Volatilization and partitioning of residual electronic cigarette emissions to particulate matter

Abstract The growing prevalence of electronic cigarettes (e-cigs) prompts investigation of their impact on indoor air quality. While the investigation of residual cigarette smoke, also known as third-hand smoke (THS), has been increasing, there is minimal literature on similar residue from e-cig emissions. Lab generated ammonium sulfate aerosol was introduced into a stainless-steel chamber which previously contained JUUL electronic cigarette vape. Aerosol composition was monitored using a high-resolution aerosol mass spectrometer. This was performed in the same chamber 3 times over 5 days. Species from electronic cigarette residue (nicotine and vegetable glycerin) were found to partition to particles with decreasing concentration over the multiple experiments while maintaining a consistent chemical signature. This signature matches closely to that of the primary electronic cigarette vape with notable exclusions of propylene glycol, benzoic acid, and cadmium. Electronic cigarette vape can deposit onto surfaces and act as a long-term source of gas phase nicotine which can concentrate on aerosols and increase human exposure. The chemical processing of vape residue observed here is less substantial than has been previously presented for residual cigarette smoke. While our experiments were performed under controlled conditions, they highlight the importance of understanding indoor electronic cigarette use and unintended routes of exposure. Copyright © 2023 American Association for Aerosol Research


Introduction
Third-hand smoke (THS), or residual tobacco smoke, has been recognized as an unintended exposure route to toxins in conventional cigarette (CC) smoke (Winickoff et al. 2009;Matt, Quintana, Destaillats, et al. 2011). Secondhand smoke (SHS) is a more commonly understood unintended route of exposure; however, unlike SHS, THS does not require the presence of a smoker or recent smoking activity for exposure and has a much longer lifetime (Kraev et al. 2009;Matt, Quintana, Zakarian, et al. 2011;Matt et al. 2017). THS consists of CC smoke particles that have deposited onto surrounding surfaces. Once deposited on these surfaces, THS constituents can undergo chemical reactions to form harmful products such as tobacco specific nitrosamines, isocyanic acid, and more (Sleiman et al. 2010;Schick et al. 2014;Borduas et al. 2016;Jacob et al. 2017). Semi-volatile chemical species from THS including nicotine have been shown to evaporate from surfaces and partition to aqueous aerosol particles via an acid-base mechanism, resulting in a route for inhalation exposure and transport through the indoor environment (Collins, Wang, and Abbatt 2018;DeCarlo, Avery, and Waring 2018). About 30% of aerosol mass in an empty, nonsmoking university classroom had a THS chemical signature, suggesting that exposure to THS can occur in public spaces regardless of smoking regulations (DeCarlo, Avery, and Waring 2018). Additionally, measurements in a nonsmoking movie theater in Mainz, Germany demonstrate that these THS species can be transported indoors by smokers (Sheu et al. 2020).
Electronic cigarettes (e-cigs), a cigarette alternative, allow liquid containing nicotine, flavorings, and more (known as e-liquid), to be inhaled as a gas/aerosol mixture (commonly called vape) by using a high temperature filament for vaporization. This nicotine delivery system is thus significantly different from combustion used for CCs, but much like THS, residue from e-cig vape has been found deposited on surfaces and provides a new avenue for human exposure to the chemicals in e-liquids, including nicotine and nicotine derivatives (Goniewicz and Lee 2015;Khachatoorian et al. 2019;Son et al. 2020). Compared to CCs, these devices have fewer indoor use regulations in the United States (ANRF 2022) which raises concern for the accumulation and aging of potentially hazardous residues associated with frequent indoor use. Some surveys have reported on types of indoor spaces and frequency of e-cig use and indicate that nearly 70% of Spanish e-cig users self-report use in restaurants or bars and 32% use in their workplaces (Matilla-Santander et al. 2017) while in the United States over 50% of employees who observed e-cig use around the workplace reported indoor use (Romberg et al. 2021). Although e-cig vape is understood to provide a much simpler chemical matrix than CC smoke, there remain potential health hazards associated with their use. While various studies have sought to characterize direct emissions of e-cigs (Khlystov and Samburova 2016;Klager et al. 2017;Pankow et al. 2017;Behar et al. 2018;Omaiye et al. 2019), this study is concerned with understanding the ultimate fate of e-cig vape and how surface and gas-phase chemistry results in its uptake to aerosols.
Nicotine is dibasic but exists mainly in two forms, free-base (semi-volatile) and mono-protonated (nonvolatile), and increased concentrations of free-base nicotine relative to monoprotonated have been associated with increased harshness of smoke/vape inhalation (Chen 1976). Because of the synthetic nature of e-liquids, manufacturers can adjust the concentration of nicotine and the pH of the liquid to produce a desired sensation. For example, the e-cig company JUUL Labs, Inc. adds benzoic acid (BA) to their liquid to increase the amount of protonated nicotine in an attempt to mimic the feel of smoking CC's (Duell, Pankow, and Peyton 2018), as protonated nicotine (or nicotine salt) is the naturally abundant form in tobacco plants (Seeman et al. 1999) and CC smoke (Pankow et al. 2003). Protonated nicotine will not likely evaporate from particles even through the vaporization process and is more likely than gas-phase free-base nicotine to be exhaled in particles from the e-cig user with subsequent deposition onto surfaces. The very high initial concentrations of nicotine in JUUL e-liquids (59.2-66.7 mg/mL) compared to many other e-liquids (1.6-34.4 mg/mL) (Omaiye et al. 2019) and a sufficiently low pH to render >90% of nicotine mono-protonated (Duell, Pankow, and Peyton 2018;Talih et al. 2019) will by extension increase the mass deposition of nicotine to indoor surfaces compared to other e-liquid formulations. It is anticipated that constituents of deposited e-cig vape undergo a similar acid-base partitioning mechanism to particles as THS (Pankow et al. 1997;DeCarlo, Avery, and Waring 2018), though fewer nicotine degradation products are expected due to the relative simplicity of e-cig aerosol matrix (Crosswhite et al. 2021;Dusautoir et al. 2021). Here, we focus on the evaporation and subsequent partitioning of nicotine and other compounds from e-cig residue with the aim of determining (1) if chemicals from e-cig residue partitions to aerosol particles and (2) the extent to which chemical processing occurs for deposited e-cig vape over time. The investigation of these aims provides new understanding of the long-term impact of e-cig usage to indoor air quality and exposure.

Materials and methods
Experiments were primarily conducted on JUUL e-cig vape from JUUL-brand cartridges (JUUL Pods) that contain e-liquid and a filament which can be easily inserted into a battery unit for vaporization. These pods contain vegetable glycerin (VG) and propylene glycol (PG) as solvents in a 70:30 ratio (Reilly et al. 2019;Talih et al. 2019), 5% nicotine by weight ($59 mg/mL) (Omaiye et al. 2019;Talih et al. 2019), benzoic acid ($44 mg/mL assuming a 1:1 molar benzoic acid/nicotine ratio) (Duell, Pankow, and Peyton 2019), and added flavors (Omaiye et al. 2019). Other experiments were performed with a tank-based e-cig, which requires the user to fill a reservoir with e-liquid prior to vaporization. The e-liquid used for this e-cig also contained VG and PG in a 70:30 ratio, 3 mg/mL of nicotine, and added flavors (Agua Vita flavor; produced by Fate). Methods for the tank-based e-cig experiments are presented in the Supplementary Information.
The JUUL experiments began on day 0 with 4 puffs ($60 mL each, $3-4 s draw time) of mango flavored vape introduced into a cylindrical stainless-steel chamber (37.1 L, 36 cm Â 36 cm; Eagle Stainless, Warminster, PA) via a sterile 60 mL syringe (BD, pn: 309653) over 4 min. This puff regime was performed by hand but was based on the CORESTA recommended method N 81 (CORESTA 2015). The vape was mixed in the sealed chamber for 5 min using a 3 inch diameter internal fan and was followed by flushing with filtered air to remove aerosols and leave only residue remaining which was then left undisturbed until the next day. The chamber was sampled at a combined flow rate of 0.4 LPM (air exchange of 0.65 h À1 ) using a high-resolution time-offlight aerosol mass spectrometer (DeCarlo et al. 2006) (HR-AMS, Aerodyne, Inc. Billerica, MA) to determine aerosol size and composition and a Picarro G2401 cavity ring-down spectrometer (Picarro Inc. Santa Clara, CA) to monitor gas phase CO, CO 2 , H 2 O, and CH 4 concentrations. The HR-AMS was operated in V mode with a resolving power of 2000 (DeCarlo et al. 2006) and calibrations for ionization efficiency were performed using dry 400 nm ammonium nitrate particles. Standard relative ionization efficiencies were applied to all other measured ions since none of the major components contain long alkyl functionalities (Katz et al. 2021). All data was analyzed in Igor Pro 7.08 (WaveMetrics, Inc.) and mass spectral data was processed using SQUIRREL and PIKA software packages for unit resolution and high-resolution analysis, respectively. The experimental layout is displayed in Figure 1 and a chamber diagram is displayed in Figure S1. All experiments were performed at a lab temperature of 19-22 C.
The following procedure was performed 1, 2, and 5 days after vape introduction to the chamber while the AMS and Picarro instruments sampled continuously. First, laboratory air was sampled with a high efficiency particle air (HEPA) filter directly on the AMS inlet for approximately 10-15 min to establish reference airbeam values for subsequent data analysis. Next, the HEPA filter was placed on the intake of the chamber and air was sampled through the chamber for 10 min to confirm there were no leaks in the system and that initial aerosol concentrations were effectively zero. Afterwards, the filter was removed from the inlet and laboratory generated ammonium sulfate (AS) seed aerosol was introduced into the chamber while sampling continued. AS aerosol was generated using a constant output atomizer (TSI model 3086, Inc. Shoreview, MN), dried using a silica dryer, and size selected at 300 nm using a differential mobility analyzer (DMA model 3080, TSI, Inc.). The 300 nm particle size for AS particles was selected to mimic typical particle sizes observed indoors (Avery, Waring, and DeCarlo 2019) and falls in the center of the accumulation mode of particles. Introduction of the laboratory-generated aerosol continued for 20 min into the well-mixed chamber. The chamber inlet was then switched back to the HEPA filter for particle free air introduction and the instruments sampled filtered laboratory air through the chamber until aerosol concentrations reached zero. The chamber was then sealed until the next experimental day. This procedure is displayed pictorially in Figure S2.
In a subsequent experiment, primary JUUL aerosol was also measured. This experiment introduced 60 mL puff of mango flavored JUUL vape to a $2 L Pyrex chamber with a sealed lid and 2 opposing orifices with stopcocks. Approximately 1 min after introduction of the vape, the HR-AMS was connected to the chamber to sample the primary vape. The mass spectra of the primary vape provided a comparison point for the mass spectra of third-hand species in the other chamber experiments.

Results
Total particle mass concentration decrease in the chamber is the combination of sampling loss rate via volumetric replacement with particle free air, and particle deposition to surfaces within the chamber. Assuming first order loss rates, the total particle mass loss rate ranges 1.09 ± 0.07 h À1 for the three residual vape experiments. With the volumetric replacement of  Figure  S1). (g) HR-AMS. (h) Picarro G2401 cavity ring-down spectrometer.
particle free air for a calculated loss rate of 0.65 h À1 we can determine that the particle deposition/impaction rate is 0.44 ± 0.07 h À1 . Example calculations for particle loss in the chamber are presented in the supplementary information. The impacts of surface/gas/ particle interactions in the sampling line were expected to be minimal since the chamber residence time of approximately 90 min was significantly longer than time spent in sampling lines between the chamber and AMS (<30 s) however, the potential impact cannot be excluded from our results. Results of the gas phase species monitored by the Picarro (CO, CO 2 , CH 4 , and H 2 O) are presented in supplementary figure  S4 and show consistency for all experiments.
The mass spectra of directly emitted, or primary, JUUL e-cig vape particles and the mass spectra from residual JUUL vape evaporation and recondensation on laboratory generated AS particles 1 day after deposition are presented in Figure 2a, b, respectively. Note that for ease of presentation the high-resolution analysis performed identifying the intensity of each ion will be presented as the nominal mass of ion rather than the ion identity. Ion identities corresponding to each nominal m/z are given here and in Table S1 for reference.  . Mass spectra of average particle composition over a single day experiment and normalized by total concentration. (a) Signature of primary JUUL vape particles. The inset spectra shows the relative intensities of cadmium isotopes also detected in primary vape. (b) Signature of species partitioned to AS aerosol from JUUL vape residue after 1 day of dwell in the chamber. Average mass spectra from days 2 and 5 are represented in Figure S5. Aerosol mass spectral signal from the AS seed particles (H x S y O z þ ) is not shown.
Nicotine is identified by a characteristic peak at m/z 84 (C 5 H 10 N þ ) in both primary vape and residue experiments (Figure 2a, b). Notable peaks observed in the primary vape but not in the residue signature include m/z 105 (C 7 H 5 O þ ) and 77 (C 6 H 5 þ ), which both correspond to BA. Primary JUUL vape particles also demonstrated the on-line detection of cadmium by Aerosol Mass Spectrometry (Figure 2a). This primary vape particle measurement experiment was repeated with new mint, cr eme, and tobacco flavor pods, and cadmium was detected in the vape particles each time. Cadmium was not detected in the tank-based e-cig experiments. The time series for JUUL vape residue experiments is presented in Figure 2a-c which displays the sum of oxygen-containing organic ions (C x H y O z þ ) and the nicotine tracer (C 5 H 10 N þ ), along with the signal from the laboratory generated AS aerosol (H x S y O z þ , also called sulfate) for days 1, 2, and 5 after e-cig vape deposition. On each sampling day, the sulfate concentration increased and peaked once introduction of AS aerosol was stopped. The concentration then followed an exponential decay function as the aerosol was sampled and replaced with filtered air. The C x H y O z þ species shows a similar increase and decrease over time as the sulfate concentration, indicating rapid equilibrium establishment with the sulfate aerosol on all sample days. Importantly, the absolute and relative concentration of the C x H y O z þ and C 5 H 10 N þ species decreased with each sampling day (Figure 3a-c). Figure 3d-f shows the relationship between nicotine and sulfate and the solvent and sulfate. On the initial day of experiments, VG showed a spike in relative concentration similar to what is observed for nicotine. Excluding the spike, ratios of VG:Sulfate show stability over each day of sampling, and the slopes for each day gradually decrease from positive to slightly positive on days 1 (2.19 Â 10 À4 ± 1.52 Â 10 À5 ) and 2 (5.15 Â 10 À5 ± 8.88 Â 10 À6 ) to a negative slope on day 5 (À1.18 Â 10 À4 ± 4.59 Â 10 À6 ) and slopes similarly decrease for nicotine over each day with 2.22 Â 10 À4 ± 1.58 Â 10 À5 on day 1 to 5.03 Â 10 À5 ± 1.08 Â 10 À5 and À1.07 Â 10 À4 ± 5.14 Â 10 À5 on days 2 and 5, respectively.

Primary vape vs. partitioned aerosol composition
Experiments with both the JUUL and tank-based e-cigs demonstrated re-volatilization and partitioning to AS aerosol particles in all tests performed, indicating the importance of this mechanism for secondary exposure to e-cigarette effluent in the indoor environment. This is true despite the low fraction of free-base nicotine in JUUL vape. For JUUL experiments, both the primary vape signature and that of the partitioned aerosol showed similar chemical distributions, although the relative magnitude of several m/z differ (Figure 2). The two species clearly absent in the partitioned spectra are BA and cadmium. It was expected that BA would deposit to and remain on surfaces because of its negligible vapor pressure in the ionic benzoate form and thus would not be detected in the evaporated residue. Metals species are difficult to measure due to the use of thermal vaporization at 600 C and many metal species will not have sufficient vapor pressure at that temperature to be measured in the gas phase. Cadmium, on the other hand, has relatively low melting and boiling points (321 and 767 C, respectively) and is the likely reason it can be detected. Attempts to identify other metal species were inconclusive due to either mass spectral interferences or the refractory nature of the metal species. Other studies focused on metal detection have found varying concentrations of metals in e-cig (including JUUL) vape (Gaur and Agnihotri 2019;Zhao et al. 2019;Neu et al. 2020). The absence of cadmium from the evaporated residue is also expected since it is solid at room temperature and will remain on surfaces to which the primary particles deposit.
The m/z 45 (C 2 H 5 O þ ) signal intensity was substantially lower during the residual experiments compared to the primary vape sampling whereas the other PG and VG signals were more consistent (Figure 2). Since m/z 45 is the primary fragment of PG (Wallace, NIST Mass Spectrometry Data Center), analysis was performed to determine the relative amount of PG present on the AS aerosol. Pure VG was vaped and sampled by the AMS and a numeric m/z 45 to m/z 61 (C 2 H 5 O 2 þ , primary VG fragment (Wallace, NIST Mass Spectrometry Data Center)) intensity ratio was calculated (pure VG m/z 45/61 ratio ¼ 0.0847). This ratio was used to identify the fraction of the m/z 45 signal from VG, with the remainder assumed to come from PG. This calculation indicated that all m/z 45 signal was due to VG and no detectable amounts of PG were present on the AS aerosol ( Figure S6). Further description and equations are available in the Supplementary Information.

Seed aerosol composition
Over the three days of experiments, the decreased concentration of oxygenated organic compounds and nicotine indicated that the reservoir of deposited residue was depleted over time (Figure 3a-c). On each sampling day the nicotine tracer rapidly peaked before the sulfate concentration, it decreased quickly, then maintained a relatively low but consistent concentration throughout the sampling period. This pattern is consistent with rapid uptake of gas-phase nicotine onto particles, which was likely from an established equilibrium between the surface residue and the gas phase after sealing the chamber, followed by slow revolatilization from the surface reservoir, thus maintaining a lower gas-phase concentration. This is in line with results from Collins, Wang, and Abbatt (2018) who described rapid partitioning of nitrogen containing organic compounds from THS to AS and ammonium bisulfate aerosol. Given that JUUL vape provides nicotine predominantly in the protonated form (Duell, Pankow, and Peyton 2018), volatilization to the gas phase will be limited until conversion of the protonated form to freebase nicotine which is governed by acid-base equilibrium. We will explore factors leading to the conversion of protonated to free-base nicotine in future work, this is important since the aging of surface-bound nicotine can produce toxins (Sleiman et al. 2010;Schick et al. 2014;Borduas et al. 2016;Jacob et al. 2017). Partitioning of nicotine to AS aerosol was also observed in the tank-based e-cig experiments, where the degree of nicotine protonation was unknown but expected to be primarily free-base ( Figure S7).
Because VG is semi-volatile, it evaporates slowly and provides steady partitioning to the AS aerosol (Figure 3d-f). The decrease in slopes over time is most likely due to the decrease in total residual VG in the sampling container after each experiment.
A similar VG/sulfate ratio pattern is seen in the tankbased e-cig experiments with an absent initial spike which was expected because it was not the first experiment using that chamber ( Figure S8). Subsequent decay experiments for the tank-based e-cig were not performed so the change of the ratio over multiple experiments was not analyzed.
Regarding the extent of chemical processing, minimal mass spectral changes observed over the course of the experiments indicate that chemical reactions and changes in the residual e-cigarette vape were not significant, in contrast to what was observed in DeCarlo, Avery, and Waring (2018) for THS, indicating that the e-cigarette residue is much less chemically reactive than that of smoke from CC's. Average mass spectra for each day of JUUL residue experiments confirm this conclusion by showing a consistent signature each day and only minor changes in relative intensities ( Figure S5). The steady loss of nicotine across the vape experiments are less drastic than the loss of nicotine seen in the THS experiments; however, neither e-cig nor CC experiment sought to resolve other minor nicotine loss pathways, such as oxidization or reaction with other residual species. Persistence of protonated nicotine on surfaces due to less reactive chemical processing suggests vape residue provides a surface reservoir for evaporation of nicotine over longer time periods.

Impact and future work
Fundamentally, this work demonstrates potential secondary exposure pathways for chemicals associated with e-cig usage. The experiments presented here were idealized lab studies with simple components (clean steel surfaces, unadulterated primary vape, and homogeneous monodispersed lab generated aerosol) to observe volatilization and uptake of e-cig associated chemicals from surfaces to aerosol with the potential for transport in the indoor environment. These results suggest additional research opportunities for future work regarding e-cig residues and their long term effects on indoor air quality including the impact of different surfaces, indoor conditions, and seed aerosol composition (Collins, Wang, and Abbatt 2018). In particular, we expect the presence of basic gases such as ammonia to have a substantial effect on the gas phase concentration of nicotine. With much of the nicotine in JUUL vape in the protonated form, we expect that most of the nicotine introduced to the system was still on the chamber surfaces after the day five experiment. The differing levels of ammonia indoors will impact the timescale for deprotonation of nicotine and hence it's lifetime on surfaces indoors. While we did not analyze concentrations of surface bound nicotine in these experiments, our results indicate that surface bound nicotine is present after vape introduction and deposition, and that this is a potentially important route of exposure for nicotine and nicotine reaction products in real-world environments. Some studies of indoor settings with e-cig use have found small or negligible amounts of nicotine on surfaces (Bush and Goniewicz 2015;Liu et al. 2017). The low levels of surface nicotine observed may due to a number of variables including: degree of nicotine protonation in the e-liquid (Duell, Pankow, and Peyton 2019), PG/VG ratio of e-liquid (Mulder et al. 2019;Li et al. 2020), e-cig type and power settings (Zhao et al. 2018;Mulder et al. 2019), size and rate of deposition of vape particles to surfaces (Li et al. 2020), room air exchange rate, respiratory retention of nicotine containing particles (Pankow 2001;Armitage et al. 2004;Khachatoorian et al. 2022), puff topography (Talih et al. 2015), and frequency of use should be considered. Other studies in vape shops and indoor spaces with months of regular use indicate that surface bound nicotine from e-cig residue can be present in large quantities, even larger than nicotine content of THS (Khachatoorian et al. 2019;Son et al. 2020). We expect that frequent use of e-cigs with high fractions of protonated nicotine in spaces with low ventilation will yield the highest surface concentrations of nicotine and negatively impact indoor air quality over time even when e-cig use is not occurring. Additional research on e-cig residue accumulation and more information about frequency of indoor use could clarify exposure and public health implications of this work and related works.