Measurement of the Free-Base Nicotine Fraction (α_fb) in Electronic Cigarette Liquids by Headspace Solid-Phase Microextraction

Abstract

A method for determining the fraction of free-base nicotine (α_fb) in electronic cigarette liquids (“e-liquids”) based on headspace solid-phase microextraction (h-SPME) is described. The free-base concentration c_e,fb = α_fbc_e,T, where c_e,T is the total (free-base + protonated) nicotine in the liquid. For gas/liquid equilibrium of the volatile free-base form, the headspace nicotine concentration is proportional to c_e,fb and thus also to α_fb. Headspace nicotine is proportionally absorbed with an SPME fiber. The fiber is thermally desorbed in the heated inlet of a gas chromatograph coupled to a mass spectrometer: the desorbed nicotine is measured by gas chromatography–mass spectrometry. For a second h-SPME measurement, an adequate base is added to the sample vial to convert essentially all protonated nicotine to the free-base form (α_fb → 1.0). The ratio of the first h-SPME measurement to the second h-SPME measurement gives α_fb in the initial sample. Using gaseous ammonia as the added base, the method was (1) verified using lab-prepared e-liquid solutions with known α_fb values and (2) used to determine the α_fb values for 18 commercial e-liquids. The measured α_fb values ranged from 0.0 to 1.0. Increasing measurement error with decreasing α_fb caused modestly lower method precision at small α_fb. Adding a liquid organic base may be more convenient than adding gaseous ammonia: one of the samples was examined using triethylamine as the added base; the measurements agreed well (with ammonia, 0.27 ± 0.01; with triethylamine, 0.26 ± 0.04). Other workers have proposed examining the nicotine protonation state in e-liquids using three steps: (1) 1:10 dilution with CO₂-free water; (2) measurement of pH; and (3) calculation of the resulting values for α_fb,w,1:10, the free-base fraction in the diluted mostly aqueous phase. As expected and verified here, because of the generally greater abilities of organic acids to protonate nicotine in water versus in an e-liquid phase, α_fb,w,1:10 values can be significantly less than actual e-liquid α_fb values when α_fb is not close to either 0 or 1.

Graphical Abstract

1. INTRODUCTION

1.1. Nicotine Forms and Inhalation Harshness.

Nicotine is a dibasic alkaloid and so can exist in three forms: (1) free-base (fb, aka unprotonated), Nic; (2) monoprotonated (mp), NicH⁺; and (3) diprotonated (dp), NicH₂²⁺ (Figure 1). The fraction of free-base nicotine in a liquid is defined as¹

$${\alpha }_{\text{fb}}\equiv \frac{[\text{Nic}]}{[\text{Nic}]+\left[{\text{NicH}}^{+}\right]+\left[{{\text{NicH}}_2}^{2+}\right]}$$ (1)

Figure 1.

Three forms of nicotine.

The bracketed terms are concentrations in the liquid with units of mols per mass (or volume) of liquid. Free-base nicotine is neutral, lipophilic, and modestly volatile. Protonated nicotine is charged, not lipophilic, and not volatile.

The e-liquid free-base nicotine concentration c_e,fb (μg free-base nicotine/μg e-liquid) is related to the total e-liquid nicotine concentration c_e,T [μg total nicotine (as free base) per μg e-liquid] through

$${\alpha }_{\text{fb}}=\frac{c_{\text{e},\text{fb}}}{c_{\text{e},\text{T}}}$$ (2)

There is much utility in α_fb as a parameter for characterizing electronic cigarette liquids (“e-liquids”) given that free-base and protonated nicotine exhibit markedly different chemical² and physiological^3,4 behaviors. Because of these differences, acids are being added to e-liquids to reduce α_fb values and the associated inhalation harshness during vaping.³ (Inhalation harshness in tobacco smoke is also associated with high levels of free-base nicotine.^4,5) The following excerpt from a 2015 WIRED magazine interview with personnel at Pax Labs, Inc. (now Juul Labs, Inc.) makes clear the ab ovo design intent of JUUL e-cigarettes:

‘Atkins tries to explain: “In the tobacco plant, there are these organic acids that naturally occur, and they help stabilize the nicotine in such a way that makes it...” He pauses. “I’ve got to choose the words carefully here: appropriate for inhalation. ”Steve Christensen, a design engineer, pipes in. “Smoother,” he says. Atkins goes with that. “Yeah, it’s smoother.”⁶

The sensory harshness caused by free-base nicotine has been attributed to chemesthetic activation of the trigeminal nerve system,⁷ which has nerve endings throughout the oral cavity⁸ and the nasal epithelium.⁹ Inhaled menthol, CO₂ from carbonated beverages, and free-base ammonia (NH₃) also activate this nerve system.^7,10,11 The chemesthetic effect of free-base nicotine is sufficiently strong that Schuh et al. (1997)¹² concluded that the abuse liability of nicotine replacement therapies (NRTs) involving nasal sprays and vapor inhalers delivering free-base nicotine is low because of the consequent “burning throat and nose, watery eyes, runny nose, coughing, and sneezing”.

It has been reported that protonated nicotine does not activate the trigeminal nerve:¹³ a US patent for a protonated-nicotine (aka “nicotine salt”) solution in NRT nasal sprays was given in 2003.¹⁴ A 1989 letter from D. E. Creighton of the British American Tobacco Co. to P. G. L. Oteri of the Nigerian Tobacco Co. states the following:

“Air-cured tobaccos... will release nicotine into the smoke... [that nicotine is] in the free-base form and is irritant [sic], if compared with the same amount of nicotine in the bound [protonated] or salt form...” Creighton (1989).¹⁵

For e-cigarettes, it was Pullicin et al. (2020)³ that confirmed a correlation between inhalation harshness and free-base nicotine levels.

1.2. Determination of α_fb in E-Liquids.

Two types of methods have been proposed for measuring α_fb values in e-liquids and their associated aerosol droplet phase: (1) diluted solution methods (DSMs) and (2) native solution methods (NSMs).¹⁶ In the former, a propylene glycol/glycerol (PG/GL)-based e-liquid is diluted with a solvent (usually water); then, a measurement is carried out (usually pH). DSMs alter an e-liquid and so cannot inherently access α_fb values in an e-liquid phase.¹⁶

In NSMs, a measurement is carried out directly on an unaltered e-liquid phase. Duell et al. (2018, 2019)^17,18 determined α_fb values in e-liquids by proton nuclear magnetic resonance (¹H NMR) spectroscopy. The method can be applied to samples that lie between two bookend conditions, giving (1) nearly complete concentration dominance for the free-base Nic form at one extreme (α_fb ≈ 1, α_mp ≈ 0, α_dp ≈ 0) and (2) nearly complete concentration dominance for the monoprotonated NicH⁺ form (α_fb ≈ 0, α_mp ≈ 1, α_dp ≈ 0). The method determines where, proportionally, the ¹H NMR spectral peaks for nicotine in a measurement for a sample fall between those for the bookend spectra. This is an ¹H NMR analogue of the well-known use of UV–vis spectroscopy for measuring pH-dependent acid/base species distributions (e.g., Martinez and Dardonville¹⁹). The ¹H NMR method cannot be used when a sample is (1) so acidic that the dp form contributes significantly to the ¹H NMR spectrum or (2) high in one or more flavor chemicals that add to the ¹H NMR signal in the nicotine spectral regions.

1.3. Headspace SPME.

An NSM alternative for determining nicotine α_fb values in e-liquids can be developed based on headspace solid-phase microextraction (h-SPME). h-SPME has been used with condensed phases that include water,²⁰ food products,²¹ petroleum products,²² and tobacco smoke particulate materials.²³ For a given sample, the headspace is allowed to equilibrate so that the analyte can acquire a gas-phase concentration that is proportional to its concentration in the condensed phase. The analyte in the headspace is sorbed into/onto the fiber. For reproduced exposure conditions (time, temperature, etc.), the amount sorbed is proportional (within a linear range) to the concentration in the condensed phase. Equilibrium for a given analyte between the fiber coating and the headspace is not required; if not sought, the exposure conditions must be reproduced carefully.

2. MATERIALS AND METHODS

For sampling free-base nicotine in the headspace above an e-liquid, we assume

$${\text{h-SPME signal}|}_{\text{sample}}\propto c_{\text{e},\text{fb}}={\alpha }_{\text{fb}}{c}_{\text{e},\text{T}}$$ (3)

By analogy with the ¹H NMR method, with h-SPME, a “base-added” condition signal is obtained for t he α_fb = 1 bookend

$$\text{h-SPME signal}|{}_{\text{sample +base:}{\alpha }_{\text{fb}}\to 1}\propto c_{\text{e},\text{T}}$$ (4)

Measurement of an α_fb ≈ 0, acid-added bookend is not required because the signal will be ~0. Equations 3 and 4 share the same proportionality constant. Dividing eq 3 by eq 4 gives

$$\frac{\text{h-SPME}\text{signal}{|}_{\text{sample}}}{\text{h-SPME}\text{signal}{|}_{\text{sample}+\text{base:}{\alpha }_{\text{fb}}\to 1}}=\frac{c_{\text{e},\text{fb}}}{c_{\text{e},\text{T}}}={\alpha }_{\text{fb}}$$ (5)

Application of eq 5 does not require a calibration of the h-SPME signal versus actual gas-phase free-base concentrations. The fact that h-SPME is applicable as an NSM is discussed further in the Supporting Information. The α_fb method for tobacco smoke particulate matter (PM) of Pankow et al. (2003)²⁴ is analogous (in that effort, the actual equilibrium gas-phase concentrations were also measured).

For increasingly acidified e-liquids, mathematically α_fb values continually decrease but never reach zero (...0.1, 0.01, 0.001, 0.0001, etc.) Very low e-liquid α_fb values will become impossible to distinguish analytically from 0. For example, for all samples with values α_fb = 0.001, unless the total nicotine concentration values are very high, h-SPME signal|_sample will be ~0, giving α_fb ≈ 0. This is not a problem if the only goal is to evaluate an e-liquid in terms of its inhalation harshness due to free-base nicotine: in this context, α_fb values of 0.001 and 0.0001 are substantively the same. With that said, we add that at a given nicotine level, the amount of acid added is greater when α_fb = 0.0001 versus when α_fb = 0.001. This aspect of the chemistry would remain of interest especially when the mequiv/mL of added acid ≫ mequiv/mL of nicotine, so that the e-liquid would contain free acid, untitrated by nicotine.

2.1. Samples.

For method verification, three simulated, laboratory e-liquids were prepared at known α_fb values of ~0.25, ~0.50, and ~0.75 in 50:50 by weight PG/GL. The total nicotine level in each was 6 mg/mL as free-base nicotine. The strong acid HCl was used to stoichiometrically protonate the nicotine. If x = formula weights/mL of added HCl, y = formula weights/mL of initial free-base nicotine, and x < y (so that formation of NicH₂²⁺ is not important), x = mols/mL NicH⁺ will be produced, giving α_fb = (y − x)/y.

For use of the method with actual samples, 18 commercial e-liquids were purchased online between January 2016 and June 2019; the years of purchase are given in Table S1 in the Supporting Information. The e-liquids were stored at 4 °C and analyzed for this work in 2020 and 2021. The nicotine levels on the product labels are given in Table 2; the identities and levels of any associated acid additives were neither given on the labels nor were they determined here.

Table 2.

Values of α_fb by h-SPME for Selected Commercial E-Liquids at ~20 °C along with pH_1:10 and the Corresponding Inferred α_fb,w,1:10 Values for 1:10 Dilution with CO₂-Free Water at ~20 °C^a

		e-liquid by h-SPME
			ammonia as a base				1:10 water dilution
brand, flavorb	nicotine label concn clabel (mg/mL)	${\alpha }_{\text{fb}}^{\text{est}.}$	vammonia (mL)	αfb (mean ± 1s) (final)	CV (%)c	m (final)	pHw,1:10d	α fb,w,1:10 d,e
Sour Vape, “Atomic Apple”	6	0.007	3	0.003 ± 0.00	26	3.4	3.37	0.000
OKAMI, “Bubble Gang, Sour Menace Green Sour Apple”	6	0.006	3	0.012f	NAf	3.4	3.55	0.000
Artist Collection, “Hedon’s Bite”	12	0.03	2 × 3	0.032 ± 0.00	6.8	3.5	4.90	0.001
OMG, “WTF Ice”	3	0.072	3	0.11 ± 0.01	10	7.6	4.76	0.001
Cloud Nurdz, “Watermelon Apple”	3	0.13	3	0.27 ± 0.01	2.6	9.3	7.34	0.16
Bombies, “Tiger Style”	3	0.47	3	0.40 ± 0.03	6.8	11	7.62	0.26
The Milkman, “Churrios”	6	0.31	3	0.40 ± 0.03	8.6	5.6	6.48	0.03
ANML, “Carnage”	6	0.34	3	0.51 ± 0.01	1.8	6.9	7.79	0.34
Glas, “Pound Cake”	6	0.47	3	0.51 ± 0.01	2.8	6.9	6.96	0.07
NicQuid, “Midnight Express”	6	0.48	3	0.59 ± 0.03	5.6	8.3	8.12	0.53
The Standard, “Cell Block Four”	6	0.50	3	0.59 ± 0.05	8.9	8.3	7.68	0.29
Beard Vape, “No. 5”	18	0.42	2 × 3 + 2	0.69 ± 0.02	2.9	9.8	7.86	0.38
Charlie Noble, “Charlie’s Custard”	12	0.76	2 × 3	0.69 ± 0.01	1.2	11	7.67	0.28
Sengoku, “Kendo Man”	6	0.95	3	0.77 ± 0.02	2.6	14	8.66	0.80
Nerdz, “Premium Blend”	6	0.66	3	0.83 ± 0.02	1.9	20	8.56	0.76
Teleos, “Delta Cloud Science”	6	0.90	3	0.84 ± 0.05	5.7	21	8.85	0.86
The Mad Alchemist, Twice in a Blue Moon”	12	0.85	2 × 3	0.93 ± 0.01	0.8	50	8.70	0.81
Caesar, “Seduce Juice”	18	0.90	2 × 3 + 2	1.03 ± 0.03	3.3	>100	NA	NA
		e-liquid by h-SPME
			triethylamine as a base				1:10 water dilution
brand, flavorb	nicotine label concn clabel (mg/mL)	${\alpha }_{\text{fb}}^{\text{est}.}$	vPOLB (μL)	αfb (mean ± 1s) (final)	CV (%)c	m (final)	pHw,1:10d	α fb,w,1:10 d,e
Cloud Nurdz, “Watermelon Apple”	3	0.21	12	0.26 ± 0.04	17.0	6.3	7.34	0.16

^aNumber of replicates N = 3 except as noted.

^bFrom the product label.

^cCV = coefficient of variation.

^dAssumes calibration with aqueous buffers is sufficiently accurate.

^eAssumes pK_a = 8.07 at ~20 °C for NicH⁺ and that α_fb,1:10 = 10^−pK_a/(10^−pK_a + 10^−pH), which neglects NicH₂²⁺.

^fNumber of replicates N only 2: standard deviation and CV not available (NA).

2.2. Vial Silanization for h-SPME.

Amber glass “VOA” vials (40 mL) and Teflon-backed septa were obtained from Restek (Bellefonte, PA). Gaseous nicotine is notorious for adsorbing strongly to surfaces, especially to glass and its acidic surface silanol (–Si–OH) groups. Deactivation of sorptive glass sites is possible by silanization;²⁵ preliminary results indicated that silanization of the sample vials was necessary. The vials were washed, rinsed with deionized water, and then dried overnight at 100 °C. They were silanized by (1) rinsing with ~5% dimethyldichlorosilane in heptane [“silanization solution 1” from MilliporeSigma (Burlington, MA)]; (2) rinsing with deionized water; and (3) drying overnight at 100 °C. Silanized vials can, alternatively, be purchased from chemical supply companies.

2.3. Neutralizing Base Estimates for Moving α_fb → 1.

The second h-SPME determination (in which α_fb → 1) requires addition of enough sufficiently strong base to convert essentially all protonated nicotine to the free-base form. A caveat is that the base addition should not substantively change the physical properties of the solvent medium so as not to change the gas/liquid partitioning coefficient. We investigated two bases for this purpose: (a) pure (anhydrous) ammonia gas and (b) an organic liquid base (LB), triethylamine (TEA). The choice of gaseous ammonia for most of the measurements was due to our familiarity with it from measuring α_fb in tobacco smoke PM²⁴ (mixing an LB with tobacco smoke PM on a filter substrate or Teflon bag wall would be difficult and could alter the PM phase). Here, use of an LB may be significantly more convenient, subject to some precautions about dissolution/mixing (see below). An organic LB should be stable and exhibit good reasonable solubility in PG/GL mixtures. In this work, optimization of the LB was not sought; the range of possibilities includes numerous organic compounds (strong bases such as tetramethylammonium hydroxide and NaOH carry the disadvantage that they can react with very weak bases such as the common phenolic flavor chemicals vanillin and ethyl vanillin).

The minimum required mequiv of added base depends on the concentration of acid in a subject e-liquid and the volume of the e-liquid aliquot tested. A total of 1 mg of nicotine corresponds to 0.00616 mmol nicotine. For a 1:1 acid/nicotine “salt”, this indicates 0.00616 mequiv of added acid. Pure ammonia gas at 20 °C/1 atm holds 0.0416 mequiv base/mL. Liquid TEA (density = 0.728 g/mL, MW = 101.19 g/mol) holds 0.00719 mequiv base/μL. An estimate of α_fb can be obtained by an initial dummy run and then used to estimate the minimum amount of base needed to achieve α_fb → 1, as in the example given below.

2.4. Dummy Conditioning Run and ${\mathbf{\alpha }}_{\mathbf{f}b}^{\mathbf{e}st\mathbf{.}}$

For each triplicate h-SPME sample set, at least one dummy run was first carried out. A dummy run pre-conditions the analytical system with nicotine and can provide ${\alpha }_{\text{fb}}^{\text{est.}}$, a preliminary estimate of α_fb. The free-base e-liquid nicotine concentration for a dummy run is denoted as c_dummy (mg/mL); here, it was 6 mg/mL for a lab-prepared e-liquid (50:50 by weight PG/GL + nicotine). The system response for c_dummy is r_dummy. For a sample e-liquid with a labeled nicotine level of c_label (mg/mL), if the fraction of nicotine in the free-base form was 1.0, the expected response from the sample would be ~r_dummy × (c_label/c_dummy). If the actual response is r_sample, then

$${\alpha }_{\text{fb}}^{\text{est.}}\approx \frac{r_{\text{sample}}}{r_{\text{dummy}}\times \left(c_{\text{label}}/c_{\text{dummy}}\right)}\le 1$$ (6)

2.5. Estimating the Base Factor (m) Amount.

As an example, for c_label = 6 mg/mL and sample aliquot volume v_sample (mL) = 1 mL, if ${\alpha }_{\text{fb}}^{\text{est.}}\approx 0.5$, the amount of acid that has reduced α_fb from 1.0 to that value equals ~(1.0 − 0.5) × 6 mg/mL × 1 mL × 0.00616 mequiv/mg = 0.0185 mequiv. This is a lower bound because organic acids of the type used in e-liquids do not quite stoichiometrically protonate nicotine.^17,18 A lower bound for the estimated 1× stoichiometrically required volume (represented here as the m = 1 volume) of pure ammonia gas at 20 °C and 1 atm for this example is then $v_{\text{ammonia}}^{1\times ,\text{est.}}=1\times 0.0185$ mequiv × 1 mL/0.0416 mequiv = 0.445 mL; for 100% excess (m = 2), $v_{\text{ammonia}}^{2\times ,\text{est.}}=0.889$ mL. Similarly, a lower bound for the stoichiometrically required (m = 1) volume of TEA as an LB is $v_{\text{TEA},\text{LB}}^{1\times ,\text{est.}}=1\times 0.0185$ mequiv × (1 μL/0.00719 mequiv) = 2.57 μL; for 100% excess, $v_{\text{TE},\text{LB}}^{2\times ,\text{est.}}=5.15$μL. In general

$$\begin{gathered} v_{\text{ammonia}}^{m\times ,\text{est}.}(\text{mL})=m\times \left(1-{\alpha }_{\text{fb}}^{\text{est}.}\right)\left(c_{\text{label}}\text{mg}/\text{mL}\right)\left(v_{\text{sample}}\text{mL}\right) \\ \times \frac{0.00616\text{mequiv}}{\text{mg}}\times \frac{1\text{mL}}{0.0416\text{mequiv}} \end{gathered}$$ (7)

$$\begin{gathered} v_{\text{TEA,LB}}^{m\times ,\text{est}.}(\mu \text{L})=m\times \left(1-{\alpha }_{\text{fb}}^{\text{est}.}\right)\left(c_{\text{label}}\text{mg}/\text{mL}\right)\left(v_{\text{sample}}\text{mL}\right) \\ \times \frac{0.00616\text{mequiv}}{\text{mg}}\times \frac{1\mu \text{L}}{0.00719\text{mequiv}} \end{gathered}$$ (8)

In the event that ${\alpha }_{\text{fb}}^{\text{est}.}=0$ (because r_sample = 0), further analysis may not be needed: α_fb in the sample is likely ~0. For method assurance and to reduce the likelihood that base dissolution mass transfer would be problematic (see below), the estimated m values used were significantly greater than 1. The final actual values of m used were calculated using the final measured α_fb values and amounts of base used, giving an m of at least 3.4 (Table 2).

2.6. Base Addition Comments.

2.6.1. Ammonia Gas—Vial Pressurization/Venting.

When using ammonia gas, limiting v_ammonia to ≤10% of the sample headspace volume will limit initial pressurization of the vial. With v_ammonia = 3 mL, initial pressurization of a ~40 mL headspace will be <10%. The headspace may or may not be vented prior to the base-added h-SPME sampling step. If the headspace is vented, adequate time must be provided for re-establishment of nicotine gas/liquid equilibrium (the venting will temporarily lower the gas-phase nicotine concentration). The challenge posed here when using ammonia gas can be mitigated by lowering the sample aliquot volume.

2.6.2. Ammonia—Dissolution and Mixing.

Neutralization of acidic species by adding ammonia gas requires dissolution in the e-liquid. Near the interface with the gas, this will begin immediately. A fully homogeneous liquid phase is not required for proceeding with the ammonia-added h-SPME step; all that is required is passage of adequate time for liquid- and gas-phase diffusion to establish c_e,fb = c_e,T at the interface.

2.6.3. LB—Dissolution and Mixing.

Using an LB will require full mixing with the sample aliquot. Some organic LB choices may have limited solubilities in PG/GL mixtures, and the considerable viscosity of e-liquids makes them somewhat difficult to mix using a magnetic stir bar. Checking for good method precision can be used to evaluate whether proper mixing was achieved.

2.7. Analysis Sequence.

After the dummy run(s), the replicate vials were designated as a, b, and c. Sample runs (including gas chromatography (GC)/mass spectrometry (MS) analysis) proceeded in the sequence a, b, c, a, b, and c.

2.8. Dummy Run and Pre-base h-SPME Sample Measurement.

On the day of analysis, a dummy run measurement was first carried out. For each sample, 1 mL was placed at room temperature (~20 °C) into the bottom of a vial which was then capped (in triplicate). Vials were stored in the dark at ~20 °C for 1 h to establish nicotine gas/liquid equilibration. For most samples, each initial measurement by h-SPME was carried out after 20 h, immediately preceded by the dummy h-SPME run(s). The SPME fiber assembly with 1.5 cm-long fiber and 65 μm-thick polydimethylsiloxane/divinylbenzene coating was obtained from Supelco/Sigma-Aldrich (Bellefonte, PA). The fiber assembly with a metal sleeve was mounted in a manual SPME holder (Supelco/Sigma-Aldrich). Each cap septum was pre-pierced; a side-port needle was used to prevent coring of the septum. The distance from the bottom of the manual holder to the end of the assembly sleeve was set at 0.6 cm. The metal sleeve was passed through the cap septum until the holder reached the top of the septum. The 0.6 cm distance minimized the length of the potentially sorptive metal sleeve that was exposed to the headspace. The fiber was then extended out of the sleeve and into the headspace. After an exposure time of 20 min, the process was reversed.

2.9. Thermal Desorption of the SPME Fiber.

With the distance between the bottom of the manual holder and the end of the assembly sleeve set at 2.8 cm, the metal sleeve was manually introduced into the 225 °C injector of an Agilent 7890 gas chromatograph. The carrier gas was helium. The fiber was then extended out of the sleeve and into the gas that was flowing through the injector and onto the GC column. Split injection was used with a column/split ratio of 1:10 or 1:20 in the case of high-nicotine samples. The flow rate onto the column was ~1 mL/min. After the desorption time of 5 min, the fiber was retracted into the sleeve, and the holder/assembly was withdrawn from the injector. Quantitative desorption was verified by means of periodic blank runs between samples.

2.10. Post-base h-SPME Sample Measurement.

For the second (post-base) h-SPME measurement, the total for the base amount used ensured that excess base had been added relative to the possible acid additive levels, based on the dummy run. When ammonia was used, v_ammonia was composed of one or more aliquots of ammonia gas (see Table 2). The aliquot(s) was (were) added to each of the three vials using a syringe with a side-port needle. Any intervening time between additions was ~2 min. At ~40 min, a side-port needle was then used to vent pressure from the vial. The 20 min h-SPME exposure began 75–120 min after the final base addition. In the analyses using TEA as an LB, 12 μL was added using a syringe to the e-liquid. Mixing was by means of a mini magnetic stir bar. The h-SPME exposure began at 75 min after base addition (stir bar mixing continued through the 20 min h-SPME exposure).

2.11. GC/MS Conditions.

Analyses were performed with GC as interfaced with an Agilent 5975 mass spectrometer for mass selective detection. The quantitation ion used for nicotine was 84. The GC column type used was Agilent J&W DB-HeavyWax column (30 m × 0.25 mm × 0.25 μm coating thickness). The GC temperature program was 60 °C for 2 min, 20 °C/min to 100 °C, 15 °C/min to 250 °C, and then hold at 250 °C for 2 min. The GC/MS transfer line temperature was 230 °C; the MS electron impact (70 eV) ionization source temperature was 226 °C; and the MS quadrupole region temperature was 150 °C.

2.12. Measurement of pH of 1:10 Water Dilution of an E-Liquid Using CO₂-Free Deionized Water.

Deionized (18.2 MΩ) water was obtained from a MilliporeSigma (Burlington, MA) system. Dissolved CO₂ was removed by sonicating for 30 min and then sparging with helium for 30 min. Under a protective flow of N₂ gas, 9 mL of CO₂-free water was used immediately to dilute 1 mL of e-liquid in a 25 mL microtitration beaker (Mettler Toledo, Columbus, OH), after which the pH was measured using a 100% aqueous buffer-calibrated pH electrode at ~20 °C.

2.13. Calculation of α_fb,w,1:10 for 1:10 Water Dilution of an E-Liquid Using CO₂-Free Deionized Water.

When the diprotonated species can be neglected (which is true except in very acidic solutions), eq 1 becomes

$$\begin{gathered} {\alpha }_{\text{fb}}=\frac{[\text{Nic}]}{[\text{Nic}]+\left[{\text{NicH}}^{+}\right]} \\ =\frac{1}{1+\frac{\left[{\text{NicH}}^{+}\right]}{[\text{Nic}]}} \end{gathered}$$ (9)

For acid/base equlibrium between NicH⁺ and Nic, if K_a is the infinite dilution acidity constant for NicH⁺ in water and solution-phase activity corrections can be neglected²⁶

$$K_{\text{a}}=\frac{\left[{\text{H}}^{+}\right][\text{Nic}]}{\left[{\text{NicH}}^{+}\right]}=\frac{{10}^{-\text{pH}}[\text{Nic}]}{\left[{\text{NicH}}^{+}\right]}$$ (10)

In a 1:10 dilution of an e-liquid with water, if a calibration with 100% aqueous solutions is assumed to hold

$${\alpha }_{\text{fb},\text{w},1:10}=\frac{{10}^{-\text{p}{K}_{\text{a}}}}{{10}^{-\text{p}{K}_{\text{a}}}+{10}^{-{\text{pH}}_{\text{w},1:10}}}$$ (11)

where pH_w,1:10 is the pH value returned by a 100% aqueous buffer-calibrated pH electrode. In water at ~20 °C, it has been measured that pK_a = 8.07.²⁷ If for an aqueous solution, the concentrations of nicotine and added acid(s) are known, the pH can be calculated using methods described by Pankow.²⁶

3. RESULTS AND DISCUSSION

3.1. h-SPME Method Precision.

Table 1 summarizes the results of the method verification tests using lab-prepared nicotine-containing e-liquids that were partially protonated with HCl to give three α_fb values (0.25, 0.50, and 0.75). The method was found to be accurate and precise. The fact that the coefficient of variation (CV, %) value was largest at α_fb = 0.25 is likely due to the fact that distinguishing α_fb from 0 becomes increasingly difficult as α_fb → 0. For example, for α_fb = 0.75, 0.50, and 0.25, an absolute error of 0.02 in α_fb corresponds to CV values of 2.7, 4.0, and 8.0%, respectively. Evidence supporting this interpretation is found in Table 2, which presents the h-SPME values at 20 °C for α_fb and CV for 17 of the 18 commercial e-liquids tested here; a plot of CV versus α_fb is found in Figure 2, showing a significant negative slope, with CV values again generally highest at low α_fb.

Table 1.

Method Verification Values of α_fb Measured by h-SPME at ~20 °C in 50:50 PG/GL Solutions (by Weight); Nicotine at 6 mg/mL Plus Varying Levels of Strong Acid (HCl) to Partially Protonate the Nicotine

αfb as prepared	αfb by h-SPME (N = 3, mean ± 1s)
0.25 ± 0.01	0.23 ± 0.011
	CV = 4.8%a
0.50 ± 0.01	0.50 ± 0.009
	CV = 1.8%
0.75 ± 0.01	0.78 ± 0.019
	CV = 2.5%

^aCV = coefficient of variation = s × 100%/mean.

Figure 2.

CV (%) vs α_fb for measurements on commercial e-liquids by h-SPME. CV = (s/mean) × 100%, s = standard deviation (see also Table 2). The linear regression line is y = −12.5x + 12.54; the negative, non-zero slope is highly significant (99.5%).

3.2. α_fb Value Range in the E-Liquid Sample Set.

The α_fb values by h-SPME at 20 °C in Table 2 for the 18 commercial e-liquids ranged from 0.003 ± 0.00 to 1.03 ± 0.03; the data thus demonstrate the presence in the market of e-liquid products that effectively span the whole of the possible α_fb range.

3.3. h-SPME Versus 1:10 Dilution with Water.

The α_fb,w,1:10 values obtained using pH_1:10 at 20 °C covered the same range as the α_fb values by h-SPME at 20 °C but were biased low against the latter from 0.1 ≤ α_fb ≤ 0.75 (Table 2 and Figure 3). The agreement at the very low portion of the range was presumably a consequence of the presence of considerable mequiv of acid relative to the mequiv of nicotine, so that whether in the native e-liquid or after a 1:10 dilution with water, protonation was extensive. At the upper portion of the range, the agreement was presumably a consequence of significant excess of mequiv of nicotine over acid so that whether in the native e-liquid or after a 1:10 dilution with water, the fraction of free-base nicotine is near 1. It is for intermediate values of acid/nicotine that the greater ability of acids to protonate nicotine in a 1:10 dilution in water versus in the native e-liquid will cause α_fb,w,1:10 to be significantly less than α_fb in the e-liquid itself; the problem will be exacerbated if the water used in the dilution is not CO₂-free.

Figure 3.

α_fb,w,1:10 (with CO₂-free water) vs α_fb in e-liquids by h-SPME. For investigating e-liquid α_fb values, the α_fb,w,1:10 suffers from significantly low bias when α_fb in the e-liquid is not near 0 or 1.

Overall, the h-SPME method has been found to be a reliable and accurate method for determining α_fb values in e-liquids; the method is easily automatable, especially if an LB is used. The α_fb,w,1:10 method was confirmed as seriously flawed.

ABBREVIATIONS

CV

coefficient of variation

DSM

diluted solution method

e-cigarettes

electronic cigarettes

e-liquid

electronic cigarette liquid

EP

equivalence point

GC

gas chromatography

GL

glycerol

h-SPME

headspace SPME

MS

mass spectrometry

NMR

nuclear magnetic resonance spectroscopy

NSM

native solution method

PG

propylene glycol

PM

particulate matter

SPME

solid-phase microextraction