Comprehensive two-dimensional gas chromatography–time-of-flight mass spectrometry for the forensic study of cadaveric volatile organic compounds released in soil by buried decaying pig carcasses

(1)

JournalofChromatographyA,xxx (2012) xxx–xxx

ContentslistsavailableatSciVerseScienceDirect

Journal

of

Chromatography

A

j o ur na l ho me p a g e :w w w . e l s e v i e r . c o m / l o c a te / c h r o m a

Comprehensive

two-dimensional

gas

chromatography–time-of-ﬂight

mass

spectrometry

for

the

forensic

study

of

cadaveric

volatile

organic

compounds

released

in

soil

by

buried

decaying

pig

carcasses

Catherine

Brasseur

a

_,

_Jessica

_{Dekeirsschieter}

b

_,

_Eline

_M.J.

_Schotsmans

c

_,

_Sjaak

_de

_Koning

d

_,

Andrew

S. Wilson

c

_,

_Eric

_Haubruge

b

_,

_{Jean-Francois}

_Focant

a,∗

a_CART,_Organic_and_Biological_Analytical_Chemistry,_Chemistry_Department,_University_of_Liège,_Allée_du₆_août,_B6c,_Sart-Tilman,_B-4000_Liège,_Belgium b_Department_of_Functional_and_Evolutionary_Entomology,_Gembloux_Agro-Bio_Tech,_University_of_Liege,₂_Passage_des_Déportés,_B-5030_Gembloux,_Belgium c_Forensic_&_{Archaeological}_Sciences,_School_of_Life_Sciences,_University_of_Bradford,_West_Yorkshire_BD7_1DP,_United_Kingdom

d_LECO_Instruments_GmbH,_{Marie-Bernays-ring}_31,₄₁₁₉₉_{Mönchengladbach,}_Germany

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Available online xxx Keywords:

Comprehensivetwo-dimensionalgas chromatography(GC×GC)

Time-of-ﬂightmassspectrometry(TOFMS) Volatileorganiccompound(VOC) Cadaverdecomposition Forensic

Geotaphonomy Scripting

a

b

s

t

r

a

c

t

Thisarticlereportsontheuseofcomprehensivetwo-dimensionalgaschromatography–time-of-ﬂight

massspectrometry(GC×GC–TOFMS)forforensicgeotaphonomyapplication.Gravesoilsampleswere

collectedatvariousdepthsandanalyzedfortheirvolatile organiccompound(VOC)proﬁle. Adata

processingprocedurewasdevelopedtohighlightpotentialcandidatemarkermoleculesrelatedtothe

decompositionprocessthatcouldbeisolatedfromthesoilmatrix.Some20speciﬁccompoundswere

speciﬁcallyfoundinthesoilsampletakenbelowthecarcassand34othercompoundswerefoundatall

depthsofthegravesoilsamples.Thegroupofthe20compoundsconsistedofketones,nitriles,sulfurs,

het-erocycliccompounds,andbenzenederivativeslikealdehydes,alcohols,ketones,ethersandnitriles.The

groupofthe34compoundsconsistedofmethyl-branchedalkaneisomersincludingmethyl-,dimethyl-,

trimethyl-,tetramethyl-,andheptamethyl-isomersrangingfromC12toC16.Atrendintherelative

pres-enceofthesealkanesoverthevariouslayersofsoilswasobserved,withanincreaseintheamountof

thespeciﬁcalkaneswhencomingfromthecarcasstothesurface.Basedonthespeciﬁcpresenceofthese

methyl-branchedalkanesingravesoils,wecreatedaprocessingmethodthatappliesaspeciﬁcscript

tosearchrawdataforcharacteristicmassspectralfeaturesrelatedtorecognizablemassfragmentation

pattern.Suchscreeningofsoilsamplesforcadavericdecompositionsignaturewassuccessfullyapplied

ontwogravesoilsitesandclearlydifferentiatessoilsatproximityofburieddecayingpigcarcassesfrom

controlsoils.

1. Introduction

Forensicgeotaphonomyis abranchofforensicanthropology focusingonthegeologicalandbotanicalalterationofthe surround-ingenvironmentofaburieddeadbody.Thestudyofbiochemical changesinsoilrelatedtothegravedigging,tocadaver decompo-sitionprocessesand todecompositionstates themselvescanbe usedin forensicinvestigationtoestimatepost-mortem interval andpossiblylocateclandestinegravesorhiddenremainsofa vic-tim.Despitethisinterest,theknowledgeoftherelationbetween cadaverdecompositionandsoilchemistryisstilllimited[1].For

∗ Correspondingauthorat:UniversityofLiège,CART,BiologicalandOrganic Ana-lyticalChemistry,Alléedu6août,B6c,B-4000Liège,Belgium.

Tel.:+32043663531;fax:+32043664387. E-mailaddress:JF.Focant@ulg.ac.be(J.-F.Focant).

this reason, gravesoilis normally only used as associative evi-denceratherthanasmatrixthatcanpotentiallygiveinformation oncadavericdecomposition.It hasbeenknownfora longtime thatcadaverdecomposition caninfluencegravesoilbiologyand chemistry[2].Eachstageofcadaverdecompositionisknownto havesomespecificeffectonsoilparameters (microbialactivity, carbondioxideproduction,pH,electricalconductivity,etc.)[3,4]. Duringthevariousdecaystagesofaburiedcadaver,alotof dif-ferent and specific volatile organiccompounds (VOCs) are also emitted [5]and releasedinto thesoil.Someof those VOCsare responsiblefor thetypical“smell ofdeath”that weare ableto sense,althoughsomeareonlydetectedbysensitiveorganismslike scavengerinsectsorcadaverdogs[6–8].Severalgroupsare study-ingthisspecificVOCemission[9–14]butmostofstudiesconcern abovegrounddecomposition.Additionally,despiterecentadvances in VOCcharacterizationandtheestablishment ofa decomposi-tionodoranalysisdatabase[9],thereisstilla lotofworktodo tobettercharacterizetheolfactivesignatureofadecayingbody. 0021-9673/$–seefrontmatter © 2012 Elsevier B.V. All rights reserved.

(2)

2 C.Brasseuretal./J.Chromatogr.Axxx (2012) xxx–xxx

Suchacharacterizationisessentialtobetterunderstand mecha-nismsresponsibleforinsectattractionandpossiblybevaluablefor forensicentomology[15].

Analyzingmammalian decompositionproducts is an analyt-icalchallengeas complex reactionstakeplace, resultingin the chemicalbreakdownofthebody’smainconstituents(lipids, pro-teinsandcarbohydrates)[16].FortheanalysisofVOCsproduced duringdecompositionprocesses,gaschromatography (GC) cou-pledtomassspectrometry(MS)isthetoolofchoice.Solid-phase extraction(SPE) [17], solid-phase micro extraction (SPME) [9], thermaldesorption (TD)[11–13,18] have beenusedfor sample collectionandtransfertoGC–MS.Alargerangeofvolatileand semi-volatilelowtomediumpolarcompoundscaneasilybeanalyzed byGC–MS, althoughpolaranalytesrequireaderivatizationstep prioranalysis.BecauseofthelargenumberofVOCspresentinthe chemicalprofileofdecompositionodor[10],GChowevereasily suffersfrompeak capacitylimitations. Theuseof non-scanning time-of-flight (TOF) MS analyzers that offer constant ion ratio overGCpeakscanresolvesomeoftheGCco-elutionsintheMS domainbyspectraldeconvolution.However,suchanalytical res-olutionmightsufferinefficiencyasmanycompoundsareissued fromthesamechemicalfamiliesandarethereforecharacterizedby similarmassspectraldata,whichcomplicatesthedeconvolution of the MS signal. Such a complex situation is further compli-catedinthecaseofgravesoilanalysisasthesoilmatrixitselfis madeofvariousgroupsofchemicalcomponents,organisms,and debris[19].

Comprehensive two-dimensional gas chromatography (GC×GC) has been developed to meet an increasing need for complex sample analysis and to address limitations such as peak capacity, dynamic range and restricted specificity of one-dimensional(conventional)GCsystems(1D-GC)(i.e.toimprove theefficiencyoftheseparation).GC×GCcanbedefinedasa chro-matographictechniqueduringwhichasampleissubjectedtotwo differentseparationprocessescoupledon-line[20].Inpractice,the endofthefirstdimension(1_D)_column_is_placed_in_a_temperature

controlled interface called ‘the modulator’ and further serially connectedtotheseconddimension(2_D)_column._The _cryogenic

modulatorensureshighsamplingratesandtransferofthesample to2_D_column_[21]_._Modulation_also_acts_as_a_signal_enhancer_by

zonecompression[22].Theentire1_D_chromatogram_is_thus_‘sliced’

followingamodulationperiod(PM)ofafewsecondsandsentinto 2_D_for_a_fast_GC-type_separation_[23]_._By_ﬁne_tuning_of_the_GC_phase

combination,compoundspotentiallystillco-elutingattheendof the1_D_can_be_separated_on_the_basis_of_their_different_behaviors

asregardofthe2_D_phase._The_separation_power_is_increased_and

thesensitivityisenhanced[24].Forthedetectorresponsiblefor recordingthesignal,everythinghappensasinclassicalGCanda traceismonitoredcontinuously.Inpractice,seriesofhighspeed secondary chromatogramsof a lengthequal to PM (3–10s) are

recorded one after another. They consist of slices that can be combinedto describethe elutionpattern by meansof contour plotsinthechromatographicseparationspace[25].Asoftwareis responsibleforprocessingthecollectedrawdataandextractthe multi-dimensionalinformation.AcompletedescriptionofGC×GC instrumentalsetupisavailableinapreviousreport[26].GC_×GC, oftencoupledtofastacquisition TOFMS,hasthusbeenusedto analyzecomplexsamplesinvariousﬁelds,includingVOCanalyses

[27–29].

Tothebestofourknowledge,despitethecomplexityof sam-ples,GC×GC–TOFMShasneverbeenusedinthegeoforensicﬁeld. Here,wereportpreliminaryresultsonusingGC×GC–TOFMSfor thestudyofVOCsreleasedinsoilbyburieddecayingpigcarcasses. Theuseofbothretentiontimesfromtheﬁrstdimensioncolumn,

1_t

R,andfromtheseconddimensioncolumn,2tR,peakintensity,

deconvolutedmassspectraldataaswellasclassiﬁcation,scripting

andsimplechemometricsarecombinedtotrytoextractspeciﬁc informationaboutthecompositionandpotentialvariationsofthe VOCmixture.

2. Experimental

2.1. Fieldsite

Thestudysitewasaforestbiotope,locatedinBelgium (Lambert-coordinates:172800.00/167150.00).Thefieldsitewasaresearch facilityareadevotedtoforensicresearchmanagedbythe Disas-terVictimIdentification(DVI) oftheBelgianFederalPolice.The tree layerof thefield site was dominated by oak trees (Ouer-cusrubra, Quercus robur) and beechtrees (Fagus sylvatica).The shrublayerwasabsent.Thesoilvegetationwasscatteredandthe herb layerwasmainlyconstituted bybracken (Pteridium aquil-inum),blackberry(Rubusfructicosus),lily-of-the-Valley(Convallaria maialis)andMayLily(Maianthemumbifolium)duringspringand summer.Concerningthemosslayer,thereweresomesparespots ofPolytrichumsp.TheBelgiansoilmapindicatedadrysandyloam withstrongdrainage.Thesoilprofileconsistedofahumus top-soilA-horizon(pH2.1),which overlaidasubsoil B-horizon(pH 3.9),withcleareluviationandilluviationlayers.The environmen-talparametersweremonitoredusingtwodatarecordingmethods, semi-localweatherdataandmicroclimatedata.Semi-localweather data(atmospherictemperaturesandprecipitations)wereobtained fromthenearestweatherstation(at6.5kmoftheexperimental site)oftheRoyalMeteorologicalInstituteofBelgium(KMI-IRM). Microclimatedatawererecordedusingatinytagdualchannel tem-peraturelogger(TGP1520®_,_Gemini_Dataloggers,_Chichester,_UK).

One probe wasused tomeasure temperature variationsin the core(anus)ofthepig’sbodyandtheotherprobewasplaced5cm beneaththegroundsurfaceofthegraveﬁll.

2.2. Animalmodel

Two pig (Sus domesticus L.) carcasses were used to model thehumandecompositionprocess[30–33].Bothcarcasseswere obtainedwithin10daysofdeath.InFebruary2008,thepig car-casseswereburiedinshallowgravesdugbyhand,atadepthof 40cmandontheirleftside(aspartofanothercadaver decompo-sitionforensicstudy[34]).Thepiggraveswerebackﬁlledwithin 3.5hwiththeexcavatedsoilinrandomfashionandcompactedby gentletrampling.Thesamewasperformedwithasinglecontrol pitusedforcomparisonwithbothcarcasses.Thespacingbetween thegraveswasjustover1m.Toavoidanyvertebratescavenging, ametalfencewasplacedaroundthegravesite.Aftersixmonths ofburial,thegraveswereexcavatedbyhand(atthebeginningof August2008)byaforensicinvestigationteam(Fig.1).

2.3. SamplingprocedureandcollectionofVOCs

Aftersixmonthsofburial,gravesoilsamplesandcontrolpit samples(300gofsoil)werecollectedatdifferentdepthsduring thepigexcavation.Soilsamplesweretakenat5cm,10cm,20cm, 30cm(sampleabovecarcass)and40cm(samplebelowcarcass) fromboththepiggravesandcontrolpitforblankcollection.Soil sampleswerehermeticallypackedinziplockplasticbags, trans-ferredtothelaboratoryandconservedat−80◦_C_until_collection_of

thevolatilefractionwascarriedoutinlaboratoryconditions(21◦C). Thesizeofsamplingbagswasselectedtominimizetheheadspace. AdynamicsamplingwasusedtocollectVOCsreleasedbygravesoil samplesplacedinsideaclosedvolatilecollectiondevice(Fig.2). Thedevicewasmadeofapump(MSA,EscortElfpump)pullingair throughthesamplingdeviceataconstantrateof0.5L/minfor1 or2hafter15minofequilibrationtime.Gravesoilsamples(50g)

(3)

C.Brasseuretal./J.Chromatogr.Axxx (2012) xxx–xxx 3

Fig.1.Theforensicexcavationteamremovinglayersofsoil(top),andoneofthe pigcarcassesaftersixmonthsofburialat40cmdepth(bottom).

wereplacedinaclosedglasscontainer(250mL).Itwasconnected tothetrappingsystembymeanofglassandTeﬂon®_tubing._VOCs

weretrappedonacartridgecontaininga60␮g SuperQ®

adsor-bentﬁlter(80–100mesh,AlltechAssociates,Inc.,Deerﬁeld,IL,USA). Aftersampling,VOCswereelutedfromtheSuperQ®_adsorbent_with

200␮Lofdiethylether(HPLCgrade,Sigma–AldrichSA,Bornem, Belgium)andcappedinGC-typevialspriorGC–MSinjection.The airpulledinsidethesamplingdevicewasfilteredthroughacarbon filter.Anextra60␮gSuperQ®_adsorbent_filter_was_placed_between

thecarbonﬁlterandthesampletofurtheravoidsample contami-nationandbeusedasblank.

2.4. GC×GC–TOFMSanalyses

SampleswereanalyzedwiththeLECOGC×GC–TOFMS Pega-sus4D system(LECO corp., St Joseph, MI, USA) equipped with a 100%dimethylpolysiloxane 30m_×0.25mm_×0.25_␮mVF-1ms

Fig.2.SchematicdiagramoftheVOCdynamicsamplingfromsoilsamples.Airwas pumpedataconstantrateof0.5L/minfor1or2hafter15minofequilibrationtime. Thedevicewashomemadeandkepttoitssimplestdesign(asamplevial,apump, glassandTeﬂon®_tubing,_and_removable_cartridges).

phaseas1_D_column_and_a_50%_phenyl,_50%_{dimethylpolysiloxane}

1m× 0.20mm×0.10␮mVF-17msas2_D_column_(Agilent,

Wald-bronn,Germany).Heliumwasusedasthecarriergasataconstant ﬂowrateof1mL/min.Twomicrolitersofthesampleindiethylether wereinjectedintoasplit/splitlessinjectorheldat270◦Cinsplitless mode.TheprimaryGCovenramptemperaturewasprogrammed from40◦Cto220◦Cwitharateof1.5◦C/min.ThesecondaryGC ovenwassetwithaniso-rampingtemperaturemodeandan off-settemperatureof20◦C.Themodulator systemwasa quad-jet withtwo permanent cold nitrogenjetsand two pulsed hot-air jets.Liquidnitrogenwasusedtocreatethecoldjets.The tempera-tureoffsetofthemodulatorwas30◦CwithaPMof10sandahot

pulseduration of600ms.TheMStransferlinetemperaturewas 250◦C.Theionsourcetemperaturewas200◦CwithEIenergyof 70eV.Thecollectedmassrangewas20–600amuwithan acquisi-tionrateof100spectra/sanddetectorvoltageof1800V.Thedata wereprocessedwiththeLECOChromaTOF4.22automatedpeak findandspectral deconvolutionsoftwarewithadefined signal-to-noisethreshold.Wiley(2008)andNIST (2008)massspectral librarieswereusedfortemptativeidentification.Analkane stan-dard (C8–C20) solutionwasrun for thecalculationof retention

indices,basedonthefollowingformula.

1_DRI _(target)₌₁₀₀

₁

tR(s)−1tR(n) 1_t_R₍_n₊₁₎₋1_t_R₍_n)+n

For dataanalyses,thecomparison algorithmavailablein the ChromaTOF4.22softwarewasused.Referencepeaktableswere createdbyprocessingofthegravesoilsamplesofalldepths,based on minimum signal to noise ratio (S/N) and specified library matches as minimum requirements for a peak to be included in peak tables. Further classificationprocesses wereapplied to remove the chromatographic noise (column bleed) and poten-tial peak tailing issues. Each control samples were processed againstthereferencetablestoidentifynondecompositionspecific compoundsandremovethemfromthelistofcandidatespecific analytes.Furthercomparisonswererealizedbetweenthelistsof candidate specificanalytestopotentially identifydifferencesin theVOCcontentofsoilstakenatdifferentdepths.Comparisons werebased ondeconvolutedpeak retentiontime (1_t

R and 2tR),

apexmassspectraandrelativearea.Aretentionreference com-poundwasusedtoaccountforpotentialretentiontimeshiftsdue to GCcolumn related factors.Retention time variations of20s (twicePM)and0.2sweresetfor1Dand2D,respectively.Tohelpin

thenon-targetanalyticalapproach,theclassiﬁcationofthe chro-matographicpeakswasalsobasedonrecognizablefeatureinthe massfragmentationpattern[35,36].Thisapproachispossibleby creatingandapplyingscriptstoevaluatethemassspectraldata forthesecharacteristicmassspectralfeatures.Scriptswere writ-tenin MicrosoftTM _VBScript_language,_set_in _different_functions

and incorporatedinclassificationproceduresoftheChromaTOF 4.22software.Chromatographicpeaksmatchingthemass spec-tralpatternofcompoundsofinterestwerelabeledandtherefore highlightedamongthelargequantityofpeaksfoundineach sam-ple.Theapplicationofscriptsreducedthenumberofcompounds formanualreviewandallowedtoconfirmthespecificityofsome compoundsfoundinourgravesoilsamples.

2.5. Statisticalanalyses

Multivariateanalysiswasperformed ondatausingprincipal component analysis(PCA)(TheUnscrambler®_,_v_10.0). _PCA_is _a

projectionmethodusedtogetabetteroverviewandavisual rep-resentationofrelationshipsbetweensamplesandvariables.The informationfrominitialvariablesiscombinedandreduced gen-erally to a bi- or tri-dimensional system represented with the

(4)

Fig.3.Comparisonbetweenminimumandmaximumaverageairtemperatures inBelgiumduringthestudy periodandtemperaturesrecordedatthe exper-imentalsite forair,at subsurface,and atcarcasslevel. (graylines) KMI-IRM archivesminimum–maximumairTduringthestudy,(blacklines)averagerecorded minimum–maximumairTforthestudyarea,(blackdots)subsurfacerecordedT duringthestudy,(blackdashes)pigcorerecordedTduringthestudy.

principalcomponents(PCs)andtakinginaccountthemost impor-tantinformation.In thisstudy,thematrixwassetwiththepig gravesoilsamplesasobjectsandtheVOCs(usingchromatogram areavalues) asvariables inorder toshowa potentialstructure withinthedataand throughdepth.TheStudent’s t-distribution statisticalhypothesistestwasusedtoshowsigniﬁcantdifferences betweenpiggraveandcontrolsamplesobtainedwiththescript ﬁlteringmethod.

3. Resultsanddiscussion

3.1. Environmentalparameters

Piggravesandthecontrolpitwerelocatedclosetoeachother (1m)toreducethesoilcompositionvariability.Shorterdistance wouldpotentiallyhavebeenproblematicintermsofcross leach-ingofdecompositionproducts.Theforestbiotopeofthestudyis representativeoftypicalBelgianforest.The6 monthsperiod of burialrangedfromFebruarytoAugust2008.Localatmospheric temperatureandprecipitationvalueswereclosetoseasonal aver-ages.Meanprecipitationsover the6monthsstudyperiodwere 62.9mm/month.MeanprecipitationsfromtheBelgianKMI-IRM archivesforthesame6monthsover therange 1981–2010was 66.0mm/month (4.7%deviation). The maximum and minimum air temperature values were recorded daily and the calculated monthlyaverages(rangeof3.0–22.2◦C)wereclosetoaverage tem-peraturesreportedforthelast20years(rangeof0.7–23.0◦C)with 2–3◦C higherforFebruaryandMay(Fig.3).Lowertemperature variationswereobservedforsubsurface andcarcass.Subsurface temperaturerangewas3.9–15.5◦C,andthecarcasstemperature rangewas5.3–16.3◦C (Fig.3).Theseenvironmentalparameters canthereforebeconsideredastypicalforthegeographicalarea. Additionaldetailedinformationonsoilparametersisavailable else-where[34].

3.2. Datatreatmentstrategytoselectcandidatespecificanalytes Fortheestablishmentofalistofcandidatespecificanalytes,the completesamplessetofoneofthepiggraves(pigA)was inves-tigatedand compared tothecontrol pitsamples.Thegravesoil sampleswereprocessedwithaS/Nthresholdvalueof200inorder tolimitthenumberofhitstothemostsignificantcompounds.For furtherrefinementofthelist,alibrarymatchvalueof 700was appliedandchromatographicnoiseandpeaktailingwereremoved

fromthehittable.Thisapproachconductedtotheselectionofa limitednumberofanalytespersample(Table1).Areferencetable wasbuiltforeachseparategravesoilsample atdifferentdepths andeverysinglecontrolpitsoilsamples,alsotakenatdifferent depths,wereprocessedversusthosereferencetables.According tothelibrarysearch,hydroxybutylatedtoluenewasfoundinevery sampleatsigniﬁcantlevelandwasusedasretentionreference com-poundforretentiontimeadjustmentinboth1_D_and2_D._In_practice,

compoundsthatwerenotfoundinanyofthecontrolpitsoil sam-plesorwithanareavaluelowerthan10%,relativelytothesame analytefoundinthecarcasssamplewerekeptascandidatespeciﬁc analytes.Compoundsthatwerefoundinatleastoneofthedifferent depthscontrolpitsoilsampleswererejectedfromthelistof candi-dates.Alistofspeciﬁccompoundswasestablishedforeachdepth ofthepiggrave(Table1).Thehighestnumbersofcompoundswere foundinsamplestakenatthesurface(depth5cm,n=31)andbelow thecarcass(n=30).Amongstallsamples,severalcompoundswere observedinmultiplelayersofsoils.

3.3. Differentiationbetweentwogroupsofcompounds

In order to highlight possible differences betweenthe vari-ousdepthsofgravesoils,basedonthecandidatespecificanalytes reportedinTable1,allgravesoilsampleswerereprocessedwitha thresholdS/Nvalueof5tocheckifsomecompoundsofaspecific depthwerepresentintheotherdepthsamplesatalowerS/Nthan 200,potentiallyimprovingthespecificity.Allreprocessedreference tablewerecrosscompared todeterminewhat werethe depth-specificandrecurrentanalytes.Itappearedthatalargemajority ofanalyteswerepresentateverydepthofthepiggrave,exceptfor somecompoundsthatshowedtobespecifictothesampletaken belowthepigcarcass.

Practically,20speciﬁccompoundswerespeciﬁcallyfoundinthe soilsampletakenbelowthecarcass(Table2).Theidentityofthe compoundsresultedfromthelibrarysearchingbasedonmass spec-traldataandfromfurthercomparisonof1_D_retention_indices₍1_DRI)

valuescalculated againstreferencealkanesand valuesavailable intheliterature.Additionalstructuralanalysisinvestigationofthe fragmentationpatterncompletedtheidentificationprocesswhen morethanonecandidatewaspossible.Thisgroupof20compounds consistedofketones(n=7),nitrilesandsulfurs(n=4),heterocyclic compounds(n=3),andbenzenederivativeslikealdehydes, alco-hols,ketones,ethersandnitriles(n=6).Someofthesecompounds werealreadyreportedintheliterature(seeTable2).Someothers werenotbutcouldbelinkedtotypicaldegradationproductsissued fromsofttissues[11].Surprisingly,severalfamiliesofcompounds, such as acids, esters, aldehydes and alcohols, usually detected when monitoringdecomposition processes werenot identified. Asothertypicalaminoacidsorbonefatmicroorganisms metab-olizationproductsweredetected(indole,phenol,benzaldehyde, benzonitrile,acetophenone,dimethyldisulfide,dimethyltrisulfide, and2-nonanone[5]),andasthebodydidclearlyundergoall decom-positionstagesuptoskeletonization(Fig.1),theabsenceofthese specificVOCswasmostprobablytoberelatedtoeitherthe analyt-icalsamplingprocedureortheremovalofthesoilsignatureduring dataprocessing.Theliquidextractionoftheanalytesfromthe sam-plingcartridgesaswellasthestorageoftheliquidfractionprior analysesisnotadequateforthemostvolatilepolarcompounds, comparedtotheuseofdirectthermaldesorptiontotheGCcolumn. Ontheotherhand,thedynamicVOCcollectionand concen-trationbeforeliquidextractioncouldalsoberesponsibleofthe detection of a second group of candidate specific compounds. Thatgroupconsistedof34compoundsfoundatalldepthsofthe gravesoilsamples.Thisgroupincludedalkaneisomersrangingfrom C12toC16,madeofmethyl-branchedalkanesincludingmethyl-,

(5)

Table1

Summaryofthedataprocessingresultsforthedeterminationofcandidatespeciﬁccompoundsforthevariouslayersofgravesoil(pigA).

Gravesoilsamples Automatedpeakﬁnding

WithS/Nthresholdat 200(a)

With(a)andlibrarymatch>700 bleedandtailingremoved(b)

With(a)and(b)andcleanedfrom analytespresentincontrols

Depth5cm 545 170 31

Depth10cm 519 129 17

Depth20cm 401 103 9

Abovecarcass 442 108 10

Belowcarcass 611 153 30

exactidentificationofthesecompoundswasdifficultbecauseofthe complexityoftheMSfragmentationpatterns.Thisnotonly pre-ventedtheclearlocalizationofthemethylgroupsonthealkane chains,butalsosignificantlyreducedtheefficiencyofthelibrary searching.Eventheuseof1_DRI_calculation_was_not_conducive_to

uniqueidentiﬁcations.Verygoodlibrarymatcheswerequestioned by1_DRI _values_that _highlighted_{discrepancies} _between_the

pro-posednumberofcarbonatomsandthepredictedpositioninthe chromatographicspace.

Thesemethyl-branchedalkaneswereneverthelessspecificto thegravesoilsamples.Onlyafewreportsmentionthedetectionof thesecompounds,especially whenusingasimilardynamic col-lection technique based on anextended time of accumulation, potentiallyleadingtoenrichmentinthesevolatiles[11,12,14]. Fur-thermore,theuseofthesensitiveGC×GC–TOFMSinstrumentand boththeassociatedenhancedpeakcapacityandspectral decon-volutionalsocontributedtohighlightthepresenceofthespecific alkanefraction.Fig.4showstheapexplotforthe54compoundsof interestandthealkaneretentionstandardmixture.The segrega-tionbetweenthetwogroupsofanalytesoverthechromatographic retentionareaissignificantwithacharacteristicelutionbandat short 2_t

R for methyl-branchedalkanes, and a larger dispersion

inthechromatographicspaceforthemorefunctionalized com-pounds.

Theemissionof methyl-branchedalkanes hasbeenreported instudiesinrelationwithspeciﬁcoxidativestresses[40,41].The hypothesiswasthatvolatilemethyl-branchedalkaneswere pro-ducedbythelipidperoxidationofpolyunsaturatedfattyacidsof cellmembranes,itselfinitiatedbyreactiveoxygenspecies(ROS) issuedfromoxidativestresses.Suchamechanismcouldlogically occurindecayingcarcassesasitiswellknownthattheproduction

Fig.4. Apexplotshowingthespatialdistributionofthetwogroupsofspeciﬁc analytes.ThenumberingsystemreferstoTable2.

ofROSisincreasedafterdeathwhenantioxidantdefense mecha-nismsend[5].Anothersourceofmethyl-branchedalkanestoalso considerwouldberelatedtoants,astheseanalytesareknownto beoneofthemainconstituentoftheantcuticularhydrocarbon mixtureusedinsignaldiscrimination[42],andhavebeenalready investigatedtoForensicapplications[43].Despite thefactsthat verylowconcentrationsaretobeexpectedandthatlongercarbon chaincompoundsarereported,thecontributionofthesecuticle hydrocarbonstotheVOCspatternispossibleandalreadyearlier suggested[5].

Table2

Listofthespeciﬁccompoundsidentiﬁedinthegravesoilsamplebelowthecarcass(1_t

Rand2tRforﬁrstandseconddimensionretentiontimes,respectively).

Compounds Retentiontimes(s) Molecularformula Literaturereport

Number Identity 1_t R 2tR 1 Pentane-2,3-dione 370 1.48 C5H8O2 2 Tetrahydro-2,5-dimethylfuran 390 1.30 C6H12O 3 2,2,4-Trimethyloxetane 420 1.46 C6H12O 4 2-Methylbutanenitrile 420 1.95 C5H9N 5 3-Methylbutanenitrile 430 2.06 C5H9N 6 Dimethyldisulﬁde 450 1.99 C2H6S2 [9–14,32,38,39] 7 Hexane-2,3-dione 560 2.09 C6H10O2 8 Cyclopentanone 570 2.97 C5H8O 9 Methoxybenzene 1040 3.97 C7H8O 10 Benzaldehyde 1270 5.31 C7H6O [9,13,14,38] 11 Dimethyltrisulﬁde 1290 4.91 C2H6S3 [9–12,14,32,39] 12 Phenol 1370 4.75 C6H6O [10,12,14] 13 Benzonitrile 1400 5.92 C7H5N [9,14] 14 Cyclohept-4-en-1-one 1510 5.58 C7H10O 15 1-Methoxy-4-methylbenzene 1580 4.45 C8H10O 16 Acetophenone 1830 6.05 C8H8O [12,14] 17 Nonan-2-one 1960 3.59 C9H18O [10,12,14,37] 18 Undecan-2-one 3060 3.66 C11H22O 19 Indole 3110 7.94 C8H7N [13,14,18,32,38] 20 Tridecan-2-one 4070 3.64 C13H26O

(6)

Fig.5.Variationinlevelsofspeciﬁcmethyl-branchedalkanes(C12–C16)withthedistancefromthedecayingcarcass.Numbersrefertothe34analytes.

3.4. Apparentdepthtrendinsoil

In addition to the separation between the group of func-tionalizedmolecules(n=20)present belowthecarcass andthe methyl-branchedalkanegroup(n=34),atrendintherelative pres-enceofthesealkanesoverthevariouslayersofsoilswasobserved. Amainreasonforthepresenceofthespecificcompoundsbelow thecarcassistheleachingandpercolationofdegradationfluids bygravity.Thesecompoundscanlaterbarelymigrateupwardsto thesurfaceofthesoilastheyareshieldedbythecarcassitself. Additionally,consideringthattheadsorptionofVOCsonsoil parti-clesincreaseswiththepolarityofthecompounds,mainlybased onion–dipole interactions and ␲electrons donations,different behaviorsweretobeexpectedbetweenfunctionalizedandalkane molecules[44].Aliphaticlowpolarcompoundscouldthusundergo weakersoiladsorptionwithinteractionsofLondon–vanderWaals forcetype(induceddipole-induceddipole),possiblyresultingin highermigrationrate.Suchaneffectwouldevenbemore signif-icantfor methyl-branchedalkanesas theyarecharacterizedby lowervanderWaalsforces.Thisiscurrentlyonlyhypothesisthat stillhavetobeverified.Furthermore,ashumidityinfluencesthe adsorptionofcompoundsontosoilduetocompetitionwithsoil interactionwithwatermolecules[44],thewatercontentofthe soilwouldadditionallyweakenpossibleinteractionbetweensoil andaliphaticcompounds.

Concerning thedistribution of themethyl-branched alkanes betweenthedifferentlayersofsoils,afternormalizationofthe rela-tiveareaofeachofthe34compoundsfoundatalldepths,tothearea valuesofthecompoundsinthesurfacesample(depth5cm),some concentrationtrendsappeared.AsillustratedinFig.5,anincrease intheamountofthespecificalkanesishighlightedwhencoming fromthecarcasstothesurface.Itwasdifficulttorelatethis obser-vationwithotherreports.Anexternalcontaminationofthesoil byoiloranyotherpossiblesourceofbranchedalkanesisunlikely ascontrolsamplesweresubtractedtogravesoilsamples. Further-more,suchdepthtrendwasnotobservedforothernon-specific alkanesincontrolsoils.

PCAanalysesidentiﬁed a similartrend,highlightingthe dis-tributionofthegravesoilsamplestakenatdifferentdepthsin a scoreplot,obtainedafterloadingtheareavalues ofthesimilar methyl-branchedalkanesasvariablesinthematrix.Thesoil sam-plesclosertothecarcass(belowcarcass,abovecarcassandata depthof 20cm) expressedsimilar spatialdistribution although samplestakenclosertothesurface(depthsof5cmand10cm)were clusteredtogether(Fig.6).Theﬁrstcomponent(PC-1,93%)covered mostoftheinformationandshowedtheseparationofthe5cmand 10cmdepthsfromtheothers.

3.5. Utilizationofscriptstoscreenforcadavericdecomposition signatureinsoilsamples

Following the idea of using GC×GC–TOFMS in forensic geotaphonomyforscreeningofsoilsamplesforcadaveric decom-positionsignature,aprocedureforﬁlteringofpeaktablesgenerated fromsuspectedsoilusingscriptingwasinvestigated.Basedon rec-ognizable features in themass spectraof the methyl-branched alkanesfoundathighlevelintheupperlayersofgravesoils,we wroteascriptthatincludesthemajorspectralspeciﬁcities(Fig.S1). Straight-chainalkanesspectraarecharacterizedbyCnH2n+1+and

lessabundantCnH2n−1+ chargedfragmentsseries,withmaxima

around C3 and C4. Branched-chain alkanes show a decrease in

molecularionsandCnH2n+1+alongwithCnH2n+ionseriesincrease

throughcleavageandchargeretentionatthebranchedcarbon[45]. Thesamplessetofthetwopiggraves(pigAandB)andcontrolpit weretestedwithscripting.Thesecondgravesoilexperiment(pig B)wasusedtoconfirmtheobservationobtainedfromthepigA samples.Tosimplifythesamplingprocess,thegravesoilsamples ofpigBweretestedwith1hofextractiontime(incomparisonwith 2hforpigA).Thecontrolpitsoilsampleswerealsotestedwith1h ofextractiontimetoincludetheinfluenceoftheextractiontime ontheresults.Thescriptwasappliedonrawpeaktables gener-atedbyprocessingsamplesusingdifferentpeakfindingprocesses withS/Nthresholdsof200,50and10.ThevariouslevelsofS/N

Fig.6. PCAofthedifferentlayersofgravesoilsbasedonrelativeareaofspeciﬁc methyl-branchedalkanes.

(7)

Fig.7.Comparisonofthepercentageofpeaksselectedbyapplyingthe alkane-speciﬁcscripttogravesoilandcontrolsamplesprocessedatvariousS/Nvalues. SingleanddoublestarsrepresentStudent’st-distributiontestwithvalue<0.05 andvalue<0.01,respectively.

Fig.8.Apexplotforbothgravesoil(pigA)andcontrolresultsafterscriptingofsoils at10cmdepth(S/Nof200).Diamondsareforcontrolsoil(n=16hits),andtriangles areforgravesoil(n=47hits).Gravesoil2_t

Rhavebeenshiftedof2sforclarity.

thresholdswereinvestigatedtofindacompromisebetweenpeak tablesizesandspecificity.Themajorissuewasthepotential gen-erationofverylargepeaktablesrequiringlongandtediousdata processingprocedures.Theaveragenumbersofhitsinthepeak tableswere500,2000and13,000forS/Nthresholdsof200,50and 10,respectively.Aftercompoundswiththerecognizablealkaneor methyl-branchedalkanefragmentationpatternwereisolatedby scriptingthroughthepeakstable,thenumbersofhitswerereduced tovaluesbetween0.5%and10%oftheinitialnumbers.Fig.7shows themeanpercentagevaluesforthenumberofhitsobtainedafter thescript wasappliedtopeaktablesgeneratedatdifferentS/N thresholdlevelsforthethreeupperlayersofthesoils(depths5,10 and20cm)andforthetwodifferentgravesoilexperiments(pig Aand B) andtheirrespectivecontrol samples.TheStudent’s t-distributiontestshowedsignificantdifferences(value<0.05or 0.01)betweengravesoilsamplesandcontrolsamplesforeveryS/N thresholdlevelstested.AS/Nof200appearedtobesufficientto highlightthedifferencebetweengravesoilandcontrolsamples.

Fig.8illustratestheapexplotforbothgravesoilpigAandcontrol resultsafterscriptingofsoilsat10cmdepthprocessedatS/Nof 200.Thecadavericdecompositionsignatureisclearlyhighlighted byapplyingthescript.

4. Conclusions

TheanalysisofgravesoilsamplesbyGC×GC–TOFMSconducted totheidentificationof two main groups ofspecific analytes.A first groupof 20 functionalized compounds specificallylocated belowthecarcass,andasecondgroupofmethyl-branchedalkanes

presentinallgravesoilsamples.Atrendintherelativepresence of these alkanesover thevarious layers of soilswasobserved. The useof a speciﬁc script permittedto propose a data treat-mentstrategytopotentiallyscreenforcadavericdecomposition signatureinsoilsamples.Thisstrategycouldcomplement other approaches [46] and contribute to aid forensic geotaphonomy investigators.

Acknowledgements

J. Dekeirsschieter was the recepient of a PhD grant from F.R.I.A.(FondspourlaFormationàlaRecherchedansl’Industrie etl’Agriculture).DonaldC.HiltonandYvesBrostauxare acknowl-edgedfortheirhelpinwritinginMicrosoftTM_VBScript_language

andstatisticaldatatreatment,respectively.Theauthorswouldalso liketothanktheDisasterVictimIdentiﬁcationteam,theAirSupport UnitandtheNationalTechnicalandTacticalSupportUnitofthe BelgianFederalPolice.Thankstoallindividualswhohelpedwith ﬁeldwork,especiallythestudentsfromtheUniversityofApplied Sciences(HvA)inAmsterdam.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,atdoi:10.1016/j.chroma.2012.03.048.

References

[1] M.Tibbett,D.O.Carter,in:K.Ritz,L.Dawson,D.Miller(Eds.),Criminaland EnvironmentalSoilForensics,Springer,NewYork,2009,p.317.

[2]J.A.Payne,E.W.King,G.Beinhart,Nature219(1968)1180.

[3]L.A.Benninger,D.O.Carter,S.L.Forbes,ForensicSci.Int.180(2008)70. [4] K.L.Stokes,S.L.Forbes,L.A.Benninger,D.O.Carter,M.Tibbett,in:K.Ritz,L.

Dawson,D.Miller(Eds.),CriminalandEnvironmentalSoilForensics,Springer, NewYork,2009,p.357.

[5]S.Paczkowski,S.Schütz,Appl.Microbiol.Biotechnol.91(2011)917. [6] G.S.Anderson,S.L.VanLaerhoven,J.ForensicSci.41(1996)617. [7] D.T.Hudson,A.M.Curran,K.G.Furton,J.ForensicSci.54(2009)1270. [8]A.M.Curran,P.A.Prada,K.G.Furton,J.ForensicSci.55(2010)50.

[9]A.A.Vass,R.R.Smith,C.V.Thompson,M.N.Burnett,D.A.Wolf,J.A.Synstelien, N.Dulgerian,B.A.Eckenrode,J.ForensicSci.49(2004)760.

[10] A.A.Vass,R.R.Smith,C.V.Thompson,M.N.Burnett,N.Dulgerian,B.A.Eckenrode, J.ForensicSci.53(2008)384.

[11]M.Statheropoulos,C.Spiliopoulou,A.Agapiou,ForensicSci.Int.153(2005)147. [12] M.Statheropoulos,A.Agapiou,C.Spiliopoulou,G.C.Pallis,E.Sianos,Sci.Total

Environ.385(2007)221.

[13]E.M.Hoffman,A.M.Curran,N.Dulgerian,R.A.Stockham,B.A.Eckenrode, Foren-sicSci.Int.186(2009)6.

[14] J.Dekeirsschieter,F.J.Verheggen,M.Gohy,F.Hubrecht,L.Bourguignon,G. Lognay,E.Haubruge,ForensicSci.Int.189(2009)46.

[15]C.vonHoermann,J.Ruther,S.Reibe,B.Madea,M.Ayasse,ForensicSci.Int.212 (2011)173.

[16]L.M.Swann,S.L.Forbes,S.W.Lewis,Anal.Chim.Acta682(2010)9.

[17]S.J.Notter,B.H.Stuart,B.B.Dent,J.Keegan,Eur.J.LipidSci.Technol.110(2008) 73.

[18]A.A.Vass,S.-A.Barshick,G.Sega,J.Caton,J.T.Skeen,J.C.Love,J.A.Synstelien,J. ForensicSci.47(2002)542.

[19]A.R.Conklin,IntroductiontoSoilChemistry:AnalysisandInstrumentation, Wiley,NewJersey,2005.

[20]J.C.Giddings,Anal.Chem.56(1984)1258A.

[21]E.B.Ledford,C.A.Billesbach,J.HighResolut.Chromatogr.23(2000)202. [22]D.G.PattersonJr.,S.M.Welch,W.E.Turner,A.Sjödin,J.-F.Focant,J.Chromatogr.

A1218(2011)3274.

[23]J.-F.Focant,E.Reiner,K.MacPherson,T.Kolic,A.Sjödin,D.G.PattersonJr.,S. Reese,F.Dorman,J.Cochran,Talanta63(2004)1231.

[24] J.B.Phillips,J.Beens,J.Chromatogr.A856(1999)331. [25]J.-M.D.Dimandja,Anal.Chem.76(2004)167A.

[26]G.Semard,M.Adahchour,J.-F.Focant,in:L.Ramos(Ed.),ComprehensiveTwo DimensionalGasChromatographyin:D.Barcelo(Ed.),WilsonandWilson’s ComprehensiveAnalyticalChemistry,Elsevier,TheNetherlands,2009,p.15. [27]Z.L.Cardeal,P.P.deSouza,M.D.R.GomesdaSilva,P.J.Marriott,Talanta74(2008)

793.

[28] I.Silva,S.M.Rocha,M.A.Coimbra,P.J.Marriott,J.Chromatogr.A1217(2010) 5511.

[29]C.Cordero,E.Liberto,C.Bicchi,P.Rubiolo,P.Schieberle,S.E.Reichenbach,Q. Tao,J.Chromatogr.A1217(2010)5848.

(8)

[30]D.L.France,T.J.Grifﬁn,J.G.Swanburg,J.W.Lindermann,G.C.Davenport,V. Trammell,J.ForensicSci.37(1992)1445.

[31]B.Turner,P.Wiltschire,ForensicSci.Int.101(1999)113. [32]B.B.Dent,S.L.Forbes,B.H.Stuart,Environ.Geol.45(2004)576.

[33]K.G.Schoenly,N.H.Haskell,D.K.Mills,C.Bieme-Ndi,K.Larsen,Y.Lee,Am.Biol. Teach.68(2006)402.

[34]E.M.J.Schotsmans,J.Denton,J.Dekeirsschieter,T.Ivaneanu,S.Leentjes,R.C. Jan-away,A.S.Wilson,ForensicSci.Int.(2011),doi:10.1016/j.forsciint.2011.09.025. [35]D.C.Hilton,Curr.TrendsMassSpectrom.(2007)28.

[36] D.C.Hilton,R.S.Jones,A.Sjödin,J.Chromatogr.A1217(2010)6851. [37]D.Ercolini,F.Russo,A.Nasi,P.Ferranti,F.Villani,Appl.Environ.Microb.75

(2009)1990.

[38]N.Lorenzo,T.Wan,R.Harper,Y.Hsu,M.Chow,S.Rose,K.Furton,Anal.Bioanal. Chem.376(2003)1212.

[39]B.Kalinova,H.Podskalska,J.Ruzicka,M.Hoskovec,Naturwissenschaften96 (2009)889.

[40]M.Phillips,K.Gleeson,J.M.B.Hughes,J.Greenberg,R.N.Cataneo,L.Bager,W.P. McVay,Lancet353(1999)1930.

[41]M. Phillips, R.N. Cataneo, J. Greenberg, R. Gunawardena, A. Naidu, F. Rahbari-Oskoui,J.Lab.Clin.Med.136(2000)243.

[42]S.Martin,F.Drijfhout,J.Chem.Ecol.35(2009)1151.

[43] G.H.Zhu,X.H.Xu,X.J.Yu,Y.Zhang,J.F.Wang,ForensicSci.Int.169(2007)1. [44]J.Ruiz,R.Bilbao,M.B.Murillo,Environ.Sci.Technol.32(1998)1079. [45] F.W.McLafferty,F.Turecek,InterpretationofMassSpectra,4thed.,University

ScienceBooks,California,1993.