52 LCGC NORTH AMERICAVOLUME 22NUMBER 1JANUARY 2004 www.chromatographyonline.comtemperature-risingelutionfractionationovercomesthelimitationsofpreparativetemperature-risingelutionfractionationbyreducingthecolumnandsamplesizescon-siderablyandbycontinuouslymonitoringthe amount of the eluted polymer fractionsasafunctionofcolumntemperatureusingan on-line detector.Forpolyethylenes,themajorfactorthatdeterminesmeltinginasolutionisthedegreeofshort-chainbranching(3).Eachbranchingpointinachainisrejectedfromthe growing crystal, thereby limiting crystalthicknessandhencereducingthemeltingtemperature.Thus,adirectrelationshipexists between the melting temperature andthe degree of branching, which is reected inanalyticaltemperature-risingelutionfrac-tionationcurves.Moreover,NaganoandGoto(3)andMirabella(4)havereportedthatalinearrelationshipexistsbetweentheelution temperature and the degree of short-chainbranching(numberofbranchesper1000carbonatoms).Therefore,analyticaltemperature-risingelutionfractionationcurvesdoinfactdescribetheshort-chainbranching distribution (5).Theshort-chainbranchingdistributionprovides some information about the mech-anismofpolymerizationandthenatureofthe catalysts. As an example, when chemistsuseconventionalZiegler-Nattacatalystsforthepolymerizationofethylene,abroadThe authors describe a system for automated analytical temperature-risingelution fractionation that uses the built-in differential refractometer of acommercial gel-permeation chromatography instrument and a stop-owmethod for sample crystallization. They discuss the performance of thesystem and its ability to fractionate different types of polyethylenes.An Improved AnalyticalTemperature-Rising ElutionFractionation System forAutomated Analysis ofPolyethylenesAdrian G. Boborodea*,Daniel Daoust*, Alain M. Jonas*, andChristian Bailly** Catholic University of Louvain,Unit de Physique et de Chimiedes Hauts Polymres, BtimentBoltzmann, Place Croix du Sud, 1,B-1348 Louvain-la-Neuve,Belgium, e-mail boborodea@poly.ucl.ac.be Polytechnical University ofBucharest, Department ofPolymer Physics, 149 CaleaVictoriei, Sector 1, Bucharest,RomaniaAddress correspondence to A.G. Boborodea.olymersbasicallyareheteroge-neousmaterials,andheterogene-itycanbeexhibitedinvariousways such as distribution of chainlengths,differencesinchemicalcomposi-tion,stereoregularity,andarchitecture(tac-ticity or branching). Branching can be char-acterized as either long chain or short chain.Temperature-risingelutionfractionationisatechniquethatallowstheanalysisofsemicrystallinepolymersbyseparatingfrac-tionsaccordingtotheircrystallizability.Temperature-risingelutionfractionationenables analysts to evaluate structure hetero-geneityandtocharacterizethedistributionofshort-chainbranching.Elutionduringtemperature-risingelutionfractionationisgovernedbythemeltingofsemicrystallinepolymers in the presence of a solvent (1).Preparativetemperature-risingelutionfractionationcanbeperformedtoisolatefractionsforadditionalinvestigations.Dur-ing this process, a large quantity of solvent isrequired for eluting the sample, and a largeamountofnon-solvent,usuallymeth-anol, also is used for precipitating the elutedfractions(2).Thistechniqueisquitetime-consuming; it typically requires several daysto perform a fractionation.The development of analytical temperature-risingelutionfractionationhasbeenanimprovementuponpreparativetemperature-risingelutionfractionation.Analytical P54 LCGC NORTH AMERICAVOLUME 22NUMBER 1JANUARY 2004 www.chromatographyonline.combond stretch frequency. IR detection gener-ally is used for analytical temperature-risingelution fractionation because it offers betterbaselinestabilitythandifferentialrefractiveindexdetectionforbothlargevariations of temperature and pressure during the elu-tionprocess(7,1216).However,theIRdetectorhaspoorsensitivityandconse-quently requires the use of large amounts ofpolymerinanalysis.Forthisreason,eitherthe baseline presents a high level of noise (forexample,seereference7sFigure1orrefer-ences 4 and 12s Figure 1) or the peaks con-tainspikes(forexample,seereference8sFigure8orreference15sFigure9).Todecreasethenoiselevel,analystssometimesaveragethechromatogramdatapointsandsmooththecurves(seereference8sFigure11).Thepresenceofahighnoiselevelbecomes even more important when mathe-maticaltransformationsareappliedtoana-lyticaltemperature-risingelutionfractiona-tionchromatogramstoobtainmethylenesequence length distributions for polyethyl-enecopolymers(17)ortopredictdifferen-tialscanningcalorimetrythermograms(4,16,18).Furthermore,IRdetectorsareunavailableoncommercialequipmentandarecomplicatedtoinstall.Forthisreason,theresultsobtainedusingtheanalytical temperature-risingelutionfractionationtechniqueareverydifficulttoreproducefrom one laboratory to another.Inthisarticle,wedescribeanimprovedanalyticaltemperature-risingelutionfrac-tionationsystemthatusesacommercialinstrumentforhigh-temperaturegel-permeation chromatography connected withan oil bath. Our goal was to keep the modi-cation to a minimum by using the existingparts of the chromatograph as much as pos-sible: namely, the automatic injector and thedifferential refractometer, which is more sen-sitivethananIRdetector.Thissystemallowedtheuseofalowerquantityofinjectedpolymer,whichenabledthepossi-bilityofcompletelyautomatingtheprocessby reducing the injected volume of solutionto 200500 L.ExperimentalAnalyticaltemperature-risingelution fractionationapparatus:Theanalyticaltemperature-rising elution fractionation wasperformedusingamodifiedWatersGPC150CVgel-permeationchromatography(GPC)system(WatersCorp.,Milford,Massachusetts) using greater than 99% pure1,2,4-trichlorobenzene(Sigma-AldrichCo.,St. Louis, Missouri) as the solvent. The sol-bimodal distribution of polymer chains withdifferentshort-chainbranchingcontents ispresentinthefinalproductbecauseof themultisitenatureofthesecatalysts(6), ascomparedwithmetallocenecatalysts,which produce a much narrower short-chainbranchingdistribution(7).Short-chainbranchingdistributionandthemolecularmassdistributionhaveamarkedinuenceupon the polymer properties. The shapes ofthesedistributionshaveaprofoundeffectupon the end-use properties of the material,and their determination can generate a morecompleteunderstandingofthebehaviorofpolyethylenes in different applications (8).Several researchers have described systemsfor analyzing a solution of a crystalline or asemicrystalline polymer sample to determineitsshort-chainbranchingdistribution(911).Asageneralmethod,thepolymersamplerstisprecipitatedfromasolutionduring a decreasing temperature gradient toproduce a precipitated polymer sample. Theprecipitatedpolymersamplethenisredis-solved and eluted during an increasing tem-perature gradient to produce successive frac-tionsofthepolymersample,whicharemeasuredtoproduceconcentrationdatafromwhichtheshort-chainbranchingdis-tributioniscalculated.Theconcentrationdetector is a built-in differential refractome-ter (9) or a separate infrared (IR) or nuclearmagneticresonance(NMR)detector(10).In both cases the polymer solution is intro-duced into the fractionation column, whichis placed in an oil bath, using a complex pro-cedurethatrequireshighquantitiesofsol-vent and polymer sample.On the other hand, Yau and Gillespie (11)describeananalyticaltemperature-risingelutionfractionationsystemwithanauto-maticinjectorandanIRdetector.Thisdetectorwastunedto3.48m,theCHventwasstabilizedwith2%ofantioxidantIrganox1010(CibaSpecialtyChemicals,Inc.,Basel,Switzerland).Weplaceda100mm 10 mm stainless steel column packedwith aluminum oxide (70-230 mesh, MerckKGaA,Darmstadt,Germany)inaJulaboF32HPoilbath(JulaboLabortechnikGmbH, Seelbach, Germany).Intherststepoftheanalysisprogram,400 L of solution was injected in the col-umnautomaticallyusingthebuilt-ininjec-toratalowowrateof0.2mL/min.Thislowowratewasusedtoavoidapressure-inducedleakofthepolymersolutionthroughthecolumnafterinjection.Inthesecondstep,theowwasstoppedandthecolumnwascooledintheoilbathat0.5C/min from 135 C to 35 C. In the thirdstep, the polymer was eluted at a ow rate of0.8mL/min,whilethetemperaturewasincreasedat1C/minfrom35Cto135C. The concentration of the eluted specieswas measured using the instruments built-indifferentialrefractiveindexdetector.Boththe chromatogram and the temperature wererecorded using two data channels and storedusing Millennium 32 software (Waters). Weallowedanadditional40minforcompletecleaning of the column before the next injec-tion.Therefore,thetotaltimeforanalysiswas approximately 6 h, and the system couldperformfiveanalyticaltemperature-risingelutionfractionationsinanautomatedmode.Samplepreparation:Tocheckrepro-ducibility and to calibrate the area versus theconcentration,weusedpolyethylenestan-dardsfromtheNationalInstituteofStan-dards and Technology, denoted polyethyleneNBS1475(linearpolyethylene)andpoly-ethyleneNBS1476(branchedpolyethyl-ene).Thepolymersolutionswerepreparedat 145 C in 1,2,4-trichlorobenzene at 115mg/mLconcentrationswiththeinstru-ments built-in procedure.Results and DiscussionChromatograminterpretation:Todemon-strate how the variation of column tempera-tureaffectsthebaselinestability,Figure1shows two chromatograms of the polyethyl-ene NBS 1475 sample obtained for two con-centrations.Thebaselinesoverlapoutsidetheelutiondomain.Thebaselinedriftiscausedprimarilybythetemperatureandpressure variations in the column during theheating step. To eliminate the drift, we sim-ply used a linear interpolation of the baselineunder the peak.Optimization of injection time: In previ-ouslydescribedmethods,theanalytical Figure1: Analyticaltemperature-risingelu-tion fractionation chromatograms of polyethyl-ene NBS 1475 recorded for two concentrations(3 mg/mL: black line; 6 mg/mL: pink line) show-ingthecontinuousdriftofthebaselinewithelution time.0.200.180.160.14Detector response (mV)Time (min)3 mg/mL6 mg/mL40 50 60 70 80 9056 LCGC NORTH AMERICAVOLUME 22NUMBER 1JANUARY 2004www.chromatographyonline.comtemperature-risingelutionfractionationcolumnwasfilledwithapolymersolu-tionandthenclosedandcooledwitha controlled-temperatureprogram(911).Afterthecrystallizationstep,thecolumnwas reintroduced in the solvent stream.The use of an autoinjector is essential forthepossibilityofacompletelyautomatedprocess,whichimpliesthatthecolumnremains coupled with the high performanceliquidchromatography(HPLC)systemduringbothcoolingandheatingsteps.Therefore, the objective of the injection stepistointroducethesamplecompletelyintothe column and to avoid any spilling of thesamplefromthecolumnwhenthesolventowisstopped.Becauseoftheserequire-ments,itisnecessarytomeasuretheopti-mal injection time in the column.For highly crystalline samples, shorter orlonger injection times will produce the pre-cipitation of the polymer in the inlet or out-let tube, respectively. Because the polyethyl-ene NBS 1476 sample has a lower meltingtemperatureandalsopresentsabroaderrange of elution temperatures, we selected itformeasuringtheoptimalinjectiontime.Thesefeaturesareessentialtopreventobstructionofthecapillarytubeswhenusing short or long injection times. Figure 2illustrates the variation of the ratio betweenthechromatographicareaandtheinjectedmass versus the injection time.Forinjectiontimeslessthan10min,weobserved a domain of almost constant values.For 11 min, a portion of the injected samplespilledintheoutlettubeandwaselutedimmediatelyaftertheowwasrestarted.Because this portion was masked by the sol-ventpeak,theanalyticaltemperature-risingelutionfractionationpeakpresentedasmaller area than expected. Based upon datafrom Figure 2, we selected an injection timeof 9 min.Calibration curve: For quantitative mea-surements of the eluted fractions, we gener-atedacalibrationcurveofthedetectorresponse versus injected mass for polyethyl-eneNBS1475.Figure3presentsthedataobtainedfor117mg/mLconcentrations.The linearity of the data is satisfactory.Reproducibility of elution temperature:Animportantparameterforthequalityofananalyticaltemperature-risingelutionfractionationmethodisthereproducibilityoftheelutiontemperature(volume)foracharacteristicpeak.Theelutiontempera-ture (Tpeak) was taken as the temperature ofthebaththatcorrespondedtothemaxi-mumsignalofthedetector.Tpeakwasrecordedusingtheseconddatachannel.Figure4showstheelutiontemperaturesthat correspond to the peaks of the samplesinjected for the calibration curve presentedin Figure 3.The elution time is reproducible for con-centrationsgreaterthan3mg/mL.Averysmalldecreasecanbedetectedonlyforlowerconcentrations.Figure5illustratesthechromatogramsforthreeinjectedcon-centrations after baseline subtraction.Methodresolution:Thepowerofananalyticaltemperature-risingelutionfrac-tionation method is its ability to resolve dif-ferent polyethylenes that differ by only theirshort-chainbranchingdistribution.Baseduponthelinearrelationshipbetweentheelutiontemperatureandthedegreeofbranching, we can represent the short-chainbranchingdistributionastheweightfrac-tion of the eluted species versus elution tem-peratures.Forthispurpose,wesubtractedthe baseline and normalized the peak.Figure 6 shows the short-chain branchingdistributionsobtainedusingthedescribedanalyticaltemperature-risingelutionfrac-tionation method for the two polyethylenestandards.ThetemperaturedomainsinwhichthepolyethyleneNBS1475andpolyethyleneNBS1476standardsareelutedaresimilarwiththereporteddata forhighlycrystallinepolyethyleneandFigure6: Short-chainbranchingdistribu-tionsfortwotypesofpolyethylenes:linear(polyethyleneNBS1475)andbranched(poly-ethylene NBS 1476).0.250.200.150.100.050.00Short-chain branching (%)Temperature (C)NBS 1476NBS 147560 70 80 90 100 110Figure5: Analyticaltemperature-risingelu-tion fractionation chromatograms of polyethyl-ene NBS 1475 for three injected concentrations(1.01mg/mL:blue;5.37mg/mL:pink;16.38mg/mL: black).0.0400.0300.0200.0100.000Detector response (mV)Temperature (C)1.01 mg/mL5.37 mg/mL16.38 mg/mL60 70 80 90 100 110Figure 4: Variation of the peak elution tem-peratureforpolyethyleneNBS1475asafunc-tionoftheconcentrationofthesamplesolu-tion.97.597.096.596.0Elution temperature (C)Concentration (mg/mL)0 5 10 15 20Figure 3: Calibration curve for polyethyleneNBS1475in1,2,4-trichlorobenzenewithcon-centrations between 1 mg/mL and 17 mg/mL.12.08.04.00.0ADRI ([mVs] 106)Injection mass (mg)R2 0.99920 2 4 6 8Figure2: Variationoftheratiobetweenchromatographic area and injected mass versusinjectiontimeforpolyethyleneNBS1476at 5 mg/mL.1.501.451.401.351.30ADRI/Minj ([mVs/mg] 106)Injection time (min)6 7 8 9 10 11 12JANUARY 2004 LCGC NORTH AMERICAVOLUME 22NUMBER 1 57 www.chromatographyonline.combranchedpolyethylene,respectively(11,18,19).It is important to mention that the reso-lutiondependsuponthetemperaturepro-gram (18). The resolution can be improvedforcomplexsamplesbyusingslowertem-peraturegradientsforbothcrystallizationand elution steps.ConclusionsTodevelopanautomatedanalytical temperature-risingelutionfractionationmethodweneededtwoequallyimportantthings:asuitablecommercialsystem(oraminormodificationofacommercialsys-tem)andareproduciblemethodwiththissystem.In this article, we have presented a simplemethod to obtain an analytical temperature-risingelutionfractionationsystemusingaminor modication of a commercially avail-ableGPCinstrumentandatemperature-programmingsystem.Wealsohaveestab-lishedseveralguidelinesfortheanalyticaltemperature-risingelutionfractionationsys-temqualicationthatusetheacceptedpro-cedure for the validation of chromatographicsystems.An important parameter for the identi-cation of compounds using HPLC systemsis the reproducibility of the elution time fordifferent analyzed compounds. For the ana-lyticaltemperature-risingelutionfractiona-tionmethod,thisreproducibilityistrans-lated into the reproducibility of the elutiontemperature.Usingthesystemdescribedabove,weobtainedareproducibilityof0.05 C for polyethylene NBS 1475 con-centrationsgreaterthan3mg/mLandasgreat as 20 mg/mL.ForquantitativeanalysesofmixturesusingtheHPLCmethods,agoodcalibra-tion curve peak area versus injected mass is essential. In the analytical temperature-risingelutionfractionationmethod,themixtureiscomposedoffractionsthathavedifferentmeltingtemperaturesinsolution.These fractions are recorded as a single chro-matogram, which is used to obtain a short-chainbranchingdistribution.Forthisrea-son,alinearcalibrationcurvemustbeveriedforabroaddomainofconcentra-tionsbeforedeterminationofshort-chainbranchingdistribution.Usingoursystem,wefoundanexcellentlinearcorrelationbetween peak area and injected mass of thepolymerforadomainof0.117mg/mLconcentrations, which provides the possibil-ity of a quantitative determination of short-chain branching distribution.References(1) P.J. Flory, Principles of Polymer Chemistry (Cor-nell University Press, Ithaca, New York, 1953).(2) P.Viville,D.Daoust,A.M.Jonas,B.Nysten,R.Legras,M.Dupire,J.Michel,andG.Debras, Polymer 42, 19531967 (2001).(3) S. Nagano and Y. Goto, J. Appl. Polym. Sci. 26,42174231 (1981).(4) F.M.Mirabella,J.Polym.Sci.BPolym.Phys.39, 28192832 (2001).(5) J. Xu and L. Feng, Eur. Polym. J. 36, 867878(2000).(6) L.J.D. Britto, J.B.P. Soares, A. Penlidis, and B.Monrabal,J.Polym.Sci.BPolym.Phys. 37,539552 (1999).(7) M. Zhang, D.T. Lynch, and S.E. Wanke, Poly-mer 42, 30673075 (2001).(8) A.FaldiandJ.P.B.Soares,Polymer 42,30573066 (2001).(9) L.G. Hazlitt and D.G. Moldovan, U.S. patentnumber 4,798,081, 17 January 1989.(10) F.C.Stehling,U.S.patentnumber5,008,204,16 April 1991.(11) W.W.YauandD.Gillespie,Polymer 42,89478958 (2001).(12) G. Gabriel and D. Lilge, Polymer 42, 297303(2001).(13) A.M.E.BakerandA.H.Windle,Polymer42,651665 (2001).(14) C. Wang, M.C. Chu, T.L. Lin, S.M. Lai, H.H.Shih,andJ.C.Yang,Polymer42, 17331741(2001).(15) I.A.Hussein,K.Ho,S.K.Goyal,E.Kar-bashewski, and M.C. Williams, Polym. Degrad.Stab. 68, 381392 (2000).(16) S.A.KaroglanianandI.R.Harrison,Ther-mochem. Acta 288, 239245 (1996).(17) J.G. Bonner, C.J. Frye, and G. Capaccio, Poly-mer 34, 35323534 (1993).(18) C.A.FonsecaandI.R.Harrison,Thermochem.Acta 313, 3741 (1998).(19) J.P.SoaresandA.E.Hamielec,Polymer36,16391654 (1995).