JournalofChromatographyA,1581–1582(2018)105–115
ContentslistsavailableatScienceDirect
Journal
of
Chromatography
A
jou rn al h om ep a g e : w w w . e l s e v i e r . c o m / l o c a t e / c h r o m a
Comprehensive
comparative
compositional
study
of
the
vapour
phase
of
cigarette
mainstream
tobacco
smoke
and
tobacco
heating
product
aerosol
夽
Benjamin
Savareear
a,
Juan
Escobar-Arnanz
a,
Michał
Brokl
b,
Malcolm
J.
Saxton
b,
Chris
Wright
b,
Chuan
Liu
b,
Jean-Franc¸
ois
Focant
a,∗aCentreforAnalyticalResearchandTechnologies(CART),UniversityofLiege,Belgium bResearchandDevelopment,BritishAmericanTobacco,Southampton,UK
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received20August2018
Receivedinrevisedform12October2018 Accepted16October2018
Availableonline19October2018 Keywords:
Mainstreamtobaccosmoke Vapourphase
Tobaccoheating(heat-not-burntobacco) product
Thermaldesorption
Comprehensivetwo-dimensionalgas chromatography
Highresolutiontime-of-flightmass spectrometry
a
b
s
t
r
a
c
t
Asimpledirectsamplecollection/dilutionandintroductionmethodwasdevelopedusingquartzwooland Tenax/sulficarbsorbentsforthermaldesorptionandcomprehensivetwo-dimensionalgas chromatogra-phy(TD-GC×GC)analysesofvolatileorganiccompoundsfromvapourphase(VP)fractionsofaerosol producedbytobaccoheatingproducts(THP1.0)and3R4Fmainstreamtobaccosmoke(MTS). Analy-seswerecarriedoutusingflameionisationdetection(FID)forsemi-quantificationandbothlowand highresolutiontime-of-flightmassspectrometry(LR/HR-TOFMS)forqualitativecomparisonandpeak assignment.Qualitativeanalysiswascarriedoutbycombiningidentificationdatabasedonlinear reten-tionindices(LRIs)withamatchwindowof±10indexunits,massspectralforwardandreverselibrary searches(fromLRandHRTOFMSspectra)withamatchfactorthresholdof>700(bothforwardand reverse),andaccuratemassvaluesof±3ppmforincreasedconfidenceinpeakidentification.Usingthis comprehensiveapproachofdatamining,atotalof79outof85compoundsandatotalof198outof202 compoundswereidentifiedinTHP1.0aerosolandin3R4FMTS,respectively.Amongtheidentified ana-lytes,asetof35compoundswasfoundinbothVPsampletypes.Semi-quantitativeanalyseswerecarried outusingachemicalclass-basedexternalcalibrationmethod.Acyclic,alicyclic,aromatichydrocarbons andketonesappearedtobeprominentin3R4FMTSVP,whereaslargeramountsofaldehydes,ketones, heterocyclichydrocarbonsandesterswerepresentinTHP1.0aerosolVP.Theresultsdemontsratethe capabilityandversatilityofthemethodforthecharacterizationandcomparisonofcomplexaerosol samplesandhighlightedtherelativechemicalsimplicityofTHP1.0aerosolincomparisontoMTS.
©2018TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Forthepurposeoftobaccoharmreduction,newgenerations
oftobacco heating(heat-not-burn- HnB)products(THPs)were
introducedinthemarket[1,2].Suchnewelectronicallycontrolled
heatingdevicessignificantlyimpacttheglobalchemical
composi-tionoftheaerosols,comparedtoconventionalcigarettes.Despite
thelargeamountofknowledgeandstandardisedmethodsexisting
fortheanalysisofconventionalcigarettes,verylittleinformation
夽 Selectedpaperfromthe42ndInternationalSymposiumonCapillary Chromatog-raphyand15thGCxGCSymposium,13–18thMay2018,Italy.
∗ CorrespondingAuthorat:UniversityofLiège,ChemistryDepartment–CART, Organic&BiologicalAnalyticalChemistry,Alléedu6AoûtB6c,B-4000,Liège, Belgium.
E-mailaddress:JF.Focant@uliege.be(J.-F.Focant).
is currently available for THPs. In addition, some studies have
reportedthescientificassessmentofTHPbasedontarget
analy-sesofcompoundstypicallyfoundincombustibleproducts[1,2]
andonlyafewmethodshavebeendevelopedfortheuntargeted
analysisofTHPandcombustiblesamples[3,4].
Themainstreamtobaccosmoke(MTS)thatexits thefilterof
a cigarette is an extremely complex chemical mixture [5,6]. It
consistsofliquid/soliddropletscalledtheparticulatephase(PP),
suspended in a mixture of gases and semi-volatiles called the
vapourphase(VP).Apartfromthebulkgases(nitrogen,oxygen,
carbonoxides,nitrogenoxides,ammonia),VPalsoconsistsofVOCs
andtheirimportanceonproductcytotoxicityandcarcinogenicity
hasbeendemonstratedinseveralcellularandanimalsystems[7].
AlthoughitcanbeexpectedthatTHPaerosolislesscomplexthan
MTS,atpresentthechemicalcompositionoftheVPofTHPaerosol
hasnotbeenfullydescribed[8].Developingnewqualitativeand
https://doi.org/10.1016/j.chroma.2018.10.035
0021-9673/©2018TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4. 0/).
106 B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115
quantitativeanalyticalmethodologiescapableofVPanalysesfor
bothTHPsandMTSisthereforeimportanttoallowpractical
com-parison.
MTSVPanalysesrequirespecificsamplingstrategiestoensure
representativecollectionofthewholechemical profileacrossa
verylargedynamicrange[5,9].Variousapproaches,suchasgas
samplingbags[5],solidphasemicroextraction(SPME)[7],
solvent-filledimpingertrains[10],andcoldtraps[11]havebeenreported
forthispurpose.DirectsamplingofVPonspecificadsorbentshas
alsobeenreported[12–14].Inthelastfewyears,theuseof
adsor-bents has increasedin significance as thermal desorption (TD)
capabilityhasbecomemorewidelyapplicableandrobust.Sucha
solvent-freetechniqueoffersseveraladvantagesoverothersolvent
extractionmethodsthatinvolvemoremanualstepsandoften
suf-ferfromreducedsensitivityduetomultipledilutionstepsbefore
finalmeasurement[15,16].Amongstotherapplications[17,18]TD
hasrecentlybeensuccessfullyusedfortheanalysisofMTSVPof
differentcigarettetypes[5].
Comprehensive two-dimensional gas chromatography
(GC×GC)coupledtotime-of-flightmassspectrometry(TOFMS)is
theestablishedmethodofchoicefordetailedanalysisofVOCsin
medium-to-highcomplexitysamples[17–19].Advantagesofthe
state-of-artGC×GC-TOFMSinstrumentationhavebeendescribed
in severalreports [17,20,21]. In the context ofidentification of
unknowns, an important advantage of GC×GC is the spatial
coordinationofchemicalclassesin2Dchromatogramsthat
pro-videsafurtheridentificationpointinadditiontolinearretention
indices (LRIs), fragmention and accuratemass MSdata during
thepeakassignmentprocess.Inordertoperformqualitativeand
semi-quantitative untargetedanalyses, the simultaneous useof
a flameionisation detector(FID) and aTOFMS (dualdetection)
hasbeenreported[21,22].Despitethepotentialbenefitsof this
approachintermsofspeedandefficiency,relativelyfewreported
applicationsinvolvesuchdualdetectionGC×GCinstrumentation
[23–28]. As GC×GC can now be more easily coupled to fast
acquisitionHRTOFMS,whichoffersmassaccuracyatthesub-ppm
level consistently on deconvoluted mass spectral signals, an
extradimensionofelementalcompositionismadeavailablefor
compound identification [29]. The utility of GC×GC-HRTOFMS
ishoweverconstrainedbythesizeoftheacquireddatafilesand
associateddataprocessingrequirements[30,31].Nevertheless,a
GC×GC-HRTOFMSapproachwassuccessfullyusedforcollecting
accuratemassvalues(±15ppm)forincreasedconfidenceinpeak
identificationofdifferenthopessentialoils[32].
Inthisstudy,wedevelopedanoriginalanalyticalmethodfor
the qualitative and semi-quantitative comparison of VOC
con-stituentsintheVPfractionofTHPaerosolandcombustibleMTS.For
thispurpose, TD-GC×GC-TOFMS/FIDand TD-GC×GC-HRTOFMS
wereusedtomaximizetheidentificationconfidenceforthe
non-targetedscreeningapproach.
2. Materialsandmethods
2.1. Analyticalreagentsandsupplies
Saturatedalkanestandardsolutions(C5–C30)werepurchased
fromSigmaAldrich(Diegem,Belgium). Acustom standard mix
made of benzene, toluene, 2-hexanone, p-xylene, styrene,
bro-mobenzeneando-cresol,waspreparedatconcentrationsof0.25
and0.5g/LfortheoptimizationofthesplitratiobetweenFID
andTOFMSandthetuningoftherecollectionprocess.Acustom
standardmix(SupplementaryTableS1)waspreparedfor
build-ing thecalibration curves for semi-quantificationpurposes. All
standardswerepurchasedfromSigma-Aldrich(Diegem,Belgium),
purity was >99.5%. All solutions were prepared volumetrically
usingmethanol(Sigma-Aldrich)andstoredat4◦C.Quartzwool
andTenax/sulficarbstainlesssteelTDtubeswerepurchasedfrom
Markes(Pontyclun,UK).
2.2. Blankanalysisprocedure
Blank analysiswere performedtoensure analytical systems
were free from contaminations and that possible interferences
wereundercontrol.AllTDtubeswereconditionedpriortheiruse
accordingtomanufacturer’sinstructions.Ablanktestwasalways
performedbeforeuseforeachTDtubetocheckforpossible
carry-over.Thesmokingmachineswerelocatedinseparateroomsto
avoidanyinterferencesontheVOCprofiles.Priortosampling,air
blankanalyseswereconductedoneachsmokingmachine.Forthisa
blankTDtubewasplacedinthesmokingmachineandthesmoking
procedurewasrun withouttobacco consumables.For each
rec-ollectionprocess,ablankrunwasalwaysperformedusingblank
TDtubes intheUnity-2and recollectionunit tocheckthatthe
instrumentwasfreeofcontamination.Afterthisachromatographic
runsequencewasmadeasfollows:instrumentalblank,smoking
machineairblank,recollectionprocedureblank,andan
instrumen-talblankprioranalysesofunknownsamples.Thisprocedurewas
repeatedforeachsampleandreplicateanalyses.
2.3. Samplingprocedure
2.3.1. Samplesandsamplecollection
The THP1.0 (gloTM) device and consumables were provided
byBritishAmericanTobacco(Southampton,UK).Thedescription
ofTHP1.0deviceandsamplingproceduredetailsareillustrated
inFig.S1(SupplementaryInformation).3R4Fresearchreference
cigaretteswereacquiredfromtheUniversityofKentuckyCollege
ofAgriculture(KentuckyTobaccoResearch&DevelopmentCenter,
USA).THPconsumablesandreferencecigaretteswereconditioned
inseparatehumidifierforatleast48hat60%relativeair
humid-ityand22◦C[25].ForTHP1.0samples,VPaerosolsampleswere
generatedusingalinearsyringedrivesystemA14(BorgwaldtKC
GmbH,Germany).Asnostandardpuffingregimehasbeendefined
forTHPsofar,allsamplecollectionswereconductedaccordingto
modifiedHealthCanadaIntense(HCI)puffingregimeforcigarettes
thatconsistedofbell-shapedpuffs,eachof55mLwithpuff
dura-tionof2sandwith30sintervalsbetweenpuffs[33]withairinlet
zonenotoccluded.Thenumberofpuffscorrespondstotheheating
cycleavailableforTHP1.0usedinthepresentstudy(8puffs).One
THPconsumablewasusedforeachanalysis.Forthe3R4Freference
cigarette,MTSVPfractionwasgeneratedusingaBorgwaldtRM20D
smokingmachine(BorgwaldtKCGmbH,Germany).Smokingwas
conductedaccordingtotherelevantISOstandardsapplyinga35mL
puffof2sdurationtakenevery60swithnoblockingoffilter
venti-lationholes[34].Onereferencecigarettewasusedforeachanalysis.
SamplingproceduredetailsareillustratedinsupplementaryFig.S1.
Routineairflowandpuffvolumemeasurementswereperformed
priortothesmokerunsforeachsmokingmachines.ForTD
sam-pling,twoprepackedtubes(1stlevel)wereconnectedinseries,
quartzwoolTDtubesfortrappingthetotalparticulatephase(PP)
fractionandTenax/sulficarbTDtubesfortrappingtheVOC
com-ponentofthevapourphase(VP)fraction.TDtubeswereplacedin
betweentheconsumableandthecorrespondingsyringepumpof
thesmokingmachines.Thetotalvolumeofgasdrawnthroughthe
sorbentsforTHPsamplingwas440mLforeachanalysis.Thetotal
volumeofgasdrawnthroughthesorbentsforreferencecigarette
samplingwas280mLforeachanalysis.TDtubeswerecappedwith
DiffLok® caps(MarkesLtd)directlyafterthesamplingprocedure
B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115 107
2.3.2. VPfractionrecollection/dilutionprocess
TheVP fractionofaerosol/smoke trappedona TDtube was
splitacrossthree2ndlevelTDtubesfilledwiththesamesorbent
priortothechromatographic injectionstoavoid overloadingof
chromatogramsandcontaminationoftheTDunitandGCcolumn.
For thispurpose, a Recollect-10device(SepSolveAnalytical Ltd
(Peterborough,UK)connectedtoUnity-2TDunitthroughaheated
transferlinewasused.Thisinstrumentiscapableofrecollecting
gaseoussamplesacrossuptotenTDtubesat atime. The
dilu-tionproceduredetailsareillustratedinsupplementaryFig.S2.1st
levelTDtubeswereplacedintheUnity-2thermaldesorberfortwo
stagesofdesorptionprocess.Thepre-desorptionstageconsisted
of2minofpre-purgeofthesampletubeatambienttemperature
withaflowrateof50mLmin−1.Duringthetubedesorptionstage,
thesampletubewasdesorbedat320◦Cfor 10minwitha flow
rateof60mLmin−1 andtheflowequallysplitacrossthreetubes
placedinthemanifold.Themanifold,transferlineandflowpath
temperatureweremaintainedat200◦C.TDtubeswerecappedwith
DiffLok®caps(MarkesLtd)directlyaftertherecollectionprocedure
wascompletedandanalysedimmediatelyafterusingTD-GC×
GC-TOFMS/FIDandTD-GC×GC-HRTOFMSinstruments.
2.4. Instrumentalanalysis
TD-GC×GC-TOFMS/FIDanalyseswereperformedusingaLECO
Pegasus 4D (LECO Corp., St. Joseph, MI, USA) GC×GC system
equipped with a quad jet LN2 Cooled Thermal Modulator, a
secondary GC oven, an Agilent Technologies (Santa Clara, CA,
USA)capillaryflowtechnologysplitter,athermaldesorptionunit
(TD-100xr,MarkesLtd.),aLECOPegasustime-of-flightmass
spec-trometer(TOFMS)andanAgilentFlameionizationdetector(FID).
TD-GC×GC-HRTOFMS analyses were performed using a LECO
PegasusGC-HRT4D(LECOCorp.)systemequippedwithaquadjet
LN2CooledThermalModulator,asecondaryGCoven,athermal
desorptionunit(TD-100xr,MarkesLtd.)andaLECOPegasushigh
resolutiontime-of-flightmassspectrometer(TOFMS).Aschematic
overviewoftheTD-GC×GC-TOFMS/FIDdualdetectionset-upand
theTD-GC×GC-HRTOFMSset-upisgiveninsupplementary
infor-mation,Fig.S3.
2.4.1. Thermaldesorptionprocedure
SampleTDtubesunderwentthreestagesofdesorptionprocess.
Thepre-desorptionstageconsistedof0.1minofpre-purgeofthe
sampletube atambienttemperaturewitha flow rateof50mL
min−1.Duringthetubedesorptionstage,thesampletubewas
des-orbedat300◦Cfor8minwithaflowrateof50mLmin−1.Samples
wererecollectedonacoldtrapcontainingproprietarysorbentfrom
Markes(‘sulphur/labile’carbonnumberC3-C30)atatemperatureof
25◦C.Thecoldtrapplussamplewaspurgedfor2minataflowrate
of50mLmin−1toremovepossibletracesofundesirablewaterby
divertingthemtotheventportpriortosampledesorptionintothe
GC×GCinstrument.Thecoldtrapwasdesorbedatthemaximum
heatingrateof24◦Csec−1 from25◦Cto315◦C,atwhichitwas
heldsubsequentlyforanother6minAtransferlineof1.5m
deac-tivatedfusedsilicacolumnwasconnectedbetweentheTD-100xr
andtheGCcolumninletandmaintainedat200◦C.Thesplitratio
forGC×GC-TOFMS/FIDanalysiswassetto1:50andforGC×
GC-HRTOFMSanalysiswassetto1:100.
2.4.2. GC×GC-TOFMS/FIDanalyses
For all analyses a non-polar, 5% diphenyl 95% dimethyl
polysiloxanephase(60m×0.25mmi.d.×0.5mdf)(Rtx-5SilMS,
RestekCorp.,Bellefonte,PA,USA)wasusedasthefirstdimension
(1D)column.Two30mcolumnswereconnectedtomakea60m
column.Theseconddimension(2D)wasamidpolarCrossbond®
silarylene phase column exhibiting similar selectivity to 50%
phenyl/50%dimethylpolysiloxane (1m×0.25mm i.d.×0.25m
df)(Rxi®-17SilMS,RestekCorp.).Allcolumnconnections,
includ-ing thetransferlineof TDto the1Dcolumn and two 30m1D
columnsand2DcolumnweremadeusingSilTiteTM-Unions(SGE
International,Victoria,Australia).Theeffluentof the2Dcolumn
wasdirectedtoanAgilentthree-waycapillary flowsplitterand
theeffluentwassplitviarestrictorsR1andR2totheTOFMSand
FID.Aconstantsplitpressureof3.8PSIwasusedforcalculating
the1:1splitbetweentheTOFMS/FIDsetup,corresponding
restric-tors(deactivatedfusedsilica)wereR1:2.78m×0.15mmi.d.and
R2:0.53m×0.15mmi.d.Thecarriergaswasheliumatacorrected
constantflowrateof1.2mLmin−1.Themainoventemperature
programstartedwithanisothermalperiodat40◦Cfor5min,then
arampof5◦Cmin−1upto300◦C,afinalisothermalperiodat300◦C
for3min.Thesecondaryovenwasprogrammedwitha5◦Coffset
abovetheprimaryoventemperature.Themodulationparameters
consistedofa3smodulationperiod(PM)(0.7shotpulseand0.8s
coldpulsetime)andatemperatureoffsetof15◦Cabovethe
sec-ondaryoventemperature.TheFIDtemperaturewas350◦C,using
anairflowof400mLmin−1,hydrogenflowof40mLmin−1 and
a N2 make upgasflow of 30mLmin−1. TheTOFMS was
oper-atedinelectronimpactmodeusing70eV,detectorvoltagewas
at1800V,ionsourcetemperaturewassetat230◦C,transferline
temperaturewassetat250◦Candmassscanrangem/z35–500at
theacquisitionrateof100spectras−1.Dailymasscalibrationand
autotuningwereperformedusingperfluorotributylamine(PFTBA).
SampleswereacquiredusingChromaTOF®(LECOCorp.)software
version4.50.8.
2.4.3. GC×GC-HRTOFMSanalyses
1D and 2D columns were similar to TD-GC×GC-TOFMS/FID
setupexceptthe2Dcolumnwas1.5mlongin theTD-GC×
GC-HRTOFMSsetup.Sincetherewasnosplitterconfigurationinthis
instrument,the2Dcolumneffluentwasdirectedtothemass
spec-trometer.The mainoven temperatureprogram startedwithan
isothermalperiodat40◦Cfor5min,thenarampof2.2◦Cmin−1up
to130◦C,followedbyarampof2.8◦Cmin−1to300◦Candafinal
isothermalperiodat300◦Cfor3min.Massspectrawereacquired
intherangem/z35–500attheacquisitionrateof100spectras−1,
inahighresolutionmodewitharesolvingpower>25,000FWHM
form/z218.9851.SampleswereacquiredusingLECOChromaTOF®
-HRT(LECOCorp.)softwareversion1.91.Allotherparameterswere
thesameasforGC×GC-TOFMS/FIDanalyses.
2.5. Dataprocessing
Theschemeofdataproductionandprocessingisillustratedin
supplementaryinformation,Fig.S4.TOFMS/FIDdatawereexported
as.pegand.csvfilesforbothTOFMSandFID,respectively.These
datawereseparatelyprocessedandsummarised(retentiontimes,
peak areas, library comparison etc.) using the pixel-based GC
ImageTM (ZOEXCorp.,Houston, TX,USA)softwarepackage
ver-sionR2.5.FortheanalysisofVPandthecomparisonofsamples,
chromatogramswerealignedfollowingaprocedurebasedonthe
creationofatemplatechromatogramthatrecordspeakpatterns
andcarryingoutresamplingofthedatatomatchretentiontimes
usingGCProjectTM,partoftheGCImageTMsoftwarepackage
(Sup-plementary Fig.S5). The following blob detection criteria were
used:area22,volume300,000andSN>200wereappliedoneach
individualchromatogramforTOFMSandFID,basedona
compro-misebetweenthenumberofdetectedsignalsandthequalityofthe
recordedsignals,thelaterdependingstronglyonpeakshapesover
thelargedynamicrange.ForHRTOFMSdata,peakswerematched
usingmassspectralandGC×GCstructuredchromatogrampattern
108 B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115
Fig.1. RepeatabilityoftherecollectionprocessbasedonaveragepeakareaobtainedfromdirectTDinjection(Black),recollectionoveroneTDtube(Grey),andthreetubes (White)(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle).
Qualitative analysis was carried out using linear retention
indices C5-C25 (LRI window ±10) and mass spectral
similar-ity/reverse match (similarity/reverse > 700 MS spectra match)
againstlibraryspectrafromLRTOFMSandHRTOFMSinstruments.
Inaddition,anaccuratemassvalueof<±3ppmwasappliedto
giveincreasedconfidencefortentativelyidentifiedpeaksassuch
mass accuracy values allowed to univocally attribute a
molec-ular formula to most compounds of molecular weights below
200Da [35]. Librarysearchesof blobs/compounds(both TOFMS
andHRTOFMS)wereperformedusingNIST/EPA/NIHMass
Spec-tralLibrary(NIST14)andWileyRegistryofMassSpectralData(9th
edition)withamatchfactorthresholdof>700.Furtheranalysesof
GC×GC–TOFMS/FIDresultsusinginteractiveLRIfilters(±10range)
wereperformedusingNIST/EPA/NIHMassSpectralLibrary(NIST
14)database.IfnotavailableintheNIST14database,WEBbasedRI
collections(PubChem)informationwasused[36].
3. Resultsanddiscussion
3.1. Methoddevelopment
3.1.1. Thermaldesorptionsampling
Based ona previous study[5] thermal desorption(TD) was
usedasa uniqueapproachtoaccommodatethechemical
com-plexityandthelargedynamic rangecovered bycomponentsof
MTSand THPsamples.Despitethefact thatsample production
fromTHPandcombustibleproductsrequiresdifferentapparatus,
wholesmoke/aerosolTDsamplingwascarriedoutbyplacingaset
ofTDtubesintheoutputstream(SupplementaryFig.S1).Atfirst,
asingleTDtube(Tenax/sulficarb)wasusedbutbreakthroughwas
observed(datanotshown),evenonasinglepuffbasis.Basedon
recentresearch[5],weincludedglassfiberfilterpads(Cambridge
filterpads,CFP)whichresolvedthebreakthroughissue
(Supple-mentaryInformation,Fig.S7).TosimplifytheprocesstheCFPwas
replacedbyaTDtubecontainingquartzwool,creatingasample
trainthatcomprisedafirstleveltube(Tube1)containingquartz
woolforparticulatephase(PP)trappingandanotherfirstleveltube
(Tube2)containingTenax/sulficarbforVPtrapping.Theimpacton
backpressureandairflowoftwoTDtubesbetweentheproduct
andthepuffingmachinewasminimized(<10mL/min)bylimiting
thequantityofpackedquartzwoolto700mg/tube.Theefficiency
ofsamplingwasevaluatedbymonitoringpotentialbreakthrough
ofcompoundsusingathirdtubeplacedaftertheTube2.Underthe
abovesetuptherewasnobreakthorughforbothTHP1.0VPand
3R4FMTSVPsamples,whichisasignificantfactortoconsiderfor
comparisonofthesamples(SupplementaryInformation,Fig.S8).
The emissions from a whole consumable were collected in
ordertomanageknownvariationsbetweenindividualpuffs.This
resulted in highly concentrated samples that required dilution
before TD-GC×GC-TOFMS/FID analysis to prevent overloading
of instruments and carry-over between samples. The
recollec-tion/dilutionprocedure(SupplementaryInformation,Fig.S2)was
alsooptimizedtopreventbreakthroughontherecollectiontubes
andtoallowquantitativedilutionofthesamples.Asillustratedin
Fig.1forasetofrepresentativecompounds,acceptablelosseswere
observedwhenaTDtubecontainingasamplewasrecollectedonto
oneorthreeTDtubes.Theaveragelossbasedonpeakareaswas
9%andrangedbetween4%(Benzene)and19%(2-Hexanone),with
RSDsrangingbetween4%(Toluene)and13%(o-Cresol).
The potential impact on sample integrity of the storage of
TD tubes prior to analysis was evaluated by monitoring peak
areasofasetofselectedanalytes(2-Methylbutane,2-Butanone,
Benzene, Dimethylfurane, 2-Hexanone, Pyridine, Ethylbenzene,
Styrene, Limonene) over time. Peak areas did not significantly
vary(<20%)overaperiodoffivedays.Nevertheless,sampleswere
alwaysanalysedwithin48hofcollectiononTDtubes.Tube
des-orptionconditionswereoptimizedtoensurethecompletetransfer
ofallanalytesfromtheTDtubetothe1DGCcolumn.Thiswas
sys-tematicallycontrolledbymonitoringblanklevelsofTDtubesand
ofthethermaldesorbercoldtrap.
3.1.2. Chromatography
TheTD-GC×GC-TOFMS/FIDinstrumentalset-up
(Supplemen-taryinformation,Fig.S3)wasoptimizedbyusingaconstantsplit
pressureapproach. Thedesiredsplitratio(1:1)wasmaintained
during the whole temperature program by calculating column
flowsofbothrestrictors.Thesplitratiowasvalidatedbyreplicate
(n=6)sequentialanalysisofatestmixturebysingle(FID)anddual
detection(FIDandTOFMS).Ratioofaveragepeakareasofthe
sin-gleanddualdetectionanalysisdemonstratedtheefficient1:1split
repro-B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115 109
Fig.2.Apexplotofvapourphasesamplefrom3R4FMTSandTHP1.0aerosolobtainedfromFID(top),LRTOFMS(middle)andHRTOFMS(bottom).
duciblity(averageratioof0,496±0,03;RSDvaluesfrom4,1%to
8,7%).
The chromatograms (three individual chromatograms from
three second level TD tubes) wereprocessed for both FID and
LRTOFMSseparately(Supplementary Information,Fig.S5). After
manuallycleaningtheimagesfromcolumnbleed,artifactpeaks
andmutliplehitsduetopeaktailingeffectsforhighlyabundant
compounds, theywere manually matched on a
compound-by-compoundbasisusingfirstandseconddimensionretentiontime
(1t
Rand2tR)matching.Underthisapproach,thetotalnumbersof
peaksfoundtobepresentintheVPfractionofTHP1.0andMTS
were88and207,respectively.Fig.2showsthereconstructed
chro-matograms(apexplots)obtainedforTD-GC×GC-TOFMS/FIDand
usedfordataprocessing.
InordertoalsocombinehighaccuracyMSdatatocompound
identification,theTD-GC×GC-TOFMS/FIDmethodwastransposed
toTD-GC×GC-HRTOFMS.Althoughitmightbeconsidereda
triv-ialstep,thetranspositionofchromatographicmethodsbetween
vacuumoutletinstrumentsfromonethatincludessplittingofthe
flowtoanotherthatusesasignificantlylongertransferline
(Sup-plementaryInformation,Fig.S3)wasnotstraightforward.Despite
allefforts(variationsofflows,columndimensionsandcoating,and
otherGC×GCoperationalconditions)itappearedtobe
impossi-bletoobtainchromatogramsthatalignretentiontimesbetween
LRTOFMSandHRTOFMS.
TheTD-GC×GC-HRTOFMSapexplotsshowninFig.2forTHP1.0
and 3R4F MTS represent the best compromise that could be
obtainedin terms ofrun time andchromatographic resolution.
EachoftheHRTOFMSsignalswasmatchedwiththecorresponding
FID/LRTOFMSsignal. Thisrequired significantmanual
interven-tiontolocate thecorrespondingpeak basedonfirst dimension
linearretentionindex,positionofthepeakinthesecond
dimen-sionretentionplaneandmassspectraldataforeachcompound.
Despitethetechnicalchallengesexplainedabove,theFID,LRTOFMS
andHRTOFMSsignalsof88and207compoundswerematchedfor
THP1.0and3R4FMTS,respectively.
The developed analytical method was capable of analysing
molecules in the range of C4-C25. However, a limited number
ofunresolved(SupplementaryInformation,Fig.S10)compounds
elutingbeforetheC5regionwerenotconsideredfurtherbecauseof
thelackoflinearretentionindexinformationandalackoffragment
ionsbelowm/z35givinginaccuratepeakassignmentagainstlibrary
match.Adedicatedanalyticalmethodologycoupledtotheuseof
standardswould berequiredtoconsideranalytesbelowtheC5
regionbutwasoutofthescopeofthepresentstudy.When
110 B. Savareear et al. / J. Chromatogr. A 1581–1582 (2018) 105–115 Table1
Qualitativeandsemi-quantitativeresultsofvapourphasefractionofTHP1.0aerosol.
FID LRTOFMS HRTOFMS
PeakNo. Compoundname Chemical
formula
LRI LRIlib. Amount
(g/stick) aMean±SD MS for-ward match MS reverse match MS for-ward match MS reverse match Mass accuracy (ppm) Chemical class 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 2-Propanone*
Aceticacid,methylester
Carbondisulfide Propanal,2-methyl-* 2-Propenal,2-methyl-* 2,3-Butanedione* 2-Butanone*$ Furan,2-methyl-* Furan,3-methyl-* Unknown
Propanoicacid,methylester
1,3-Pentadiene,3-methyl-,(E)-*
Butanal,3-methyl-* Butanal, 2-methyl-1-Penten-3-one 2,3-Pentanedione* Furan,2,5-dimethyl-* 2-Butanone,
3-hydroxy-Thiocyanicacid,methylester*
2,4-Dimethylfuran*
2-Vinylfuran*
Propanoicacid,2-oxo-,methylester
1-Butanol, 3-methyl-pyrazine 2-Pentanone,4-methyl-* 1H-Pyrrole,1-methyl-* Unknown Pyridine*$ Disulfide,dimethyl* 1H-Pyrrole 2-Pentenal, (E)-1,4-Dioxin, 2,3-dihydro-Toluene*$ Thiophene,3-methyl-* 3-Hexanone 2-Hexanone*$ Cyclopentanone* Hexanal 2,4-Pentanedione 3(2H)-Furanone, dihydro-2-methyl-Furan,2-ethyl-5-methyl-* 1H-Pyrrole,1-ethyl-* 2,5-Furandione 2-Vinyl-5-methylfuran* 2-Furancarboxaldehyde C3H6O C3H6O2 CS2 C4H8O C4H6O C4H6O2 C4H8O C5H6O C5H6O – C4H8O2 C6H10 C5H10O C5H10O C5H8O C5H8O2 C6H8O C4H8O2 C2H3NS C6H8O C6H6O C4H6O3 C5H12O C4H4N2 C6H12O C5H7N – C5H5N C2H6S2 C4H5N C5H8O C4H6O2 C7H8 C5H6S C6H12O C6H12O C5H8O C6H12O C6H12O C5H8O2 C7H10O C6H9N C5H6O2 C7H8O C5H4O2 519 545 555 565 573 588 595 600 609 618 626 638 652 662 685 694 706 706 713 715 725 727 734 735 739 741 743 746 748 752 757 767 771 776 785 790 794 800 804 808 810 816 831 832 835 509 536 549 561 567 595 598 606 614 – 627 643 652 662 681 698 707 706 711 711 725 722 736 736 735 743 – 746 746 755 754 680# 770 776 784 790 791 800 795 809 802 821 830 826 833 13.3±0.15 3.9±0.28 0.3±0.04 8.4±0.13 5.4±0.24 15.7±0.21 1.2±0.08 3.1±0.09 0.7±0.02 0.1±0.00 0.3±0.03 0.6±0.05 7.4±0.02 5.9±0.24 0.3±0.03 5.2±0.09 trace 1.2±0.14 1.6±0.25 0.2±0.01 0.3±0.01 0.6±0.06 0.1±0.01 0.2±0.01 0.1±0.00 1.0±0.01 trace 1.8±0.25 2.1±0.04 2.3±0.31 1.8±0.06 0.2±0.01 0.4±0.02 0.2±0.01 0.3±0.02 0.1±0.01 0.1±0.01 0.9±0.14 0.1±0.00 1.5±0.17 0.2±0.01 0.2±0.01 0.6±0.09 0.5±0.06 15.7±0.23 964 952 899 967 962 885 966 929 762 – 856 865 952 953 899 945 941 885 907 838 886 927 826 958 726 948 – 959 960 926 900 748 917 708 935 783 971 905 880 947 773 892 726 909 957 964 960 944 967 962 891 966 934 785 – 856 878 952 953 899 948 962 892 907 856 893 959 879 958 766 948 – 963 961 949 900 757 927 814 935 805 971 905 915 947 783 892 906 959 957 886 963 960 933 940 963 832 960 799 – 849 744 942 944 720 953 959 973 959 952 928 958 903 967 – 936 – 969 967 948 875 806 949 801 950 903 935 810 757 954 777 906 737 883 950 899 963 989 933 940 974 929 960 884 – 897 864 942 947 764 958 959 980 967 952 930 958 911 967 – 940 – 969 967 954 875 815 958 818 952 903 945 837 768 954 789 906 879 924 950 1.4 −1.1 −1.5 −1.0 −0.4 −1.4 −1.0 −1.0 −0.7 – −0.3 −0.3 −0.4 −0.6 0.0 −0.5 −0.4 −0.2 −0.7 −0.8 −0.1 −0.1 1.5ˆ −0.3 1.2 0.3 – −0.8 −2.5 −1.4 −0.6 −0.6 −0.2 −0.4 −0.2 1.0 0.2 2.4ˆ −0.3 −0.4 −0.8 −0.1 1.2ˆ −0.5 −1.8 8 9 11 7 7 8 8 3 3 13 9 1 7 7 8 8 3 8 9 3 3 9 6 3 8 3 13 3 11 3 7 12 4 3 8 8 8 7 8 9 3 3 9 3 7
B. Savareear et al. / J. Chromatogr. A 1581–1582 (2018) 105–115 111 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 Vinylcrotonate 1,3-Cyclopentadiene, 5-(1,1dimethylethyl)-2,5-Diethylfuran 2-Hexenal, (E)-p-Xylene*$ 1H-Pyrrole, 2,5-dimethyl-2(3H)-Furanone, 5-methyl-o-Xylene* Bicyclo[3.1.0]hexan-2-one Styrene*$ Heptanal 2-Methyl-5-isopropenylfuran 6,8-Dioxabicyclo[3.2.1]octane Oxepine, 2,7-dimethyl-1H-Pyrrole, 1-butyl-2-Heptanone, 6-methyl-Benzaldehyde* Dimethyltrisulfide* beta-Myrcene Furan, 2-pentyl-Unknown Octanal Unknown delta-3-Carene* o-Cymene* Benzene, 1,2,3-trimethyl-Benzylamine Limonene* Cyclohexanone, 2,2,6-trimethyl-trans-beta-Ocimene* Benzeneacetaldehyde 4-tert-Butyltoluene Unknown Nonanal 1-Methoxyadamantane Decanal Unknown Trimethyl-tetrahydronaphthalene Triacetin Naphthalene,1,3-dimethyl-* C6H8O2 C9H14 C8H12O C6H10O C8H10 C6H9N C5H6O2 C8H10 C6H8O C8H8 C7H14O C8H10O C6H10O2 C8H10O C8H13N C8H16O C7H6O C2H6S3 C10H16 C9H14O – C8H16O – C10H16 C10H14 C9H12 C7H9N C10H16 C9H16O C10H16 C8H8O C11H16 – C9H18O C11H18O C10H20O – C13H18 C9H14O6 C12H12 836 849 855 857 869 870 873 877 897 900 904 938 939 949 953 957 978 986 989 992 997 1008 1012 1017 1035 1035 1036 1039 1048 1052 1061 1067 1079 1113 1198 1216 1247 1279 1343 1385 783# 839 888# 854 865 867 873 881 793# 893 901 933 839# 944 975# 953 971 984 991 993 – 1003 – 1011 1026 1033 1035 1030 1047 1049 1060 1066 – 1109 1179# 1214 – 1250# 1344 1385 0.3±0.04 1.0±0.01 0.1±0.01 trace trace 0.2±0.02 0.6±0.04 0.3±0.00 0.1±0.01 0.1±0.01 0.1±0.01 0.2±0.00 0.2±0.02 0.2±0.00 0.1±0.00 0.1±0.00 0.1±0.10 0.4±0.03 0.4±0.01 0.2±0.01 trace trace trace 0.9±0.05 trace 0.1±0.00 trace 1.0±0.01 0.1±0.01 0.4±0.06 1.6±0.29 0.1±0.01 trace 0.1±0.01 0.1±0.00 0.1±0.00 trace 0.9±0.02 0.1±0.00 0.7±0.01 755 777 854 864 758 772 907 936 763 797 730 854 834 786 855 753 936 941 929 924 – 831 – 850 874 776 753 860 759 812 808 802 – 766 786 – – 906 785 – 815 845 854 969 889 828 933 947 778 911 782 854 850 786 867 832 936 941 965 924 – 851 – 891 952 871 829 885 848 891 929 816 – 779 796 – – 940 902 – 765 863 777 918 880 864 931 953 831 889 892 847 796 782 812 931 898 919 873 909 – 894 – 847 950 923 855 880 815 – 952 – – 888 766 844 – 832 915 899 850 864 856 923 880 867 935 953 835 919 911 921 796 782 849 933 919 924 877 912 – 905 – 851 950 923 855 880 905 – 956 – – 888 786 875 – 845 915 899 0.1 −0.5 −0.0 0.2 0.3 −0.0 −0.1 −0.4 0.1 −0.4 −0.5ˆ −0.8 −0.2 −0.6 −0.7 1.1ˆ −0.3 −0.2 0.7ˆ −0.4 – 0.8ˆ – −0.5 0.0 −0.0 −0.5 −0.2 0.1 2.6 −0.7 0.9 – 1.4ˆ 1.3 0.4ˆ – 1.0 −1.0ˆ −1.0 9 2 3 7 4 3 9 4 8 4 7 3 12 3 3 8 7 11 1 3 13 7 13 2 4 4 4 2 8 1 7 4 13 7 12 7 13 5 9 5
asamplemean(n=4)alongwithstandarddeviationofthemeasurements,*compoundsfoundinreferencecigarettesmokeVPsample,compoundsfoundinHoffmannlist,$compoundswerepositivelyidentifiedusingtheir
respectivestandard,-informationnotavailable,trace-concentrationbelow0.1g/stick,#estimatednon-polarretentionindexfromNISTdatabase,butsemi-standardcolumnretentionindexinformationnotavailablefrom
112 B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115
qualitativeandquantitativeinvestigationinTHP1.0VPand3R4F
MTSVP,respectively.
3.2. ComparisonsofVPaerosolfractionfromTHP1.0and3R4F
MTS
3.2.1. QualitativeaspectsusinglowandhighresolutionTOFMS
QualitativeanalysisoftheVPfractionofTHP1.0aerosolandMTS
wascarriedoutusingseveralidentificationcriteriasuchas
reten-tiontimes(1t
Rand2tR)ofavailableauthenticreferencestandards,
1tRlinearretentionindexmatches(±10window),andmass
spec-tralforwardandreversesimilaritymatches(>700)againstspectral
librariesforbothTOFMSandHRTOFMSdata.Inaddition,accurate
massinformation within±3ppm window based oneither
par-entionsorabundantfragments(limitedcases)wasemployedto
improvetheconfidenceofidentificationofcompounds.Asseenin
Tables1andS2,insomelimitedcases,eitherLRIdatawerenot
availableortheMSmatchfactorwasnotobtainedfromeitherthe
loworhighresolutionmassanalyserbecauseofinterferingions,
butpeakassignmentwasalwaysbackedupbyatleastgoodMS
match(>700)ontheotherMSanalyserandgoodmassaccuracy
(<±3ppm)values.Inaddition,thespatialcoordinationof
homolo-gousseriesofcompoundsinGC×GCchromatogramsaidedfurther
confirmationofidentities.
As expected, reverse library matches (ranged 757–971 and
764–989forLRTOFMSandHRTOFMS,respectively)wereslightly
betterthanforwardmatches(ranged 708–971and720–973 for
LRTOFMS and HRTOFMS, respectively) and HRTOFMS matches
were globally betterthan LRTOFMS matches, even though MS
librariesaremadeofLRMSdatasets.Thiscanbeexplainedpartly
bythehighermassresolutionoftheHRTOFMSinstrument,which
providesnarrowermasswindowsandreducestheimpactof
inter-feringionsfromco-elutingspeciesandbackgroundnoise[37].
BecauseoftheoperationoftheHRTOFMSinelectronimpact
mode(70eV)toprovidethedesiredfragmentationforMSlibrary
searchpurpose,forsomecompoundstheabundanceofthe
molec-ularionwasverylow.Thiswasparticularlythecaseforalkanes,
aldehydes, alcohols and nitriles, for which the most abundant
fragmention wasused to estimatemass accuracy,as reported
earlier[32].Theuseofasofterionisationtechniquesuchas
UV-basedphoto-ionisation(PI)iscurrentlyunderinvestigationinother
projectsandmayimproveMSdataqualityforfragmentionsand
molecularions[38].Inpractice,82%and94%oftheidentified
ana-lytespresentedausablemolecularionforidentificationforTHP1.0
and3R4FMTS,respectively.Amongthese,70%oftheassignedpeaks
exhibitedamassaccuracyof±1ppm,forbothsampletypeswhile
15%and25%ofassignedpeaksexhibitedamassaccuracybetween
±1and±3ppmforTHP1.0and3R4FMTS,respectively.Despitethe
highaddedvalueofaccuratemasstoMSlibrarycomparisonsfor
identificationofcompounds[39],theunequivocalidentificationof
compoundspresentindifferentisomericformsstillrequiresthe
useofLRIsandideallytheanalysisofpurestandardcompounds,
whichisdifficulttoimplementinnon-targetedstudies.Byapplying
thecomprehensivedataminingstrategiesdescribedpreviously79
(outofthe85)(Table1)and198(outofthe202)compoundswere
identifiedintheTHP1.0aerosolVPand3R4FMTSVP
(Supplemen-taryTableS2),respectively.Theremainingfewcompoundscould
notbeidentifiedatthisstage.Nevertheless,forbothTHP1.0VPand
3R4FMTSVPsamples,itconstitutesthefirstcomprehensivelistof
VOCscomposingtheirchemicalprofile.Previously,aTD-GC×
GC-TOFMSapproachwasabletotentativelyidentify127compounds
[5].Whencomparedtoourcurrentapproachthatwasableto
iden-tify181compoundsinthecarbonnumberrangeC6-C14,asetof56
compoundswerefoundtobecommontobothstudies(TableS2).
WhenweconsidertheidentifiedcompoundsofTHP1.0VPand
3R4FMTSVPsamples,35compoundswerecommontobothTHP1.0
aerosolVPand3R4FMTSVPsamplesandarehighlightedinTable1
andinSupplementaryinformation,TableS2.Theidentified
com-pounds were further grouped in chemical classes (Table 2)to
highlightmajor chemical differencesbetweenthe two typesof
aerosolVP.Itappearedthatthehigherchemicalcomplexityof3R4F
MTSsamples(202compoundsversus85compoundsforTHP1.0
aerosolVP)wasmainlyduetothelargenumberofacyclic,alicyclic
and monocyclic aromatichydrocarbons.In 3R4F MTS VP,these
threechemicalfamiliesaccountedfor71%ofthe202compounds
found.ForTHP1.0VP,thesethreechemicalfamiliesaccountedfor
16%ofthe85compoundsfound,whileheterocyclics,aldehydesand
ketonesaccountedfor53%.Thesedataillustratethechemical
dif-ferencebetweentheVPproducedbycombustionandpyrolysisof
tobaccoandtheVPofanaerosolproducedbylowertemperature
heatingoftobacco[4].Interestingly,italsoappearedthat,among
theso-called‘Hoffmannlist’of44compoundsofprimeinterestin
termsoftoxicity,sevenandsixanalyteswerepresentinVPsamples
generatedin3R4FMTS(SupplementaryInformation,TableS2)and
THP1.0aerosol(Table1),respectively.Amongthem,therewere5
commoncompounds(2-propanone,2-butanone,pyridine,toluene
andstyrene).
3.2.2. Semi-quantitativeanalysisusingFID
FID/TOFMS dual detection was intended to facilitate
semi-quantitative analysis of the samples. Tobacco-related samples
containhundredsofcompoundsoverseveralordersofmagnitude
ofconcentration[5,40].Thispresentsachallengeformass
analy-sersthathavelineardynamicrangeslimitedto3or4decades.After
robustassignmentofcompoundidentitiesbasedon(HR)TOFMS,
FID signals were used for semi-quantification [21,22]. Because
of the non-targeted nature of analysis, a generic, external
cal-ibration wasconstructedusing representativecompoundsfrom
thirteenchemicalclasses(SupplementaryInformation,TableS1).
Theselectionofstandardcompoundswasbasedontheassigned
VOCchemicalclassesinVPfractionsofTHP1.0aerosoland3R4F
MTS samples.The calibrationrange wasbased ondetectability
ofboth FIDand TOFMSdetectors (datanot shown).Thelowest
pointwassetat0.1g/L, lowerconcentrationswerereported
astraces.Thehighest point(50g/L)wassetaftermeasuring
THP1.0VPsamples.Threecompoundsin3R4FMTSVP(TableS2)
wereoutofthatrange,thusthesecompoundsconcentrationswere
estimatedfromanextentionofthecalibrationplot.Compounds
weresemi-quantifiedusingtheresponsefactorobtainedforthe
representativesubstanceofthesameclassinthecalibration
solu-tionafternormalisation betweensamplesandcalibrationusing
anotherexternalstandardtoaddresspotentialinstrumentaldrift.
Ifnorepresentativecompoundwasincludedincalibration,
non-classifiedcompounds(organosulfur,miscellaneousandunknowns)
weresemi-quantifiedusingthefactorsobtainedforbromobenzene,
acetonitrileandp-xylene,respectively.
The resultsof the external 9-pointcalibration (n=4)at the
g/Lleveldemonstratedacceptablereproducibility(RSD%<6%,
range3–6%)forallcompounds(SupplementaryInformation,Table
S1).Semi-quantitativeresultsforchemicalclassesidentifiedinthe
vapourphasefractionsofTHP1.0aerosoland3R4FMTSare
pre-sentedinTable2.
Fig. 3 illustrates the differences in chemical composition
and abundancebetweenTHP1.0VPand 3R4F MTS VP samples.
THP1.0VP generally contained lower abundances of the less
volatilecomponentsofchemicalclasses,whichwouldbeexpected
whenconsideringthetemperaturesatwhichtheaerosolswere
formed.The major THP1.0VP classes were aldehydes, ketones,
estersand heterocycliccompounds.In 3R4F MTSVP,
hydrocar-bons,ketones,andheterocycliccompoundsweremoreabundant
B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115 113
Table2
Summaryofqualitativeandsemi-quantitativeinformationforVPsamplesofTHP1.0aerosoland3R4Fcigarettesmokewiththeirrespectiverelativepercentage.
THP1.0VP 3R4FMTSVP
Chemical class
Chemicalclassname No.of
analytes Relative %of analyte Amount (g/stick) Relative %of amount No.of analytes Relative %of analyte Amount (g/stick) Relative %of amount 1 2 3 4 5 6 7 8 9 10 11 12 13 Acyclichydrocarbons Alicyclichydrocarbons Heterocycliccompounds Monocyclicaromatic hydrocarbons Polycyclicaromatic hydrocarbons Alcohols Aldehydes Ketones Esters Nitriles Organosulfurcompounds Miscellaneouscompounds Unknowns 3 3 18 9 2 1 14 14 9 ND 3 3 6 4 4 21 11 2 1 16 16 11 ND 4 4 7 1.4 2.8 11.9 1.1 1.6 0.1 48.3 37.9 9.5 0.0 2.8 0.2 0.1 1 2 10 1 1 0 41 32 8 0 2 0 0 70 41 17 33 3 ND 7 14 1 8 2 2 4 35 20 8 16 1 ND 3 7 0 4 1 1 2 496.8 173.1 69.4 119.0 2.4 0.0 57.6 264.4 2.5 36.8 4.7 1.2 1.0 40 14 6 10 0 0 5 22 0 3 0 0 0 ND-notdetected.
Fig.3.TD-GC×GC-LRTOFMStwo-dimensionalapexbubbleplotillustratingthedifferenceinrelativechemicalcomplexitybetween3R4FMTS(top)andTHP1.0aerosol (bot-tom).Bubblesizesarerelatedtolevelsofanalytes.Fordatavisualisationpurpose,bubblesizeswererescaledbasedonconcentrationintervals:<0.1g=1;0.1-0.25g=2.5; 0.25-0.5g=5;0.5–1g=10;1–2.5g=15;2.5–5g=20;5–10g=25;10–25g=30;25–50g=35;50–100g=40;>100g=60(seeTables1andS2fordetailedvalues).
114 B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115
Fig.5.Differencesofchemicalconcentrationof35overlappingcompoundsfoundinboth3R4FMTSVPandTHP1.0VPsamples.Y-axisinsertedaslogarithmicscalefordata visualisationpurpose(actualvaluesareprovidedinTables1andS2).
The total abundance of compounds present in the VP
frac-tionof3R4FMTSwasestimatedtobetentimesgreaterthanin
THP1.0aerosolVP.In3R4FMTSVP,acyclic,alicyclicand
mono-cyclicaromatichydrocarbonsaccountedfor64%intermsofthe
totalestimatedconcentration.ForTHP1.0VP,thesethreechemical
familiesrepresentedlessthan4%ofthetotalestimated
concen-tration.Aldehydes,ketones,andheterocyclicsaccountedfor41%,
32%and10%respectivelyofthetotalestimatedconcentrationof
analytesinTHP1.0VP.Table1(andSupplementaryinformation,
TableS2)providesindividualsemi-quantitativevaluesingper
consumable(stick)foreachassignedcompound.Fig.4illustrates
thechemicaldistributionofthe35compoundspresentinboth
sam-pletypes.Thesummedconcentrationofthese35compoundswas
sixtimeshigherin3R4FMTSVP(457.5g/stick)thaninTHP1.0VP
(73.7g/stick).Whenconsideringrelativeamounts,ofthe35
com-moncompoundsfoundinTHP1.0and3R4FMTSproducts,higher
concentrationsweresystematicallymeasuredinMTS,onawhole
productbasis,exceptforpyridineanddimethyltrisulfide(Fig.5).
Inthesetwocases,thelevelsinTHP1.0appearedtobemarginally
higherbutconfirmationofanydifferencewouldrequire
quantita-tiveanalysis.Finally,intermsofsemi-quantification,2-propanone
(oneofthe‘Hoffmannlist’toxicants)waspresentatsignificantly
lowerconcentrationsinTHP1.0VPsamples(13.3g/stick)thanin
3R4FMTSVP(152g/stick).Similarly,theconcentrationsofother
Hoffmannlistcompounds(e.g.toluene,2-butanoneandstyrene)
weresignificantlyreducedinTHP1.0VPcomparedto3R4FMTSVP
(seeTables1andTableS2fordetailedvalues).Resultswerealso
comparedtoLietal.data[3],despitethefactthattheyuseda
tar-getapproachusingdifferentTHP2.2producttocomparetoMTS
VP.Theystudiedthepercentagereductionofanalytes
concentra-tionbetweenTHP2.2andMTSsamples.Thepercentagereduction
rateoftolueneand2-propanonewere97%and87%,respectively.
Almostsimilarresultswerefoundinthepresentworkfortoluene
(99%)and2-propanone(91%).Thereduction rateof2-butnaone
wasfoundtobe41%intheirstudy[3]althougha97%reduction
ratewasobservedforTHP1.0VPsample.Thepresentworkwas
basedonapreliminarysemi-quantitativeapproachthatrequires
furthervalidtionusingspecificcalibrationsolutions,ideallyusing
stableisotopedilution.
4. Conclusions
A novel characterization approach based on the use of
independentandcomplementaryTD-GC×GC-TOFMS/FIDand
TD-GC×GC-HRTOFMSmethodshavebeendevelopedfortheanalysis
oftheVPaerosol fractionofTHP1.0and 3R4FMTS.Compounds
wereidentifiedusingLRIs,MSmatchesagainstspectrallibraries
usingbothLRTOFMSandHRTOFMSdata,andaccuratemassvalues.
Thiscomprehensivedataminingapproachpermittedthe
assign-ment of chemical identity for more than 90% of the detected
constituentsforbothsampletypes.Thechemicalcompositionof
theVPofTHP1.0aerosolwasobservedtobemuchlesscomplex
than3R4FMTSVP.TheGC×GC-FIDsemi-quantitativedata
indi-catedthatthetotalabundanceofanalytesintheTHP1.0VPwasten
timeslowerthaninthe3R4FMTSVP.Inconclusion,thepresent
studyprovidesdatathatcontributetoamoreextensive
chemi-calcharacterisationoftheaerosolsgeneratedbytobaccoheating
products.
AppendixA. Supplementarydata
Supplementarymaterialrelatedtothisarticlecanbefound,in
theonlineversion, atdoi:https://doi.org/10.1016/j.chroma.2018.
10.035.
References
[1]J.-P.Schaller,D.Keller,L.Poget,P.Pratte,E.Kaelin,D.McHugh,G.Cudazzo,D. Smart,A.R.Tricker,L.Gautier,M.Yerly,R.R.Pires,S.LeBouhellec,D.Ghosh,I. Hofer,E.Garcia,P.Vanscheeuwijck,S.Maeder,Evaluationofthetobacco heatingsystem2.2.Part2:chemicalcomposition,genotoxicity,cytotoxicity, andphysicalpropertiesoftheaerosol,Regul.Toxicol.Pharmacol.81(2016) S27–S47.
[2]M.Forster,S.Fiebelkorn,C.Yurteri,D.Mariner,C.Liu,C.Wright,K.McAdam,J. Murphy,C.Proctor,AssessmentofnoveltobaccoheatingproductTHP1.0.Part 3:Comprehensivechemicalcharacterisationofharmfulandpotentially harmfulaerosolemissions,Regul.Toxicol.Pharmacol.93(2018)14–33.
[3]X.Li,Y.Luo,X.Jiang,H.Zhang,F.Zhu,S.Hu,H.Hou,Q.Hu,Y.Pang,Chemical analysisandsimulatedpyrolysisoftobaccoheatingsystem2.2comparedto conventionalcigarettes,NicotineTob.Res.(2018)1–8.
[4]B.Savareear,R.Lizak,M.Brokl,C.Wright,C.Liu,J.-F.Focant,Headspace solid-phasemicroextractioncoupledtocomprehensivetwo-dimensionalgas chromatography-time-of-flightmassspectrometryfortheanalysisofaerosol fromtobaccoheatingproduct,J.Chromatogr.A1520(2017)135–142.
B.Savareearetal./J.Chromatogr.A1581–1582(2018)105–115 115 [5]B.Savareear,M.Brokl,C.Wright,J.-F.Focant,Thermaldesorption
comprehensivetwo-dimensionalgaschromatographycoupledtotimeof flightmassspectrometryforvapourphasemainstreamtobaccosmoke analysis,J.Chromatogr.A1525(2017)126–137.
[6]M.Brokl,L.Bishop,C.G.Wright,C.Liu,K.McAdam,J.F.Focant,Multivariate analysisofmainstreamtobaccosmokeparticulatephasebyheadspace solid-phasemicroextractioncoupledwithcomprehensivetwo-dimensional gaschromatography-time-of-flightmassspectrometry,J.Chromatogr.A1370 (2014)216–229.
[7]Q.Ye,Developmentofsolid-phasemicroextractionfollowedbygas chromatography-massspectrometryforrapidanalysisofvolatileorganic chemicalsinmainstreamcigarettesmoke,J.Chromatogr.A1213(2008) 239–244.
[8]E.Simonavicius,A.McNeill,L.Shahab,L.S.Brose,Heat-not-burntobacco products:asystematicliteraturereview,Tob.Control0(2018)1–13.
[9]J.-Z.Dong,J.N.Glass,S.C.Moldoveanu,AsimpleGC–MStechniqueforthe analysisofvaporphasemainstreamcigarettesmoke,J.MicrocolumnSep.12 (3)(2000)142–152.
[10]T.Miyake,T.Shibamoto,Quantitativeanalysisbygaschromatographyof volatilecarbonylcompoundsincigarettesmoke,J.Chromatogr.A693(1995) 376–381.
[11]K.G.Darrall,J.A.Figgins,R.D.Brown,G.F.Phillips,Determinationofbenzene andassociatedvolatilecompoundsinmainstreamcigarettesmoke,Analyst 123(1998)1095–1101.
[12]G.Barrefors,G.Petersson,Assessmentofambientvolatilehydrocarbonsfrom tobaccosmokeandfromvehicleemissions,J.Chromatogr.A643(1993) 71–76.
[13]S.M.Charles,S.A.Batterman,C.Jia,CompositionandemissionsofVOCsin main-andside-streamsmokeofresearchcigarettes,Atmos.Environ.41 (2007)5371–5384.
[14]S.Uchiyama,T.Tomizawa,Y.Inaba,N.Kunugita,Simultaneousdetermination ofvolatileorganiccompoundsandcarbonylsinmainstreamcigarettesmoke usingasorbentcartridgefollowedbytwo-stepelution,J.Chromatogr.A1314 (2013)31–37.
[15]E.Woolfenden,Sorbent-basedsamplingmethodsforvolatileand
semi-volatileorganiccompoundsinair.Part1:sorbent-basedairmonitoring options,J.Chromatogr.A1217(2010)2674–2684.
[16]P.-H.Stefanuto,K.Perrault,S.Stadler,R.Pesesse,M.Brokl,S.Forbes,J.-F. Focant,Readingcadavericdecompositionchemistrywithanewpairof glasses,ChemPlusChem79(2014)786–789.
[17]S.Stadler,P.-H.Stefanuto,M.Brokl,S.L.Forbes,J.-F.Focant,Characterization ofvolatileorganiccompoundsfromhumananaloguedecompositionusing thermaldesorptioncoupledtocomprehensivetwo-dimensionalgas chromatography–time-of-flightmassspectrometry,Anal.Chem.85(2013) 998–1005.
[18]S.L.Forbes,K.A.Perrault,P.-H.Stefanuto,K.D.Nizio,J.-F.Focant,Comparison ofthedecompositionVOCprofileduringwinterandsummerinamoist, mid-latitude(Cfb)climate,PLoSOne9(2014).
[19]M.Brokl,L.Bishop,C.G.Wright,C.Liu,K.McAdam,J.F.Focant,Analysisof mainstreamtobaccosmokeparticulatephaseusingcomprehensive two-dimensionalgaschromatographytime-of-flightmassspectrometry,J. Sep.Sci.36(2013)1037–1044.
[20]J.Dallüge,R.J.J.Vreuls,J.Beens,U.A.Th.Brinkman,Optimizationand characterizationofcomprehensivetwo-dimensionalgaschromatography withtime-of-flightmassspectrometricdetection(GC×GC-TOFMS),J.Sep.Sci. 25(2002)201–214.
[21]R.Shellie,P.Marriott,P.Morrison,Comprehensivetwo-dimensionalgas chromatographywithflameionizationandtime-of-flightmassspectrometry detection:qualitativeandquantitativeanalysisofwestAustralian sandalwoodoil,J.Chromatogr.Sci.42(2004)417–422.
[22]Z.L.Cardeal,M.D.R.GomesdaSilva,P.J.Marriott,Comprehensive
two-dimensionalgaschromatography/massspectrometricanalysisofpepper volatiles,RapidCommun.MassSpectrom.20(2006)2823–2836.
[23]T.Dijkmans,M.R.Djokic,K.M.VanGeem,G.B.Marin,Comprehensive compositionalanalysisofsulfurandnitrogencontainingcompoundsinshale oilusingGC×GC–FID/SCD/NCD/TOF-MS,Fuel140(2015)398–406.
[24]L.Nicolotti,C.Cordero,D.Bressanello,C.Cagliero,E.Liberto,F.Magagna,P. Rubiolo,B.Sgorbini,C.Bicchi,Paralleldualsecondarycolumn-dualdetection: afurtherwayofenhancingtheinformativepotentialoftwo-dimensional comprehensivegaschromatography,J.Chromatogr.A1360(2014)264–274.
[25]D.Bressanello,E.Liberto,M.Collino,S.E.Reichenbach,E.Benetti,F.Chiazza,C. Bicchi,C.Cordero,Urinarymetabolicfingerprintingofmicewithdiet-induced metabolicderangementsbyparalleldualsecondarycolumn-dualdetection two-dimensionalcomprehensivegaschromatography,J.Chromatogr.A1361 (2014)265–276.
[26]P.Q.Tranchida,S.Salivo,I.Bonaccorsi,A.Rotondo,P.Dugo,L.Mondello, Analysisoftheunsaponifiablefractionoflipidsbelongingtovarious milk-typesbyusingcomprehensivetwo-dimensionalgaschromatography withdualmassspectrometry/flameionizationdetectionandwiththesupport ofhighresolutiontime-of-flightmass,J.Chromatogr.A1313(2013)194–201.
[27]J.Krupˇcík,R.Gorovenko,I. ˇSpánik,P.Sandra,D.W.Armstrong,
Flow-modulatedcomprehensivetwo-dimensionalgaschromatographywith simultaneousflameionizationandquadrupolemassspectrometricdetection, J.Chromatogr.A1280(2013)104–111.
[28]D.Turner,G.H.Morgan,B.White,Amatteroftaste....flavourprofilingby GCxGC-qMS/FID,Chromatogr.Today(2012)40–43,
flavourprofilingbygcxgc-qmsfid/1136/.(Accessedonlineon12/04/2018)
http://www.chromatographytoday.com/articles/gc-mdgc-gc-ms/32/diane turnerdrg.h.morganbryanwhite/amatteroftaste.
[29]P.-H.Stefanuto,K.A.Perrault,R.M.Lloyd,B.Stuart,T.Rai,S.L.Forbes,J.-F. Focant,Exploringnewdimensionsincadavericdecompositionodour analysis,Anal.Methods7(2015)2287–2294.
[30]Y.Duan,F.Zheng,H.Chen,M.Huang,J.Xie,F.Chen,B.Sun,Analysisof volatilesindezhoubraisedchickenbycomprehensivetwo-dimensionalgas chromatography/highresolution-timeofflightmassspectrometry,LWT– FoodSci.Technol.60(2015)1235–1242.
[31]J.D.Byer,K.Siek,K.Jobst,DistinguishingtheC3vsSH4masssplitby comprehensivetwo-dimensionalgaschromatography-highresolution time-of-flightmassspectrometry,Anal.Chem.88(2016)6101–6104.
[32]D.D.Yan,Y.F.Wong,L.Tedone,R.A.Shellie,P.J.Marriott,S.P.Whittock,A. Koutoulis,Chemotypingofnewhop(HumuluslupulusL.)genotypesusing comprehensivetwo-dimensionalgaschromatographywithquadrupole accuratemasstime-of-flightmassspectrometry,J.Chromatogr.A1536 (2017)110–121.
[33]J.Gee,K.Prasad,S.Slayford,A.Gray,K.Nother,A.Cunningham,E.
Mavropoulou,C.Proctor,AssessmentoftobaccoheatingproductTHP1.0.Part 8:studytodeterminepuffingtopography,mouthlevelexposureand consumptionamongJapaneseusers,Regul.Toxicol.Pharmacol.93(2018) 84–91.
[34]ISO,3308,RoutineAnalyticalSmokingMachine-DefinitionandStandard Conditions,InternationalOrganisationforStandardisation,Geneva,2000.
[35]T.Kind,O.Fiehn,Metabolomicdatabaseannotationsviaqueryofelemental compositions:Massaccuracyisinsufficientevenatlessthan1ppm,BMC Bioinform.7(2006)1–10.
[36]LinearRetentionIndicesdatabaseonlines:https://www.ncbi.nlm.nih.gov/ pccompound?cmd=search(Accessed20March2018).
[37]P.Ausloos,C.L.Clifton,S.G.Lias,A.I.Mikaya,S.E.Stein,D.V.Tchekhovskoi,O.D. Sparkman,V.Zaikin,D.Zhu,Thecriticalevaluationofacomprehensivemass spectrallibrary,J.Am.Soc.MassSpectrom.10(1999)287–299.
[38]A.Giri,M.Coutriade,A.Racaud,K.Okuda,J.Dane,R.B.Cody,J.-F.Focant, Molecularcharacterizationofvolatilesandpetrochemicalbaseoilsby photo-ionizationGC×GC-TOF-MS,Anal.Chem.89(2017)5395–5403.
[39]E.Rathahao-Paris,S.Alves,C.Junot,J.-C.Tabet,Highresolution massspectrometryforstructuralidentificationofmetabolitesin metabolomics,Metabolomics12(2016)10.
[40]J.Dallüge,L.L.P.vanStee,X.Xu,J.Williams,J.Beens,R.J.J.Vreuls,U.A.Th. Brinkman,Unravellingthecompositionofverycomplexsamplesby comprehensivegaschromatographycoupledtotime-of-flightmass spectrometry,J.Chromatogr.A974(2002)169–184.