Comprehensive on-line two-dimensional liquid chromatography × supercritical fluid chromatography with trapping column-assisted modulation for depolymerised lignin analysis

Lignin depolymerisation produces a large variety of low molecular weight phenolic compounds that can be upgraded to value-added chemicals. Detailed analysis of these complex depolymerisation mixtures is, however, hampered by the lack of resolving power oftraditional analysis techniques.

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

Mingzhe Sun, Margareta Sandahl, Charlotta Turner∗

Lund University, Department of Chemistry, Centre for Analysis and Synthesis, P.O Box 124, SE-22100 Lund, Sweden

Article history:

Received 14 November 2017

Received in revised form 29 January 2018

Accepted 5 February 2018

Available online 6 February 2018

Keywords:

Lignin

Phenolic compound

Supercritical fluid chromatography

Trapping capacity

Two-dimensional chromatography

Lignindepolymerisationproducesalargevarietyoflowmolecularweightphenoliccompoundsthatcan

beupgradedtovalue-addedchemicals.Detailedanalysisofthesecomplexdepolymerisationmixturesis, however,hamperedbythelackofresolvingpoweroftraditionalanalysistechniques.Inthisstudy,anovel onlinecomprehensivetwo-dimensionalreversed-phaseliquidchromatography(RPLC)×supercritical fluidchromatography(SFC)methodwithtrappingcolumninterfacewasdevelopedfortheseparation

ofphenoliccompoundsindepolymerisedligninsamples.Thetrappingcapacitiesofdifferenttrapping columnswereevaluated.Theinfluenceoflargevolumewater-containinginjectiononSFCperformance wasstudied.Therelationbetweenpeakcapacityandfirstdimensionflowrateandgradientwas inves-tigated.Theoptimized methodwasappliedfortheanalysisofadepolymerisedligninsample.The RPLC×SFCsystemexhibitedhighdegreeoforthogonality.Comparedwithtraditionalloopbased inter-face,trappingcolumninterfacecansignificantlyshortentheanalysistimeandofferhigherdetectability, withthedisadvantageofmoresevereundersamplinginthefirstdimension

©2018TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND

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

1 Introduction

Biomass holds potential to partly replace petroleum as a

rawmaterialfor productionoffuels andvaluable fineand bulk

chemicals [1 Lignin is a naturally occurring polymer made

up from phenyl-propenoid units, decorated by hydroxy- and

methoxy-groups.Theseareconnectedthroughradical

polymerisa-tionmechanisms,resultinginseveraltypesofbondsbetweenthe

constituentbuildingblocks.Terrestrialbiomassconsistsprimarily

ofthemacromoleculeslignin,celluloseandhemicellulose.Lignin

couldbeanimportantsourcefortheproductionofaromatic

chemi-calsifwellharnessed[2 Oneofthekeystepsofconversionoflignin

tovalue-addedchemicalsisthedepolymerisation−byselective

bondcleavage–ofligninintolowmolecularweightphenolic

com-pounds,whichinturncanbeusedforchemicalproduction.Many

techniqueshavebeendevelopedandarecontinuouslyimproved

forthepurposeofdepolymerisation[3 Duetothestructural

com-plexityof lignin and thelarge variance in themonomeric unit

compositionofdifferentkindsoflignin,thedepolymerisation

prod-uctsareusuallyverycomplexmixturesofaliphaticandphenolic

∗ Corresponding author.

E-mail address: charlotta.turner@chem.lu.se (C Turner).

compounds,thecompositionsofwhichvarygreatlydependingon thelignintypeanddepolymerisationmethodapplied[3

Gaschromatography coupledwithmassspectrometryis the mostwidely adoptedmethodfor theanalysisof lignin depoly-merisationproducts[4 However,thistechniquerequiresarather complicatedsamplepreparation,usuallyincludingextractionand derivatisation,whichmayhavepreferencetowardscertainclasses

ofcompoundsifnotcarefullyperformed.Also,therelativelylow selectivityhinderstheusageofthistechnique,ifprofilingallmajor componentsistheaiminsteadofanalysisofonlyafewselected compounds.High-performanceliquidchromatography(HPLC)or ultrahighperformanceliquidchromatography(UHPLC)havealso beenusedintheanalysisoflignin-derivedcompounds[5,6 How-ever, these techniques also lack the separation power to fully resolveallmajorcomponentsforprofilingstudy

Multidimensionalchromatographyhasbeenwidelyappliedin variousfieldsasitofferstremendouslyimprovedresolvingpower comparedwiththeirone-dimensionalchromatography counter-parts Lignin pyrolysis bio-oil analysis by two-dimensional gas chromatography(2DGC)hasbeenreportedinrecentyears[7,8

As a complementary technique to 2DGC, two-dimensional liq-uid chromatography (2DLC) also holds a large potentialin the analysis of complex samples like lignin depolymerization mix-tures.Itcanovercomethedisadvantageof2DGCintheanalysis

https://doi.org/10.1016/j.chroma.2018.02.008

0021-9673/© 2018 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.

IngeneralSFCdisplaysaseparationmechanismresemblingthatof

NPLC[14].However,theselectivitycanbetunedinawiderrange

asbothpolarandnon-polarcolumnscanbeusedandmobilephase

temperatureandpressureplayimportantroles[15,16].The

poten-tialofSFCasthefirstdimensioninanonlineSFC×RPLCsystem

hasbeendemonstratedbyFrancoisandco-workers[17–19].To

betterharnessthehighefficiencyandfastseparationofSFC,efforts

havealsobeenmadetouseonlinemultipleheart-cuttingLC-SFC,

inparticularforpharmaceuticalachiral-chiralanalysis[20].Online

comprehensiveRPLC×SFCwithtraditionalcollectionloopsasthe

interfacehasalsobeeninvestigatedfortheanalysisoftraditional

Chinesemedicine and processed biomass oil [21,22] However,

compared withRPLC usedas seconddimensionin a 2D online

LC×LCsystem,SFCimposesstricterlimitationontheamountof

eluent transferredtothe seconddimensionin a 2D RPLC×SFC

design.Transfervolumeswithinthenormalrangeofa2DLC×LC

systemcancausesignificantpeakbroadeningandmisshaping,as

wellaspressurespikesatthebeginningofaseconddimensionrun

andpressurebuild-upwithcontinuous2DrunsinaRPLC×SFC

sys-tem[21].Consequently,averylowflowrate(<0.02mL/min)needs

tobeusedinthefirstdimensionLCwithtraditionalloop

collec-tionintheinterface,consideringthefeasiblemodulationlength

Alowfirstdimensionflowrateleadstorelativelylonganalysis

timeandhighdilutionfactor.Onewaytoaddressthisissueisto

introduceananalyteretentionandconcentrationstepbetweenthe

twodimensionalseparations.Theuseoftrappingcolumnsinthe

interfaceof 2DLCsystems hasbeenreportedinrecentyears by

severalauthors[18,23,24],andverywellsummarisedasa

solu-tionforbothcircumventingsolventincompatibilityandfocusing

thetransferredfractionintworeviews[25,26].Thisapproachdoes

notonlyincreasethesignaltonoiseratioasanalytesinthefirst

dimensioneluentsareconcentratedintonarrowbands,butalso

allowshigherflowratetobeusedinthefirstdimension

How-ever,thoroughevaluation and comparisonof differenttrapping

columnsbeforemethoddevelopmentandtheinvestigationof

trap-pingcolumnassistedmodulationonmethodrepeatabilitywerenot

performedinanyofthepublishedstudies.Also,two-dimensional

liquidchromatographyhastoourknowledgeneverbeenapplied

fortheanalysisofdepolymerisedligninsamples

Inthepresentstudy,acomprehensive2DRPLC×SFCmethod

wasdeveloped fortheseparation ofsmallphenolic compounds

indepolymerisedligninsamples.Differenttrappingcolumnswere

evaluatedandcomparedwithcollectionloopsastheinterface.The

effectof thelargevolumewater-richfirst dimensioneluenton

theseconddimensionseparationwasstudied,andtheadvantages

anddisadvantagesofreplacingthecollectionloopswithtrapping

kindlyprovidedbyAssociateProfessorJosephS.M.Samecfrom StockholmUniversity

2.2 Apparatus Theexperimentswereconductedonahome-built comprehen-sive2DRPLC×SFCsystemconsistingofaSFCcontrollingmodule (G4301A,AgilentTechnologies), twobinarypumps (G4302Afor SFC, G4220A for LC, Agilent Technologies), two diode array detectors (G1315C for SFC, G4212B for LC, Agilent Technolo-gies),anautosampler(G4303A,AgilentTechnologies),adegasser (G4225A, Agilent Technologies), a thermostated column com-partment (G1316C, Agilent Technologies) and a flexible cube module (G4227, Agilent Technologies) A 2-position/4-port-duo valve(5067–4214,AgilentTechnologies) iscontrolledbyoneof thetwodrivesinsidetheflexiblecubemodule.Theschemeofthe 2DRPLC×SFCsystemisshowninFig.1.Thesystemwascontrolled

byanAgilentOpenlabCDSChemstationC.01.07software.Allthe onedimensiondatawereevaluatedwithAgilentChemstation soft-ware.Allthe2DdatawereprocessedusingGCImagesoftware(GC Image,Lincoln,NE,USA)

2.3 Trappingcolumnperformanceevaluation 2.3.1 Trappingcapacity

Threetrappingcolumnsweretested:AgilentEclipsePlusC18, 2.1*5mm, 1.8␮m (EC-C18), Agilent Eclipse Plus phenyl-hexyl, 2.1*5mm, 1.8␮m (EC-PH) and Waters VanGuard Torus Diol, 2.1*5mm,1.7␮m(VG-Diol).Notemperaturecontrolwasapplied

onthetrappingcolumns.Theseparationcolumnusedinthefirst dimensionRPLCwasanAgilentZorbaxEclipsePlusC18column (2.1*100mm,1.8␮m).Fortheevaluationofthethreedifferent trap-pingcolumns,threeflowrates0.03,0.05and0.08mL/minwere appliedinthefirstdimension.ThemobilephaseconsistedofH2O (A)andacetonitrile(ACN)(B)inagradientthatwassettostart from10%(vol.)ofBto50%(vol.)ofBindifferentlengthsoftime dependingontheflowrate.Forallexperimentsinthissection,the firstdimensionLCcolumntemperaturewassetto50◦C.A stan-dardmixturecomposedof4-hydroxybenzoicacid,acetosyringone andveratraldehyde(all200mg/L)dissolvedin9:1water:ACN (vol-umetricratio)wasusedandthefirstdimensioninjectionvolume was5␮L

AWatersTorusDiol(3.0*50mm,1.7␮m)columnwasusedin theseconddimensionSFCseparation.Thecolumntemperaturewas setat50◦Candthebackpressureregulator(BPR)at130bar.The flowratewas1.5mL/min,themobilephaseconsistedofCO2 (A) andacetonitrile(B)andthegradientwassettostartfrom3%(vol.)

ofBto30%(vol.)ofBin2.0min

Fig 1.Scheme of 2D RPLC × SFC system A: the eluent from the first dimension is being collected by trapping column A and the content in trapping column B is being flushed into the second dimension column for analysis (back-flush) B: vice versa.

Oneofthetrappingcolumnsinstalledintheinterfacing

switch-ingvalvewasreplacedwithoneshorttubingconnectingtheLCflow

pathwhennocompoundpeakwaseluting.Thetrappingcolumn

wasswitchedtotheLCflowpathtocollecteluentsofinterest

Dif-ferentvolumesofthefirstdimensioneluentacrossthepeakapexes

flowthroughthetrappingcolumnandthentransferredtothe

sec-onddimensionfor analysisby settingdifferentvalve switching

times(Fig.2)

2.3.2 Theinfluenceofsamplematrixontrappingcapacity

Agilent EC-PH 2.1*5mm, 1.8␮m trapping column was used

forevaluationoftheinfluenceofsamplematrixonthetrapping

capacity The chromatographic conditions were kept the same

as described in the previous section, except that 0.05mL/min

was used as the 1st dimension flow rate For the assessment

oftheimpactof samplematrix,thesame depolymerisedlignin

matrix wasspiked with4-hydroxybenzoic acid, acetosyringone

andveratraldehydeofdifferentconcentrations.Standardmixtures

composed of 4-hydroxybenzoic acid, acetosyringone and

vera-traldehydeofdifferentconcentrationsdissolvedin9:1water:ACN

(volumetricratio)werealsousedforcomparison

2.3.3 Repeatabilitytest

Thesamemixtureusedintheprevioussectionwasusedfor

repeatability tests,the injection volume was5␮L The column

usedinthefirstdimensionwasanAgilentZorbaxEclipsePlusC18

column(2.1*100mm,1.8␮m).0.05mL/minwasappliedasflow

rateinthefirstdimension.The columntemperaturewassetto

50◦C.ThemobilephaseconsistedofH2O(A)andacetonitrile(B)

andthegradientwassettostartingfrom10%(vol.)ofBto70%

(vol.)ofBin45min.Theseconddimensionconditionswerethe

sameasdescribedintheprevioussection.OneAgilentEclipsePlus

phenyl-hexyl,2.1*5mm,1.8␮mtrappingcolumnwasinstalledin

theinterfacewithashorttubinginstalledontheotherpositionof

theinterfaceconnectingtheLCflowpathwhennocompoundpeak

waseluting.ThetrappingcolumnwasswitchedtotheLCflowpath

tocollect0.8minofeluentacrosstheanalytepeaks,andthenitwas switchedbacktotheSFCflowpathforanalysisoftheanalytes col-lected.Therepeatabilitytestsweredoneby6successiveinjections

ofthemixtureinonedayandinjectionsofthesamemixturein3 non-consecutivedays

2.4 SFCinjectionvolumestudy

One-dimensionstand-aloneSFCwasusedinthissection Dif-ferentvolumes(1–10␮L)ofaveratraldehyde(200mg/L)solution dissolvedindifferentcompositionsofacetonitrile:water(9:1,1:1, 1:9,allvolumetricratios)wereinjected.Theseparationcolumn usedwasaWatersUPC2TorusDiol(3.0*50mm,1.7␮m).The col-umntemperaturewassetto 50◦C and theBPRat 130bar.The mobilephaseconsistedofCO2(A)andacetonitrile(B)andthe gra-dientwassettostartfrom10%(vol.)ofBto35%(vol.)ofBin1min andtheflowratewassetat3.0mL/min

2.5 FirstdimensionRPLCpeakcapacitystudy The40ligninmodelcompoundsstandardsmixturewasusedfor thepeakcapacitystudyofthefirstdimensionRPLC.Thecolumn usedinthefirstdimensionwasanAgilentZorbaxEclipsePlusC18 column(2.1*100mm,1.8␮m).ThemobilephaseconsistedofH2O (A)andacetonitrile(ACN)(B).Thestudyconsistsoftwoparts:linear gradientsfrom10 to70%Batdifferentgradienttime/voidtime ratios(tg/t0),usingaflowrate0.05mL/min;andlineargradients from10to70%Bwithaconstanttg/t0 of12wereappliedwith variedflowrates

2.6 SeconddimensionSFCcolumnselection The 40 standard mixture was used for column screen-ing Four columns were screened for the second dimension

Fig 2.Chromatogram showing the collection time of different volumes of eluent across the peak apex Analyte: acetosyringone (dissolved in 90:10 water:ACN, 200 mg/L).

SFC separation: Waters UPC2 Torus 2-PIC (2-Picolylamine,

3.0*50mm, 1.7␮m); Waters UPC2 Torus Diol (High

den-sity diol, 3.0*50mm, 1.7␮m); Waters UPC2 Torus 1-AA

(1-Aminoanthracene,3.0*50mm,1.7␮m);WatersUPC2BEH(Silica,

3.0*50mm,1.7␮m).Thecolumntemperaturewassetto50◦Cand

theBPRwassetat130bar.Themodulationtimewas0.8min.The

seconddimensionwasrunat 3.0mL/min.Gradientelutionwas

appliedforallexperiments.Themobilephaseusedinthesecond

dimensionconsistedofCO2(A)andacetonitrile(B).Thegradient

wassettostartfrom3%(vol.)ofBto30%(vol.)ofBin0.5min;30%

(vol.)ofBto3%(vol.)ofBin0.01minandthenholdat3%Bfor

0.29min.ForthefirstdimensionRPLCseparation,anAgilent

Zor-baxEclipsePlusC18column(2.1*100mm,1.8␮m)wasused.The

columntemperaturewassetto50◦C.Themobilephaseconsisted

ofH2O(A)andacetonitrile(B)andthegradientwassettostarting

from10%(vol.)ofBto70%(vol.)ofBin45min.Theflowratewas

setat0.05mL/minandtheinjectionvolumewas5␮L

2.7 RPLC×SFCofligninsampleusingtrappingcolumns

AnAgilentZorbaxEclipsePlusC18(2.1*100mm,1.8␮m)was

usedinthefirstdimension.Thecolumntemperaturewas50◦C

ThemobilephaseconsistedofH2O(A)andacetonitrile(B)andthe

gradientwassettostartingfrom10%(vol.)ofBto90%(vol.)ofB

in60min.Theflowratewassetat0.05mL/minandtheinjection volumewas5␮L

AWatersUPC2TorusDiol(3.0*50mm,1.7␮m)columnwasused

intheseconddimensionSFCseparation.Thecolumntemperature wassetat55◦CandtheBPRat130bar.Theseconddimensionwas runat3.0mL/min,themobilephase consistedofCO2 (A)and a mixtureofmethanol(25vol.%)andacetonitrile(75vol.%)(B)and thegradientwassettostartfrom7%(vol.)ofBto32%(vol.)ofB

in0.49min;32%(vol.)ofBto2%(vol.)ofBin0.01min,thenhold

at2%for0.20minandincreasesfrom2%(vol.)ofBto7%(vol.)of

Bin0.1min.Thesecond-dimensiongradientstartedalreadyinthe previousmodulation,inordertocompensateforthecomparatively largegradientdelayvolumeofthesystem.TwoAgilentEclipsePlus phenyl-hexyl,2.1*5mm,1.8␮mtrappingcolumnswereusedinthe interfaceandthemodulationtimewas0.8min

2.8 RPLC×SFCofligninsampleusingcollectionloops ThesameAgilentZorbaxEclipsePlusC18(2.1*100mm,1.8␮m) wasusedinthefirstdimension.Thecolumntemperaturewas50◦C ThemobilephaseconsistedofH2O(A)andacetonitrile(B)andthe gradientwassettostartingfrom10%(vol.)ofBto90%(vol.)ofB

in60minandmaintain90%(vol.)ofBforanother20min.Theflow ratewassetat0.012mL/min

Section2.7.Twoshortcapillariesof10␮Lwereusedintheinterface

andthemodulationtimewas0.8min

2.9 Chromatographiccalculations

Experimentalpeakcapacitiesforbothdimensionswere

calcu-latedbasedon:

jn=tn−t1

jrepresentsthenumberofthedimension(1or2).tn andt1 are

theretentiontimesofthelastandfirstpeakrespectively,wisthe

average4peakwidth

Conditional2Dpeakcapacitywascalculatedaccordingto[27]:

nco=ˇ×f×1n×2n (2)

ˇistheundersamplingfactor,whichwascalculatedfrom[27]:

ˇ=  1

1+0.21t

m⁄1

tmisthemodulationtime.1isthetimevarianceofthefirst

dimen-sionpeak

Thecoveragefactor fwascalculated usingtheratiobetween

occupied retention space and theoretically available retention

space,whichhasbeenintroducedindetailinreference[27]

3 Results and discussion

SFChasbeenappliedinourpreviousworktothetargeted

analy-sisoflignin-derivedphenoliccompounds[28].Asthecomplexityof

depolymerisedligninsamplesisfarbeyondtheresolvingpowerof

anyone-dimensionaltechnique,acomprehensive2D

chromatog-raphysystemisneededinordertoseparatethemajorcomponents

insuchcomplexsamplesfornon-target profilingpurposes.The

highefficiencyandspeedofSFCmakesitsuitabletobeplacedin

theseconddimension.Inaddition,ahighdegreeoforthogonality

mayalsobegeneratedbya2DRPLC×SFCdesign.Theuseof

trap-pingcolumnsintheinterfacecanpotentiallyretainandtransferthe

analytewithoutintroducingexcessiveamountoffirstdimension

mobilephaseintotheseconddimensionseparation

3.1 Trappingcolumncapacity

In general, compounds are presented as bands of different

lengthsaftertheyarepartiallyorcompletelytrappedinashort

column.Thelengthofthebandlargelydependsonthewidthofthe

compoundpeakinthefirstdimension,theelutionstrengthofthe

firstdimensioneluentandthespecificinteractionsbetweenthe

compoundandthetrappingstationaryphase[29]

Thethreetrappingcolumnsinvestigatedherehavesimilar

inter-nalvolume(∼10␮L),butwillinteractwithlignin-derivedphenols

differently.TheC18trappingcolumnretainsphenoliccompounds

mainlybydispersionforce.Theretentionofthephenyl-hexyl

trap-ping column is a mixtureof hydrophobic interactionand ␲-␲

interactionofthebenzenerings.TheDIOLtrappingcolumnoffers

hydrogenbondinganddipole-dipoleinteractionwiththephenolic

compounds.Anotherimportantfactorthatdeterminesthe

reten-tionofthecompoundsistheelutionstrengthandtransfervolume

ofthe1stdimensioneluentafteracompoundistransferredinto

thetrappingcolumnuntilthevalveswitches.Thethreephenolic

compoundsselectedforassessingthetrappingcapacityrepresents

three categoriesof lignin-derivedphenols thatare prevalentin

depolymerisedligninsamples:phenolicacids(4-hydroxybenzoic

acid),phenolicketones(acetosyringone)andphenolicaldehydes

Fig 3.Trapping capacity of trapping column interface for representative lignin-derived phenolic compounds.

(veratraldehyde),whichwerewellspreadinthefirstdimension chromatogramwithagradientelution

Ingeneral,allthreetrappingcolumnsshowedonlylimited trap-pingcapacityforthethreecompounds,whichcanbeattributedto thesmallvolumesofthecolumns.However,thethreetrapping columnsshowedquitedifferentretainingbehavior ofthethree phenoltypes.Fig.3showstheretentionofthecompoundswhen thefirstdimensionflowratewassetat0.05mL/min.Underthisfirst dimensionconditions,4-hydroxybenzoicacid,acetosyringoneand veratraldehydeelutedatapproximately24%,34%and43%(vol.)of

veratraldehydetrapped in theshort C18 trappingcolumn kept

increasinguptoacollectiontimeof0.8minand1.0min

respec-tively However, 4-hydroxybenzoic acid appeared to be poorly

retainedbytheC18column,significantbreakthroughoccurredat

collectiontimes>0.6min.Thiscanbeattributedtothefactthat

4-hydroxybenzoicacidlacksinalkylmoietiesthatcaninteractwith

C18stronglyandasignificantpartofthe4-hydroxybenzoicacid

molecules(pKa=4.54)wereinanionicformandpreferredstrongly

tostayinthemobilephase

Compared with the other two trapping columns, the DIOL

columnshowedapparentlyalowerretentionofveratraldehyde

Fig.3(c) showedthat apparentbreakthrough of veratraldehyde

startedtooccur(0.8min)shortlyaftermostofitwastrappedin

thecolumn(∼0.6min).Veratraldehydeelutedatthelowestwater

contentofthe1stdimensioneluentcomparedwiththeothertwo

compoundstested.AswaterhasgoodH-bondingandisregarded

asastrongsolventwithaDIOLcolumn,thelowretentionof

ver-atraldehydecan only beexplained bythe absence of hydroxyl

or carboxyl groups in the veratraldehyde molecular structure,

whichmakesitlessfavoredtoberetainedbytheDIOL

station-aryphase, compared with4-hydroxybenzoic acid( COOH) and

acetosyringone( OH)

The phenyl-hexyl column exhibited good trapping capacity

ofallthreecompoundswithveratraldehydebeingslightlymore

retainedthantheothertwo(Fig.3(a)).Thisindicatedthatthe␲-␲

interactionofthebenzeneringscombinedwiththehydrophobic

interactionswiththehexylchaincaneffectively retainphenolic

compoundsregardlessoftheirdifferencesinsubstitutedfunctional

groups.Thetrappingcapacityofthecompoundslargelydependson

theaccessibilityofthebenzenering,thenumberofalkylmoieties

andthenumberofdoublebondsonsubstituents.4-hydroxybenzoic

acidelutedin 1st dimensionC18column atcomparatively low

percentofACNowningtothetwopolarsubstituentsonthe

ben-zenering.ItsretentionintheEC-PHtrapcanbeattributedtothe

higheraccessibilityofthebenzeneringcomparedwiththeother

twocompoundswithmorethantwosubstitutiongroups.For

ace-tosyringoneandveratraldehydethatelutedin1stdimensioneluent

withrelatively higherelutionstrength,theirretentioncouldbe

fromthecombinationaleffectof␲-␲interactionwiththebenzene

ringsandhydrophobicinteractionswiththehexylchain.Itwasalso

foundoutthattheretentiontimeof4-hydroxybenzoicacidinthe

seconddimensionSFCwithEC-PHtrappingcolumnisslightlylower

thanwiththeothertwotrappingcolumns.Thiscanberegardedas

beneficialforperformingonlineRPLC×SFCoflignin-derive

com-pounds,asthelengthofthemodulationtimeislargelydecidedby

theelutionofthemostretainedphenolicacidsincertain

modula-tions.Thus,theEC-PHtrappingcolumnwasselectedforfurther2D

methoddevelopment

ligninmatrixwhenthecollectiontimeexceeded1.0min.Thistrend

isespeciallyapparentfor4-hydroxybenzoicacid,whichis com-parativelytheleastretainedcompoundonthiscolumnamongthe threecompoundstested.Thiscanbecorrelated tothepresence

ofothercompoundsinthematrix competingwiththetargeted analytesonthelimited trapping sites,causing faster and more significantbreakthroughatlongercollectiontimes

Asisshown inFig.S2,thematrixalsohad aslightnegative impactonthetrappingofcompoundspresentatdifferent concen-trations,resultinginarelativelyhigherstandarddeviationforthe 2nddimensionpeakareasofthetrappedcompoundscompared withthestandardmixture.Whenpeakareasareplottedagainst compoundconcentration,thelinearrelationshipisslightlybetter withthestandardmixturepointsforallthreecompoundsincluded

inthestudy

Apossiblewaytoimprovethetrappingcapacitycanbeactive modulation[23],whichinvolvesmixingofthefirstdimension elu-entandaflowofweaksolventbeforethetrappingcolumninterface

todecreasetheeluentstrengthandincreasethetrappingcapacity Oneofthemajorconcernswhenusingtrappingcolumnassisted modulationisrepeatability.AscanbeseenfromTable1,the reten-tiontimesof4-hydroxybenzoicacidandacetosyringoneexhibited comparable RSD values (within-day and between-day) in both dimensions.TheretentiontimeRSDvaluesofveratraldehydein theseconddimensionwerehigherthanthoseinthefirst dimen-sion.Thisislikelyexplainedbytheearlyelutionofveratraldehyde

intheseconddimension(k<1).However,theRSDvaluesarestillin theacceptablerange.TheseconddimensionpeakareaRSDvalues

of4-hydroxybenzoicacidwerehigherthanthoseoftheothertwo compounds.Thiscanbetheconsequenceofthelimitedtrapping capacityof4-hydroxybenzoicacidontheEC-PHtrappingcolumn

Aslightshiftofthefirstdimensionpeakretentiontimecancause eitherincompletetrappingorbreakthroughoftheanalyte,which leadstoseconddimensionpeakareafluctuation

3.2 Theinfluenceoftransfereluentvolumeandwatercontenton theseconddimensionSFCseparation

Itisapparentthatlarger-volumetrappingcolumnscanprovide highertrappingcapacity,whichwouldbebeneficialinaRPLC×SFC interfaceset-up.However,alargetransfervolumecanhave detri-mentaleffectsontheseconddimensionSFCseparationatthesame time,sincealargevolumeoftransferredorganic-richeluentcan leadtodistortedearlyelutingpeaksintheSFC.Furthermore,the transferoflargevolumesofwater-richeluenthasbeenprovento causesignificantSFC pressureincreaseand accumulationatthe beginningofthemodulations[21]

Fig 4. Influence of injection volume and water content on SFC performance.

a) veratraldehyde in ACN:H 2 O 1:9, different injection volumes; b) veratraldehyde in ACN:H 2 O 1:1, different injection volumes; c) veratraldehyde in ACN:H 2 O 9:1, different injection volumes; d) veratraldehyde in different ACN:H 2 O mixture, 5 ␮L injection; e) pump pressure curves of different injections See experimental section for specific chromatographic conditions.

Inordertoinvestigatetheinfluenceofinjectionvolumeand

watercontentonSFCseparationperformance,differentvolumes

of veratraldehydein samplediluents ofhigh, medium and low

amountofwaterwereinjected.Veratraldehydewaschosenasit

elutesearlyinallfourSFCcolumnsscreenedinthisstudyandis

morepronetopeakdistortioncausedbythesamplediluent[30]

AscanbeseeninFig.4,injectionscontainingrelativelyhighamount

ofwaterleadtogoodpeakshapeupto10␮L(Fig.4a).However,the

peakshapestartedtodeteriorateat10␮Land5␮Lrespectivelyfor

injectionofmediumandlowpercentagesofwater(Fig.4 andc)

Anothergeneralobservationisthattheretentiontimeofthe

ana-lyteseemedtoincreasewithincreasinginjectionvolumebetween

1and5␮Lforallthreesamplediluents.Furthermore,forthesame

injectionvolume,increasingwatercontentinthesamplediluent

seemstoprolongtheanalytesretention.Interestingly,sample

dilu-entscontainingwaterhasbeenproventocausepeakdistortionfor

relativelylargeinjectionvolumesinSFC,whichhasbeenattributed

todemixingeffectofwaterwithmobilephase,strongsolventeffect

andviscousfingering[30].Aplausibleexplanationforthegood

peakshapewithlargevolumeofhighwatercompositionsample

injectionsinthisstudycouldbethatforthisshortcolumnwith

relativelyhighmobilephaseflowrate,thewatercontaining

injec-tionsolventplugtravelsthroughthecolumninaveryshorttime,

sothattheextentofdemixingofwaterinthemobilephaseand

viscousfingeringisminimal.Anotherveryimportantfactorthat

hastobeconsideredfor2DRPLC×SFCapplicationisthepressure

increasecausedbylargevolumeinjectionofawatercontaining

solution.Fig.4 showsthatimmediatelyaftertheinjection,a

pres-sureincreaseoccursbeforethenormalpressurerampcausedbythe

gradient(redcircle).Andthemorewatertheinjectionsolvent

con-tains,thehigherthepressureincreaseis.Moreover,Fig.4 shows

thatwiththesameinjectionvolume,retentionisdecreasedwith

lesswaterinjected.Ourhypothesisonthesefindingsisthata

tem-porarywaterlayerisformedonthediolstationaryphaseatthe

frontendofthecolumnafterthesampleplugreachestheinletof

thecolumn.Thethicknessofthewaterlayerisdeterminedbythe

amountofwaterinjected.Thissuggestedwaterlayercould

perti-nentlyincreasetheretentionpowerofthestationaryphasewhich mightbeoneofthereasonswhyretentionincreaseswithincreasing injectionvolumeforthesamesamplediluentandwith increas-ingwatercontentinsamplediluentofthesameinjectionvolume Also,themobilephasechannelsinthecolumnarenarrowedby thewaterlayer,whichcausesapressurespikerightafterinjection Thesystempressuredifferenceamongdifferentinjectionsmight

beanothercauseofdifferentretentiontimesobserved, consider-ingthecompressibilityofsupercriticalCO2.Althoughitisnotthe focusofthisstudy,theeffectofwaterassamplediluentonSFC certainlydeservesathoroughandsystematicinvestigationinthe future

Basedontheseresults,trappingcolumnswithlargervolume than 10␮Lare not favored in the 2D RPLC×SFC system,even thoughitmayofferhighertrappingcapacities.Also,theusageof theselectedtrappingcolumninthisstudyrequiresthatthefirst dimensionLCseparationshouldbecontrolledinawaythatmost

oftheanalytesareelutedbeforethegradientreachesveryhigh organicsolventpercentage.Furthermore,thegradientinthefirst dimensionshouldbesetasgradualaspossibletoavoidmismatch

ofretentiontimesofdifferentfractionsofthesamepeak

3.3 FirstdimensionRPLCpeakcapacitystudy

Theinfluencesofflowrateandgradienttime/voidtime(tg/t0)

on the first dimension RPLC peak capacity were investigated

As can be seen in Fig 5(a), the first dimension peak capacity showsaseeminglypositivelogarithmicrelationshipwiththeratio betweengradienttimeandvoidtime,whentheflowratewaskept unchangedatarelativelylowvalue.Thisindicatedthatwhenthe firstdimensionisonlyrunningatarelativelylowflowrate,the increase of gradienttime is unnecessaryfor elevatingthepeak capacityafteracertainpointasonlylimitedincreaseofpeak capac-itycanbeachievedwithmuchlengthenedanalysistime.Although thefirstdimensionLCstill hastoberunatlowflow rates con-sideringthelimitedtrappingcapacityofthetrappingcolumn,the importanceofincreasingtheflowrateasmuchaspossibleis

Althoughhigherflowratescanleadtohigherfirstdimension

peak capacity, this impact canbe graduallyevened outby the

increasingundersamplingeffect.Thisisbecausethefirst

dimen-sionpeakswillbecomenarrowerwithhigherflowrates,butthe

modulationtimehastobelongenoughtomakesureallthe

com-poundstransferredinonemodulationareelutedbeforethenext

modulation.Takingthisandthetrappingcolumnretention

capac-ityintoconsideration,0.05mL/minwasfinallyselectedasthefirst

dimensionflowrateandthefirstdimensionLCmethodwasthen

improvedaccordingly

3.4 Seconddimensioncolumnscreeningand2Dmethod

development

Preliminaryexperiments(datanotshown)revealedthat

phe-noliccompoundsindepolymerisedligninsamplesshowedawide

varietyofretentionbehavior inSFCanalysisdue totheir

struc-turalcomplexity.Inordertoachievecompleteelutionofanalytes

ineverymodulationandtakingintoconsiderationtheresultsofthe

trappingcolumntest,0.8minwaschosenasthemodulationtime

LCseparationinthefirstdimension,␲-␲andH-bonding interac-tionintheseconddimensionSFCcanbothprovidehighdegreesof orthogonality.ThecomparativelylowercoveragefactoroftheBEH columncouldberelatedtotheethylenebridgedhybridparticle whichdecreasestheH-bondingcapabilityofthesilicastationary phase.However,oneadvantageoftheBEHcolumnisthatithad thelowest pressure spikeat thebeginning of each modulation (20barcomparedwith30to40barfortheotherthreecolumns) Thisobservationcouldalsoberelatedtothehypothesisofwater layerformationonthestationaryphase.Asethylene-bridgedsilica

isfarlessacidicthannormalsilica,theweakenedhydrogenbonds betweenthestationaryphaseandH2Ocanreducethethickness

ofthewaterlayer.Thehighlyorthogonalseparationdemonstrated

bythecombinationofRPLCandSFCindicatedthathydrophobic interactionwhichdetermines theelutionof thefirst dimension RPLCplaysaweakroleindecidingthecompoundelutionofthe seconddimensionSFC.TheDIOLcolumnwaspickedforfurther experimentsasitprovidedrelativelybetterseparationoftheearly elutingpeaksfromthefirstdimension,mediumpressurespikeand

Fig 6.RPLC × SFC second dimension column screening Sample: 40 lignin phenolic compound standard mixture Chromatographic conditions can be found in the Experimental

Fig 7. 2D separation of a lignin depolymerised sample with the final RPLC × SFC methods with interface using A: two trapping columns; B: two collection loops Chro-matographic conditions can be found in the Experimental section See Table S1 for peak identities Due to the software limitation of the home-built instrument, only limited number of modulations could be programmed and performed for one analysis Consequently, 2D analysis was only applied with the maximum number of modulation allowed

by the software during the periods of time when most compounds eluted in both A and B, including the same fractions of peaks eluting from the 1st dimension column.

comparativelybetterseconddimensionpeakshapeandlessnoisy

background

Fortheimprovementofthe2Dseparation,differentco-solvents,

columntemperatures,backpressuresandgradientsweretested

for thesecond dimensionSFC Additionally,the second

dimen-siongradientwastunedsothatitstartedalreadyintheprevious

modulation,inordertocompensateforthecomparativelylarge

gradientdelayvolumeofthesystem.ThefinalRPLC×SFCmethod

wasthenappliedfortheseparationofalignindepolymerised

sam-ple(Fig.7A)

3.5 ComparisonoftrappingcolumnsandloopsintheRPLC×SFC

system

Forthecomparisonofusingtrappingcolumnsandcollection

loopsasinterfaceintheRPLC×SFCsystem,thedevelopedmethod

wasmodifiedbyreplacingtrappingcolumnswithloopsof

simi-larvolume(∼10␮L).Eventhoughthemodulationtimewaskept

thesame,thefirstdimensionLCflowratehadtobesignificantly

decreased(4times)consideringtheamountofeluenttransferred

intotheseconddimension

Thecomparison of2Dchromatogramsgenerated fromusing

collectionloopsandtrappingcolumnscanbevisualizedinFig.7

Ingeneral,thecombinationofRPLCandSFCshowshighdegree

oforthogonality Thecoverage factor wascalculatedto be0.79

and0.77fortrappingcolumninterfaceandloopinterface

respec-tively.Theusageoftrappingcolumnintheinterfacereducedthe

totalanalysistimeby2times,ascomparedtointerfacewithloops

Fourcompoundscouldbedetectedandtentativelyidentifiedboth

withthetrappingcolumninterfaceandcollectionloopinterface

basedontheseparationofthemixtureofthe40standards

How-ever,vanillicacid(peak11)canonlybedetectedwiththeusage

oftrappingcolumnsintheinterface.Thisdemonstratedanother

advantageofusingtrappingcolumns:someanalytesofrelatively

lowconcentrationcanbedetectedastheyareconcentratedinthe

trappingcolumnbeforetransferredintotheseconddimension.This

isnotpossiblewithtraditionalloopinterfaces.However,thetotal

conditionalpeakcapacityobtainedwithtrappingcolumnis277,

whichislowerthanthatwithloops(340).Thisiscausedbymore

severeundersampling(ˇ=0.26fortrappingcolumn;ˇ=0.57for

loops)whentrappingcolumnisusedasinterface.Whatisalso

wor-thytobepointedoutisthateventhoughnowrap-aroundpeaks

wereobservedduringthe2nddimensionoptimizationwiththe

standardmixture,asthevarietyofcompoundsinthereal

depoly-merisedligninsampleisverywide,afewpeakselutedattheendof

oraftertheendofonemodulation.Ifthesewrap-aroundsaretobe eliminated,thegradienthastoendwithahighpercentofco-solvent

oranextrahold-uptimeatarelativelyhighpercentofco-solvent hastobeadded.Eitherway,the2nddimensionseparationorthe trappingofthecompoundswouldhavetobesignificantly compro-mised.Therefore,inthisstudythe2nddimensiongradientwasset

inawaythatthenumberofwrap-aroundpeakswerereducedto minimumandtheremainingwrap-aroundpeaksallelutedvery earlyinthenextmodulationbeforethesolventpeakwithout over-lappingwiththepeaksinthefollowingmodulationperiod

Inourpreviousresearchwehavedevelopedanultra-high per-formancesupercriticalfluidchromatographymethod(UHPSFC)for theanalysisoflignin-derivedphenoliccompounds[31].Thesame depolymerisedligninsampleinthisstudywasalsoanalysedusing theUHPSFCmethod(seechromatograminFig.S1).In compari-son,theRPLC×SFCmethoddevelopedhadamorethanthreetimes higherpeakcapacitythantheUHPSFCmethod(277forRPLC×SFC and81forUHPSFC).ThedisadvantageoftheRPLC×SFCmethodis thelongeranalysistime,whichwas6timeslongerthanthatofthe UHPSFCmethod.Inordertoshortenthe2Danalysistimewithout

asacrificeofthepeakcapacity,futureworkshouldfocuson devel-opingmoreeffectiveinterfacesamplefocusingtechniquestoallow higherflowratetobeusedinthefirstdimensionandachievefaster SFCseparationtoremedytheundersamplingissue

4 Conclusion

A novel comprehensive online two-dimensional RPLC×SFC methodwasdevelopedfortheanalysisofdepolymerisedlignin sampleswithtrappingcolumnassistedmodulation.APhenyl-hexyl trappingcolumnwaspickedbasedonatrappingcapacity eval-uationof3differenttrappingcolumns.AlthoughtheRSDvalues werewithinacceptableranges,therepeatabilityofthetrapping columnassistedRPLC×SFCsystemwasshowntobeundermined

bythelimitedtrappingcapacity.Activemodulationoramore thor-oughsearchforcolumnsspecificallydesignedfortheretentionof smallphenoliccompoundscouldpotentiallyimprovethesystem

Nodemixingeffectswereobservedwheninjectinglargevolumes

ofwatercontainingsamplesinSFCwhenusingahighflowrate anda shortcolumn,possiblyduetofasttravelofintactsample diluentplugthroughthecolumn.Waterassamplediluentturned outtoenablegoodpeakshapes,evenataslargeinjectionvolumes

as10␮L.WhenthefirstdimensionLCflowiskeptrelativelylow (e.g.≤0.05mL/mininthisstudy),anincreaseinthefirst dimen-sionflow ratecanraise thefirstdimensionpeak capacitymore

Appendix A Supplementary data

Supplementarydataassociatedwiththisarticlecanbefound,in

theonlineversion,atdoi:10.1016/j.chroma.2018.02.008

References

[1] P Gallezot, Conversion of biomass to selected chemical products, Chem Soc.

Rev 41 (2012) 1538–1558.

[2] O.Y Abdelaziz, D.P Brink, J Prothmann, K Ravi, M.Z Sun, J Garcia-Hidalgo, M.

Sandahl, C.P Hulteberg, C Turner, G Liden, M.F Gorwa-Grauslund, Biological

valorization of low molecular weight lignin, Biotechnol Adv 34 (2016)

1318–1346.

[3] M.P Pandey, C.S Kim, Lignin depolymerization and conversion: a review of

thermochemical methods, Chem Eng Technol 34 (2011) 29–41.

[4] J.S Lupoi, S Singh, R Parthasarathi, B.A Simmons, R.J Henry, Recent

innovations in analytical methods for the qualitative and quantitative

assessment of lignin, Renew Sust Energ Rev 49 (2015) 871–906.

[5] X.P Ouyang, Z.L Chen, X.Q Qiu, Determination of monophenolic compounds

from lignin oxidative degradation using ultra performance liquid

chromatography/high resolution mass spectrometry, Chin J Anal Chem 42

(2014) 723–728.

[6] T.M Jarrell, C.L Marcum, H.M Sheng, B.C Owen, C.J O’Lenick, H Maraun, J.J.

Bozell, H.I Kenttamaa, Characterization of organosolv switchgrass lignin by

using high performance liquid chromatography/high resolution tandem mass

spectrometry using hydroxide-doped negative-ion mode electrospray

ionization, Green Chem 16 (2014) 2713–2727.

[7] T Michailof, S Sfetsas, K Stefanidis, G Kalogiannis, A Theodoridis,

Quantitative and qualitative analysis of hemicellulose, cellulose and lignin

bio-oils by comprehensive two-dimensional gas chromatography with

time-of-flight mass spectrometry, J Chromatogr A 1369 (2014) 147–160.

[8] M Windt, D Meier, J.H Marsman, H.J Heeres, S de Koning, Micro-pyrolysis of

technical lignins in a new modular rig and product analysis by GC–MS/FID

and GC x GC-TOFMS/FID, J Anal Appl Pyrol 85 (2009) 38–46.

[9] A Le Masle, D Angot, C Gouin, A D’Attoma, J Ponthus, A Quignard, S.

Heinisch, Development of on-line comprehensive two-dimensional liquid

chromatography method for the separation of biomass compounds, J.

Chromatogr A 1340 (2014) 90–98.

[10] M Gilar, P Olivova, A.E Daly, J.C Gebler, Orthogonality of separation in

two-dimensional liquid chromatography, Anal Chem 77 (2005) 6426–6434.

[11] I Francois, K Sandra, P Sandra, Comprehensive liquid chromatography:

fundamental aspects and practical considerations-a review, Anal Chim Acta.

641 (2009) 14–31.

(2010) 1504–1512.

[20] C.J Venkatramani, M Al-Sayah, G.N Li, M Goel, J Girotti, L.S Zang, L Wigman, P Yehl, N Chetwyn, Simultaneous achiral-chiral analysis of pharmaceutical compounds using two-dimensional reversed phase liquid chromatography-supercritical fluid chromatography, Talanta 148 (2016) 548–555.

[21] M Sarrut, A Corgier, G Cretier, A Le Masle, S Dubant, S Heinisch, Potential and limitations of on-line comprehensive reversed phase liquid

chromatography x supercritical fluid chromatography for the separation of neutral compounds: an approach to separate an aqueous extract of bio-oil, J Chromatogr A 2015 (1402) 124–133.

[22] L.A Gao, J Zhang, W.B Zhang, Y.C Shan, Z Liang, L.H Zhang, Y.S Huo, Y.K Zhang, Integration of normal phase liquid chromatography with supercritical fluid chromatography for analysis of fruiting bodies of Ganoderma lucidum, J Sep Sci 33 (2010) 3817–3821.

[23] A.F.G Gargano, M Duffin, P Navarro, P.J Schoenmakers, Reducing dilution and analysis time in online comprehensive two-dimensional liquid chromatography by active modulation, Anal Chem 88 (2016) 1785–1793 [24] M Verstraeten, M Pursch, P Eckerle, J Luong, G Desmet, Thermal modulation for multidimensional liquid chromatography separations using low-thermal-mass liquid chromatography (LC), Anal Chem 83 (2011) 7053–7060.

[25] B.W.J Pirok, A.F.G Gargano, P.J Schoenmakers, Optimizing separations in online comprehensive two-dimensional liquid chromatography, J Sep Sci 41 (2018) 68–98.

[26] P Cesla, J Krenkova, Fraction transfer process in on-line comprehensive two-dimensional liquid-phase separations, J Sep Sci 40 (2017) 109–123 [27] D.R Stoll, X.L Wang, P.W Carr, Comparison of the practical resolving power

of one- and two-dimensional high-performance liquid chromatography analysis of metabolomic samples, Anal Chem 80 (2008) 268–278.

[28] M Sun, G Liden, M Sandahl, C Turner, Ultra-high performance supercritical fluid chromatography of lignin-derived phenols from alkaline cupric oxide oxidation, J Sep Sci 39 (2016) 3123–3129.

[29] J De Vos, S Eeltink, G Desmet, Peak refocusing using subsequent retentive trapping and strong eluent remobilization in liquid chromatography: a theoretical optimization study, J Chromatogr A 1381 (2015) 74–86 [30] V Desfontaine, A Tarafder, J Hill, J Fairchild, A Grand-Guillaume Perrenoud, J.L Veuthey, D Guillarme, A systematic investigation of sample diluents in modern supercritical fluid chromatography, J Chromatogr A 2017 (1511) 122–131.

[31] J Prothmann, M Sun, P Spegel, M Sandahl, C Turner, Ultra-high-performance supercritical fluid chromatography with quadrupole-time-of-flight mass spectrometry (UHPSFC/QTOF-MS) for analysis of lignin-derived monomeric compounds in processed lignin samples, Anal Bioanal Chem 409 (2017) 7049–7061.