Potential for release of pulmonary toxic ketene from vaping pyrolysis of Vitamin E acetate

Article abstract of DOI:10.1073/pnas.1920925117

A combined analytical, theoretical, and experimental study has shown that the vaping of vitamin E acetate has the potential to produce exceptionally toxic ketene gas, which may be a contributing factor to the upsurge in pulmonary injuries associated with using e-cigarette/ vaping products. Additionally, the pyrolysis of vitamin E acetate also produces carcinogen alkenes and benzene for which the negative long-term medical effects are well recognized. As temperatures reached in vaping devices can be equivalent to a laboratory pyrolysis apparatus, the potential for unexpected chemistries to take place on individual components within a vape mixture is high. Educational programs to inform of the danger are now required, as public perception has grown that vaping is not harmful.

Full text of DOI:10.1073/pnas.1920925117

Potential for release of pulmonary toxic ketene from
vaping pyrolysis of vitamin E acetate
Dan Wua,1 and Donal F. O’Sheaa,1

aDepartment of Chemistry, Royal College of Surgeons in Ireland (RCSI), Dublin 2, Ireland
Edited by Scott H. Randell, The University of North Carolina at Chapel Hill, and accepted by Editorial Board Member Brigid L. Hogan February 11, 2020
(received for review December 3, 2019)
A combined analytical, theoretical, and experimental study has shown
that the vaping of vitamin E acetate has the potential to produce
exceptionally toxic ketene gas, which may be a contributing factor to
the upsurge in pulmonary injuries associated with using e-cigarette/
vaping products. Additionally, the pyrolysis of vitamin E acetate also
produces carcinogen alkenes and benzene for which the negative long-
term medical effects are well recognized. As temperatures reached in
vaping devices can be equivalent to a laboratory pyrolysis apparatus,
the potential for unexpected chemistries to take place on individual
components within a vape mixture is high. Educational programs to
inform of the danger are now required, as public perception has grown
that vaping is not harmful.
and the mode of user usage (9, 10). In reality, a vaping device should
be considered as a crude uncontrollable minipyrolysis apparatus ca-
pable of causing changes to the chemical composition of a vaping
mixture in an unpredictable manner. Specifically, the aryl acetate
functional group present within 1 attracted our attention as a func-
tional group sensitive to pyrolysis. The same functional group is
present in phenyl acetate 3, the simplest aryl acetate, which is known
to liberate the highly toxic gas ketene 4 under pyrolysis (Fig. 1). In this
article, we describe our initial efforts to illustrate and highlight the
potential of vitamin E acetate to release ketene under pyrolysis con-
ditions, and that such conditions are achievable using a vaping device.
Pyrolysis of phenyl acetate 3 undergoes a thermal unimolecular
elimination reaction to form phenol and ketene 4 at temperature
of 625 °C, with ketene obtained in an 84% yield (Fig. 1) (11). This
pyrolysis has been studied using electron ionization (EI) mass
spectrometry to generate conditions to promote the elimination of
ketene (12). Ketene is manufactured by pyrolysis of acetic acid at
700–800 °C and is used for synthetic acetylation reactions and for
the production of bulk reagents (13). It is a colorless gas (boiling
point −56 °C) with a penetrating odor above 12 ppm. Due to its
extreme reactivity, the public extremely rarely encounters ketene,
as when industrially produced it is not isolated but reacted in situ.
However, it has very high pulmonary toxicity, and is lethal at high
concentrations. At lower concentrations in animal studies, minor
irritation during exposure and central nervous system impairment
were observed (14, 15). It can acetylate nucleophilic components
of proteins in aqueous solution and the pulmonary effects of
vaping pyrolysis ketene vitamin E acetate lung injury
|
|
|
|
he Centers for Disease Control and Prevention (CDC) and
Food and Drug Administration have recently reported an in-
T
creasing number of clinically reported cases of pulmonary injury
following use of e-cigarette/vaping products (1). The cause(s) of this
growing epidemic of vaping associated pulmonary injury remain
unidentified, although studies in mice have shown that chronic ex-
posure to e-cigarette vapor abnormally alters the physiology of lung
epithelial cells with changes in lipid homeostasis being independent
of nicotine (2). Vitamin E acetate 1 is used as an additive in some
vaping products and was recently reported as one possible causative
agent (Fig. 1). Results from a multicentered study have identified it
in bronchoalveolar lavage fluid specimens from 48 of 51 patients
examined (3). As the synthetic O-acetylated analog of vitamin E 2, it
is widely used in cosmetic skincare products and as a dietary sup-
plement. The acetylated version of the vitamin is preferred due to
its more facile purification during production, its increased formu-
lation stability, and as ester hydrolysis in vivo generates vitamin E
(4). As it is a viscous lipid oil, 1 is added to some vape mixtures,
especially to those containing tetrahydrocannabinol and cannabidiol
oils. These mixtures are growing in use for nonmedical recreational
purposes, which appears to be closely linked to the current medical
crisis (1).
Significance
The Centers for Disease Control and Prevention has recently
reported an increasing number of clinically cases of lung injury
following use of vaping products. The cause(s) of this growing
epidemic of vaping-associated pulmonary injury remain un-
identified, although vitamin E acetate has been identified as
one possible causative agent. In this research, a combined an-
alytical, theoretical, and experimental study has shown that
the vaping of vitamin E acetate has the potential to produce
exceptionally toxic ketene gas, which may be a contributing
factor to the upsurge in lung injuries associated with vaping
products. Our important results will help inform medical prac-
titioners, research scientists, and the public, which may assist in
bringing this medical crisis under control.
In the first preliminary reports, lipid accumulation in the lungs
as a result of vaping was proposed to play a role in causing ex-
ogenous lipoid pneumonia (5, 6). In a subsequent report of six
patients suffering from vaping-associated lung injury, computed
tomography scans did not show known features consistent with
exogenous lipoid pneumonia (7). Most recently, pathology of 17
patients with vaping-associated lung injury revealed patterns more
consistent with airway-centered chemical pneumonitis from in-
haled toxic substance(s) rather than exogenous lipoid pneumonia
(8). To date, the causative agent(s) remain unidentified and with
so many parameters and variables involved, it appears unlikely that
vitamin E acetate alone would have such acute and severe toxicity.
However, the chemical stability and reactivity of 1, under high-
temperature pyrolysis conditions, has not been investigated. Crit-
ically, the temperature of heating coils within vaping devices is
highly variable and has been measured to range between 110 and
1,008 °C. Several factors can contribute to the actual temperature(s)
applied to the complex vape mixture, such as the heating coil re-
sistance, voltage applied, composition of the mixture being vaporized,
Author contributions: D.F.O. designed research; D.W. and D.F.O. performed research;
D.W. and D.F.O. analyzed data; and D.F.O. wrote the paper.
The authors declare no competing interest.
This article is a PNAS Direct Submission. S.H.R. is a guest editor invited by the Editorial Board.
This open access article is distributed under Creative Commons Attribution-NonCommercial-
NoDerivatives License 4.0 (CC BY-NC-ND).
See online for related content such as Commentaries.
1To whom correspondence may be addressed. Email: donalfoshea@rcsi.ie or danwu@
rcsi.ie.
This article contains supporting information online at https://www.pnas.org/lookup/suppl/
doi:10.1073/pnas.1920925117/-/DCSupplemental.
First published March 10, 2020.
www.pnas.org/cgi/doi/10.1073/pnas.1920925117
PNAS
|
March 24, 2020
|
vol. 117
|
no. 12
|
6349–6355

in various animal species and these studies indicate that it has
similar clinical effect and mode of action as phosgene (COCl2) (16).
In toxicity studies with primates, the minimum lethal in-air con-
centration was determined to be 200 ppm, causing death after a
single 10-min exposure (17).
As vitamin E acetate contains a similar acetate functional found
in phenyl acetate, it appeared plausible to us that it too may pro-
duce ketene upon pyrolysis (Fig. 1). To investigate this possibility, a
threefold investigation was adopted utilizing mass spectrometry,
density-functional theory (DFT) calculations, and analysis of iso-
lated products from vaped vitamin E acetate.
Mass Spectrometry Study. Phenyl acetate 3 was analyzed using
atmospheric pressure chemical ionization (APCI) mass spec-
trometry (MS) at a series of increasing source cone voltages
(CVs) from 5 to 30 eV. At 15 eV the parent ion [M+H]+ was
identifiable as the mass-to-charge ratio (m/z) of 137.1 (10%) with
fragmentation ion at 95.2 (100%) (Fig. 2A and SI Appendix, Fig.
S1). This corresponds with a mass loss of 42 from the parent ion,
which is equivalent to ketene 4 and consistent with previous studies
that used EI MS instrumentation (Fig. 3A) (12). An identical
analysis of vitamin E acetate 1 also showed the elimination of ke-
tene (m/z = 42) from the parent ion (as seen for 3) and from a
fragment with m/z of 207.2 (Fig. 2B and SI Appendix, Fig. S2). A
representative spectrum taken with CV of 20 eV shown in Fig. 2B
reveals the parent ion m/z of 473.6 (100%) with fragmentation ions
at 431.5 (15%), 207.2 (65%), and 165.2 being the most prominent
features. Increasing the CV to 25 eV decreased the abundance of
the parent ion and increased the fragmentation such that (m/z)
abundances were 473.6 (10%), 431.5 (10%), 207.2 (100%), and
165.2 (10%) (SI Appendix, Fig. S2). Clearly two fragmentation
patterns can occur for 1 with fragmentation pattern A due to
Fig. 1. Chemical structures of vitamin E acetate 1, vitamin E 2, phenyl ac-
etate 3, and ketene 4.
inhalation exposure to 4 may be manifested in the absence of direct
irritation by ketene. Severe damage to the lungs at the alveolar level
may manifest as long as 24 h postexposure. Thus, it is unknown
whether odor detection and minor irritation would provide ade-
quate warning of ketene exposure, especially given the potential for
delayed pulmonary toxicity. The toxicity of ketene has been assessed
Fig. 2. Positive-mode APCI-MS of (A) phenyl acetate 3 with source voltage of 15 eV showing loss of ketene; (B) vitamin E acetate 1 with source voltage of
20 eV showing loss of ketene from both the parent ion and a fragmentation species of m/z 207.2; (C) vitamin E 2 with source voltage of 20 eV with sole
fragmentation to [Ar3OH2]+.
6350
|
www.pnas.org/cgi/doi/10.1073/pnas.1920925117
Wu and O’Shea

Fig. 3. APCI-MS fragmentation patterns of A phenyl acetate 3 and B vitamin E acetate 1 each of which are observed as +H+ masses in the mass spectra.
elimination of ketene 4 from the molecular ion producing vitamin at 69.5 kcal/mol, indicating comparable pyrolysis reactivity for
E 2 ([Ar1OH2]+, m/z 431.5) whereas fragmentation pattern B has
both (Fig. 4, blue energy profile). As the MS fragmentation data
for 1 indicated ketene elimination from fragment 6 can also occur,
its energy barrier to release ketene was calculated and found to be
loss of prist-1-ene 7 producing 6 ([Ar2OH2]+, m/z 207.2) from
which elimination of 4 generates [Ar3OH2]+ (m/z 165.2) (Fig. 3B).
As a control, vitamin E 2 was also included in the MS study
showing, as expected, no m/z loss of 42 and showing the same
[Ar3OH2]+ (m/z 165.2) species observed as the lowest molecular
weight fragment from 1 and attributable to structure 8 (Fig. 2C
and SI Appendix, Fig. S3). As both 1 and 3 showed the same MS
fragmentation pattern of ketene elimination, this provides the
chemical basis by which the pyrolysis of vitamin E acetate would
produce ketene.
As the APCI-MS for 1 and 3 both showed loss of ketene from
the molecular ion but 1 also showed a loss from an intermediate
fragment, we sought to further explore this possibility using DFT
calculations. This would allow a direct comparison of activation
energy barriers between the known ketene producer 1 and vi-
tamin E acetate. The energy profiles and transition states (TS)
for the elimination of ketene from 1, 3 and 6 were calculated in
the gas phase at the M06-2X/6–311G (d,p) level using the
Gaussian 09 package (18).
DFT Theoretical Study. The elimination of ketene from 3 has been
reported to proceed via four-membered TS with a concerted
[1,3] hydrogen shift from methyl to oxygen as shown in Fig. 4
Fig. 4. Calculated energy profiles for the pyrolysis of the three aryl acetate
substrates 1, 3, and 6 via concerted [1,3] hydrogen shift mechanism. Relative
(19, 20).
Gas-phase M06-2X/6–311G (d,p) calculations from 3 gave an
activation energy of 65.4 kcal/mol via this pathway, confirming
that it would be energetically feasible under pyrolysis conditions
(Fig. 4, black energy profile). At this level of theory, the barrier
for ketene elimination from 1 was only 4.1 kcal/mol higher than 3
free energies (kcal mol−1) for starting materials, TS, and products formed for
the pyrolysis of phenyl acetate, vitamin E acetate, and 6 leading to the
elimination of ketene 4. For details of calculation data see SI Appendix, Figs.
S4–S6. Simplified structures used for calculation of 1 and 2 without C15H32
alkyl group.
Wu and O’Shea
PNAS
|
March 24, 2020
|
vol. 117
|
no. 12
|
6351

Fig. 5. Optimized geometries [M06-2X/6–311G(d,p) level] for the TS corre-
sponding to the pyrolysis of phenyl acetate 3 (Left) and vitamin E acetate 1
(Middle) and 6 (Right), leading to the elimination of ketene. Simplified
structures used for calculation of 1 and 2 without C15H32 alkyl group. For
details of calculation data see SI Appendix, Figs. S4–S6.
65.9 kcal/mol, which is lower than 1 and comparable with 3 (Fig. 4,
green energy profile). The optimized geometry of the four-membered
cyclic TS for the unimolecular decomposition of 1, 3, and 6 showed
similarity for each and that ortho-methyl substituents of 1 and 6 ac-
etate did not impede ketene elimination (Fig. 5 and SI Appendix,
Figs. S4–S6). Additionally, intrinsic reaction coordinate calculations
confirmed the TS structures associated with the products and reac-
tants along the minimum energy pathway.
Fig. 6. (Top) 1H NMR spectrum of pure 1. (Bottom) 1H NMR spectrum of
entire isolated vaped mixture from 1. For clarity the peak at 7.36 ppm is
not shown in cropped view, see SI Appendix, Fig. S9 for full spectra. Inset
box (i) indicates VC; box (ii) indicates mixture of VC and NVC; box (iii)
primarily NVC.
As two pathways are operable for the elimination of ketene
from 1, DFT calculations were carried out to determine if either
one was more favorable. The competitive first steps of the two
pathways are either loss of ketene from 1 forming 2 or loss of
prist-1-ene 7 from 1 forming 6, from which subsequent elimi-
nation of ketene would occur (Fig. 3B). Conversion of 1 into 6 +
7 is akin to the pyrolysis of a chroman ring. Chroman pyrolysis
occurs through a retro Diels–Alder reaction to produce ortho-
quinone methide and ethane by C–C and C–O bond cleavage in
the dihydropyran ring (21). Employing gas-phase M06-2X/6–
311G (d,p) calculations, the TS energy barrier for the formation
of 6 from 1 was determined to be 58.8 kcal/mol (SI Appendix, Fig.
S7). This value is lower than conversion of 1 into 4 + 2 by 10.7
kcal/mol. As such, it seems probable that both ketene-producing
pathways may be in operation, i.e., 1 → 4 + 2 and 1 → 6 → 4 + 8
but that the latter is more favorable.
1
Separate H NMR spectra were obtained for both NVC and
VC. Close examination of the NVC spectrum and comparison
with 1 and 2 provided important insights (Fig. 7 and SI Appendix,
Fig. S10). This showed that NVC comprised vitamin E acetate 1 (key
singlet peaks at 2.09, 2.03, 1.98 ppm), 2,3,5,6-tetramethylcyclohexa-
2,5-diene-1,4-dione, (duroquinone) 8 (key singlet peak at 2.01 ppm),
and prist-1-ene 7 (key doublet peak at 4.68 ppm). The assignments of
7 and 8 were further confirmed by 13C spectra, proton-carbon cor-
relations using heteronuclear single quantum coherence spectros-
copy, and heteronuclear multiple bond correlations (SI Appendix, Fig.
S11). Comparison with a spectrum of pure 2 showed that, perhaps
surprisingly, no vitamin E was contained in the mixture (key singlet
peaks at 2.18, 2.13 ppm, two overlapping singlets) (Fig. 7). Integration
of a methyl singlet of 1 versus that for duroquinone 8 gave an esti-
mate of vape-induced pyrolysis between 15–20%. This amount of
pyrolytic conversion was consistent across six independent experi-
ments. The formation of 8 was confirmed by high-performance liquid
chromatography (HPLC) and gas chromatography (GC)-MS exper-
iments (SI Appendix, Figs. S12 and S13).
Experimental Analysis of Vaped 1. With experimentation and the-
oretical evidence for the elimination of ketene from 1, we felt
compelled to directly investigate the effect, if any, on vaping
vitamin E acetate. However, it should be noted that these ex-
periments were not designed to exactly replicate a user’s expe-
rience; rather, the goal was to determine the vaping pyrolysis
effect on 1 as a single pure substance at the chemistry molecular
level. Extrapolation of the exact relevance of these results to the
direct cause of lung injury is beyond the scope of this preliminary
report due to the diversity in vaping devices, mixtures, and their
modes of use. As described in Methods, the isolated mixture of
vaped 1 was collected and analyzed as either the total vaped
mixture or following separation into its volatile (VC) and non-
volatile components (NVC). APCI-MS analysis was inconclusive,
as spectra closely resembled that of pure 1, although it was
suspected that changes may be masked by the fragmentation
pattern of remaining 1 and the formation of hydrocarbon frag-
ments that are not APCI-MS responsive (SI Appendix, Fig. S8).
As such, NMR data were collected for the total vaped material
and individually for the VC and NVC of the vaped mixture. Com-
parison of the 1H NMR spectrum of the entire vaped material with
that of pure 1 was revealing as it showed that significant chemical
transformations had occurred. Numerous peaks with complex split-
ting patterns, not observed for vitamin E acetate, between 4.6 and 7.4
ppm were recorded in the vaped material and the spectral region
between 0.8 and 2.1 ppm showed increased complexity (Fig. 6 and SI
Appendix, Fig. S9).
Fig. 7. Portion of the 1H NMR spectrum of vitamin E acetate 1, vitamin E 2,
and the isolated NVC of vaped mixture. A red asterisk marks the aryl methyl
signal for duroquinone 8 (full spectra in SI Appendix, Fig. S10).
6352
|
www.pnas.org/cgi/doi/10.1073/pnas.1920925117
Wu and O’Shea

The routes to forming the spectrum of compounds generated
by the vape pyrolysis of 1 are illustrated in Fig. 9. In pathway A,
ketene is released from 1 to produce vitamin E which undergoes
C–C and C–O bond breaking of the dihydropyran ring to pro-
duce 8 and 7. Pathway B generates the same mixture of end
products but by a different sequence such that first the dihy-
dropyran ring cleaves followed subsequently by loss of ketene
from 6. Long-aliphatic-chained alkene 7 undergoes further de-
composition to produce the lower molecular weight alkenes
ethene, propene, butadiene, and aromatic benzene. Importantly,
many of these alkenes and benzene are known carcinogenic
constituents within tobacco smoke (22, 23) and elevated amounts
of their oxidative metabolites have been found in a study of
adolescent e-cigarette users (24). Both pathways produced
duroquinone 8, which is an organic oxidant although insufficient
data are currently available of its inhalation hazards (25). The
high ketene reactivity prevented its NMR detection within the
isolated complex volatile mixture. Cumulatively, theoretical DFT
calculations, analytical APCI-MS, and experimental results point
toward pathway B being the predominate route of pyrolysis
under vaping conditions.
Fig. 8. Portion of the 1H NMR spectrum of the isolated VC following vaping
vitamin E acetate 1. See SI Appendix, Fig. S14 for full spectrum including
benzene peak at 7.36 ppm. Red asterisk, 1-methyl-1-alkyl alkenes; blue as-
terisk, propene; black asterisk, butadiene.
As direct isolation of ketene under low-temperature conditions
proved challenging, chemical trapping experiments were next
carried out to verify its formation. Benzylamine 9 was chosen as a
representative amine nucleophile as the chemical shift of its
methylene group (s, 3.85 ppm) appears in a spectral region devoid
of peaks in the vape mixture (between 4.6 and 3.8 ppm). Experi-
mentally, the vaped mixture of 1 was passed through a trap con-
taining a solution of benzylamine in CDCl3 at room temperature.
Subsequent NMR analysis clearly showed an additional doublet at
4.36 ppm (not present in the vape mixture alone) which was
consistent with an authentic sample of the acetylation benzylamide
product 10 (Fig. 10 and SI Appendix, Figs. S15 and S16). In-
tegration of the benzylamide methylene peak, from three in-
dependent experiments, with that of the methyl signal for 8
approximated that 30% of ketene produced had reacted with 9.
The formation of 10 was confirmed by HPLC and GC-MS ex-
periments (SI Appendix, Figs. S17 and S18).
NMR analysis identified the VC in the mixture as benzene
(key singlet peak at 7.36 ppm), butadiene (key multiplet peak at
6.36 ppm), propene (key d,d,q peaks at 5.83 ppm), ethene (key
singlet peak at 5.41 ppm), 2-methylprop-1-ene (key singlet peak
at 4.66 ppm), and lower molecular weight 1-methyl-1-alkyl-
alkenes (key doublet peak at 4.67 ppm) (Fig. 8). Trace amounts
of tetrahydrofuran, formaldehyde, and short-chain aliphatic al-
dehydes were also observed (SI Appendix, Fig. S14). Propene was
the most abundant volatile material with a three- to fourfold
stoichiometric ratio greater than that of duroquinone 8. This
indicates that upon release of prist-1-ene 7 from 1, it fragments
into numerous propene units and the other identified alkenes.
Trace aldehydes would be oxidation products of these alkenes
and benzene a downstream byproduct from them.
Fig. 9. Rationale of experimental and theoretical results from vape pyrolysis reactions of vitamin E acetate 1.
Wu and O’Shea
PNAS
|
March 24, 2020
|
vol. 117
|
no. 12
|
6353

Methods
DL-α-tocopherol acetate (vitamin E acetate), ( )-α-tocopherol (vitamin E), phenyl
acetate, duroquinone, benzylamine, and chloroform d1 were purchased from
Sigma-Aldrich. Benzylamide was synthesized following a literature procedure
(27). Vaping device (NEXUS P-1 mini) was purchased in a retail outlet in Dublin,
Ireland. Vaping device (30–50 W, 129 A, 0.25 Ω Kanthal coil) was used fully
charged and at 50 W power setting. Mass spectra of each sample were recorded
using an Advion Expression mass spectrometer, with each sample analyzed at a
series of source voltages of 5, 10, 20, and 30 eV. Other instrument parameters
including capillary temperature: 250 °C; capillary voltage 180 eV; source voltage
span 30 eV; source gas temperature 350 °C; APCI corona discharge 5 were
used consistently for all samples. MS solvent system of MeOH/IPA/H2O/
CHO2H in a ratio of 80:10:10:0.1. Positive mode was superior to negative
mode and was selected for use throughout the study. Positive-mode ESI mass
spectra of vitamin E acetate were complicated by the appearance of sig-
nificant amounts of sodium adducts which were not prevalent in the APCI
spectra, as such APCI was selected as the preferred mode for analysis. NMR
spectra were obtained on a 400-MHz instrument in CDCl3. HPLC analysis
Fig. 10. Chemical trapping of vaped produced ketene with benzylamine 9
forming benzylamide 10. Portion of the 1H NMR spectra showing vape
mixture from 1 isolated in CDCl3 containing 9 and an authentic sample of
benzylamide 10. Black asterisk, methylene peak of 9; red asterisk, methylene
peak of 10. See SI Appendix, Figs. S15 and S16 for full spectra.
The current worrying trend of increasing vaping-associated
lung injuries is due to complex and multifaceted issues encom-
passing social, physical, biological, and medical sciences. When
viewed from a physical science standpoint, medical complica-
tions from vaping a diverse set of substrates is perhaps not un-
expected as the pyrolytic chemistry of single pure compounds is
complex, so what occurs within ill-defined mixtures is a risky
venture into the unknown. The temperatures obtainable within
vaping devices places them in the category of a small-scale lab-
oratory pyrolysis apparatus which if not used with precision and
care can have unforeseen outcomes. In this article, we have
highlighted the chemical basis for concern over vitamin E ace-
tate, just one known component of vape mixtures. Our primary
concern initially lay with the ability of the aryl acetate functional
group to eliminate as ketene, although other reactive and known
carcinogen (alkenes, benzene) are also produced for which the
negative long-term effects are well recognized. Thermal activa-
tion of aryl acetates above their decomposition temperatures
leads to the formation of highly toxic ketene, which if inhaled
into the lungs, even in small quantities, can cause severe pul-
monary injury. Evidence for ketene formation from vitamin E
acetate was obtained from APCI-MS, DFT calculations, and
trapping experiments. In closing, it is important to note that it is
most likely that other aryl O-acetates would eliminate ketene in a
similar manner to the two substrates investigated in this article.
Additionally, published data on the pyrolysis of flavor ingredi-
ents and additives used in vape products verified the generation
of a spectrum of known carcinogens (26). Considering the con-
tinuing evolving and large number of natural and synthetic
substances used in recreational vaping and the unknown chem-
istries that may occur under vaping pyrolysis conditions, urgent
research into this topic is now required. The most efficient ap-
proach would appear to be the development of pyrolysis pre-
diction software, cross-referenced with toxicity data, allowing the
rapid identification of individual compounds and compound
mixtures that pose the most serious risks.
Fig. 11. Apparatus setup for trapping vaped products of vitamin E acetate
1. Setup A: Apparatus for low-temperature trapping of all vaped material.
Setup B: Apparatus for low-temperature trapping of all vaped material
followed by separation of VC from NVC. Setup C: Apparatus for trapping of
vape-produced ketene by benzylamine in CDCl3.
6354
|
www.pnas.org/cgi/doi/10.1073/pnas.1920925117
Wu and O’Shea

used an Atlantis dC18 reverse-phase column (5 μM, 4.6 × 250 mm with UV-vis
detector at 254 and 220 nm, mobile phase ACN/H2O 30:70 to 90:10 from
0.01–30 min, flow rate 1 mL/min).
isolate the traps from the vaping device. Trap 1 was removed from its liquid N2
bath and allowed to warm to room temperature. Once trap 1 reached room
temperature, it was isolated from trap 2 by closing stopcock 2. Following
isolation of traps 1 and 2 from each other, CDCl3 (1.0 mL) was added to each
trap. Trap 2 was removed from its liquid N2 bath and allowed to warm to room
temperature. The CDCl3 was removed from both traps and immediately ana-
lyzed by NMR. Spiking vape samples with pure benzene and duroquinone
confirmed their assignments. The formation of 8 was confirmed by HPLC, GC-
MS experiments, and accurate mass measurement.
Apparatus for Vape Pyrolysis Trapping Experiments. Glass traps were purchased
from Rettberg GmbH, catalog number 137082015. Rubber tubing was used to
connect vaping device to the glass trap 1, the glass traps together, and glass trap
to the vacuum source. Low-form Dewar flasks (Sigma-Aldrich catalog number
Z150398) were used for liquid nitrogen baths. Schematics of the apparatus
configurations are shown in Fig. 11 and photographs showing the apparatus
used are shown in SI Appendix, Fig. S19. The average weight of isolated vaped
material was measured from three independent experiments at 2.4 mg
per puff.
Procedure for Trapping Vape Pyrolysis-Produced Ketene 4 with Benzylamine 9.
All vaping experiments were conducted in a fumehood using the apparatus
setup as shown in Fig. 11C. The vaping device containing 1 was connected via
rubber tubing to the stopcock which in turn was connected to the glass trap,
the outlet of which was connected by rubber tubing to the vacuum source.
Benzylamine 9 (0.02 mL, 0.183 mmol) in CDCl3 (2.0 mL) was added into the glass
trap. The vaping device was turned on for 5 s with an airflow (65 mL/min) ap-
plied, generated by a vacuum, to draw the vapor (puff volume 35 mL) into the
trap. This was repeated 30 times at 30-s intervals and the airflow was stopped by
closing the vacuum. The CDCl3 was removed from the trap and immediately
analyzed by NMR and reverse-phase HPLC (Atlantis dC18 reverse-phase col-
umn, mobile-phase ACN/H2O 30:70 to 90:10 from 0.01–30 min, flow rate 1 mL/min).
Benzylamide 10 was identified in vape mixture NMR by a doublet peak at
4.36 ppm and confirmed by spiking the vape sample with pure 10. HPLC
retention time for benzylamide in the vape mixture was 5.0 min, which was
confirmed by spiking the vape mixture with an authentic sample of 10. The
formation of 10 was confirmed by GC-MS experiments and accurate mass
measurement. Excess of benzylamine was removed by aqueous acid ex-
traction prior to GC-MS analysis. A control experiment showed no formation
of benzylamide 10 following mixing 1 and 9 in CDCl3 for 24 h.
Procedure for Vape Pyrolysis with Low-Temperature Trapping. Vaping exper-
iments were conducted in a fumehood using the apparatus setup as shown in
Fig. 11A. The vaping device containing 1 was connected via rubber tubing to
the stopcock which in turn was connected to the glass trap, the outlet of
which was connected by rubber tubing to the vacuum source. Once assem-
bled the glass trap was placed into a liquid N2 bath for 2 min and vaping
then commenced. The vaping device was turned on for 5 s with an airflow
(65 mL/min) applied, generated by a vacuum, to draw the vapor (puff vol-
ume 35 mL) into the trap. This was repeated 8 times at 30-s intervals and the
airflow was stopped by closing the vacuum. To analyze the vaped material,
the trap was isolated from the vaping device by closing the stopcock, CDCl3
(1.0 mL) was added to the trap, the trap removed from the liquid N2 bath,
and allowed to warm to room temperature. The CDCl3 was removed from
the trap and immediately analyzed by NMR.
Procedure for Vape Pyrolysis with Low-Temperature Trapping and Separation
of VC from NVC. All vaping experiments were conducted in a fumehood using
the apparatus setup as shown in Fig. 11B. The vaping device containing 1 was
connected via rubber tubing to stopcock 1 which was connected to glass trap
1. The outlet of glass trap 1 was connected to stopcock 2, which was connected
to trap 2 which was connected by rubber tubing to the vacuum source. Once
assembled, glass traps 1 and 2 were placed into individual liquid N2 baths for
2 min and vaping then commenced. The vaping device was turned on for 5 s
with an airflow (65 mL/min) applied, generated by a vacuum, to draw the
vapor (puff volume 35 mL) into trap 1. This was repeated 8 times at 30-s in-
tervals, the airflow stopped by closing the vacuum, and stopcock 1 closed to
Data and Materials Availability. All data are available in the main text or SI
Appendix. Raw data files are available from the corresponding author upon
request.
ACKNOWLEDGMENTS. D.W. acknowledges the Synthesis and Solid State
Pharmaceutical Centre for postdoctoral funding support. Thanks to the Irish
Centre for High-End Computing and Dr. G. Sanchez for use of computational
hardware and software resources. Thanks to Emmet Campion (RCSI) for
assistance with mass spectrometry and NMR experiments.
1. H. Ledford, Scientists chase cause of mysterious vaping illness as death toll rises.
Nature 574, 303–304 (2019).
15. R. M. Mendenhall, H. E. Stokinger, Tolerance and cross-tolerance development to
atmospheric pollutants ketene and ozone. J. Appl. Physiol. 14, 923–926 (1959).
16. L. S. Hardison, Jr, E. Wright, A. F. Pizon, Phosgene exposure: A case of accidental
industrial exposure. J. Med. Toxicol. 10, 51–56 (2014).
17. J. F. Treon et al., Physiologic response of animals exposed to air-borne ketene. J. Ind.
Hyg. Toxicol. 31, 209–219 (1949).
18. M. J. Frisch et al., Gaussian 09 (Revision A.1, Gaussian, Inc., Wallingford, CT, 2009).
19. R. Taylor, The mechanism of thermal eliminations, part 25. Arrhenius data for py-
rolysis of isochroman-3-one, benzyl methyl ether, 2-hydroxyethylbenzene, phenyl
acetate, and 3,4-dihydro-2H-pyran. J. Chem. Soc. Perkin Trans. 2, 183–189 (1988).
20. E. Ghibaudi, A. J. Colussi, Very low pressure pyrolysis of phenyl acetate. Int. J. Chem.
Kinet. 16, 1575–1583 (1984).
21. J. Bland, G. da Silva, A detailed chemical kinetic model for pyrolysis of the lignin
model compound chroman. AIMS Environ. Sci. 1, 12–25 (2014).
22. K. A. Persson, S. Berg, M. Törnqvist, G. P. Scalia-Tomba, L. Ehrenberg, Note on ethene
and other low-molecular weight hydrocarbons in environmental tobacco smoke. Acta
Chem. Scand. B 42, 690–696 (1988).
2. M. C. Madison et al., Electronic cigarettes disrupt lung lipid homeostasis and innate
immunity independent of nicotine. J. Clin. Invest. 129, 4290–4304 (2019).
3. B. C. Blount et al.; Lung Injury Response Laboratory Working Group, Vitamin E acetate
in bronchoalveolar-lavage fluid associated with EVALI. N. Engl. J. Med. 382, 697–705
(2020).
4. M. A. Keen, I. Hassan, Vitamin E in dermatology. Indian Dermatol. Online J. 7, 311–
315 (2016).
5. J. E. Layden et al., Pulmonary illness related to E-cigarette use in Illinois and Wis-
consin–Preliminary report. N. Engl. J. Med., 10.1056/NEJMoa1911614 (2019).
6. S. D. Maddock et al., Pulmonary lipid-laden macrophages and vaping. N. Engl. J. Med.
381, 1488–1489 (2019).
7. T. S. Henry, J. P. Kanne, S. J. Kligerman, Imaging of vaping-associated lung disease. N.
Engl. J. Med. 381, 1486–1487 (2019).
8. Y. M. Butt et al., Pathology of vaping-associated lung injury. N. Engl. J. Med. 381,
1780–1781 (2019).
9. W. Chen et al., Measurement of heating coil temperature for e-cigarettes with a “top-
coil” clearomizer. PLoS One 13, e0195925 (2018).
23. K. G. Darrall, J. A. Figgins, R. D. Brown, G. F. Phillips, Determination of benzene and
associated volatile compounds in mainstream cigarette smoke. Analyst (Lond.) 123,
1095–1101 (1998).
24. M. L. Rubinstein, K. Delucchi, N. L. Benowitz, D. E. Ramo, Adolescent exposure to toxic
volatile organic chemicals from e-cigarettes. Pediatrics 141, e20173557 (2018).
25. A. G. Siraki, T. S. Chan, P. J. O’Brien, Application of quantitative structure-toxicity
relationships for the comparison of the cytotoxicity of 14 p-benzoquinone conge-
ners in primary cultured rat hepatocytes versus PC12 cells. Toxicol. Sci. 81, 148–159
(2004).
10. O. Geiss, I. Bianchi, J. Barrero-Moreno, Correlation of volatile carbonyl yields emitted
by e-cigarettes with the temperature of the heating coil and the perceived sensorial
quality of the generated vapours. Int. J. Hyg. Environ. Health 219, 268–277 (2016).
11. C. D. Hurd, F. H. Blunck, The pyrolysis of esters. J. Am. Chem. Soc. 60, 2419–2425
(1938).
12. H. Nakata, A. Tatematsu, Mechanism of mass spectral ketene elimination from phenyl
acetates. Org. Mass Spectrom. 5, 1343–1344 (1971).
13. R. Miller, C. Abaecherli, A. Said, B. Jackson, “Ketenes” in Ullmann’s Encyclopedia of
Industrial Chemistry (Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany,
2012), vol. 20, pp. 171–183.
26. Zs. Czégény et al., Thermal behavior of selected flavour ingredients and additives
under simulated cigarette combustion and tobacco heating conditions. J. Anal. Appl.
Pyrolysis 121, 990–204 (2016).
14. National Institute for Occupational Safety and Health (NIOSH), Ketene. https://www.
cdc.gov/niosh/idlh/463514.html. Accessed 24 February 2020.
27. X. Wang et al., Synthesis and biological evaluation of phenoxyacetic acid derivatives
as novel free fatty acid receptor 1 agonists. Bioorg. Med. Chem. 23, 132–140 (2015).
Wu and O’Shea
PNAS
|
March 24, 2020
|
vol. 117
|
no. 12
|
6355