Development and performance evaluation of a novel dynamic headspace vacuum transfer “In Trap” extraction method for volatile compounds and comparison with headspace solid-phase

Headspace in-tube extraction (HS-ITEX) and solid phase microextraction (HS-SPME) sampling, followed by gas chromatography-mass spectrometry (GC–MS), are widely used to analyze volatile compounds in various food matrices.

j ou rn a l h om ep a ge :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

Pascal Fuchsmann∗, Mireille Tena Stern, Patrick Bischoff, René Badertscher,

Katharina Breme1, Barbara Walther

Agroscope, Schwarzenburgstrasse 161, CH-3003 Berne, Switzerland

a r t i c l e i n f o

Article history:

Received 25 February 2019

Received in revised form 10 May 2019

Accepted 10 May 2019

Available online 22 May 2019

Keywords:

In-tube extraction

Vacuum-transfer in trap

Reduced pressure sampling

Dairy matrix

Response surface methodology

GC–MS

a b s t r a c t Headspacein-tubeextraction(HS-ITEX)andsolidphasemicroextraction(HS-SPME)sampling,followed

bygaschromatography-massspectrometry(GC–MS),arewidelyusedtoanalyzevolatilecompoundsin variousfoodmatrices.Whiletheextractionefficiencyofvolatilecompoundsfromfoodstuffsiscrucialfor obtainingrelevantresults,theseefficiencyoftheseextractionmethodslimitedbytheirlongextraction timesandrequirementsforlargesamplequantity.Thisstudyreportsonthedevelopmentand appli-cationofanewextractiontechniquebasedonHS-ITEXhardware,whichimprovestheextractionrate andcapacitybyoperatingunderreducedpressure,calledDynamicHeadspaceVacuumTransferIn-Trap Extraction(DHS-VTT).TheresultsofthestudyindicatethatDHS-VTTimprovestheextractionofthe targetcompounds.Theareaofthemassspectrometersignalforeachcompoundcanbeupto450times moreintensethantheHS-SPMEandHS-ITEXtechniquesperformedinthesameexperimentalconditions

ofextractiontemperatureandtime.DHS-VTTrunsinautomatedmode,makingitpossibletoworkwith smallersamplequantityandalsofavorstheHSextractionofallvolatilecompounds.Inaddition,the nec-essarymodificationstotheinstallationwerecheapandthelifeofanITEXtrapisupto10timeslonger thananSPMEfibre

©2019TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND

license(http://creativecommons.org/licenses/by-nc-nd/4.0/)

1 Introduction

In food and flavor analytics, maintaining sample quality is

a significant challenge throughout the analytical workflow A

commonlyused methodfor performing samplingof foodis by

headspace(HS)techniques.Althoughtheyare“clean”techniques

withrespecttotheworkingenvironmentandinstallations,from

ourexperiencetheyhave severaldrawbacks dependingonuse

suchas:Extendedheattreatmentandextractiontime,whichcan

resultinartifactformation,changeofthemolecularstructure,and

evendegradationofthesample.AutomatedHS-microextraction

samplingtechniquesperformextractionandinjectionintothegas

chromatograph(GC)inasinglestep,butoftenresultina

discrimi-nativetransferofcompoundsfromthesampleintotheheadspace

∗ Corresponding author.

E-mail address: pascal.fuchsmann@agroscope.admin.ch (P Fuchsmann).

1 Current address: ELSA-Mifroma Group, CH-1470 Estavayer-le-Lac, Switzerland.

Phasepartitioncoefficientsair-water(Kaw),sorbent-air(Ksorb-a), andsorbent-water(Ksorb-w)arethethreefactorsthatinfluencethe phasedistributionduringtheextractionforasystemin equilib-rium[1 Headspacesolid-phasemicroextraction(HS-SPME)isa cheap,simple andsensitiveautomatedtechnique forextracting volatileorganiccompounds(VOCs)fromacomplexmatrix with-out specific sample preparation However, ourown experience showsthatit isverydifficulttomakerobustandaccurate ana-lyticalmethodsusingHS-SPME.Toovercomecertaindrawbacksof HS-SPME,andadddynamicandautomatedfeaturesatthesame time,‘in-tubeextraction’ (ITEX)wasintroducedin 2006byCTC AnalyticsAG(Zwingen,Switzerland).Thetechniqueisgenerally operatedusingamultifunctionalautosampler.HS-ITEXisa solvent-lessdynamicHSmicroextractiontechniquederivedfromseveral othersimilartechniques(suchasSPMEandstirbarsorptive extrac-tion(SBSE))listedanddescribedbyJochmannetal.[2 HS-ITEXis

asequentialextractionbasedonprogressivedilutionand extrac-tionoftheheadspaceonthesample.Thisextractiontechniqueis relatedtothemultipleheadspaceextractionmethod(MHE)[3–6]

https://doi.org/10.1016/j.chroma.2019.05.016

0021-9673/© 2019 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4 0/ ).

GC/MS Detection Parameters for the Investigated Compounds and Validation Parameters of the DHS-VTT Method.

NA: Not available.

andprovidesaninterestingalternativetoHS-SPME; ithasbeen

[2,7–10].AnothersimilartechniquedevelopedbySmartNoseSA

(Marin-Epagnier,Switzerland),InsideNeedleDynamicExtraction

(INDEX),usesthesameconceptofextractionasHS-ITEX,and

fur-therHSextractiontechniqueshavebeenextensivelyreviewedin

theliterature[11].Improvementofthedifferentparametersofthe

HS-ITEXisrequiredtoachieveadequatesignalsthroughoptimized

extractionandthermaldesorption[7,12].TheHS-ITEXkey

param-etersarethenatureofthesorbentmaterial,thenumberofstrokes

(accordingtotheliterature:between20and120[2,3,8,10,13]),the

extractionspeed,thedesorptionspeed,thesample,thetrapand

syringetemperature,andtheheadspacevolumeextracted.Many

commercialextractionpolymerswereevaluatedbyLaaksetal.[14]

andareavailableonthemarket.Thetrapcanalsobefilledwith

noncommercialsorbents,suchasmultiwalledcarbonnanotubes

orpolystyrene-divinylbenzene[1,9,15].Thechoiceofthesorbent

materialismadebasedonthetargetmoleculetobeextracted.An

innovativetechniquedevelopedbyBarajasetal.[16]alsomakes

itpossibletoevaluateandcharacterize sorbentsbyinversegas

chromatographycolumn

Dependingonthematrixanddesiredresults,improving

differ-entparametersmayberequired,makingtheoptimizationofthe

technologyverycomplexcomparedtoHS-SPME[17]

Thegoalofthisstudywastodevelop andoptimizea

repro-ducible,robust,andsensitiveextractionmethodthatreducesthe

drawbacksoftraditionalheadspaceextractiontechniques.Anew

techniqueisproposedthatcombinesfeaturesofHS-ITEXand

prin-ciplesofvacuum-HS-SPMEandvacuumdistillationcoupledwith

gaschromatography[18–21]:DynamicHeadspaceVacuum

Trans-ferInTrapExtraction(DHS-VTT).Thenewmethodaimedtoavoid

strokes,thelimitedinjectionvolumeandprovidedynamic

extrac-tionundervacuumconditions.Severalpublicationshavereporteda

significantimprovementoftheextractionunderreduced-pressure

usingso-calledvacuumHS-SPME[22–24].However,thistechnique

canuptonowonlybeusedmanually.Toaddressthislack,the

currentmethodwasdesignedtofunctionwithanautomaticmode

Thispaperdiscussesthedevelopmentandapplicationofour

newextractiontechnique,DHS-VTTusingamodelofanartificially

constructedmatrix(ACM).Weassessthemethod’ssensitivity,and

itssuitabilityforthequalitativeandquantitativeanalysisofabroad

rangeofvolatilecompounds,smallsamplevolume,andlarge

sam-pleseries.ThesuitabilityandefficiencyofDHS-VTTwereverified

bycomparisontoHS-SPMEandHS-ITEX,aswellasanalysisofACM

andplainyoghurtusingparameterscommonlyfoundinthe

litera-tureforHS-ITEXandHS-SPME[25]

2 Experimental

2.1 Materialsandmethods

2.1.1 Chemicalsandsamples

Therelevantchemicalcompoundswerepurchasedfrom

Sigma-AldrichChemieGmbH(Buchs,Switzerland),R.C.Treatt&Co.Ltd

(Suffolk, UnitedKingdom), and Caesar&Loretz GmbH(Hilden,

Germany):Hexan-1-ol(Aldrich471402),2-Phenylethanol(Aldrich

77861),Butan-2-one(Aldrich360473),Butane-2,3-dione(Aldrich

B85307), 3-Hydroxybutan-2-one (Aldrich W200808),

Heptan-2-one (Aldrich W254401), Undecan-2-one (Aldrich W309311),

Ethyl butanoate (Aldrich W242705), Ethyl 3-methylbutanoate

(Aldrich W246301), Ethyl hexanoate (Aldrich W242906), Ethyl

decanoate (Aldrich W243205),␥-Butyrolactone (Aldrich 90970),

␦-Decalactone (Aldrich W236101), (Z)-6-Dodecen-4-olide (R.C

Treatt 62185), Butanal (Aldrich W221902), 3-Methylbutanal

Table 2

Two-level full factorial design of experiment for the method validation.

Changed parameters: sample temperature (37 ◦ C–80 ◦ C) and extraction time (5–60 min).

seeTable1 alldilutedtoaconcentrationof10mgL−1in20mL

ofpolyethyleneglycol200.Thealkaneswereseparatelydilutedin Miglyol® 812atthesameconcentration.Polyethyleneglycol200 waschosenbecauseitisagoodsolventfortheselectedcompounds anddoesnotreactwiththevolatilecompoundspresentintheACM Thecompoundsthatarenotsolubleinpolyethyleneglycol200(the alkanes)weredissolvedinMiglyol®812whichalsosimulatesthe lipidpartofadairymatrix.Thesemixtureswerestoredat−40◦C

untilanalysis

2.1.4 Samplepreparation ACMstandardsolutions(25␮Lofeachstandard)wasdiluted withH3PO44%indeionizedwateratpH1.3(totalsamplevolumeof twomillilitres)andplacedin20mLHSvials(Interchim,Montluc¸on, France) The acidification of the sample helps to promote the volatilityofcarboxylicacids.Phosphoricacidisnotvolatileand willnotinterfereintheanalyseseither.Thevialswerehermetically sealedusingabluesilicone/Teflonseptum(Interchim).Calibration curvesweremadebydilutingthestocksolutionbyafactoroftwo foraminimumofsevenpoints.Tofacilitatedataprocessing,the43

Comparison between DHS-VTT with different sorbent materials, HS-ITEX TTA and HS-SPME for the investigated compounds.

C13andC20(Tables1and3)

Forrealdairy analyses, two millilitresofhomogenizedplain

yoghurtwereplacedin20mLHSvials(Interchim)andtheywere

hermetically sealed using a bluesilicone/Teflon septum

(Inter-chim)

2.1.5 Modificationofthegasdistributionblock

Theoriginaltwo-wayaluminumgasdistributionblockofthe

autosamplerMPS2(Gerstel,Sursee,Switzerland)wasreplacedby

a three-wayblock printed ona 3DTouch Pegasus printer(Full

SpectrumLaser,LasVegas,US)with“UniversalClearResin”(Full

spectrumLaser)(Fig.1).ThepolymerwassolidifiedusingUV

expo-sureafterprinting.Thismodificationmakesitpossibletoswitch

thesolenoidvalvebetweenthevacuumpumpandtheinertgasto

extractvolatilecompoundsfromthesamplewiththevacuumand

desorbthemintotheinjectorusinganinertgasflow

2.1.6 Pressuremeasurement

In order to verify the hermeticityof thevial, a pressure of

10mbarwasappliedinanemptyvialusingavacuumpumpBuchi

V-300equippedwith a pressurecontrol interfaceI-300 (Büchi,

Flawil,Switzerland).Thevialwasconnectedtoapressuresensor

PAA-27(Keller,Winterthur,Switzerland)andameasuringdevice

Almemo2590(Ahlborn,Holzkirchen, Germany).Thepumpwas

thenstoppedandthesignalwasrecordedsimultaneouslyinacsv

fileandprocessedwithExcel

EvaluationofthepressureinthevialduringHS-ITEXand

DHS-VTT extraction was carried out using the same equipment as

describedabove.Themeasuringdevicewasconnectedtoa

nee-dlewhichpiercestheseptumofthevialsatthesametimethatthe

extractionwasperformed.InstrumentationforGC–MSanalysis

TheanalyseswerecompletedusinganMPS2autosampler

(Ger-stel)onanAgilent7890BGCsystemcoupledtoanAgilent5977A

massselectivedetector(MSD)(AgilentTechnology,SantaClara,

CA,USA).Theheadspacewasextractedaccordingtothe

experi-mentaldesign(Table2).BoundvolatilesweredesorbedinaPTV

oftypeCIS4(Gerstel).Volatilecompoundswereseparatedona

TRB-FFAPfusedsilicacapillarycolumn(100%PEGwith

nitrotereph-thalicacid,bondedandcrosslinked,60m×0.32mm×1.0␮mfilm;

Teknokroma,Barcelona,Spain)withheliumasthecarriergasata

constantflowof2.5mLmin−1(30cmsec−1)

Theoventemperatureprogramwasasfollows:fiveminutesat

40◦C,thenheatedto220◦C ata rateof5◦Cmin−1,witha final

holdtimeof34mintomakeatotalruntimeof75min.TheMS

settingswereasfollows: thetransferlineat 230◦C,thesource

temperatureat230◦C,thecompoundsmonitoredinSCANmode

between29amuand250amuwithoutsolventdelay,andinSIM

modeforthemethoddetectionlimits(MDLs)andthecalibration

curves.TheautosamplerwascontrolledusingtheCycleComposer

V.1.5.4(CTCAnalytics,Zwingen,Switzerland)andthePTV

injec-torwithMaestro1softwareV.1.4.8.14/3.5(Gerstel).Thedetector

responsesignalswereintegratedusingMasshunterquantitative

analysissoftwareversionB.08.00(Agilent).TheNIST/EPA/NIHmass

spectrallibrary(NIST14)version2.2(NIST,Gaithersburg,MD,USA)

wasusedforpeakidentification

2.2 Headspacein-tubeextraction(HS-ITEX)

TheITEX-2 option (Brechbühler,Schlieren, Switzerland)was

usedontheMPS2autosampler.Theheadspacewasextractedusing

aTTAtrapwith70extractionstrokes(volumeof1.3mL/stroke,

extractionflowrateof100␮Ls−1).ACMandyoghurtsampleswere

notconditionedforthisexperiment.Thesyringetemperaturewas fixedat100◦C,theITEXtrapat35◦C,andthesampleat55◦C.The boundvolatilesweredesorbedfromthesorbentmaterialat240◦C

ina CIS4injectorequippedwitha glasslinerfilledwithTTAat

10◦C.Avolumeof1.3mLwastransferredintotheinjectorata des-orptionflowrateof100␮Ls−1,afterhavingwaited30s(plunger down)intheventmode(50mLmin−1,setpressure:0kPa).The injectorwasthenheatedatarateof12◦Cs−1to240◦C.Thepurge flowtosplitventwassetat300mLmin−1after5min.After injec-tion,theITEXneedlewasreconditionedaccordingtothesupplier’s temperaturerecommendationfor15minunderanitrogenflowof

220mLmin−1(measuredat35◦C)

2.2.1 Headspacesolid-phasemicroextraction(HS-SPME)

ToobtainarepresentativeevaluationoftheVOCsofthesamples, the VOCs were extracted using a 1cm 50/30␮m divinylben-zene/carboxen/polydimethylsiloxane(DVB/CAR/PDMS)StableFlex fibre (Supelco, Bellefonte, PA, USA) The fibre was conditioned accordingtothesupplier’srecommendations(270◦Cfor60min) ACMandyoghurtsampleswerenotconditionedforthis experi-ment.Thesampleheadspacewasextractedfor30minat55◦Cwith

anagitationrateof250rpm.Theboundvolatilesweredesorbedfor

60sat240◦Cintheinjector,whichwasinthesplitlessmodefor60s beforethesplitvalvewasopened(splitflow=80mLmin−1)[25] 2.2.2 Dynamicheadspacevacuumtransferintrapextraction (DHS-VTT)

DHS-VTTrequiresthesameequipmentasHS-SPMEandHS-ITEX methods.Avacuumpumpwithpressurecontrolinterfaceandthe newgasdistributionblockcompletetheinstallation(Fig.2A)

ApieceofcleanedswabTopper®810×10cm(Systagenix,North Yorkshire,UnitedKingdom)wasaddedtothevialtoavoid boil-ingandfoaming.Thesyringetemperaturewasfixedat100◦C,the ITEXtrapat35◦C,andthesampleaccordingtotheexperimental design(Table2).Theextractiontakesplaceasillustrated(Fig.2B) Sampleconditioning:ACMandyoghurtsampleswerenot condi-tionedforthisexperiment.Extraction:Thetrappiercestheseptum and theplungerrises totheupperposition.When theplunger

isintheupperposition,theHSwasextractedundercontinuous reducedpressureusingavacuumpumpBuchiV-300equippedwith

apressurecontrolinterfaceI-300(Büchi)andthegasdistribution blockaccordingtotheexperimentaldesign(Table2)without agi-tation.Drying:Thetrapwasthenremovedfromthesampleandthe solenoidvalveswitchestonitrogen.Sorbentandsyringeweredried underanitrogenstreamfor17minat220mLmin− 1[26] Desorp-tion:ThetrapwasthenintroducedintothePTVinjectorandthe boundvolatilesweredesorbedfromthesorbentfor120switha nitrogenflowof220mLmin−1(fixedat35◦C)at240◦C.ThePTV injectorwasintheventmodeat50mLmin−1and0kPafor120s TheinjectorwasequippedwithaglasslinerfilledwithTTAand cooledwithliquidnitrogenat10◦C.Thetrapwasremovedfrom theinjectorandtheinjectorwasthenheated,atarateof12◦Cs−1,

to240◦C.Thepurgeflowtosplitventwassetat300mLmin−1after

5min.TrapCleaning:Thereconditioningofthetrapwasachieved

inthesamewayasfortheHS-ITEXmethodbycirculatingaflowof nitrogenthroughthetrapattherecommendedtemperaturewith theplungerintheupperposition

2.3 Experimentaldesignandstatistics

Tooptimizeexperimentalparameters,a two-levelfull facto-rial design of theexperiment was performed.Only thesample temperatureandthetimeofextractionwereinvestigatedbecause thesearetheparametersthathavethemosteasilymodifiableand havethegreatestimpactontheextractionofvolatilecompounds Thetemperature of thetrapshouldbekeptas low aspossible

Fig 1. Design of the modification of the gas distribution block in polymer (dimension in mm) The block was designed with three ways, one for the nitrogen supply (A), one for the vacuum (B), and the last way was directly connected to the autosampler.

Fig 2. (A) Diagram of the GC–MS instrument with the autosampler (MPS2) and ITEX-2 hardware Connection of the original nitrogen line to the new distribution block (in orange) and the solenoid valve The vacuum and nitrogen lines are coloured green and blue respectively (B) Stages of the extraction process using the DHS-VTT method Sample conditioning, extraction, drying, desorption and trap cleaning steps.

Fig 3.Parameter optimization according to the experimental design for DHS-VTT (TTA/CSIII) X-Axis: Sample temperature [ ◦ C] Y-Axis: Extraction time [min] Z-Axis: Total ion count normalized [-] The color variations represent blue for a low signal and red for a high signal.

(withintheconstraintsoftheinstrumentspecificationsandthe

laboratory environment) and the syringe temperature kept at

100◦Ctoexcludewatercondensationduringtheextraction.Three

centrallycomposedDHS-VTTexperimentalplansweredeveloped

Therespectiveresultswereevaluatedandpresentedtoshowthe

areasofeffect.Mosaicplots(Fig.3 whichcanbeinterpretedusing

color,werechosenfor thepresent work,and eachpeakareaof

effectwassmoothedsquarely(smoothingmethod).Intheplots,

theintensityofthepeakareaobservedisdisplayedincolor,from

blue (no signal), through dark and light green, yellow-orange,

and,finally,tored,thelargestandmostintensepeakarea,which

is equivalentto themaximal signal.The mean values of these

measurement rangeswerethecenter of theCentral Composite

design The remaining levels were placed radially around the

centerina2*22experiment,whichcorrespondsgeometricallyto

twosquares,rotatedby45◦intheplanearoundtheircenter.This

designallowsfortheinfluenceoffivelevelsperparameter.Height

points are distributed around the center at the same relative

distance.Intotal,12measuringpointsaremade,thecenterpoint

forthevariabilitybeingmeasuredfourtimes

3 Results and discussion

3.1 Extractionparameters

3.1.1 Pressureintheheadspace

Theeffectivenessofbluesilicone/Teflonseptuminthe

hermetic-ityofvialshasshownthatthepressureremainedstableforaperiod

of10minafterthepumpwasstoppedandthuscanbeconsidered

assufficientlyeffectivefortheanalyses(resultsnotshown)

Thepressuremeasurementsmadeduringtheextraction

accord-ingtotheHS-ITEXmethodshowthattheextractionswerecarried

out above atmospheric pressure due to air expanding in the

vialduringextractionwhenitwasheated.Thesamplewas

pre-paredatatmospheric pressure beforeextractionthen placed in

theautosamplercoolingdevice(4◦C).Thepressureimmediately

decreaseddue tothevolumecontractionoftheairinthevials

Thenthevialwasplacedintheheatingagitator(55◦C),andthe

pressureincreasedabovetheatmosphericpressure(Fig.4,red

sig-nal).Throughoutextraction,thehighpressureremainedconstant

Asaresult,therateofevaporationofthecompoundsdecreasesand

theextractionefficiencyofvolatilecompoundswasthuslower[27]

ThisphenomenonwassimilarduringanHS-SPMEextraction

withoutthestrokeseffect.Withthechangeofthegasdistribution

block,itwaspossibletoautomatetheextractionatreduced

pres-sureinthevialthroughouttheextraction(Fig.4,greensignal).The

Fig 4.The pressure inside the vials during extraction using HS-ITEX (red) and DHS-VTT (green) method at set pressure: 100 mbar.

vacuumobtainedinthevialsgreatlyinfluencedtheextractionof volatilecompounds[22].Apressureabove100mbarwasnolonger sufficientforextractingthecompoundsefficiently(Fig.5).Reduced pressureacceleratedthereleaseofthevolatilecompoundinthe headspace,butalsohasanimpactontheefficiencyofthesorbent Lowpressureoveralongperiodoftimedesorbsthemostvolatile compoundsofthesorbent.Consequently,itwasnecessaryto deter-minetheoptimalconditionsforachievingasuitablecompromise betweenthepressureinthevialandtheextractionrate.The pres-sureeffectsdependedonthevolatilityofthemolecules.Thehighly volatilecompoundsasthepentanewerebetterretainedonthe polymerwhen thepressure inthevialwasnot lessthanabout

50mbar(datanotshown),whileothervolatilecompoundswere moreeasilyextractedatthelowestpossiblepressure.Theeffect

ofthereducedpressureduringextractioncausedanissuealready knownfromthepurgeandtraptechnology—theaccumulationof watervaporinthesystem(needle,syringe,tubes)andartifact for-mation[12,26].Toavoidthis,thetrapandthesyringeweredried underanitrogenflowforatimedeterminedpriortoinjectioninto theinjector

3.1.2 SamplevolumeandHSvolume Testswereconductedusingsamplevolumesof0.5,1, 2and

4mLin20mLHSvials.TheDHS-VTTparameterswerethesame

Fig 5. The sum of the peak areas of the 43 target compounds divided into three groups (Y-Axis: Total ion count) as a function of the pressure in the vials, expressed in mbar (set pressure) (X-Axis).

Fig 6. Influence of the sample amount on the total ion count (TIC) MS signal of the

target compounds.

foreachtest.Theamountofthesample,and,consequently,the

numberofvolatilemoleculesavailableinthevialsinfluencedthe

extraction.Whentheamountofthesampledoubled,thetotal

sig-nalincreasedproportionallybyafactorof1.7foragivenextraction

time(Fig.6).Thisdemonstratesthattheextractioncapacityofthe

trapwasnotreached,allowingforquantitativestudiesusinga2mL sample[28,29]

3.1.3 PTVinjectortemperature Lowtemperatureinthelinerhadapositiveimpactonthe recov-eryofvolatilecompounds.Thechangefrom10◦Cto−50◦Cdoes

notshowasignificantimprovementinrecoveryofvolatile com-poundsforgroups2and3.Forreasonsofcostandefficiency,we usedatemperatureof10◦CforthePTV(Fig.7)

3.1.4 Sampletemperatureandextractiontime Theeffectofthesampletemperaturewasevaluatedfor tem-peraturesbetween30and80◦Cusingthedescribedexperimental design(Table2).Increasedtemperatureledtoanincreaseinthe signal for all compounds due toa higher partitioning of com-poundsfromtheaqueousphasetotheheadspace[1 Thehigher thetemperature, thehigher theextraction efficacyfor theless volatilecompounds.Atcertainhightemperatures,thecompounds degradedandtendedtocontaminatetheautosampler,i.e.,water vapor condensed in thetrap,the syringe,and theautosampler tubes.DespitethehydrophobicnatureofTTA,anaccumulationof waterinthetrapcouldresultintheproblemofinjectingwater dropletsintotheGCwiththeinjectionofthecompounds[26]

Fig 7.Influence of the PTV injector temperature on the recovery of the investigated compounds after injection The injector temperature was tested between 20◦C and

−50 ◦ C.

Fig 8. Influence of the thermodesorption temperature on the nitrogen flow inside

the ITEX trap.

Witha reduced pressure in thevials, we observed that the

extractionrateaccelerated.For themostvolatilecompounds, a

prolongedextractiontimehadanegativeeffectontherecovery

ofthemolecules[1 Indeed,whenextractiontimewastoolong,

themostvolatilecompoundswerereleasedfromthesorbent.For

themajorityofthemoleculesstudiedinACM,theresultsofthe

experimentshowedanoptimaltimeof30minat55◦C.The

extrac-tionparameterscanbeadaptedaccordingtothemoleculestobe

studied(Fig.3)

3.1.5 Nitrogenflowdesorption

Thenitrogenflowwasoptimizedto220mLmin−1 at35◦Cto

allowforproperdesorptionofthecompoundsfixedtothepolymer

andextractiontocounteractagainstthepressureintotheinjector

duringdesorption.Anexponentialdecreaseinthenitrogenflow

ratewasnotedwhenthetemperatureofthetrapincreased(Fig.8)

Atoo-highflow(>300mLmin−1)sometimescloggedthetrapby

sealingthesorbentmaterial

3.2 ComparisonofHS-SPME,HS-ITEX,andDHS-VTT TheresultsindicatethatDHS-VTTclearlyimprovesthe extrac-tionof volatilecompoundsfroma complex matrixas theACM (Table 3) The peak area can be increased by a factor of 450 for some compounds in comparison with using the HS-SPME method.Incomparison,HS-SPMEextraction,widelyusedin com-mon headspace applications, achieves lower efficiency for all compoundsmeasuredunderthesametemperatureandextraction timeconditions.TheDHS-VTTtechniqueenriches thegasphase rapidlybylimitingthematrixeffectsbecauseofthereduced pres-sureinthevial.Moreover,theconventionalHS-ITEXmethodcannot extracttheleastvolatilecompounds.Theheadspaceisrapidlyina stablestateinHS-ITEX,andthestrokesareinadequatetoextract morevolatilecompounds.ThechoiceofpolymerintheDHS-VTT extractiontrapwasalsoveryimportantforoptimalextraction.The bestcompromiseforthe43compoundsusedinthisstudywasthe TenaxTA/CarbosieveSIIIblend

Theprofilesofthevolatilefractionobtainedforaplainyoghurt using theDHS-VTT,HS-ITEX, and HS-SPME extractionmethods showthattheDHS-VTTmethodwassuitableforthistypeofsample (Fig.9).Theanalysesperformedontheyoghurtconfirmtheresults obtainedwiththeACM.TheHS-ITEXmethodwaseffectiveforthe extractionofhighlyvolatilecompoundssuchasacetaldehydebut didnotextractheaviercompoundsunderthedescribedconditions WiththeHS-SPMEallcompoundsexceptthedecanoicacidcould

beidentified.However,thesignalsweremuchweakerthanwith theDHS-VTTmethod.Thesurfaceareaofthepeakswashigherwith DHS-VTTandahighernumberofcompoundscouldbedetectedin SCANmodeusingthismethod

3.3 PerformanceevaluationofDHS-VTT Performanceevaluationwascarriedoutbydeterminingthe fol-lowingparameters:

• Thelinearity(R2)of themethodwasevaluatedby measuring thetargetcompoundsinsixdifferentconcentrationsintheACM correspondingtotherankscommonlyfoundindairymatrices

Fig 9.Chromatograms corresponding the volatile fraction of plain yoghurt extracted by DHS-VTT, HS-ITEX and HS-SPME methods The chromatographic conditions were given in Section 2.2 Instrumentation for GC–MS analysis (sample temperature 55◦C, extraction time 30 min) Peak numbers correspond to : 1:Acetaldehyde, 2:Acetone, 3:Ethylacetate, 4:Butan-2-one, 5:Butane-2,3-dione, 6:Pentane-2,3-dione, 7:Hexanal, 8:Heptan-2-one, 9:Octanal, 10:3-Hydroxy-butan-2-one, 11:2-Methylpentan-3-ol, 12:2-Hydroxy-3-pentanone, 13:Nonan-2-one, 14:Nonanal, 15:Acetic acid, 16:Propanoic acid, 17:2-Methylpropanoic acid, 18:Undecan-2-one, 19:Butanoic acid, 20:2-Phenylacetaldehyde, 21:Pentanoic acid, 22:Hexanoic acid, 23: 6,10-dimethylundeca-5,9-dien-2-one, 24:2-Phenylethanol, 25:Octanoic acid, 26:Nonanoic acid, 27:

␦–Decalactone, 28:Decanoic acid, 29: ı-Dodecalactone A, Artifact.

Fig 10.850 injections of the ACM sample over two weeks in DHS-VTT Y-axis: Sum of the peak areas of the 43 molecules selected Standard deviation 9.6%.

TheR2wasapproximately0.99+/-forallcompoundsexceptfor

␥-butyrolactonewith0.98

• TheMDLofthetargetmoleculeswasestimatedaccordingtothe

USEnvironmentalProtectionAgencyprocedureandEq.(1):

• wheretisthestudent’stvaluewithaconfidencelevelof99%

andsixdegreesoffreedomandScthestandarddeviationofseven

replicatesataconcentrationlevelfeaturingasignal-to-noiseratio

(S/N)ofthreetoone.TheMDLsmeasuredforalkanesrangedfrom

0.036␮g L−1 to0.822␮gL−1,forestersfrom0.003␮gL−1 to

0.071␮gL−1,foraldehydesfrom0.050␮gL−1to2.89␮gL−1,

forketonesfrom0.145␮gL−1to11.9␮gL−1,forlactonesfrom

0.130␮gL−1to0.195␮gL−1,foralcoholsfrom0.032␮gL−1to

0.127␮gL−1,andforthecarboxylicacidsfrom0.741␮gL−1to

104␮gL−1(Table1)

• Thereproducibility(RSD%)wasobtainedbydeterminingthe

rela-tivestandarddeviationcalculatedontheanalysesofthesamples

eight times over a period of two weeks Reproducibility was

measuredbetween 2.2 and 33% for allcompounds (Table1)

Reproducibilitywaslowforthecarboxylicacids;theextraction

conditionswerenotoptimizedforthesecompounds

• Theextractionratio(E)ofthemethodwasevaluatedbyextracting

thesamesamplefivetimes.TheEvaluewascalculatedaccording

toZimmermannetal.,byplottingthelogarithmicalpeakareas

againstthenumberofextractionandasimplifiedEq.(2)[30]

Theslopeofthelinearregressionisrepresentedintheequation

bythelog(1-E)

lognf,x=log

ns,0E

TheEvaluerangedfrom1.0%fordodecanoicacidto77.8%for

butan-2-one

• 850injectionsofACMweremadeusingthesameITEXtrap.The

resultsshowthereproducibilityandstabilityoftheanalyses,with

amaximumvariationof9.6%betweenthetwoextremeresults

(Fig.10)

4 Conclusions

ThisnewlydevelopedDHS-VTTtechniqueimprovesthe

extrac-tion of volatile compounds from an ACM by a simple and

inexpensivemodificationoftheautosamplerandbyusing

com-mercialITEXhardware.UsingtheprincipleofDHS-VTTprovidesa

rapidextractionoftargetcompoundswithminimaldamagetothe

sampleandlimitedartifactformation.Workingatreducedpressure increasestheevaporationrateofthecompounds.Extractionunder continuousreducedpressureavoidscreatingasystemin equilib-rium,and,thus,extractionremainsdynamic.Ourresultsshowgood repeatabilityandsensitivityforthemajorityofthetargetmolecules assessed

ComparingtheresultsofextractionbyDHS-VTT(TTA,TGR,CSIII, andTTA/CSIII),HS-ITEX(TTA)andHS-SPME(DVB/CAR/PDMS),we observethatDHS-VTTshowsbenefitsforbothtimeand sensitiv-ityofextractionforvolatilecompounds,includinggroup1,group

2,andgroup3(Table3).Themuchlargeramountsofcompounds extractedwiththeDHS-VTTmethodwouldfacilitatemoreefficient olfactometricanalyses.However,furtherresearchisnecessaryto comparehowrepresentativethedifferentextractionmethodsare forodorstovalidatethedevelopedmethodforolfactometry pro-filing Thiswould also requiredetailed odorrepresentativeness studies[31,32]

Totheauthors’knowledge,nootherpublicationhasreported

onthedevelopmentofthisnewDHS-VTTmethod,whichallowsfor betterextractionoftargetvolatilesusingshorterextractiontimes andlowertemperatures

Notes

Theauthorsdeclarenocompetingfinancialinterest

Acknowledgements

WethankDr.Jacques-OlivierBosset,Dr.BarbaraGuggenbühl, andDr.KathrynJBurtonforhelpfulcommentsonthemanuscript

References

[1] T Hüffer, X.L Osorio, M.A Jochmann, B Schilling, T.C Schmidt, Multi-walled carbon nanotubes as sorptive material for solventless in-tube microextraction (ITEX2)- as factorial design study, Anal Bioanal Chem 405 (2013)

8387–8395, http://dx.doi.org/10.1007/s00216-013-7249-7 [2] M.A Jochmann, X Yuan, B Schilling, T.C Schmidt, In-tube extraction for enrichment of volatile organic hydrocarbons from aqueous samples, J Chromatogr A 1179 (2008) 96–105, http://dx.doi.org/10.1016/j.chroma.2007 11.100

[3] J Zapata, R Lopez, P Herrero, V Ferreira, Multiple automated headspace in-tube extraction for the accurate analysis of relevant wine aroma compounds and for the estimation of their relative liquid-gas transfer rates, J Chromatogr A 1266 (2012) 1–9, http://dx.doi.org/10.1016/j.chroma.2012.10.

015 [4] M.T Suzuki, S Tsuge, T Takeuchi, Gas chromatographic estimation of occluded solvents in adhesive tape by periodic introduction method, Anal Chem 42 (1970) 1705–1708, http://dx.doi.org/10.1021/ac50160a035