Confined DART-MS for rapid chemical analysis of electronic cigarette aerosols and spiked drugs

Abstract

A confined direct analysis in real time mass spectrometry (DART-MS) system and method were developed for coupling directly with commercial electronic cigarettes for rapid analysis without sample preparation. The system consisted of a confining heated glass T-junction, DART ionization source, and Vapur interface to assist aerodynamic transport. Suction generated by positioning the electronic cigarette at the junction inlet allowed for direct chemical analysis of aerosolized electronic liquids from both automatic devices powered by drag and manual button-operated devices, which is unachievable with traditional DART-MS. Parametric analyses for the system investigated Vapur suction flow rate, junction heating, puff duration, and coil power levels. Using this method, rapid chemical analyses of electronic cigarette aerosols from electronic liquids, spiked illicit drugs, and polymeric or plasticizer contaminants were performed in < 30 s. The confined DART-MS method provides a streamlined tool for rapid screening of illicit and hazardous chemical profiles emitting from electronic cigarettes.

Graphical Abstract

Introduction

The past decade has seen a widespread surge in electronic cigarette usage, production, and custom alteration. Initially developed as a disposable alternative nicotine delivery system, electronic cigarettes have expanded to encompass a wide spectrum of electronic liquid (e-liquid) formulations and rechargeable modular device designs with user control over operating parameters and component customization.^{1, 2} In general, electronic cigarettes (e-cigarettes) use battery power to drive current through a heating coil in contact with an e-liquid soaked wick material.³ Heating of the coil atomizes the enclosed e-liquid. Activation of the battery and initiation of each ‘puff’ can be achieved by manual button-operation or automatically induced by drag (i.e., the user inhales and the battery-coil heating element is automatically activated). Though adjustable, e-cigarette puff durations have been previously measured around 2 s.⁴ E-liquids are most commonly composed of a propylene glycol (PG) and vegetable glycerol (VG) base with nicotine and flavoring compounds. However, the ease in customization (e.g., after-market or 3^rd party liquid formulations and refillable formats of e-liquid cartridge, tanks, or pods) has led to the inclusion of other compounds. Most notably as seen in the popular media, is the use of recreational drugs in electronic cigarettes. Research has investigated inclusion of compounds from marijuana extract,^5–8 natural plant products,⁹ methamphetamine,¹⁰ tetrahydrocannabinol (THC)¹¹, and other drugs.⁸ Studies and methodologies for investigating the aerosolization and vaporization of drugs as part of e-liquids will play a role in future health management issues and potentially policing or regulation of this delivery route.

The speed with which electronic cigarette usage has increased, especially among younger users,¹² has fueled a significant amount of research into e-liquid compositions,^13–16 aerosolized particle sizes and characteristics,^{4, 17–20} and potential release of toxic compounds.²¹ In addition, reports and identification of e-cigarette, or vaping, product use associated lung injury (EVALI) initiated a flurry of investigations into the underlying causes.^{11, 22} Though much of the published research has focused on determining the health impacts from electronic cigarette usage,^23–25 there is value in evaluating the direct vapor from e-cigarettes to capture the complete health implications through chemical toxicity, misuse, degradation products, or even secondary contact.

Numerous traditional analytical techniques have been employed for determining the chemical composition and quantification of electronic liquids and their associated aerosols. These include liquid chromatography (LC)- and gas chromatography (GC)-mass spectrometry (MS),^{13, 14} differential ion mobility spectrometry (DMS)-tandem mass spectrometry (MS/MS),²⁶ direct analysis in real time (DART)-MS,^{11, 27} solid phase microextraction (SPME)-GC-MS and SPME-DART-MS,¹⁰ Fourier transform infrared (FTIR) spectroscopy,¹¹ and nuclear magnetic resonance (NMR).^{11, 21} Many of these rigorous techniques have provided chemical quantification of e-liquids and collected aerosols (i.e., cold trap or similar condenser). However, vaping machines, peristaltic pumps, or manual triggering must be used for ‘puff’ initiation and aerosol generation from e-cigarettes for analysis. Chemical characterization by these techniques often requires lengthy sample preparation steps that include vapor traps for initial collection of the aerosols. This collection step hinders rapid and direct analysis of the e-cigarette emitted aerosol in situ and may lead to sample losses. Studies incorporating aerosol differential mobility spectrometry, secondary electrospray ionization MS, and proton transfer reaction MS have sought to provide rapid analysis of atomized electronic liquids.^28–30

Here, we expand upon this trend with a method that employs DART ionization confined by a heated glass T-junction and a Vapur aerodynamic-assist interface for direct mass spectrometric chemical analysis. DART-MS has become a widely used rapid screening tool for the forensic community.^{31, 32} Likewise, when used in combination with the NIST DART-MS Forensics mass spectral library³³ and associated search tools (e.g. NIST MS Search³⁴), identification and spectral matching capabilities beyond existing techniques are enabled. Confined DART-MS configurations similar to the one introduced here have been used with various thermal desorption components for analysis of a range of compounds from explosives and drugs to metabolites and rodenticides.^35–38 The direct measurement capabilities of the confined DART-MS system identified chemicals attributed to not only the e-liquids, but the device itself (e.g., polymeric contaminants). The presented method may provide unique capabilities for the chemical analysis of electronic cigarettes not easily achievable by current techniques.

Methods

Materials and Atomizers.

Nicotine, Δ−9-tetrahydrocannabinol (THC), and methamphetamine were purchased from Cerilliant (Round Rock, TX, USA) at 1 mg/mL in methanol and used as additives for investigation. Deionized water, propylene glycol (PG) and vegetable glycerol (VG) mixtures (50:50 vol:vol), and commercial e-liquids were used as representative liquids for atomization. Two commercial e-liquids with 0 % nicotine were purchased from Avail Vapor (Germantown, MD, USA). These commercial products were used as demonstrative real-world e-liquids and anonymized as ELIQ 1 and ELIQ 2. Propylene glycol and anhydrous glycerol (vegetable) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Additives were diluted in various liquids (i.e., water, PG:VG, or commercial e-liquids) to concentrations ranging from 5 μg/mL to 250 μg/mL.

Three electronic cigarettes were used in this study and will be referred to as ECIG 1, ECIG 2, and ECIG 3. ECIG 1 was used for most investigations and contained a 6.5 mL reservoir and 3000 mAh battery. The liquid reservoir was rinsed repeatedly with deionized water and the 0.18 Ω, 50 W to 85 W meshed coil was replaced between each different e-liquid or additive. The ECIG 1 activation button (i.e., the button that initiated coil heating and aerosol production) was modified to enable precisely timed triggering by a custom LabVIEW code (National Instruments, Austin, TX, USA). The user interfaces of the ECIG 2 and ECIG 3 model e-cigarettes directly enabled control over coil power.

Certain commercial equipment, instruments, or materials are identified in this article in order to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by NIST, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.

Instrumentation.

Rapid mass spectrometry was achieved by electronic cigarette aerosolization directly into a confined DART-MS configuration (Figure 1(a)). A DART SVP ion source (Ionsense, Saugus, MA, USA) was coupled to a time-of-flight mass spectrometer (AccuTOF, JEOL USA, Peabody, MA, USA) through a glass T-junction and Vapur aerodynamic-assist interface (Ionsense). The custom T-junction was fabricated out of glass with a nominally 6.35 mm outer diameter and 3.18 mm inner diameter.^{35, 38} The junction was 100 mm long from the Vapur to DART source, with a flared end (12.7 mm diameter). The concentrically positioned DART source was maintained at an approximately 5 mm standoff distance (intake gap) from the flared glass end (unless otherwise noted). Each electronic cigarette was concentrically brought into contact with a 30 mm long arm, positioned 20 mm from the flared end of the glass T-junction (Figure 1(a)). Atomized liquids were pulled through the T-junction toward the mass spectrometer inlet by a combination of an auxiliary vacuum pump (6 L/min) and entrainment induced by the DART gas stream (2.5 L/min). The induced suction allowed operation with either manually triggered or automatically triggered batteries. A 450 °C DART gas temperature was set to aid in heating the glass junction. In addition, an external fiberglass tape heater was wrapped around the T-junction and controlled by a variable AC power supply unless noted otherwise. The outer surface of the T-junction was maintained at nominally (205 ± 5) °C. Nitrogen was used as the DART gas based on previous studies of the confined geometry.^{35, 38} Mass spectra were collected in positive mode with 10 V orifice 1 voltage, 5 V orifice 2 voltage, 5 V ring lens voltage, 400 V peaks voltage, 2300 V detector voltage, and m/z 40 to m/z 800 mass range.

Figure 1.

(a) Schematic representation of the confined DART-MS configuration for direct analysis of electronic cigarette aerosols. (b) Illustration of a single 2 s atomization of 250 μg/mL THC spiked in 50:50 (vol:vol) PG:VG liquid from ECIG 1, including extracted ion chronograms for select compounds and spectra at various time points (i)-(iii). Extracted ion chronograms represent m/z 151 [PG+VG-OH]⁺, m/z 313 [BBP+H]⁺, m/z 315 [THC+H]⁺, and m/z 610 [(C₂H₆OSi)₈+NH₄]⁺. Spectra display peaks associated with the PG:VG liquid (●) and polydimethylsiloxane (PDMS, ◊). Preliminary peak assignments can be found in Table 1.

Results and Discussion

The surge in electronic cigarette usage and questions related to potential health hazards has led to numerous investigations. Most studies focusing on chemical analysis by traditional techniques (e.g., LC-MS or FTIR) cannot directly sample from an e-liquid puff (i.e. aerosol plume) and have had to first collect the aerosolized liquid for subsequent analysis. The method described here provides rapid chemical analysis by directly sampling the atomized plume as it emits from an electronic cigarette. Preliminary investigations considered direct analysis of e-cigarette aerosols using a traditional DART-MS configuration with the Vapur aerodynamic-assist interface. This arrangement was only moderately successful and led to difficulties with reproducibility and e-cigarettes that required suction to activate or efficiently extract aerosols. In addition, as demonstrated in a recent work,¹¹ such an ‘open’ configuration requires appropriate safety and environmental controls to curtail any inhalation risks. Here we developed a confined geometry configuration coupling DART-MS directly with commercial electronic cigarettes. The direct suction generated at the e-cigarette mouthpiece enabled operation with both automatic e-cigarette batteries powered by drag (i.e., activated by user draw/inhalation) and manual button-operated e-cigarettes.

Figure 1(a) displays the confined geometry DART-MS configuration, which enabled direct coupling with an e-cigarette. During each experiment, the electronic cigarette mouthpiece fit intimately around the glass T-junction arm. The e-cigarette was initiated (i.e., triggering the activation button), generating an aerosol of the e-liquid while in contact with the glass junction. Following the puff duration, the e-cigarette was withdrawn from the junction. The manual activation button on ECIG 1 used for these studies was bypassed and automatically triggered with a custom LabVIEW code. The code enabled precise control over the initiation of experiments and duration of coil activation. Suction generated by the confined DART-MS system pulled the e-liquid aerosols through the junction. There, they were further vaporized (by heated DART gas and/or the heated junction) and ionized by the DART source. Ionized components were then measured by a time-of-flight mass analyzer. Figure 1(b) demonstrates an example of extracted ion chronograms and associated mass spectra output achieved from a typical experiment (i.e., a single activation/triggering of the electronic cigarette atomizer). The example consisted of 250 μg/mL THC spiked in a PG:VG base e-liquid (the organic solvent from 0.5 mL of 1 mg/mL THC standard was evaporated to dry and diluted in 2 mL PG:VG). This system enabled rapid detection of target species, background from liquid components, and contaminants from the device. Figure 1(b) displays the evolution of ions associated with the PG:VG liquid, spiked THC, and PDMS and phthalate contaminants. Here, a PDMS ion distribution³⁹ was observed at the onset of the experiment (Figure 1(b–i)). PDMS is commonly associated with lubricants or silicone-based seals and was likely leaching from the e-cigarette assembly, gaskets, and materials. The high(er) vapor pressure of PDMS (relative to e-liquid components) led to the initial spike in intensity. Preliminary peak assignments from the spectra are listed in Table 1. The ammonium adducts observed for PDMS and PEG are common to the DART ion source and have been frequently attributed to trace ammonia in the laboratory atmosphere.^{39, 40} The polymeric e-liquid, background, and contaminant species in Table 1 were not listed in the released version of the NIST DART-MS Forensics mass spectral library³³ and preliminary assignments were based on matching potential elemental compositions to exact mass of the most abundant isotope and the literature. The target THC dominated much of the mass spectra during the experiment (Figure 1(b–ii) and Figure 1(b–iii)), with additional distributions exhibiting phthalate plasticizers and the PG:VG background liquid. Phthalates are frequently found with plastic materials and were potentially emitted from the e-cigarette device during heating. The diverse PG:VG ion distribution decayed slower than the others over time as the e-liquid that had condensed on the heated glass T-junction subsequently desorbed (Figure 1b–iii)). Though the e-cigarette coil was only initiated for 2 s, the condensation of the e-liquid aerosol onto the glass junction and its subsequent desorption yielded the extended signal durations. This will be discussed further below.

Table 1.

Preliminary peak assignments related to propylene glycol (PG), vegetable glycerin (VG), polydimethylsiloxane (PDMS), phthalates [including: diethyl phthalate (DEP: C₁₂H₁₄O₄), di-n-butyl phthalate (DBP: C₁₆H₂₂O₄), diisobutyl phthalate (DIBP: C₁₆H₂₂O₄), butyl benzyl phthalate (BBP: C₁₉H₂₀O₄), dicyclohexyl phthalate (DCP: C₂₀H₂₆O₄), di(2-ethylhexyl) phthalate (DEHP: C₂₄H₃₈O₄), di(n-octyl) phthalate (DNOP: C₂₄H₃₈O₄)], and polyethylene glycol (PEG). The −OH peak assignments were distinguished by loss of a neutral water and protonation.

m/z	Assignment	m/z	Assignment	m/z	Assignment
● Propylene Glycol (C 3 H 8 O 2 ) : Vegetable Glycerin (C 3 H 8 O 3 ) Mixture
59.06	[PG-OH] +	117.10	[PG+H2O+Na] +	185.10	[2VG+H] +
75.04	[VG-OH] +	135.10	[2PG-OH] +	211.16	[3PG-OH] +
77.07	[PG+H] +	151.10	[PG+VG-OH] +	227.16	[2PG+VG-OH] +
93.06	[VG+H] +	175.14	[2PG+Na] +	243.16	[PG+2VG-OH] +
◊ Polydimethylsiloxane (C2H6OSi)n (nominal mass difference of repeating unit: +74)
388.18	[(C2H6OSi)5+ NH4]+	536.17	[(C2H6OSi)7+NH4]+	684.21	[(C2H6OSi)9+NH4]+
462.15	[(C2H6OSi)6+NH4]+	610.19	[(C2H6OSi)8+NH4]+	758.20	[(C2H6OSi)10+NH4]+
□ Phthalates
223.10	[DEP+H] +	313.23	[BBP+H] +	391.32	[DEHP+H] +, [DNOP+H] +
279.18	[DBP+H] +, [DIBP+H] +	331.25	[DCP+H] +
▼ Polyethylene Glycol (C 2n H 4n+2 O n+1 ) (nominal mass difference of repeating unit: +44)
300.21	[(C12H26O7)+NH4]+	388.24	[(C16H34O9)+NH4]+
344.22	[(C14H30O8)+NH4]+	432.24	[(C18H38O10)+NH4]+

We characterized the system response (i.e., peak area of target species) as a function of the Vapur volumetric suction flow rate. This analysis was completed for an arrangement with the traditional 5 mm standoff intake gap with the T-junction (see Figure 1(a))^{35, 38} and without a gap (i.e., DART source was in contact with the flared glass junction end). Figure 2(a) displays the peak area of 5 μg/mL THC spiked in an aqueous liquid (.05 mL of a 1 mg/mL THC standard into 9.95 mL water) as a function of Vapur flow rate. Similar to previous related confined DART-MS studies,³⁸ the Vapur flow rate demonstrated an optimal value in the range of 4 L/min to 6 L/min for the configuration with a DART-junction gap (Figure 1(a)). In an effort to maximize the suction through the e-cigarette arm of the junction, the study was completed without the gap. With the DART matted against the flared glass, no (or minimal) air could be entrained. Therefore, more of the volume of gas pulled by the Vapur pump came from the e-cigarette branch. However, this arrangement yielded lower overall peak areas (Figure 2(a)). The reduction in water vapor drawn into the DART gas stream at the intake region for Penning ionization^{32, 40} led to a reduction in available ions for further ionization of the aerosol. The reduction in charged water clusters for protonation, led to the decreased THC signal. Polyethylene glycol (PEG) clusters with THC were also observed and likely the result of carry-over from polypropylene sample storage tubes. The Vapur flow rate studies discussed here focused on a triggered e-cigarette, however, the suction generated by the confined DART-MS configuration could be used to activate automatic battery e-cigarettes as well.

Figure 2.

(a) Peak area of THC as a function of Vapur flow rate for 2 s atomization replicates of 5 μg/mL THC spiked in an aqueous liquid from ECIG 1. Representative mass spectra at 4 L/min for (a-i) a 5 mm DART-to-junction standoff gap and (a-ii) no standoff gap. (b) Peak area and clear down time of THC as a function of puff duration for 5 μg/mL THC spiked in an aqueous liquid from ECIG 1. Inset displays representative extracted ion chronograms of THC for 2 s, 5 s, and 10 s puff durations with heated junction. Representative mass spectra at 2 s puff duration for (b-i) an unheated glass T-junction and (b-ii) heated T-junction with outer surface temperature of (205 ± 5) °C. Data points and uncertainty represent the average and standard deviation of 10 replicate measurements. Base peak absolute intensity noted in insets for each spectrum. Preliminary peak assignments can be found in Table 1.

Next, we investigated the signal response for a target narcotic (i.e., THC) as a function of the puff duration. Puff duration for ECIG 1 was precisely controlled through the LabVIEW triggering code as described above. Initial studies with PG:VG and commercial e-liquids, as well as extended puff durations of any liquid (including water), resulted in aerosol condensation on the T-junction arm. Therefore, configurations with a bare glass T-junction (only heated by the DART gas stream) and the glass T-junction wrapped with a fiberglass tape heater were considered. Figure 2(b) demonstrates the peak area of a 5 μg/mL THC aqueous liquid for puff durations from 1 s up to 10 s using both unheated and heated glass junctions. The heated junction provided a minimal increase in THC signal. However, heating the glass junction consistently decreased the clear down time (i.e., time for the THC signal to drop to approximately 1 % of the peak maximum) for each experiment (Figure 2(b)). The heated glass junction sped up the desorption of liquid condensed on the glass junction during the puff duration. The Figure 2(b) inset demonstrates extracted ion chronograms for a series of puff durations with the heated junction. In this simple water base liquid, the THC peaks near the end of the puff duration and steadily decays. Heating of the glass junction to reduce aerosol condensation was critical for PG:VG based and commercial e-liquids. As demonstrated above (Figure 1(b)), the condensation and desorption of PG:VG-based liquid from the glass junction extended the observed periods of compounds. This condensation and delay in detection was a necessary trade-off of the direct and controlled aerosol plume collection/sampling. However, even with this delay, analyte peaks occurred within 30 s. Though we investigated puff durations up to 10 s for investigative purposes, typical user inhalations have been measured around 2 s.⁴ In addition, manipulation of the glass junction heating enabled a level of control over the intensity of the observed PDMS contaminant ion distribution (m/z 462, m/z 536, m/z 610, m/z 684, and m/z 758) as shown in the representative spectra of Figure 2(b–i) and Figure 2(b–ii).

Several commercial e-cigarettes offer a range of user control over aerosolization parameters, for example coil power, coil type, venting configurations, trigger durations, batteries, and modifications for plant-based materials, among others. Most e-cigarettes contain a coil, surrounded by a wick material soaked in the e-liquid. Rapid heating of this coil provided some control over the liquid atomization. The methods developed here enabled rapid measurement and targeted investigation of these parameters. The chemical profiles of aerosols generated at varying coil power levels were investigated using ECIG 2 and ECIG 3 and yielded increasing signal as a function of increasing coil power, similar to documented trends (Figure S1).⁴¹

The preliminary characterization of this method provided an avenue to investigate mass spectral signatures for a number of drugs, stimulants, and e-liquids. Similarly, the confined DART-MS platform provides a technique to detect and characterize hazardous device by-products and degradation products such as formaldehyde.⁴² The spectrum for a 50:50 PG:VG (volume fractions) mixture was dominated by a wide ion distribution of PG and VG cations, and PG-VG complexes (Figure S2(a)). An array of these cations is identified in Table 1. The commercial e-liquid, ELIQ 1 (0% nicotine), primarily composed of PG and VG, displayed a similar ion distribution (Figure S2(b)). This nicotine free e-liquid also contained various flavors and minor components. Though numerous peaks not attributed to PG or VG were observed, specific flavor compounds were not readily identified. Figure 3(a) shows a representative mass spectrum of 250 μg/mL nicotine spiked in a PG:VG e-liquid. The protonated nicotine m/z 163 [M+H]⁺, was identified with the NIST DART-MS Forensics mass spectral library^{33, 34} and dominated the spectrum. The nicotine signal decayed slowly due to condensation and subsequent desorption from the heated glass junction as described above. However, adducts with PG, m/z 239 [M+PG+H]⁺, and VG, m/z 255 [M+VG+H]⁺, as well as the dimer, m/z 325 [2M+H]⁺, were also observed during the high concentration e-liquid aerosol plume at the onset of the experiment.

Figure 3.

Representative mass spectra for (a) 250 μg/mL nicotine (M: nicotine) spiked in PG:VG e-liquid, (b) 250 μg/mL methamphetamine (M: methamphetamine) spiked in PG:VG e-liquid, and (c) 250 μg/mL methamphetamine spiked in ELIQ 2 e-liquid. Insets display extracted ion chronograms of select species. Preliminary peak assignments can be found in Table 1. Significant peaks without a preliminary assignment were labeled with and asterisk (*).

E-cigarettes have also found use as a delivery system for illicit drug delivery, including THC, methamphetamine, heroin, and synthetic cannabinoids.^{8, 10} An e-liquid sample consisting of 250 μg/mL THC spiked in PG:VG , presenting the PG:VG, THC, and PDMS peaks observed above (Figure S2(c)). Though library searching provided a preliminary identification for THC, the confined DART-MS method did not enable differentiation of potential structural isomers. Figure 3(b) displays 250 μg/mL methamphetamine spiked in a PG:VG e-liquid sample. The spectrum was dominated by the protonated methamphetamine peak, m/z 150 [M+H]⁺; and similar to nicotine, adducts with PG, m/z 226 [M+PG+H]⁺, and VG, m/z 242 [M+VG+H]⁺ were observed. Common methamphetamine degradation products and fragments present in the NIST DART-MS Forensics library^{33, 34} were also identified, including m/z 91 [M-C₃H₈N]⁺ and m/z 119 [M-(CH₃)₂]⁺. Finally, as shown in Figure 3(c), 250 μg/mL methamphetamine was detected in the commercial e-liquid ELIQ 2 (0% nicotine). The same protonated cation and adducts with PG and VG were observed, however the PG and VG adducts exhibited greater abundance than the protonated molecule. An additional peak at m/z 196 was consistently detected in ELIQ 2 samples, however, was not identified here.

Conclusions

The recent rise in electronic cigarette usage, evolution into illicit drugs, and spread of associated lung injuries have highlighted the need for rapid chemical characterization of atomized aerosols. The coupling of a confined DART-MS platform with commercial e-cigarettes delivered a system enabling direct analysis of aerosols without the need for sample collection or condensation (e.g., via vapor traps) prior to characterization. The heated confining glass junction provided intimate contact with the e-cigarette mouthpiece for immediate extraction of the aerosolized plume of e-liquid. This method enabled capabilities for rapid chemical analysis (< 30 s) of e-cigarette aerosols, including PG:VG and commercial e-liquids, illicit drugs, and polymeric and plasticizer contaminants. Adding these capabilities to existing analytical techniques may provide a powerful complementary tool for in-depth investigations into e-cigarette, or vaping, product use associated lung injury (EVALI) and other hazardous compounds such as formaldehyde and metals. Direct chemical identification capabilities may enable rapid characterization of chemicals or temporal evolution of harmful complexes directly from the commercial e-cigarettes. The DART ionization backbone of this technique also enables simple compatibility with the expanding NIST DART Forensics mass spectral library and NIST MS Search tools.