COUPLAGE LC-ICP-MS/MS

I. PRINCIPE GENERAL ICP-MS/MS

La spectrométrie de masse à couplage inductif (ICP-MS) est une technique d’analyse inorganique permettant l’atomisation, l’excitation et l’ionisation des espèces présentes, à l’aide d’un plasma induit à haute fréquence. Le système d’introduction de l’échantillon dans l’ICP-MS est constitué d’un nébuliseur et d’une chambre de nébulisation. Le nébuliseur transforme l’échantillon qui est sous forme liquide en sortie de colonne en un aérosol dense, composé de très fines gouttelettes. La chambre de nébulisation quant à elle limite la charge en solvant dans le plasma et pour cela homogénéise l’aérosol formé, en limitant la taille des gouttelettes et en réduisant la vitesse de l’aérosol. La source d’ionisation fonctionnant à pression atmosphérique et à température élevée et le quadripôle fonctionnant sous vide et à température ambiante, une interface constituée de deux cônes successifs est nécessaire. Enfin, un système de lentille focalise les ions à l’entrée du quadripôle. Les ions sont filtrés en fonction de leur rapport masse/charge et sont ensuite détectés.

II. DISPERSION LIEE AUX SYSTEMES D’INTRODUCTION

Les systèmes d’introduction d’échantillon dans l’ICP-MS commercialement disponibles sont nombreux sur le marché. Une étude approfondie a été menée sur les nébuliseurs et chambres de nébulisation constituant l’interface de couplage, en mode hydro-organique. L’objectif de ces travaux est d’évaluer la perte d’efficacité que peuvent générer de tels montages. 55 publications (de 1997 à 2017) ont servi de point de comparaison pour le couplage LC-ICP-MS.

Cette partie fait l’objet d’une publication publiée dans le journal Journal of Chromatography

A (JCA).

“Suitable interface for coupling liquid chromatography to inductively coupled plasma-mass spectrometry for the analysis of organic matrices. 1 Theoretical and experimental considerations on solute dispersion”

M. Bernardin, F. Bessueille-Barbier, A. Le Masle, C.-P. Lienemann, S. Heinisch, J. Chromatogr. A. 1565 (2018) 68–80. doi:10.1016/j.chroma.2018.06.024

ContentslistsavailableatScienceDirect

JournalofChromatography A

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

Suitableinterface forcouplingliquid chromatographytoinductively

coupledplasma-massspectrometryfor theanalysisoforganic

matrices.1 Theoreticaland experimentalconsiderations onsolute

dispersion

MarieBernardina,b,FrédériqueBessueille-Barbiera,AgnèsLeMasleb, Charles-PhilippeLienemannb,SabineHeinischa,∗

aUniversitédeLyon,InstitutdesSciencesAnalytiques,UMR5280,CNRS,ENSLyon,5ruedelaDoua,69100,Villeurbanne,France

bIFPEnergiesnouvelles,Rond-pointdel’échangeurdeSolaize,BP3,69360,Solaize,France

a r t i c l e i n f o

Articlehistory: Received4March2018

Receivedinrevisedform7June2018 Accepted9June2018

Availableonline15June2018 Keywords:

Speciation

Liquidchromatography Inductivelycoupledplasma Extra-columndispersion Sampleintroductionsystem LC-ICP-MS–flowsplitting

a b s t r a c t

Liquidchromatography(LC)hyphenatedtoaspecificdetectionsuchasinductivelycoupled plasma-massspectrometry(ICP-MS)isatechniqueofchoiceforelementalspeciationanalysis.However,various instrumentallimitationsmayconsiderablyreducetheexpectedsensitivityofthetechnique.Amongthose, wewereinterestedbythesolutedispersionintotheinterfacelocatedbetweenLCandICP-MS.The interfaceconsistsofaSampleIntroductionSystem(SIS)andapossibleflow-splitterpriortoSIS.Flow splittingcanberequiredincaseoforganicmatricestoreducetheorganicsolventamountenteringplasma whichmayleadtoplasmainstabilities.

Althoughextra-columndispersionisusuallywelltakenintoaccountwithconventionalUVdetection ithasbeenlittlestudiedinthecontextofLC-ICP-MSandmoreoverneverquantified.Ourobjectiveis toassessthelossincolumnplatesandhenceinbothseparationqualityandsensitivitywhichmaybe generatedbythecouplingofLCandICP-MSinthespecificcaseoforganicmatrices.Inthisfirststudy, thisisdone(1)fromatheoreticalapproach;(2)from55experimentalstudiesreportedinLC-ICP-MSand (3)fromourexperimentalresultshighlightingthecriticalimpactoftheflowsplitteronextra-column dispersiondependingonbothflow-rateandsplitratio.Itturnsoutbyevaluatingthe55reportedstudies bymeansoftheoreticalcalculations,thatthelossinplatesduetoextra-columndispersionwasmostof thetimebeyond50%andevenoftenbeyond90%.Moreover,fromourexperiments,ithasbeenshown thataverylowsplitratio(1:50)couldgenerateanadditionalvariancearound200␮L2whichinducesa lossintheoreticalplateof90%forultra-highperformanceLC(UHPLC)column(5cm×2.1mm,1.7␮m). ©2018ElsevierB.V.Allrightsreserved.

1. Introduction

The need for determining elemental species concentrations ratherthantotalelementconcentrationshasgrownwithinthepast decades.AccordingtoTempletonetal.[1,2]speciationanalysis con-sistsinidentifyingandquantifyingdifferentchemicalspeciesofa particularelementinagivensample.Ithasbecomeanimportant fieldofresearchoverthepastfewyearsinseveralareasincluding biochemistry, environmentalchemistry, ecotoxicology, pharma-ceuticals,petrochemicals,andnutritionscience[3].

∗ Correspondingauthor.

E-mailaddress:sabine.heinisch@univ-lyon1.fr(S.Heinisch).

NowadayshyphenatedtechniquessuchasLiquid Chromatog-raphy (LC) hyphenated with Inductively Coupled Plasma Mass Spectrometry (LC-ICP-MS) are widely used both toobtain ele-mentalinformationandtodiscriminatespeciesinagivenmatrix. LCtechniques,includingionexchangechromatography(IEC)[4], reversed-phase liquid chromatography (RPLC) [5], ion-pairing chromatography (IPC)[6],size exclusion chromatography (SEC) [7,8] andhydrophilic interactionliquid chromatography(HILIC) [9,10], have been used for speciation analysis [11]. Different hyphenatedtechniquescanbecombinedtogethertoachievemore exhaustivecharacterizationofcomplexmatrices.Forexample, LC-ESI-MScanbeassociatedtoLC-ICP-MStoobtainbothstructural informationandelementalinformation[5,12–14].

Mostspeciationanalysesareperformedinaqueousmatricesby usingion-exchangechromatographyanddealwithenvironmental

https://doi.org/10.1016/j.chroma.2018.06.024

samples[15].Inthisspecificcase,themobilephaseisnotcritical forthecouplingofbothtechniquessincetheamountoforganic solventinthemobilephaseislimited(smallpercentageoforganic modifiersometimesadded,usuallywithoutdisturbingplasma sta-bility).However,whenusingRPLC[16–28]orHILIC[9,10,29–34],a largeamountoforganicsolventisintroducedintotheplasma.The problemsinvolvedbythisintroductionwasthoroughlydiscussed byLeclercqetal.[35,36]inarecentreview.

SomekeyissueshavetobeconsideredforcouplingLCtoICP-MS incaseoforganicmatrices[3]:(i)metalcontaminationfromthe chromatographicsystemand/orthestationaryphase and/orthe mobilephase[3],(ii)plasmainstabilitiesduetothesolventload, especiallyincaseoforganicmobilephases[3],(iii)signal fluctu-ationsingradientelutiondependingonplasmaparameters[37] and(iv)solutedispersionintotheinterfacelocatedbetweenLC andICP-MS.Theinterfaceismadeofasampleintroduction sys-tem(SIS)andapossibleflowsplitterpriortoSISwhichmaybe required,incaseoforganicmatrices,toreducetheamountof sol-vententeringplasmaandhencetodecreaseplasmainstabilities[3]. Thesolutedispersionintheinterfaceunitisacriticalissuebecause thatcanresultinadditionalsolutebandbroadeningandhencein significantlossinbothsensitivityandseparationquality.Although extra-columndispersionisusuallywelltakenintoaccountwith conventionalUVdetection,ithasbeenlittlestudiedinthecontext ofLC-ICP-MS.Inthepresentstudy,wemadethereforeanattemptto assesstheextenttowhichtheinterfacecontributestosoluteband broadening.Thiswasdoneby(i)estimatingfrompublished stud-iesthelikelylossinplatesduetosolutedispersionintheinterface and(ii)showingthecriticalimpactoftheflowsplitteron extra-columndispersiondependingonbothflow-rateandsplitratio.To supportthefirstapproach,asynoptictablehasbeenbuilt(Table2) whichsummarizes55studiescarriedoutonorganicmatricesin LC-ICP-MSandgives,foreachstudy,anestimationoftheinterface contributiontosolutebandbroadening.Afurthersecondpartof ourstudywillbededicatedtothecomparisonofalargenumberof commerciallyavailableSISregardingtheextra-columndispersion.

2. Theoreticalconsiderations

Solutedispersioncanbeassessedbythepeakvariance.Thetotal solutedispersion(totalvariance,_total² )comesfromboth disper-sioninsidethecolumn(columnvariance,_col² )andextra-column dispersion(extra-columnvariance,²_ext).

Extra-column dispersion results from the injection process

(_injection² ),thedifferenttubing(_tubing² )andthedetection(_detector² )

[38].

Becausevariancescanbeaddedifthecorrespondingdispersion processareindependentofeachother,thetotalpeakvariancecan bewrittenas

_total² = _col² +_ext² (1)

Similarly,theextra-columnvarianceis thesumofindividual contributionsaccordingto

_ext² =_injection² +²_tubing+_detector² (2)

ForGaussianpeaks,thetotalpeakvarianceinvolumeunitscan begivenbythemeasuredpeakwidthathalfpeakheight(w0.5) accordingto

2

total,v= ^F

2w²_0.5

5.54 ⁽³⁾

WhereFisthemobilephaseflow-rate.

Forasymetricalpeaks,thesecondordercentralmomenthasto beusedtoprovideareliablevariancevalue.Theselatter,involume units,isgivenby _total,² _v=F² _∞ 0 (t−t_R)²I (t) dt _∞ 0 I (t) dt ⁽⁴⁾

WheretR isthemeanresidencetimeandI (t) theintensityasa functionoftime.

Extra-columnvariancecanbeapproximatedbyremovingthe columnandreplacingitbyazerodeadvolumeunionconnector.In thiscase,extra-columnvarianceiscalculatedfromEq.(3).

Thecolumnvariance,expressedinlengthunits(_col,x² )isrelated toboththecolumnlengthandthecolumnplateheight,H_col,by

_col,x² =LH_col (5)

H_colvarieswiththemobilephaselinearvelocity,u,andits vari-ationmaybefittedbythevanDeemterequation[39]orbythe Knoxequation[40]usingreducedparametersh_col(H_col/dp)and␯ (u.dp/Dm,dpbeingtheaverageparticlediameterandDmthe molec-ulardiffusioncoefficientofthesoluteinthemobilephase).Atthe minimumofthecurve,typicalvaluesforh_coland ␯are3and5 respectively.

Consideringthesolutelinearvelocityatthetimethesoluteis elutedfromthecolumn (u/(1+ke),ke beingtheretentionfactor atelution),thecolumnvariancecanbeexpressedinvolumeunits accordingto

_col,² _v=^V

2

O(1+k_e)2H_col

L ⁽⁶⁾

V0isthecolumndeadvolume,relatedtothecolumnlength,the internaldiameter,d_iandthecolumnporosity,␧tby

V₀=␲L␧t(d²_i/4) (7)

Underisocraticconditions,k_edependsontheretentionvolume andisgivenby

ke= ^VR

V0 −1 (8)

Undergradientconditions,kedependsonboththegradient con-ditionsandthesoluteproperties.Howeveraroughestimationof kecanbemadewhenthelinearsolventstrengththeory(LSST)can beappliedandthesolventstrengthparameter,Sisknown[41,42], usingthefollowingrelation

ke= ¹

2.3St0^C_t

G

(9) Witht₀ thecolumndeadtime,C thegradientcomposition range,andt_Gthegradienttime.TypicalvaluesofS(Sbeingthe absolutevalueoftheslopeofthelinearrelationshipbetweenthe logarithmoftheretentionfactorandthestrongersolventvolume fraction)are4forsmallmolecules,20forpeptidesandmuchhigher forlargermolecules[42].

Thus,FromEqs.(6)and (7),thecolumnvariancefor agiven peakinagivenchromatogramcanbeestimatedprovidedthatthe columngeometryisknown,andtheretentionfactorcanbe deter-mined.

SimilarlytoEq.(6),thetotalvarianceinvolumeunitscanbe written

_total,² _v=^V

2

O(1+ke)2H_total

L ⁽¹⁰⁾

WhereH_totalisthetotalplateheightresultingfromthetwo disper-sionprocesses.

Theratio,ˇ2,betweencolumnandtotalvariancescorresponds totheratiobetweenthetwoplateheightsandhencetotheratio betweeneffectiveandcolumnplatenumbersaccordingto

ˇ2= ^ 2 col ²_total = ^Hcol H_total = ^Neffective N_col ⁽¹¹⁾

WhereN_effectiveandN_colareeffectiveandcolumnplatenumbers respectively.

Theterm,ˇ2 representsthefractionofremainingplatesfora givensolute,ingivenchromatographic andinstrumental condi-tions.

The peak heightand theresolution between two peaks are inversely proportionalto thepeak standard deviation().As a result,thefractionsofremainingpeakheightandremaining reso-lutionaregivenbyˇ.

The mobilephase flow-rate is related tothe reduced linear velocityby F=␯^V0 L Dm dp (12) FromEqs.(6)and(12),Table1givesanoverviewofoptimum flow-ratesandcorrespondingcolumnvariances(k_e=3)depending onbothcolumninternaldiameterandparticlesize.

Depending on SIS, the flow entering the plasma source is expectedtobeacriticalparameterregardingtheperformanceof ICP-MS.AccordingtoTable1,theinternaldiameterhastobe cho-seninaccordancewiththerequiredflow-rate.Howeverasshown inFig.1,anydecreaseincolumninternaldiameterleadstoasevere decrease in column variance, and hence to a huge decrease in bothplatenumberandsensitivityifthedispersioninSISis signif-icant.ThepercentageofremainingplateswascalculatedfromEq. (11)asafunctionofextra-columnvariancefordifferentcolumn internaldiametersandtwodifferentparticlesizes(5and1.7␮m). Thecalculations wereperformedfor a ke valueof 3 anda col-umnlengthproviding10,000plates(Fig.1a.15cmwith5␮mand 5cmwith1.7␮m)and30,000plates(Fig.1b15cmwith1.7␮m). Extra-columnvariancewasconsideredintherange0.1to1000␮L2

thatcoverscurrentLC-UVinstrumentsexceptnanoLCinstruments. Withtheaimofmaintainingmorethan80%oftheplates(␤2>0.8) andhencemorethan90%ofthepeakheight(␤>0.9),itappearsthat themaximumallowablevaluesfortheextra-columnvarianceare (1)1000and100␮L2forconventionalcolumns(4.6mmi.d)packed with5␮mand1.7␮mparticlesrespectively;(2)50and5␮L2for narrowborecolumns(2.1mmi.d)packedwith5␮mand1.7␮m particlesrespectively;3and0.3␮L2formicroborecolumns(1mm i.d)packedwith 5␮mand 1.7␮mparticles respectively. These valuesareslightlyhigherfor30,000plateswith1.7␮mparticles (Fig.1b).Withcapillarycolumns(300␮mi.d.),theextra-column varianceshouldbemuchlowerthan0.1␮L2.Thisfigurealso high-lightstherangeofextra-columnvariancecoveredbycommercially availableHPLCandUHPLCinstrumentswithUVdetection.With theobjectiveofreaching10,000plates,itappearsthatHPLC-UV instrumentsarenotsuitableforconventionalcolumnspackedwith 1.7␮mparticlesandthatmostUHPLC-UVinstrumentscannotbe used withnarrow bore columns packedwith 1.7␮m particles. Furthermore,evenveryefficientUHPLC-UVinstrument(i.e. extra-columnvarianceaslowas5␮L2)arenotsuitableformicrobore columns.

Inthiscontext,itisofprimeimportancetoassesstheadditional extra-columnvariance broughtbytheinterfacebetweenLCand ICP-MS.ConsideringthecurveshapesinFig.1,itappearsthat,in anycase,theremainingplatesmightdecreasefrom80%to70%, 50%,and30%byincreasingtheextra-columnvariancebyafactor of2,5and10respectively.

Aspreviouslydiscussed,theinterestforLC-ICP-MSincaseof organicmatriceshassignificantlygrownduringthepastfewyears. In our opinion,instrument performances are often not consid-eredenough,methodsbeingdevelopedwithoutpriorevaluation ofthe interfacecontributiontosolute dispersion. Toassessthe additionaldispersionbroughtbytheinterfaceinLC-ICP-MS, 55 studiesreportedintheareaoforganicmatrixspeciationhavebeen reviewedanddiscussedintermofextra-columndispersioninthe nextsection.Withaviewtoreducingtheflow enteringSIS,an alternativetothereductionofinternaldiametermaybetheuse ofaflow-splitterpriortoSIS.AT-unionwasindeedusedfor quan-tificationwithonlineisotopicdilution[25,43–48].Howeversuch unitmayleadtosignificantadditionalextra-columnband broad-ening.Thecontributionofflowsplittingtoextra-columndispersion isthereforestudiedinthelastsection.

3. LC-ICP-MSinterface

Peakbandbroadeningisrarelytakenintoconsiderationeven when the sample matrix becomes complex. Most of the time, authorsworkwithonlyafewstandardsinsteadofacomplex mix-turetooptimizetheseparationtechniquesandevenlessoftenon a realsample.For instance,Raber etal.[49] focused oneleven arsenicspeciesusinganLCseparationpriortoICP-MSdetection but,ascanbeseen,thepeakswerepoorlyresolved.Alow mea-suredplatenumberwithrespecttothetheoreticalcolumnplate number,canbedueto(i)solutedispersioninthechromatographic system(injectionsystem,tubing,UV-detection),(ii)alossin col-umnplates(dependingonthehistoryofthecolumn),or(iii)an additionaldispersiongeneratedbytheinterfacebetweenLCand ICP-MS.Theinterfaceconsistsina SampleIntroductionSystem (SIS)andapossibleflow-splitterpriortoSISwhichcanberequired, incaseoforganicmatrices,toreducetheamountofsolvent enter-ingplasmaandhencetodecreaseplasmainstabilities.SISisusually dividedintotwodevices,thenebulizer,andthespraychamber.The nebulizerconvertstheliquidfromtheseparationtechniqueintoa heterogeneousaerosolmadeofdifferentdropletsizes.Theaerosol isthenusuallysortedoutinaspraychamber.Largerdropletsare carriedtothewastewhile smalleronesare senttotheplasma sourceforatomization/excitation/ionization[35].Nebulizers,spray chambers,andflow-splittersrepresentcriticaldevicesfor ensur-ingahigh-performancecoupling.Particularattentionmustbepaid both tothe qualityof theaerosol produced through the nebu-lizer/spray chamber [50] and tothe contribution of the whole interfacetopeakbandbroadening[3,44,45,51–53],bothfeatures beingabletosignificantlyaffectthesensitivity.Awell-documented summaryofSISdeviceswasreportedbyLeclerqetal.[36].Major advantagesanddrawbackswerediscussedintermsofsensitivity andeaseof use,but otheranalyticalperformancessuchas effi-ciency,resolution, sensitivityorextra-columnsolutedispersion, werenottakenintoaccounttocomparethedifferentdevices.Inthe presentwork,thisfeaturehasbeensubjectedtoabroadexploratory studyon55publishedstudies(from1995to2017)dealingwiththe speciationoforganicmatrices.Thecorrespondinganalytical con-ditions,includingLCconditions,typeofnebulizers,typeofspray chambersanduseornotofaflow-splitter,arelistedinTables2and S1ofSupplementaryInformation,dependingonwhetherenough datawereavailableornot.Whendatawereavailable(Table2),an evaluationofthetotalsolutedispersion(totalvariance,_total,² _v)was madefromEq.(3)byconsideringthemostretainedpeak (symmet-ricalpeaksonly).Thecolumnvariance,_col,² _v wasevaluatedfrom Eq.(6).Theextra-columnvariance,_ext,² _vwasestimatedfromEq. (1).Toeasilycomparethepercentageofremainingplates(␤2),this valuewassystematicallydeterminedforaretentionfactorof3,by columnvariancerescalingusingEq.(6).Corresponding␤2values

Table1

Optimumflow-rates(Fopt),columnplatenumber(Ncol),columnvariance(␴2

col)dependingoncolumninternaldiameter(di)andparticlediameter(dp).Calculationsperformed withke=3;Dm=10−9m2/s;L=150mm;␯=5;h=3;␧t=0.7,usingEqs.(5)–(12).

dp=5␮m dp=1.7␮m

Ncol=10,000 Ncol=30,000

di(mm) Fopt(␮L/min) ␴2

col(␮L2) Fopt(␮L/min) ␴2

col(␮L2) Conventional 4.6 698 4867 2052 1655 Narrowbore 2.1 145 211 428 72 Microbore 1 33 11 97 4 Capillary 0.3 3 0.088 9 0.03 Nanobore 0.075 0.2 0.00034 0.5 0.00012

Fig.1.Percentageofremainingplates(␤2)asafunctionofextra-columnvariancefordifferentcolumngeometriesproviding(a)10,000and(b)30,000plates.Calculations fromEqs.(6)and(7)withke=3;h=3and␧t=0.7.

aregiveninTable2.Thereasonsforlow␤2valuescanbestrongly relatedtotheinstrumentcharacteristicsincludingthetypeof inter-facebutalsothelevelofmatchingbetweenthecolumngeometry andtheLCinstrument.

Fig.2illustratestherelativedistributionofnebulizersandspray chambersusedforspeciationinLC-ICP-MSincaseoforganic sol-ventmatricesregardingthedimensionofthecolumnused.Inmany studies,itwasdifficulttodrawrelevantconclusionsbecauseofa lackofinformationaboutSISand/orthecolumnparticle diame-terand/orevensometimes,themobilephaseflow-rate.However whenthecalculationswerepossible,thefollowingcommentscan beprovideddependingonwhetherthecolumnwasconventional, narrow-bore,micro-boreorcapillary:

3.1. Conventionalcolumns

AccordingtoFig.2,threedifferentnebulizershavebeenused: pneumaticmicro-concentricnebulizer(50%)operatedwith flow-ratesintherange40–1000␮L/min,pneumaticconcentricnebulizer (10%) operated with flow-rates higher than 1000␮L/min and hydraulichigh-pressurenebulizer(10%).Withthefirstandsecond ones [17,20,22,24,25], the liquid is introduced through a hori-zontalcapillaryandthegasconductedthroughanexternaltube around theliquid capillary. Withthe third one [54], theliquid isforced throughahighlyturbulent hydraulicnozzle.Similarly, threedifferentspraychambersareusedinassociationwiththe