Free-Base Nicotine Determination in Electronic Cigarette Liquids by 1H NMR Spectroscopy.

E-liquids usually contain significant nicotine, which will exist primarily in two forms, monoprotonated and free-base, the proportions of which are alterable through the effective pH of the medium. The fraction of nicotine in the free-base form is αfb, with 0 ≤ αfb ≤ 1. When dosed via aerosol, the two nicotine forms have different mechanisms and kinetics of delivery, as well as differing implications for harshness of the inhaled aerosol, so αfb is relevant regarding abuse liability. Previous attempts to determine αfb in electronic cigarette liquids and vapor have been flawed. We employed the exchange-averaged 1H NMR chemical shifts of nicotine to determine αfb in samples of e-liquids. This method is rapid and direct and can also be used with collected aerosol material. The e-liquids tested were found to have 0.03 ≤ αfb ≤ 0.84. The αfb values in collected aerosol liquid samples were highly correlated with those for the parent e-liquids. E-liquids designed to combine high total nicotine level (addictive delivery) with low αfb (for ease of inhalation) are likely to be particularly problematic for public health.

In the United States during 2016, electronic cigarettes (e-cigarettes) were used regularly by ∼8 million adults.[1,2] For high school students, CDC surveys estimate e-cigarette use in 2013, 2014, 2015, and 2016 to have been 5, 13, 16, and 11%, respectively, and for conventional cigarettes 13, 9, 9, and 8%, respectively.[2,3] Often argued[4] though not proven to be safer than conventional cigarettes,[5,6] e-cigarettes are not, in any case, risk free. And, many e-cigarette liquids (e-liquids) contain substantial nicotine, which is addictive and can be toxic.

Nicotine has three forms: free-base (Nic, aka unprotonated), monoprotonated (NicH+), and diprotonated (NicH22+). The protonation state of nicotine can be altered by changing the acid/base conditions in the medium.[7,8] In water at 25 °C, pK1 (for NicH22+) and pK2 (for NicH+) are 3.10 and 8.01, respectively.[9] Tobacco smoke aerosols are believed to contain primarily the Nic and NicH+ forms (Figure ) because conditions in the aerosol particulate material (PM) are not considered to be sufficiently acidic to generate significant NicH22+.[7,8]

Figure 1

Distribution of nicotine in vape and tobacco aerosols primarily involves two forms: (left) Nic (free-base) which has volatility; and (right) NicH+ (monoprotonated) which is nonvolatile. The fraction of the free-base for (αfb) depends on the acid/base conditions. In water at 25 °C, pKa = 8.01.

The fraction of nicotine in the free-base form is αfb, with 0 ≤ αfb ≤ 1:where NicH22+ is neglected. The αfb can affect the kinetics and location of nicotine uptake from an inhaled aerosol because the free-base form is volatile: it can deposit from an inhaled tobacco smoke (or vape) aerosol from the gas phase and by particle deposition, whereas only particle deposition is operative for protonated nicotine.[10] It has been argued that these considerations make it likely that αfb affects nicotine addiction potential.[11,12] In addition, high αfb values have long been connected with tobacco smoke harshness upon inhalation.[13]

In water, neglecting NicH22+where Ka is the acidity constant for NicH+ in water (K2 as given above). Other than nicotine level, commercial labels on e-liquid products currently provide little compositional information, and these labels certainly do not indicate αfb values.

Historically, methods for determination of αfb in tobacco smoke PM have been flawed.[10] One method introduced a significant amount of water for subsequent measurement of the pH of the aqueous phase,[14] and a second introduced water and an organic solvent (e.g., chloroform) for what was intended to be a selective extraction of the neutral free-base form.[15] Given the disrupting effects of added liquids, neither method can give good results. Pankow et al.[16] describe a successful method for αfb determination in tobacco smoke PM that uses equilibration with a gas volume as a means to detect volatile nicotine, which is taken to be proportional to αfb. In addition, direct measurement by 1H NMR spectroscopy of αfb is possible for tobacco smoke PM[17] and for PM from the now-defunct Eclipse product[7] which gave aerosols compositionally similar to those from e-liquids. (Others attempted using NMR, but added a solvent that will perturb αfb.[18]) Our work reported here describes the development of 1H NMR spectroscopy for measurement of αfb in e-liquids and their aerosols. The materials and methods are provided in the Supporting Information.

For each sample, nicotine 1H chemical shifts (δ) were measured for different protons on the nicotine molecule (Ha through He). The assignments are in accordance with those previously made[17] and verified by the J-coupling patterns and integrations. The δ of He was subtracted from Ha through Hd to obtain the difference, Δδ, as in eq , noting that Δδ depends on its position in the molecule, that is, some of the protons shift more than others.

Nicotine standards (24 mg nicotine / mL in PG/GL mixtures; see Supporting Information) were then used to calculate Δδ for the monoprotonated and free-base states of nicotine after assessment with a variety of acids and concentrations thereof. In practice, we used only the aromatic protons Ha and Hb to avoid steric or direct charge contributions that may affect the chemical shifts of Hc and Hd; these protons being proximal to the nicotine pyrrolidine ring. Commercial e-liquid samples were then evaluated by the use of eq , with the resonances indicated in Figure :[17]

Figure 2

1H NMR spectra showing the chemical shift changes for nicotine in a propylene glycol + glycerol (PG + GL) stock mixture with the addition of acid and base, independently. (A) 1 × t-butylamine added (relative to moles nicotine). (B) PG + GL e-liquid stock (no acid or base additives). (C) 5 × acetic acid added. Stock mixture contained 54 PG:46 GL (by moles) and 24 mg/mL nicotine. Samples were prepared by isolating the e-liquid sample in an inner concentric NMR tube, with DMSO-d6 lock solvent in the outer tube, at 40 °C.

Thus, for “Taurus” (using the Ha and He chemical shifts):

Free-base fractions (αfb) for a selection of commercial e-liquids were also calculated; the results are shown in Figure , with αfb ranging from 0.03 to 0.84.

Figure 3

Free-base nicotine fraction (αfb) in commercial e-liquids as an average using aromatic protons Ha and Hb. The ranges between free base values are indicated. Nicotine amounts as indicated to the right of each name were determined by NMR integrations, relative to the PG and GL resonances.

The accuracy of the method was verified by adding acid and base, respectively, to “Zen” flavored e-liquid aliquots. The resulting free-base and protonated direct chemical shift values were used to calculate αfb = 0.83 ± 0.00 (range), which was statistically equal to the overall-calibration derived value of 0.84 ± 0.01 (range), using eq as before.

As an initial examination of how vaporization may affect αfb, e-liquids with high and low αfb values were vaporized, and the PM collected and analyzed. The “Zen” e-liquid, which had the highest free-base content of the e-liquids tested, was found to have a post-vaporization αfb of 0.80 ± 0.01 (range), which is similar to the unvaporized value of 0.84 ± 0.01. “Maui” (24 mg/mL) was determined to have a post-vaporization αfb of 0.78 ± 0.01 (range), which is comparable to the unvaporized αfb, which was 0.80 ± 0.00. The JUUL “crème brulee” flavored e-liquid was found to have a post-vaporization αfb of 0.05 ± 0.03 (range), also comparable to its unvaporized value of 0.07 ± 0.02. JUUL e-liquids are advertised to contain benzoic acid, which we verified by NMR as being present primarily in its ionic, benzoate form.

The NMR method presented here may be compared with contemporary analogs for e-liquids of the two historical methods for αfb in tobacco smoke PM. First, Stepanov and Fujioka,[19] Lisko et al.,[20] and El-Hellani et al.[8] all describe diluting an aliquot of e-liquid with water, measuring the pH, and then calculating αfb by eq . The result is that the values obtained suffer from both medium effects (water is different from an e-liquid) and dilution, though the pH values may, nevertheless, provide some useful relative indications of the overall acid/base balances in different e-liquids. However, that can be compromised if air-related CO2 is present in the added water and affects the measured pH values. This problem is likely evidenced in the data of Lisko et al.[20] (see Supporting Information). Second, El-Hellani et al.[8] describe making 6 mL aqueous solutions of e-liquids, extracting with 6 mL toluene, and then determining nicotine in the toluene solvent extract as a measure of the nicotine percentage in the water. This approach suffers from the same dilution, medium, possible CO2 incursion effects discussed above and introduces uncertainties regarding the extent to which the toluene extraction step affects the position of the NicH+ ⇆ Nic + H+ equilibrium in the aqueous dilution.

In order to confirm the above concern directly, the JUUL “crème brulee” e-liquid was diluted into D2O to determine if αfb was affected by dilution into this deuterium analog of water. The dilution (5:1, by volume) was found to result in fully monoprotonated nicotine.

Although we used a 600 MHz NMR system for this work, it is possible that these methods could be adapted for lower field NMR, and even benchtop instruments. This is a rapid and easy way to measure αfb in e-liquids accurately and may be of interest to those concerned with addiction and regulation.

In summary, αfb of e-liquids can be determined directly by 1H NMR using protonation-dependent chemical shifts for nicotine. In a small number of tests, αfb values were found to be largely unaffected by the vaping process. Of the products tested, only the JUUL liquids were found to combine high nicotine levels with low αfb values. Pharmacokinetic uptake rates for nicotine may vary among the products, and certainly tobacco company documents (e.g., Chen)[13] suggest that products with high nicotine levels but low αfb such as JUUL will yield vape aerosols of much reduced harshness as compared to products with even only moderate nicotine levels but αfb ≈ 1. This may well contribute to the current use prevalence[21] of JUUL products among youth.