power output alone and that other factors should be considered.
Among the potential other factors, flavoring agents used in e-liquids could be sources of potentially harmful carbonyl emissions. Thermal fragmentation of flavoring chemicals has been shown to form carbonyls under the burning temperatures of conventional cigarettes [18
], but the contribution of flavoring chemicals to carbonyl emission has not been sufficiently studied for e-cigarettes [19
]. In addition, some flavored e-liquids and e-vapors contain chemicals [e.g., diacetyl and acetylpropionyl] that can be toxic at sufficient levels [7
]. However, levels of the potentially harmful chemicals in e-vapor resulting from flavoring additives were not fully accessed under e-cigarette vaping topography [9
The contribution of e-liquid base materials to carbonyl emissions is also largely unknown. It has been shown that thermal degradation of vegetable glycerin (VG) and propylene glycol (PG), the main base materials used in e-liquids, generates various carbonyl compounds [15
]. However, the thermal degradation of VG and PG has not been studied across a wide range of e-cigarette coil temperatures [14
] even though coil temperature is an important determinant of e-cigarette carbonyl emissions [13
E-cigarette vaping topography (i.e., puff volume and duration) could also potentially affect carbonyl formation. Vaping topography can affect carbonyl emissions by modifying e-cigarette heating coil temperatures [23
]. However, most of the previous studies examining e-cigarette carbonyl emissions could not mimic actual e-cigarette user’s behavior [12
]. Those studies generated e-vapors similar with the ‘Health Canada Intense (HCI) Regime’ (55 mL puff volume, 2 s puff duration, every 30 s) which was developed for conventional cigarette smokers and not e-cigarette vapors [26
]. Therefore, real-world vaping topography should be used to study carbonyl emissions. In addition, most of the preceding studies only focused on formaldehyde and acetaldehyde emissions. However, other potentially harmful carbonyl compounds, such as acrolein and glyoxal, in e-vapor also need to be evaluated under real-world vaping conditions.
To address these knowledge gaps, this study evaluated the impacts of real-world vaping conditions (i.e., real-world e-cigarette heating power, vaping topography, and e-liquid components) on the emission of six potentially harmful carbonyls (i.e., glyoxal, formaldehyde, acetaldehyde, acrolein, diacetyl, and acetylpropionyl) and thirteen additional carbonyl species described below.
This study adds new evidence on the levels of carbonyls emitted from e-cigarette. In this study, a variety of carbonyls were measured under wide ranges of e-cigarette use patterns: e-liquid compositions, power outputs, and vaping topography. The formation of carbonyls using various combinations of base materials and device power outputs were explored. Thermal decomposition of VG and PG forms carbonyls during e-cigarette vaping [21
], and increased coil temperatures accelerate the decomposition rates of e-liquid base materials [42
]. PG formed higher levels of carbonyl compounds (e.g., formaldehyde, acetaldehyde, butylaldehyde, tolualdehyde) during e-cigarette vaping than VG, probably because PG has a lower thermal decomposition temperature than VG. The thermal decomposition of PG starts as low as 127 °C [43
], while VG requires at least 200 °C for its thermal decomposition reaction to begin [44
]. Kosmider et al. [14
] also reported that PG-containing e-liquids generated significantly higher amounts of formaldehyde and acetaldehyde than VG based e-liquids. Glyoxal, which was shown to cause allergic reaction [6
], was observed only with VG-based e-liquid under high power output setting. It has been reported that thermal oxidation of VG leads to the formation of glyoxal [15
Higher e-cigarette power outputs increased formaldehyde emissions for both top and bottom coils (Figure 3
). The formaldehyde levels observed in this study were within interquartile ranges of literature values [4
], but we observed wide varieties in reported formaldehyde levels. The reported e-vapor formaldehyde levels have been shown to range from 0.01 µg/puff to 342.2 µg/puff for top coil device and range from 0.02 µg/puff to 220.0 µg/puff for bottom coil e-cigarettes [4
]. The wide range of e-cigarette carbonyl emission levels reported literatures might be a factor of the coil settings [13
]. The top coils formed higher amounts of formaldehyde per puff (23.35 ± 59.68 µg/puff) with conventional cigarette smoking (12.32 ± 9.65 µg/puff) due to the limited e-liquid supply to the heating coil. A top coil is located on the top of the atomizer with long wicks dropping down into the e-liquid tank (Figure S3
A long wick cannot supply enough e-liquid to the coil, and the limited e-liquid supply can easily dry up the heating coil, leading to a rapid coil temperature increase. The dramatic increase of coil temperature is known as ‘dry puff’ or ‘dry hit’, which results in significantly increased amounts of carbonyl formation [48
]. In contrast, a bottom coil is located at the bottom of the atomizer, with a short wick contacting the e-liquid (Figure S3
). Bottom-coils, commonly used in the current generations of e-cigarettes, generally provide consistent hits without ‘dry puffs’. Consequently, a bottom coil generated 10–10,000 times less formaldehyde per puff than conventional cigarettes due to stable e-liquid supply rates and coil temperatures. Gillman et al. [13
] stated that e-cigarette devices with steady e-liquid supplies to the coil generated the lowest amounts of formaldehyde.
In addition to the device construction, variations in aldehyde emissions from flavored e-liquids might be affected by the differences in boil points, evaporation rates, and thermal decomposition rates [13
]. But, the reasons for the differential carbonyl formation patterns across different flavoring agents are cannot be explored completely because flavor manufacturers usually do not disclose the ingredients in their products [53
]. Based on partially revealed information from one vendor (The Perfumer’s Apprentice), the flavoring agents consist of PG, water, ethyl alcohol, and natural/artificial flavoring chemicals. Pyrolysis of flavoring chemicals was known to be the major source of carbonyls in e-vapor [19
]. In addition, PG in flavoring agents might also contribute to aldehyde formation. However, current knowledges on carbonyl formation from flavored e-liquids are not fully understood. Further studies of thermal degradation of flavoring chemicals are warranted to better understand the contribution of flavoring agents to carbonyl formation.
Diacetyl concentrations observed in our samples are comparable to those reported in previous studies [9
]. Even though many e-liquid manufacturers use acetoin as a safe alternative of diacetyl and acetylpropionyl, diacetyl could be found in e-vapor due to the use of natural flavors containing diacetyl and acetoin-to-diacetyl conversion during storage [10
]. Based on the diacetyl concentrations we measured, assuming 200 puffs/day of e-cigarette vaping on average, daily diacetyl exposure levels of butter-like flavored e-cigarette users (4.22–17.28 g/day) were much lower than the reported threshold level. In a rodent in vivo study, diacetyl exposures (100 ppm) to 6 h per day for 12 weeks (equivalent to 6.82 mg/day for 12 weeks assuming 0.2 mL mean tidal volume and 250 breaths/minute) caused nasal injury and peribronchial lymphocytic inflammation [55
]. However, considering uncertainties in animal-to-human extrapolation and extreme e-cigarette users (e.g., cloud chasers using sub-ohm e-cigarettes with intensive use, >1000 puffs/day), potential health impact of chronic low-level diacetyl exposures should be further accessed by the regulatory authorities.
In addition to the diacetyl, other flavoring chemicals (e.g., vanillin and cinnamaldehyde, etc.) could promote adverse health outcomes. An in vitro study demonstrated that vanilla and cinnamon flavored e-liquids had three and ten-fold lower no-observable-adverse-effect-level (NOAEL) doses (0.1–0.01% dose) than VG only e-liquid (0.3% in culture media), respectively [7
]. The cytotoxicity of cinnamaldehyde (IC50
= 0.037–0.04 mM) was approximately 100 times higher than that of vanillin (IC50
= 2.5–4 mM) [8
]. The impact of flavoring chemicals on human health need to be further studied using real-world relevant doses, such as presented in this study, because the flavoring chemicals have been identified as one of the most concerning chemicals found in e-cigarette emissions [56
Increased puff volumes with a fixed puff durations were shown not only to increase the amounts of e-vapor passing through the heating coil but also to decrease its temperature due to increased flow rates [23
]. Puff volumes and puff durations determine the volume of air and its flow rate passing through the e-cigarette heating coils. The significant differences in carbonyl emissions measured between 35 mL and 90 mL puff volumes observed in our study might be due to the increased e-vapor masses. However, the carbonyl composition might also be affected by coil temperature changes. The formation of diverse carbonyl species under different air flow regimes might indicate that the changes in coil temperature could affect the thermal degradations of the e-liquid components.
Our study makes an important contribution to the literature by using vaping topographies based on real-world user vaping patterns. As noted earlier, the puff volumes used in most previous e-cigarette vapor emissions studies were based on regular cigarette smoking topographies and were usually much lower than that of e-cigarette users. The short puff durations used in previous studies (≤2 s) might be insufficient to heat up the heating coil to evaporate e-liquid [23
] and thus may have underestimated potential exposures to toxic carbonyl emissions from e-cigarette vapor during real-world usage.
Carbonyl exposure distributions for e-cigarette and conventional cigarette were estimated using the Monte Carlo method (Figure 4
). Input parameters were the observed e-cigarette emission data in this study (Table 2
and Tables S4 and S5
), reported cigarette smoke carbonyl levels (Table S6
], and daily e-cigarette and cigarette use patterns [31
]. The distribution of carbonyl exposures associated with recent generation e-cigarettes with bottom coil setting were compared to the exposures from conventional cigarette smoking (Figure 4
and Table S7
). Daily average acetaldehyde, diacetyl, and acrolein exposures from e-cigarette were approximately 100, 125 and 21 times lower than conventional cigarette, respectively, with little to no overlap of the exposure populations. However, e-cigarette users could be exposed to 2- and 4-fold higher formaldehyde and glyoxal in a day than cigarette smokers, respectively, and near complete overlap of the distributions. Given the daily exposure estimates, e-cigarette users should aware that e-cigarette might be less effective harm reduction product when they employ vaping conditions that resulted in high carbonyl formation (e.g., top-coil device, high power output, PG e-liquid, and large flavoring additives, etc.). In addition, e-cigarette vaping is still expected to pose potential health risks due to the non-threshold characteristics of carcinogenic carbonyls (i.e., formaldehyde and acetaldehyde) and should not be considered harmless.
It is also worth mentioning that, as noticed above, e-cigarette users are prone to be exposed to more glyoxal than cigarette users. Glyoxal has been identified as an occupational allergen among health care workers who use glyoxal containing disinfectants [6
]. An in vitro study showed that glyoxal was shown to deplete glutathione, increase the production of reactive oxygen species (ROS), and induced cell damage to isolated rat hepatocytes [59
]. Higher device power outputs could increase glyoxal exposures and via the mechanisms listed above induce airway oxidative stress. Since there is no such study of health impacts of e-cigarette glyoxal exposure, it needs to be further evaluated for better harm reduction.
Moreover, carbonyl compounds in e-vapor were shown to form secondary harmful chemicals. Autoxidation of acetoin, which is a safer alternative of ‘butter’ flavoring chemicals (i.e., diacetyl and acetylpropionyl) could form diacetyl during e-liquid storage [54
]. Further, acrylamide is neurotoxic and has potency to cause cancers in the reproductive and endocrine systems [60
]. The reaction between acrolein, which is known to present in e-cigarette vapor, and amino acid or ammonia could form acrylamide [61
], but the formation of acrylamide has not been studied in e-vapor. In addition to acetoin and acrolein, other precursor chemicals may present in e-vapors. Future research needs to access the formation of secondary air toxics induced by e-vapor.
Even though we thoughtfully identified large numbers of carbonyl compounds induced by various e-cigarette vaping conditions, this study still has several limitations. First, the DNPH cartridges were designed for the gas phase carbonyl sampling rather than particle phase carbonyls [40
]. Carbonyl collection efficiencies for the e-vapors using the DNPH cartridges might be lower than the labeled efficiencies for the gas phase carbonyls because carbonyls in e-vapors are reported to be present in both gas and particle phases [15
]. Second, our analytical method might have underestimated unsaturated aldehydes and ketones. Unsaturated carbonyls, such as acrolein, crotonaldehyde, and cinnamaldehyde, and DNPH adducts could further react with additional DNPH to form side products [62
]. Further studies also need to test additional carbonyl sampling methods such as N-methyl-4-hydrazino-7-nitrobenzofurazan, 4-(2-aminooxyethyl)-morpholin-4-ium chloride or DNPH-hydroquinone methods.