E‑Liquid and Aerosol Characterization of Popular Disposable E‑Cigarettes

Abstract

Disposable electronic cigarettes are one of the most popular forms of nicotine consumption among both adults and adolescents. Yet, e-liquid composition and the inhaled aerosol emissions from these devices are under-characterized. This work investigated 25 disposable e-cigarettes of various flavors from four popular brands (Flum Pebble, Elf Bar, Esco Bars, and Geek Bar) using a combination of gas chromatography, mass spectrometry, and liquid chromatography, accurate mass spectrometry. We quantified (1) e-liquid composition, including concentration and identity of added organic acids, concentration and R/S enantiomeric identity of nicotine, and the flavor and coolant content, and (2) aerosol mass formation and aerosolized carbonyl species formed in the vaping process. Of the 25 devices tested, 10 (40%) had significantly lower nicotine concentration compared to their labeled nicotine concentration. All studied devices contained the natural (S)-(−)-nicotine enantiomer. Benzoic acid was the major organic acid in all products; however, seven products across all brands also contained notable concentrations of levulinic acid. The benzoic acid to nicotine molar ratios varied greatly (range of 0.4–4) with no clear correlation to the brand or flavor profile. All products examined featured high concentrations of the synthetic coolant WS-23, comprising between 1% and 7% of the e-liquid. The flavor chemicals vanillin, ethyl maltol, triacetin, and (3Z)-3-hexen-1-ol were found in most devices. Concentrations of six additives (WS-23, WS-3, vanillin, ethyl maltol, menthol, and limonene) exceeded the threshold for toxic effects. Both the e-cigarette design and e-liquid formulation affected carbonyl yields in the aerosol. Interestingly, aerosol mass production was inversely correlated with carbonyl yields. Carbonyl yields were not found to correlate to the flavor additive or nicotine concentration in the tested devices. The Esco Bar brand produced the greatest yield of harmful or potentially harmful carbonyls in the aerosol, while Flum Pebble produced the least, often with high variability both between and within brands.

1. Introduction

Electronic (e-) cigarettes are the most commonly used nicotine delivery products among U.S. youth. − There are hundreds of e-cigarette devices available on the commercial market, with disposable e-cigarette devices accounting for 94% of all sales. Disposable e-cigarettes are composed of a nonrefillable e-liquid cartridge consisting of a coil and wick saturated in a mixture of nicotine salts (nicotine + organic acid), solvent base (most commonly a mixture of propylene glycol (PG) and vegetable glycerin (VG)), and proprietary mixtures of coolants and flavors. They are meant to be discarded after a certain number of puffs or when the e-liquid reservoir is empty. In 2020, the U.S. Food and Drug Administration (FDA) banned flavor chemicals to reduce the appeal of e-cigarettes to youth, but only for refillable devices. Flavored disposable e-cigarette products, primarily imported to the U.S., rapidly filled this vacuum in the market and are now the most abundant products on the market. , Brands of disposable e-cigarettes Flum Pebble, Elf Bar, Esco Bar, and Geek Bar dominated market sales in the past few years and continue to grow in popularity. ,− These devices have not been fully characterized by regulatory agencies like the U.S. FDA, so there are many unknowns regarding their content and health effects.

A major concern for e-cigarettes is their capability to deliver a high dose of nicotine (20 to 60 mg/mL), which can lead to dependence. Despite the known risks, there are many factors that prevent consumers from making informed purchasing decisions. There is inconsistent enforcement of labeling regulations, and even with labeling regulations, some products deviate from the nicotine content they are advertised to contain. − Further, previous research found the inclusion of synthetic nicotine in the disposable brand Puff Bars, with a mix of the natural S-(−)-nicotine enantiomer and synthetic R-(+)-nicotine enantiomer. − Inclusion of the R-enantiomer in the device carries additional risk as its safety has not been fully characterized. Most products do not list the source or type of nicotine.

Several developments facilitate the incorporation of high nicotine concentrations in e-cigarettes. The first of which is the use of organic acids to produce nicotine salts, producing an aerosol (comprised of both gaseous and particulate constituents) that is less harsh to the user. These organic acids have been shown to significantly modify reactive oxygen species (ROS) formation during vaping. The type and concentration of organic acid used is currently not required to be labeled on the disposable products in the US. Disposable e-cigarettes also frequently have additives including synthetic coolants such as WS-23 (2-Isopropyl-N,2,3-trimethylbutanamide) and WS-3 (N-Ethyl-p-methane-3-carboxamide), which further reduce the harshness of high nicotine concentrations, often in an equal to or greater amount than the nicotine concentration itself. , The toxicological effects of these synthetic coolants through inhalation exposure have not been well studied. Research has found there could be potential cytotoxic effects upon vaping along with the production of additional ROS, , but there is no clear consensus about their inhalation safety. Furthermore, commercial e-cigarettes generally feature select flavor additives as major components, often at greater than 1 mg/mL in the e-liquid, that may also have health concerns. ,,− Common examples are vanillin, ethyl maltol, (3Z)-3-hexen-1-ol, and menthol; yet, individual flavoring chemical additives are rarely, if ever, listed on the label. Flavor chemicals can also degrade into carcinogens (e.g., formaldehyde, acrolein) and other toxic chemicals (e.g., glyoxal, butyraldehyde, propionaldehyde), , which add significant health risks. To date, the chemical composition and toxicity of disposable e-cigarette aerosols have been characterized from brands such as Puff Bar, Hyde, Ezzy, SEA, and Air Bar. − However, systematic assessments of the e-liquid ingredients and composition of the inhaled aerosols for some of the most popular disposable e-cigarettes on the commercial market are still lacking.

This work examines the composition of the e-liquids and aerosols of products from four popular e-cigarette brands: ,− Flum Pebble, Elf Bar, Esco Bar, and Geek Bar. The aerosol composition will largely mirror the e-liquid composition, with the additional contribution of aerosolized carbonyls produced during the vaping process. − We specifically aim to answer the following questions: (1) How does the labeled nicotine concentration compare with the measured concentration? (2) What enantiomers of nicotine are included in these products? (3) What types of nicotine salts are used in these products? (4) What flavor chemicals are frequently found in these products, and what concentrations of major flavor chemicals are used? (5) How much aerosol mass is produced by these devices? and (6) What is the carbonyl yield for these specific devices? Combined, these analyses provide insights into the safety of popular disposable e-cigarette products for consumers, risk assessors, and researchers.

2. Experimental Section

2.1. Product Selection

Disposable e-cigarette products of various flavor profiles were selected from four popular brands for study: ,− Flum Pebble, Elf Bar, Esco Bars, and Geek Bar. The products were purchased from various online retailers in 2024. A range of advertised puff counts from 2500 to 15,000 was studied (Table S1). Five Flum Pebble products (“Clear”, “Strawmango”, “Watermelon Lemon Mint”, “Vanilla Ice Cream”, and “White Gummy”) were advertised to deliver 6000 puffs per device. Four Elf Bar products (“Clear”, “Watermelon Ice”, “Blueberry Tobacco”, and “Blue Cotton Candy”) were advertised to deliver 5000 puffs per device. Two types of Esco Bars were examined: Esco Bars 2500 and Esco Bars 6000. Eight Esco Bars 2500 products (“Clear”, “Peach Ice”, “White Gummy”, “Tropical Rainbow Blast”, “Salted Caramel”, “Bubblegum Ice”, “Cotton Candy”, and “Blue Razz Cotton Candy”) were advertised to deliver 2500 puffs per device. The Esco Bars 2500 devices were the only nonrechargeable devices among products tested. Three Esco Bars 6000 products (“Jungle Juice”, “Peach Watermelon”, and “Spearmint”) were advertised to deliver 6000 puffs per device. Five Geek Bar products (“Strawberry Banana”, “Watermelon Ice”, “Orange Creamsicle”, “Miami Mint”, and “B Pop”) were advertised to deliver 15,000 puffs per device. The “Clear” flavor was advertised to deliver a simple “clean”, “crisp”, or “icy” flavor profile; some products label “Clear” as “Unflavored”. All devices were stored in the dark at room temperature until analyses. Given documented effects of coil aging on carbonyl formation, a new disposable device was used for each replicate analysis to minimize prolonged usage effects. For each device brand and flavor, 2–3 replicates were analyzed, depending on product availability.

2.2. Analysis of Flavor Chemicals by GC–MS

Disposable e-cigarettes were dissected to access the chemical composition of the e-liquid saturated wicks. The wicks were removed, and the e-liquid was extracted and diluted 1:100 with HPLC-grade methanol (≥99.9% from Sigma-Aldrich) and transferred to an autosampler vial for analysis on GC/MS using an Agilent 6890N gas chromatograph coupled to an Agilent 5973N quadrupole mass spectrometer (Agilent Technologies Inc., Santa Clara, CA) and a DB-Wax column (Agilent Technologies Inc., Santa Clara, CA, 30 m long, 0.25 mm i.d., 0.25 μm film thickness). A 1.0 μL aliquot was injected into the GC/MS with a spitless injection at an injector temperature of 260 °C. The GC temperature program for the analyses was as follows: 45 °C for 0.5 min, 9 °C/min ramp to 160 °C and hold for 3 min, 7 °C/min ramp to 200 °C and hold for 10 min. Flavor chemicals were identified by their mass spectrometry fragmentation pattern in the NIST 02 library based on quality scores above 75.

Major flavor chemicals were quantified by using authentic chemical standards in methanol. Vanillin, triacetin, linalool, limonene, menthol, α-terpineol, and menthone were purchased from Sigma-Aldrich (St. Louis, MO). WS-23, WS-3, ethyl maltol, and (3Z)-3-hexen-1-ol were purchased from Tokyo Chemical Industry Co., Ltd. (Portland, OR). Quantification was performed in selected ion monitoring mode (SIM) with unique ions selected for each flavor analyte (Table S2).

2.3. Collection and Analysis of Nicotine Concentration and Enantiomer

Nicotine concentration in the disposable e-cigarettes was quantified by GC/MS in selected ion monitoring mode (SIM) using the following ions: 84, 133, and 162. Analysis was performed using an HP5-MS column (Agilent Technologies Inc., Santa Clara, CA, 30 m long, 0.25 mm i.d., 0.25 μm film thickness) with an inlet temperature of 250 °C, injection volume of 1 μL, and with the following oven programing: 70 °C for 2 min, 20 °C/min ramp to 230 °C, hold for 1 min.

Evaluation of nicotine enantiomers was performed using a chiral Beta Dex 120 GC column (Supelco, Inc., Bellefonte, PA, 30 m, 0.25 mm ID, 0.25 μm film) with oven parameters modified from an existing method. Briefly, 1 μL of diluted e-liquid was injected onto the column in splitless mode with an inlet temperature of 250 °C, with a purge flow at 16.1 mL/min. The mobile phase was held at a constant flow rate of 0.7 mL/min. The oven program was as follows: 80 °C for 2 min, 1.5 °C/min to 140 °C, 15 °C/min to 200 °C, and hold for 4 min. Separation of the enantiomers was verified using a racemic analytical standard mixture of (R,S)-(±)-nicotine and (S)-(−)-nicotine (Sigma-Aldrich, St. Louis, MO) in SIM mode using the same ions as mentioned previously (Figure S1A).

2.4. Aerosol Sample Generation and Mass Analysis

Aerosols from disposable devices were generated by applying a vacuum flow of 1.6 ± 0.2 L/min to the devices, which was selected to consistently activate all tested devices during a sampling session. The puffing regimen was controlled by solenoid valves operated by a time relay controller (PTR4-SP, Changzhou Xuchuang Info., Tech., Co., Changzhou, China). A 3 s puff duration and interval of 27 s between puffs was selected for this study, which, together with the tested flow rate, produced a puff volume of 80 ± 7.5 mL. Puff topography is highly variable among e-cigarette users; however, the parameters used in this work are representative of average values observed by typical e-cigarette users within standard deviation (1.5 ± 0.6 L/min flow rate, 3 ± 1 s duration, 15 ± 22 s interval, and 74 ± 52 mL volume) based on a previous clinical study. , Devices were weighed before and after collection with a calibrated Shimadzu microbalance to assess the gravimetric mass loss of the e-liquid per 30 puffs. The total aerosol mass per puff was calculated as the difference in the mass of the disposable device divided by the number of puffs that were generated.

2.5. Collection and Analysis of Carbonyls and Small Organic Acids in Aerosols

The methods for the collection and analyses of carbonyls used in this work have been described previously. , Briefly, the total aerosol was sampled through a 2,4-dinitrophenylhydrazine (DNPH) cartridge (350 mg of DNPH, Supelco, Inc., Bellefonte, PA). Separation occurred on an Agilent 1100 HPLC using an Agilent Poroshell EC-C18 (2.1 mm × 100 mm, 2.7 μm, 120 Å) coupled to a linear trap quadrupole-Orbitrap (LTQ-Orbitrap) mass spectrometer (Thermo Corp., Waltham, MA) at a mass resolving power of 30,000 m/Δm at m/z 400. A representative HPLC–HRMS chromatogram for carbonyl-DNPH is shown in Figure S1B.

Formaldehyde–, acetaldehyde–, acetone–, acrolein–, propionaldehyde–, crotonaldehyde–, methacrolein–, 2-butanone–, butyraldehyde–, benzaldehyde–, valeraldehyde–, and hexaldehyde–DNPH hydrazones were quantified using commercial analytical standards (obtained from AccuStandard, New Haven, CT). Acetic acid and glycolaldehyde DNPH hydrazone standards were synthesized as described previously. All other target carbonyls were quantified using theoretical calculations of the Gibbs free energy of deprotonation (ΔG _d) of the carbonyl–DNPH hydrazones in the negative mode electrospray ionization (ESI). , The relative ΔG _d of targeted hydrazones was converted to ESI sensitivity by applying the linear regression relationship of ΔG _d vs measured sensitivity for synthesized and commercial standards that was updated regularly. Organic acids were captured unmodified in silica gel and quantified by HPLC–HRMS. Collection cartridges were weighed before and after to determine the total aerosol collected. Thirty puffs were collected on each cartridge such that the derivatization agent remained in excess for all products excluding the Flum Pebble Vanilla Ice Cream and Elf Bar Blue Cotton Candy, where only 10 puffs were collected to maintain the excess derivatization agent. Concentrations of carbonyls were normalized by the amount of aerosol collected on the cartridge. Blank DNPH cartridges were collected by sampling lab air for the equivalent volume of 30 puffs and subtracting carbonyl concentrations from experimental samples.

3. Results and Discussion

3.1. Nicotine and Organic Acids (Nicotine Salts) in the Disposable E-Liquid

Table compiles data on nicotine and organic acids for the disposable e-cigarette products studied (data shown in Figure A). The standard deviations obtained for duplicate (0.7–7%) and triplicate (0.5–6%) determinations for nicotine concentrations of the same device brand and flavor are relatively low. However, this error represents a precision error and not necessarily the total quantification error in disposable devices. Different e-cigarette devices have varying organic acid concentrations which change the neutralization ratio and potentially affect the quantitation of nicotine by GC/MS. To ensure accuracy of nicotine quantification across variable sample matrices, we performed a spike recovery of the same nicotine concentration in PG/VG e-liquid with a range of added benzoic and lactic acid concentrations: 0% acid (freebase nicotine) and nicotine to acid ratio ranging from 2:1 to 1:4 (Figure ), which spans the range observed in devices (expanded on below). As the acid concentration increases, nicotine recovery deviates from 7% up to 20% higher than the expected value (Figure A). This is accompanied by a decrease in e-liquid pH (Figure B), measured according to standard protocols used for e-liquids. Given that the matrix-dependent recovery uncertainty for nicotine quantification with benzoic acid salts (≤20%) was consistently larger compared to the analytical standard deviation (≤7%), we used 20% deviation as the threshold to assess the accuracy of the reported nicotine concentrations of the 25 devices.

1. A Compilation of Measured Nicotine and Major Organic Acid Concentrations in the Disposable e-Cigarette Products, Labelled Nicotine Concentration from Product Packaging, and Observed Enantiomeric Identity of Nicotine .

		nicotine conc.	labeled nicotine	benzoic acid conc.	levulinic acid conc.	nicotine enantiomer
brand	device flavor	(mg/mL)
Flum Pebble	clear	56.0 (±3.7)	50	15.8 (±0.9)	0.023 (±0.002)	S(−)
	strawmango	59.4 (±4.4)	50	37.5 (±16.7)	0.097 (±0.019)	S(−)
	watermelon lemon mint	16.2 (±1.2)	50	18.7 (±2.2)	6.41 (±1.41)	S(−)
	vanilla ice cream	27.9 (±2.0)	50	39.4 (±3.1)	0.244 (±0.110)	S(−)
	white gummy	34.7 (±2.5)	50	34.9 (±1.1)	0.045 (±0.014)	S(−)
Elf Bar	clear	53.6 (±3.9)	50	31.1 (±14.8)	0.056 (±0.004)	S(−)
	watermelon ice	51.1 (±3.7)	50	65.3 (±7.0)	0.169 (±0.021)	S(−)
	blueberry tobacco	31.6 (±4.0)	40	76.0 (±4.8)	1.14 (±0.07)	S(−)
	blue cotton candy	31.2 (±2.5)	50	5.3 (±2.4)	0.181 (±0.003)	S(−)
Esco Bar	clear	53.5 (±3.9)	50	77.8 (±23.8)	0.106 (±0.017)	S(−)
	peach ice	50.5 (±4.0)	50	45.8 (±11.7)	0.035 (±0.008)	S(−)
	white gummy	25.8 (±1.9)	50	24.9 (±2.4)	0.035 (±0.008)	S(−)
	tropical rainbow blast	25.0 (±1.8)	50	47.9 (±19.9)	0.095 (±0.005)	S(−)
	salted caramel	50.7 (±11)	50	148.9 (±39.7)	2.57 (±0.10)	S(−)
	bubblegum ice	57.8 (±4.2)	50	64.3 (±6.3)	0.071 (±0.002)	S(−)
	cotton candy	37.0 (±2.7)	50	24.2 (±2.7)	0.062 (±0.012)	S(−)
	blue razz cotton candy	61.2 (±4.2)	50	16.0 (±0.1)	0.0434 (±0.001)	S(−)
	jungle juice 6000	59.9 (±4.2)	50	18.1 (±0.5)	0.0238 (±0.001)	S(−)
	peach watermelon 6000	64.9 (±4.7)	50	25.8 (±0.2)	0.0230 (±0.001)	S(−)
	spearmint 6000	16.0 (±1.2)	30	8.0 (±0.4)	0.0269 (±0.003)	S(−)
Geek Bar	strawberry banana	36.0 (±2.7)	50	7.1 (±5.6)	0.064 (±0.002)	S(−)
	watermelon ice	21.5 (±1.7)	50	39.0(±0.5)	14.18 (±0.09)	S(−)
	orange creamsicle	27.2 (±2.0)	50	11.3 (±1.3)	0.024 (±0.002)	S(−)
	miami mint	24.8 (±3.0)	50	58.3 (±0.4)	0.014 (±0.0012)	S(−)
	B Pop	29.8 (±2.2)	50	29.9 (±6.5)	7.01 (±2.14)	S(−)

^aUncertainties represent 1 standard deviation of replicates, or, in the case of nicotine, matrix-dependent recovery errors as shown in Figure (if greater than the analytical standard deviation).

4.

Chemical characterization results for both e-liquid and aerosols from 25 disposable devices: (A) nicotine and organic acid concentration in the e-liquid, (B) concentration of major flavor chemicals in the e-liquid, (C) aerosol mass per puff produced from each device, and (D) carbonyl mass yield in the aerosol. Carbonyls labeled as “Others” in panel D are the summed yields from those not explicitly labeled in the figure but are tabulated in Table S4.

1.

Percent recovery of spiked nicotine (20 mg/mL) into e-liquids with various acid contents: (A) freebase nicotine and benzoic acid (BA) and lactic acid (LA) with acid:nicotine ratios ranging from 1:2 to 4:1. (B) The corresponding measured pH of the e-liquid with the acid:nicotine ratios specified. The horizonal dotted line in subplot A represents a 100% recovery.

Most of the e-cigarettes tested in this work were labeled as containing 50 mg/mL nicotine (equivalent to 5% by mass), except for the Elf Bar Blueberry Tobacco and the Esco Bars Spearmint 6000 which were labeled as 40 and 30 mg/mL (4% and 3% by mass), respectively. This concentration of nicotine is common for disposable e-cigarettes across different brands, including a market leader JUUL. , Our analysis of e-liquids showed significant variation in the concentration of nicotine in these products (Figure A). Using a theoretical nicotine ratio of 1 (±0.2) and the quantification uncertainty for each product, we found that 12 of the 25 products tested matched their labeled nicotine concentration when considering the uncertainty (±20%). In contrast, 13 products contained less nicotine compared to their labeled amount, and 10 out of 25 devices (40%) were significantly lower compared to their labeled nicotine content (p < 0.05, using one-way ANOVA).

2.

(A) Ratio of measured versus labeled nicotine concentration and (B) benzoic acid to nicotine ratio of Flum Pebble (yellow), Elf Bar (blue), Esco Bars (pink), and Geek Bar (teal) devices. The horizontal line in (A) denotes the theoretical unity ratio. Asterisks in (A) represent significant difference (p < 0.05) between the measured ratio (within analytical uncertainty) compared to unity within the maximum error bounds obtained from nicotine recovery (±20%) at all nicotine:acid ratios (Figure ) using 1-way ANOVA.

All four brands in this study had multiple products with a lower nicotine concentration than labeled. There is no clear pattern for the discrepancy in nicotine concentration, suggesting that quality control might be a factor. However, every Geek Bar product tested was roughly half of its labeled content. While we cannot make claims regarding manufacturer’s intent, Dawkins et al. found that products with lower nicotine concentrations lead to users purchasing more nicotine products. This phenomenon might be due to the observation that users self-regulate (“titrate”) their nicotine consumption, i.e., use nicotine products until they reach a fulfilling sensation from the nicotine exposure. ,

All products tested in this work incorporate benzoic acid to make a nicotine salt (i.e., nicotine benzoate; Figure B). In addition to benzoic acid, some devices also had low levels of levulinic acid. Flum Pebble Watermelon Lemon Mint, Esco Bar Salted Caramel, Geek Bar Watermelon Ice, and Geek Bar B-Pop had levulinic acid concentrations that exceeded 2.5 mg/mL. The concentration of benzoic acid in these products is quite diverse, ranging from approximately 8 to 150 mg/mL. Although lactic acid was found in many popular refillable e-cigarette products, we did not observe lactic acid in any disposable products in this study. ,, The benzoic acid to nicotine ratio may have health-related implications as Tran et al. observed significantly increased formation of free radicals in aerosols from benzoic acid nicotine salt e-liquids compared to freebase nicotine e-liquids, which increased proportionally with benzoic acid concentration. In contrast, levulinic acid and lactic acid nicotine salts did not produce aerosols with significantly different free-radical production compared to that of freebase nicotine. We also observed a benzoic acid to nicotine ratio that was lower than 1:1 for several products across all brands, which implies incomplete nicotine neutralization. The reason for this is not clear.

The enantiomeric identity of nicotine is important for determining the overall risk associated with using a specific device. Our analysis of these products contained solely S-(−)-nicotine in the e-liquids (Figure ). While this finding does not indicate that the nicotine in these devices came from natural sources, as synthetic procedures can produce a pure (S-enantiomer, it does clarify the risk associated with these disposable devices only associated with (S)-(−)-nicotine, which has risks that are more well-characterized than that of its R-(+) counterpart.

3.

GC–MS chromatogram of select devices’ nicotine enantiomer content. Peak sizes have been scaled to show clarity. All devices in this study had a peak eluting around 43.54 min which is indicative of S-(−)-nicotine. Dotted lines have been added to indicate the peak for each enantiomer.

Figure presents the e-liquid and aerosol chemical compositions for the disposable e-cigarette products studied in this work (tabulated data are presented in Tables , S3, and S4). Absolute concentrations of nicotine and major organic acids in each device are shown in Figure A.

3.2. Major Flavor Chemicals in the E-Liquid

The popularity of flavored disposable devices has led to a diverse range of flavor options appearing on the market. Flavor chemicals have been characterized for several brands of e-cigarettes but not yet for the brands in this study. ,,, We found that several flavor chemicals and coolants were common across numerous disposable products (Figure B, Table S3) and will qualitatively discuss their health implications at the measured concentrations. The synthetic coolant WS-23 was found in every product tested, sometimes at higher concentrations than nicotine, and the coolant WS-3 was found in 6 of the 25 devices tested (24%). This finding corroborates other reports in the literature of synthetic coolant use in e-cigarettes. , WS-3 has been described as having a similar but milder effect as menthol due to their activation of the same receptor in the body. , As such, it is plausible that the incorporation of WS-3 is used for flavor profile effects, while WS-23 is likely added for cooling purposes. Both coolants have additional health concerns. Wong et al. found that WS-23 can cause degradation of proteins associated with cell cytoskeleton and several signaling pathways. This can lead to lower wound healing and pulmonary diseases like COPD. , Additionally, doses much lower than those seen in e-cigarette aerosols were able to produce cytotoxic lesions in rats. Both WS-3 and WS-23 exhibit pharmacological activity at the TRPM8 receptor, which facilitates their cooling and anti-inflammatory effects, suggesting these coolants can mask the irritating effects of nicotine and other toxic chemicals, thus promoting chronic usage and consequently potentially leading to higher exposures.

A major recurring flavor chemical was vanillin. It was present in 19 out of 25 products (76%) and ranged in concentration from ∼0.4 to 50 mg mL^–1. Beyond imparting a vanilla flavor, vanillin can be included in e-liquids to achieve a flavor profile that is sweet, sugary, or spicy. There appears to be no clear association between a product’s flavor description and the inclusion of vanillin, which indicates that vanillin is used as a flavor enhancer or to introduce additional sweetness. There are several health concerns regarding the inclusion of vanillin in disposable e-cigarettes. Abouassali et al. found that vanillin could have significant cardiac toxicology in e-cigarette aerosols. Direct effects on human bronchial cells have also been observed including cytotoxicity, metabolic disruptions that can lead to diseases like cystic fibrosis, oxidative stress, and other acute effects. These studies tested vanillin in both the vaped and unvaped forms. The concentrations used in these studies (2.8 mM or 0.43 mg mL^–1) are smaller than the lowest concentration of vanillin observed in the disposable devices in this work; in fact, vanillin used in disposable e-cigarettes can be a factor of 100 greater in concentration than the dose that elicits in vitro toxicity responses. Vanillin also contains an aldehyde group that can react with the solvents PG and VG to form acetals capable of disrupting cellular function. − The formation of acetals and additional products is complex and depends on all other chemical species in the e-liquid, including nicotine and benzoic acid concentrations. Accordingly, the inclusion of vanillin in these disposable e-cigarettes represents a significant toxicological risk to users.

Ethyl maltol was detected in 20 out of 25 devices (80%) making it another major flavor chemical. Often used in conjunction with vanillin, ethyl maltol can impart a sweet or fruity flavor. This is supported in our work, as 18 products (72%) had both vanillin and ethyl maltol. The concentration of ethyl maltol was much lower than that of vanillin and ranged from ∼0.7 to 14 mg mL^–1. As with vanillin, there are negative health effects associated with exposure to ethyl maltol in e-cigarette aerosols. Exposure can lead to an increase in inflammation markers in lung cells, ,, formation of ROS, − and metal-mediated cancer and cytotoxicity. , This last factor is especially concerning given that metals could leach into the e-liquid − and complex with ethyl maltol, which can then be transferred in the aerosol. Ethyl maltol–copper complexes are especially toxic to lung cells. The concentration of ethyl maltol alone in these devices (4.34 to 101 mM) is sufficient to induce cytotoxicity and oxidative stress based on previous studies.

Another chemical of interest was triacetin, which was found in 13 out of 25 products (52%). The concentrations were in general much lower than those of vanillin, ranging from ∼0.2 to 33 mg mL^–1. The highest concentration of triacetin was found in the Geek Bar Strawberry Banana product, which was much higher relative to the other 12 products. Triacetin is thought to be added as an additional solvent that helps surface tension and delivery other flavor chemicals. Triacetin on its own is seemingly benign, but exposure to its vape aerosols carries significantly more risks. Vreeke et al. found that aerosols from e-liquids containing triacetin produced higher levels of aldehydes and hemiacetals. They propose that the thermal reactions in the vaping process cause triacetin to degrade into these toxic products. This is perhaps the clearest example of the aerosolization process in these products transforming safe components into toxicants. Our results do not show a clear correlation between diacetyl in triacetin in the tested products, potentially due to product heterogeneity, as diacetyl can be introduced in the e-liquid or produced from multiple chemical precursors.

Terpenes such as limonene, linalool, and menthol are commonly found in disposable e-cigarette products as additive flavoring agents. , Our study was generally in line with these results as linalool was found in 11 of 25 products (44%) and menthol was found in 10 out of 25 products (40%). The other terpenes found in this study were limonene, α-terpineol, and menthone. Each of these were found in fewer than 5 products and were all at lower concentrations (<1 mg mL^–1). Menthol was frequently detected at much higher concentrations (0.2 to 10.6 mg mL^–1) compared to the other terpenes as it is a classic ingredient in cigarette − and e-cigarette products and is used as both a coolant and a flavor additive that imparts a minty flavor. Menthol has been heavily investigated due to its frequent use in cigarette and vaping products. ,, Multiple studies have observed biomarkers of impaired respiration such as mitochondrial dysfunction, decreased fibroblast ATP production, DNA damage, and bronchial inflammation. It has also been tied to cytotoxicity at concentrations found in commercial vapes. Cellular dysfunction was also observed in exposures to JUUL products, which have similar concentrations to the devices in this study. The concentrations of menthol alone in these products carry an additional risk of lung cell damage and cytotoxicity. These documented cytotoxic outcomes and the fact that synthetic coolants target the same family of receptors as menthol raise concerns of enhanced respiratory toxicity potentially arising from coexposure to these chemicals.

Bitzer el al. found the addition of linalool could produce hydroxyl radicals in a dose-dependent manner. The disposables in our current study had concentrations of linalool much lower than those in the study of Bitzer et al., indicating the oxidative stress risk from linalool in these devices may be much lower. Additionally, linalool and limonene were found to activate the human TRPA1 receptor which can provoke an irritation response. The concentration of terpenes needed to activate this receptor is in the lower micromolar range, while all disposables tested had concentrations in the low millimolar range, with the lowest being 1.46 mM limonene in the Esco Bar White Gummy device. The exposure of limonene on its own showed cytotoxicity in BEAS 2B and A549 endothelial cells at 1 and 2.8 mM (0.14 and 0.38 g mL^–1), respectively. Terpene concentrations around these values were found in the disposable devices used in this study. Thus, while the levels of terpenes in these products were relatively low compared to those of the other chemical constituents, there is still a non-negligible risk associated with their presence in the product.

The last common flavor in these products was (3Z)-3-Hexen-1-ol. This was present in 19 of 25 products (76%). While not structurally a terpene, it is also used to impart a fruity/grassy flavor similar to limonene and linalool. To our knowledge, the safety of (3Z)-3-Hexen-1-ol has not been well studied in traditional contexts or in the context of e-cigarettes. Thus, further evaluation will be necessary to fully understand any toxicological risks that arise from the inclusion of (3Z)-3-Hexen-1-ol in these products.

3.3. Aerosol Mass Production

Aerosol mass production efficiency (Figure C) from a device is one way to compare device operation and normalize for emissions. Most rechargeable products (all products excluding the Esco Bars 2500) delivered between 11 and 12 mg of aerosol per puff under the vaping conditions in this study. The nonrechargeable Esco Bars 2500 products had the most variable and lowest aerosol masses, suggesting that variations in device design or manufacture may lead to unstable aerosolization. Composition of the e-liquid itself is not likely to be a contributing factor to aerosolization efficiency, as no other brand shows such variability, even across different e-liquid compositions. The lower sample size for other brands besides Esco Bars prevents a definite conclusion of device stability and quality control from being reached.

3.4. Carbonyl Species in the Aerosols

Figure D presents the concentration of major carbonyl species for each device, which are expected to form from degradation in the heated aerosolization process in e-cigarettes, either via direct thermal processes or via oxidative reactions involving ROS such as hydroxyl radicals. , Many factors influence the formation of these degradation products during vaping, including the construction of the device and the e-liquid composition.

Devices within a brand can have similar carbonyl yields despite different flavor profiles and formulations. All Flum Pebbles used in this study had relatively low yields of total carbonyls compared to the other devices (0.34 ± 0.09 μg mg^–1 aerosol) despite having a large diversity in flavor composition. The Elf Bar devices sampled in this study also showed a high level of consistency between products sampled. The three flavored products all had similar total carbonyl levels (0.93 ± 0.15 μg mg^–1 aerosol). The Clear (unflavored) Elf bar had significantly (p = 0.0013) lower aerosol carbonyls (0.22 μg mg^–1) despite having a higher concentration of coolant and flavor chemicals. As synthetic coolants dominated the total flavor additives, this suggests that these coolants may not be the primary precursors of carbonyls in the vaping process compared to other flavor chemicals. Future work is needed to fully understand the potential degradation of synthetic coolants and other flavor chemicals to toxic carbonyls.

In contrast, the Esco Bars devices in this study had much greater variation than those of the other brands examined. There are noticeable differences in the carbonyl yield between the nonrechargeable Esco Bars 2500 and the rechargeable 6000 devices. The Esco Bars 6000 devices had significantly (p = 0.025) lower total carbonyl yields (0.35 ± 0.11 μg mg^–1 aerosol) compared to most of the 2500 products (2.16 ± 0.89 μg mg^–1 aerosol), excluding the Cotton Candy and Blue Razz Cotton Candy flavors (0.64 ± 0.03 μg mg^–1 aerosol). Despite being from the same manufacturer, the 2500 and 6000 series products are notably different in device construction. It seems likely that this is an important factor driving carbonyl production yields. The 2500 series of devices have different product shape design compared to the 6000 and thus feature different internal features, such as heating coils and battery connectors, which result in exposure to different power levels, coil temperatures, and metals that could catalyze degradation reactions. E-liquid composition also affects thermal degradation chemistry as the different e-liquid compositions in the 2500 products showed a notable amount of variability in carbonyl yields. Future research is needed to understand which e-cigarette components, if any, have causative effects on the toxic carbonyl yields.

The Geek Bars have the most novel design compared to the other brands in this study. They have a cosmetic outer LCD screen and internally feature a unique dual mesh coil design (Figure S2). This design did not significantly alter the aerosolization of the device (Figure A). The total carbonyl yield from these devices was similar to those of the other brands; however, there was reasonable variability between the different flavors (ranging from 0.2 to 1.5 μg carbonyl mg^–1 aerosol).

Across device brands, there was a weak correlation between the total flavor chemical concentration and carbonyl yield (r ² ∼0.2). This is particularly interesting given multiple studies identified a positive correlation between flavor chemicals and aldehyde formation. − Discrepancies may be due to differences in the tested devices; i.e., previous studies were conducted on older generation, refillable e-cigarette devices that differ greatly in device construction and e-liquid composition compared to disposable devices. Similarly, there appears to be no correlation between nicotine concentration and carbonyl yields (r ² ∼0.015). The complex formulation and design of these e-liquids make it difficult to identify specific reactions that degrade e-liquid ingredients into carbonyls.

Yet, across device brands, there is a moderately strong negative correlation (r ² ∼0.68) between the carbonyl yield and the aerosol mass generated by the device (Figure ). An r ² value of 0.68 equates to Pearson’s correlation coefficient (r) of 0.82, which is widely used as a measure of sample effect size and associated with Student’s t value of 6.87 for 23 degrees of freedom (n-2, where n = 25), and a corresponding p value of 5.3 × 10^–7, likely to be significant. This provides additional evidence that device construction is a major determinant of carbonyl production, as the ability for the battery to transfer electrical energy to the coils is directly related to aerosolization efficiency. This heated aerosolization process also produces carbonyls, so a relationship is expected, but a negative correlation is not intuitive. Soulet et al. observed that devices with coil resistances >1 Ω were more efficient at using battery power to aerosolize e-liquid. However, the same type of Esco Bar devices, which should have identical coil makeup and resistances, still showed significant variations in carbonyl production and aerosolization yields. We speculate that quality control in the manufacture may play a role. As each device has a thermal capacity, energy applied to the coil that is not effectively used for aerosolization can be lost to alternative processes, such as the thermal or oxidative formation of carbonyls. Thus, it is possible that specific devices have more heat/energy dissipation from the coils due to different electrical contact or connectivity between components. A parallel observation is that a lack of efficiency in thermal combustion of hydrocarbon fuels leads to the formation of oxidative side products. It is also possible that some Esco Bars 2500 devices may use a different coil compared to other products in the same device family (e.g., a subohm coil and/or different metallic composition) despite being labeled identically, representing another quality assurance issue.

5.

Relationship between the observed total carbonyl yield and the aerosol mass generated from each disposable device tested in this study. An x- and y-error weighted linear least-squares fit applied to the data using Origin software gave an r ² coefficient of 0.68.

Acetaldehyde and formaldehyde were the most abundant carbonyl species in the aerosol across the devices in the study. Gillman et al. evaluated a select number of carbonyls and observed that only acetaldehyde increases consistently with the addition of flavor chemicals. Thus, it is possible that major flavor chemicals used in disposable e-cigarettes preferentially form acetaldehyde during the vaping process. Both aldehydes are toxicants to humans, and each carries a cancer risk.

Numerous other toxic and potentially harmful carbonyls were detected in this work, as well. Their concentrations vary greatly from device to device, even within brands. Among these, benzaldehyde is particularly interesting. Benzaldehyde itself is a flavor chemical that can impart a nutty or cherry taste into the products. , Although only detected in 3 intact e-liquids in trace amounts, benzaldehyde was found in the aerosol of most products at a variety of concentrations. The products with benzaldehyde as a flavoring agent (the Esco Bar Bubblegum Ice and Cotton Candy and Geek Bar Strawberry Banana) did not consistently have higher benzaldehyde concentrations in the aerosol compared to products without, whereas the product with the highest benzaldehyde emissions in the aerosol (Geek Bar B Pop) did not contain benzaldehyde in the e-liquid. Thus, it can be concluded that benzaldehyde is formed during the vaping process. Because there is no correlation between any specific e-liquid ingredient and the presence of benzaldehyde, it is possible that the formation of benzaldehyde in the vaping process may depend more strongly on device design (such as coil and battery connector metal, resistance, or other factors).

4. Limitations

This study has several limitations. Products were obtained from the same manufacturing batch, which did not account for the variability in any manufacturing processes with time. Additionally, any study using a single vaping topography will not be representative of the vaping pattern for all users. All devices used in this study were new devices and collections consisted of puffs from early in the device’s lifecycle. Thus, any aging effects of the device are not accounted for nor considered in this study. As the device is used more, coil degradation and dry puffing become more common and can lead to a different toxicological profile. It is expected that coil aging increases carbonyl yields. Analysis of the flavor chemicals in the e-liquids was limited to those that can be detected by the GC/MS method used in the study; it is possible that additional flavor chemicals could be present that affect the thermal degradation chemistry or overall safety of these devices. Lastly, while the collection method for carbonyls in this study is in line with the scientific standard (a cartridge packed with acidic DNPH), the efficiency of this method for its use in the diverse distribution of carbonyls (Figure D) in commercial e-cigarette products has not been validated. Thus, it should be noted that all carbonyl concentrations in this study are only the lower estimate and additional uncertainty should be considered when using the values for a risk assessment.

5. Conclusion

This study investigated the chemical composition of e-liquid and aerosol emissions from popular disposable e-cigarettes, which supported evaluations of the safety of the devices. This is a nontrivial task given the high complexity of the devices and the e-liquids. As the devices and e-liquid formulas grow increasingly more complex with a higher number of additives and electronic components, there is a greater number of factors that can evoke a toxicological effect. The Flum Pebble products tested appear to derive most of their risk from the flavor chemicals present in the e-liquid as they tended to have higher concentrations of flavor chemicals and synthetic coolants, in particular. The Vanilla Ice Cream flavor has the potential to raise toxicological concerns due to high concentrations of vanillin (∼50 mg mL^–1), along with considerable concentrations of ethyl maltol (3.5 mg mL^–1) and triacetin (0.4 mg mL^–1). The carbonyl yield for Flum Pebble devices was notably lower than any other brand in this study (e.g., <0.5 μg/mg aerosol for all Flum devices, compared to 4.1 μg/mg aerosol max for all studied devices). Interestingly, the Flum Pebble Clear (“unflavored”) device had the highest total flavor chemical concentration of any product (96.4 mg mL^–1).

The flavor concentrations in Elf Bar products were also relatively high due to the use of synthetic coolants. The Blue Cotton Candy flavor is concerning due to the high concentrations of vanillin (18 mg mL^–1), ethyl maltol (9.9 mg mL^–1), and triacetin (5.9 mg mL^–1). While the concentrations of coolants and flavor chemicals, along with the device appearance, were relatively similar to the Flum Pebble, the concentrations of carbonyls emitted from these devices were higher (up to 1 μg/mg aerosol). This points to subtle internal differences between the devices that can promote thermal degradation chemistry.

The Esco Bars tested had the greatest amount of variability in e-liquid formulation and aerosol emission and, thus, uncertainty regarding their risk. The acid concentrations in these devices range from incredibly high in the Salted Caramel flavor (∼150 mg mL^–1) to very low in the Spearmint 6000 (∼8 mg mL^–1). These products have lower flavor and coolant concentrations than the other three brands and lower aerosol formation yet unexpectedly the highest levels of carbonyl per mass of aerosol. The concentration of flavor chemicals, namely, ethyl maltol (up to 3.7 mg mL^–1) and menthol (up to 10.6 mg mL^–1), is sufficiently high to trigger adverse effects in the body. All Esco Bars 6000 and some Esco Bars 2500 flavors (“Cotton Candy” and “Blue Razz Cotton Candy”) had carbonyl levels closer to those of the other three brands, further complicating the evaluation of the Esco Bars products. The consistently high levels of acetaldehyde, formaldehyde, and glycolaldehyde contribute significantly to the risk of using these products as all are known carcinogens and pose significant risk to the lungs and other organ systems. ,− Overall, these products merit attention from regulators due to their high degree of carbonyl emissions and variability during the vaping process­(0.2–4.1 μg/mg aerosol). It should be noted that while the Esco Bars 2500 and Esco Bars 6000 are both manufactured by the same company, there are enough differences in device design and aerosol emission that they should be evaluated for safety separately, with greater focus given to the Esco Bars 2500.

The last brand, Geek Bar, was also structurally different from the other brands. The novel coil system did not appear to significantly alter aerosolization efficiency or aerosol carbonyl emissions, compared to other brands. The e-liquid composition of these products was the most unique aspect of the devices. Notably, all products had less nicotine than their 5% label, with 4 out of 5 devices significantly lower (Figure A). The devices use relatively high flavor and coolant concentrations in their formulations, with Strawberry Banana using the second most flavor chemicals of all devices tested. There was a great degree of variability in total carbonyls, which makes the risk for the Geek Bar brand more difficult to define. The Geek Bar Brand has been gaining popularity in the market, and while not as concerning in some respects as the other brands in this study for carbonyl toxicant formation, the mislabeling of the nicotine concentration and relatively high flavor concentration warrant monitoring as the brand develops.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c03167.

• GC–MS and HPLC–HRMS chromatograms for nicotine enantiomeric separation and carbonyl–DNPH separation, respectively; deconstructed devices showing coil structures; advertised number of puffs per device; SIM ions for GC–MS analysis of flavor chemicals; tabulated mass concentrations of flavor chemicals and synthetic coolants; and tabulated mass concentrations of carbonyls and organic acids (PDF)

The authors declare no competing financial interest.