ANALYSIS OF BRANCHING DISTRIBUTION IN POLYETHYLENES BY DIFFERENTIAL SCANNING CALORIMETRY

Chinese Journal of Polymer Science Vol. 21, No. 2, (2003), 231-239Chinese Journal ofPolymer Science92003 Springer-VerlagANALYSIS OF BRANCHING DISTRIBUTION IN POLYETHYLENES BYDIFFERENTIAL SCANNING CALORIMETRYRobert Shanks', Fei Chen and Gandara AmarasingheApplied Chemistry, RMIT University, GPO Box 2476V, Melbourne, 3001, AustraliaAbstract Short chain branching has been characterized using thermal fractionation, a stepwise isothermal crystallizationtechnique, followed by a melting analysis scan using differential scanning calorimetry. Short chain branching distributionwas also characterized by a continuous slow cooling crystallization, followed by a melting analysis scan. Four differentpolyethylenes were studied: Zicgler-Natta gas phase, Ziegler-Natta solution, metallocene, constrained-geometry single sitecatalyzed polyethylenes. The branching distribution was calculated from a calibration of branch content with meltingtemperature. The lamellar thickness was calculated based on the thermodynamic melting temperature of each polyethyleneand the surface free energy of the crystal face. The branching distribution and lamellar thickness distribution were used tocalculate weight average branch content, mean lamellar thickness, and a branch dispersity index. The results for the branchcontent were in good agreement with the known comonomer content of the polyethylenes. A limitation was that high branchcontent polyethylenes did not reach their potential crystallization at ambient temperatures. Cooling to sub-ambient wasnecessary to equilibrate the crystallization, but melting temperature versus branch content was not applicable after cooling tobelow ambient because the calibration data were not performed in this way.Keywords Polyethylene, Short chain branching, Lamellar thickness, Branch distribution, Thermal fractionation, DSCINTRODUCTIONEthylene copolymers are made by polymerization of ethylene with a-olefins, such as 1-butene, 1-hexene or 1-octene. These polymers contain short chain branches (SCB) along the polymer backbone and exhibit broadmelting ranges indicating the presence of broad lamellar thickness distributions. The properties and performanceof a polymer depend on the short chain branch content and its distribution in the polymer chain. The distributionof SCB depends on the conditions used for polymerization. Zeigler-Natta (ZN) catalysis gives a broad branchingdistribution, particularly when the solution or slurry method is used. This method must be used for 1-octenecomonomer, because of its relatively high boiling temperature. The ZN gas phase polymerization gives lesshighly branched molecules and is suitable for 1-butene or 1-hexene comonomers. Single site catalysts, such asmetallocene and constrained geometry types, provide narrow distributions of SCB of the polymerslPolyethylenes form a chain folded crystal structure where the lamellar thickness is determined by thenumber of methylenes between the folds. Branch points are excluded from the crystals. When the branchesconsist of less than about 20 methylenes they cannot take part in crystallization, so for most short branchedpolyethylenes the branches are excluded from the crystals. When the branches are methyl groups then they can beaccommodated in the crystals as defects, but ethyl, butyl and hexyl branches are excluded, except under veryrapid quench cooling rates4. The analysis of the thickness of the lamella is possible when a polyethylenecrytallization is limited by the SCB distribution. Another way of considering the branching distribution is themean sequence length MSL) of methylenes between branches. There will be a minimum MSL for a chain'Corresponding author: Robert Shanks, E-mail: Robert.shanks@rmit.edu.aPresented at the Intermational Symposium on Polymer Plhysics, 2002, Qing中国煤化工Received August 31, 2002YHCNMHG232R Shanks et al.segment to join a particular lamella.The widely used method to analyze short chain branch distibution of ethylene copolymers is temperaturerising elution fractionation (TREF)P. Since TREF requires long fractionation times with solvent, an altermativefaster and more definitive method is preferred. Thermal fractionation (TF) (segregation) as an alternative tool forqualitative SCB distribution analysis of ethylene copolymers has recently been reported4. Thermal fractionationcan segregate molecules with intramolecular branch distributionsl. 5, whereas TREF can only separateintermolecular branch distributions. TF provides a visual way of studying the branch distribution, whereas foranalysis of the distribution it is not necessary to have segregated fractions. In the present research, a simplifednew method, very slow continuous cooling, that can be used as an altermative for comonomer distributionanalysis with enough accuracy and minimum time using differential scanning calorimetry (DSC) is proposed. Inaddition, this method can be used as an alternative method to obtain lamellar thickness distribution for somebranched polyethylenes. Another DSC method for segregation of the crystals of a polymer is successive self-nucleation/annealinglo. Other experimental methods, which have been used to determine the lamellar thicknessdistribution, are transmission electron microscopy", small angle X-ray sattering and atomic force microscopy'8.DSC has previously been used to measure lamellar thickness distributions using the Thomson-Gibbsequation9. A constraint is that the polyethylene crystals must not rearrange during the DSC scan. In thisresearch, the polyethylene has been cooled very slowly to ensure that the formed crystals are stable. Anotherlimitation is that copolymers cannot reach the equilibrium melting temperature. They are restricted by the molefraction of branches in addition to the need to use realistic crystallization temperatures. The thermodynamicmelting temperature of the ethylene copolymers is the quantity applicable to copolymers that is used in themethod presented.Polyethylenes show both a molar mass distribution and a branching distribution. A two-dimensionaldistribution analysis can be performed by using TREF followed by size exclusion chromatography on the TREFfractionslo. The molar mass distribution can be displayed as a curve, but it is usual to provide averages to definethe distribution. Typically, the number average, weight average and polydispersity are calculated, though otherhigher moment averages can be obtained. This type of analysis should also be possible for the branchingdistribution since an analogous set of averages will completely define the SCB in any polyethylene. It is the aimof this research to provide a set of parameters that quantitatively define polyethylene branching and crystalstructure.Table 1. Characteristics of the ethylene copolymersPropertiesC6-BPE-ZNgas'Cg-BPE-ZNson’C4-BPE-M°Cg-BPE-SVLDPE2ComonomerHexeneOcteneButeneCatalyst type'ZN gasZN solnSMFI (dg min^)0.780.9427.01.0Density (g cm^ )0.9350.9200.9010.908138,00058,00096,700MJM。2.652.86 .Comonomer content (mol %)2.36.32.4Tm(C)'121.9110.4, 118.7, 121.992.7105.4T(C)105.9103.176.690.3'Data were taken from chemical data sheets published by the manufacturer; 'Refs. [, 5]; Ref.[]; Ref []; ZN=Ziegler-Natta catalyst, gas phase or solution; M = Metallocene; S = Single-site catalyst, constained geometry;Cirsallization (T) and melting (Tm) temperature were obtained at scanning rates of 10 K/minEXPERIMENTALThe characteristics of the four branched polyethylenes are listed in Table 1. The copolymers were chosen to beZiegler-Natta (ZN) or single-site metallocene (M) or constrained geometry (S) catalyzed copolymers with butene,hexene or octene comonomers. The ZN polyethylenes were|中国煤化工5) and a solution (solm)MYHCNMHGBranching Distribution in PE233or slurry polymerized system.A Perkin-Elmer Pyris1 DSC was used for the experiments. All samples were held in the melt at 180C for5 min, cooled from 180C to 10C and again heated to 180C from 10°C at the rate of 10 K/min. For the slowcontinuous cooling experiments, the sample was first heated at 180C for 5 min, rapidly cooled to the initialcrystallization temperature for the sample (Ti) and held for 1 min. After holding, the sample was continuouslycooled from Tci to 25C at the very slow cooling rate of 0.08 K/min, which is the same as the average rate forthermal fractionation and of the order of TREF rates. The Ti values were 127, 122, 105 and 122C for C6-BPE-ZNgas, Cg-BPE-ZNsoln, C.-BPE-M and Cg-BPE-S, respectively. Thermal fractionation was performed byheating to 180C and holding for 5 min, rapidly cooling to 124'C at 200 K/min and then holding for 50 min. Thesample was again rapidly cooled by 4C at 200 K/min then held for another 50 min, and then this step wasrepeated sufficient times for the sample to be cooled to about 20°C. The step intervals of 4C and theequilibration period of 50 min were chosen after testing other combinations. If larger temperature steps arechosen, then the polymer will take a longer time to equilibrate at each temperature, because equilibration is alogarithmic process. The total time for cooling was about 22 h. Finally, both treated samples were heated from10- 150C at a rate of 10 K/min to observe their melting behavior.The melting temperatures (Tm) were converted into branch compositin (B) using equations derived from the .data of Hosodal from TREF analyses (Eqs. 1, 2 and 3). The peak areas in the case of thermal fractionation, orpartial areas in the case of continuous cooling, were converted into total amount of polymer present at each levelof branching by the equations derived from the data of Hosodalu for crystallinity (2x) at each branch content(Eq. 4, 5 and 6). The TREF analysis was limited to lower temperatures in the ambient range, whereas the DSCcould be used at sub-ambient temperatures.1-butene copolymers: Tm =-1.55B+ 134 (1)x=-0.0132B+0.82 (4)1-hexene copolymers: Tm = -1.69B+133 (2)x=-0.0134B +0.77 (5)1-octene copolymers: Tm = -2.18B+134(3)xo=-0.0251B+0.86 (6)Sub-ambient temperatures were found to be necessary to convert the DSC curve into a specific heat curvesince the sample baseline must be at equilibrium before starting a scan. If the sample is changing, that is stillcrystallizing, then the initial specific heat will deviate from the expected value and the start of the DSC curve willnot be suitable for analysis. All DSC curves were converted into specific heat curves so that the curves fordifferent samples and treatments could be quantitatively compared.Lamellar thickness () was calculated from the Thomson-Gibbs equation using the surface free energy perunit area of the crystal growth face, σc= 90x 10~ J m", the enthalpy of melting per unit volume OHv = 286 x10° J m' and Tm° is the thermodynamic melting temperature of ethylene copolymer and Tm is the measuredmelting temperaturel2]:2σ。(7)0HV (Tm -Tm)The thermodynamic copolymer melting temperature (Tm) is obtained from the equilibrium meltingtemperature (Tm° = 418.7 K) by considering the mole fraction of crystallizable units in a random copolymer (x)where 0H is the molar heat of fusion per repeat unit in the crystal,△H = 8.284 kJ mor113).11R.Inx(8)TTi0HThe methylene sequence length (MSL) was calculated from Eq. (9), which derived from the calibration curvegenerated by Keating et al!4.中国煤化工MHCNMHG234R Shanks et al.-Im(X)=a135.5-0.331(9)Tm +273)where X is the methylene mole fraction.The statistics used to characterize each distribution were the weight average values (SCBw, MSLw and Lw),the standard deviation (0) and cofficient of variation (CV).Results and DiscussionFigure 1 shows DSC heating curves obtained for LDPEs after normal cooling (10 K/min), very slow continuouscooling (0.08 K/min), and thermal fractionation (4 K steps with 50 min isothermal). The melting temperatures(Tm, 0.08) obtained after slow continuous cooling were higher than those obtained with normal cooling and theextent of shift of melting temperature was different for each LLDPE. The melting curves after fast coolingshowed final melting at a much lower temperature for the ZN polymers, and in the case of the solution Cg-BPE-ZNsoln the higher temperature melting region exhibited a double melting endotherm, indicative of crystalrearrangement. The fast cooling is distinguished from the very slow cooling crystallization curves. Since thecooling rate used for slow cooling is 0.08 K/min, this slow continuous cooling method allows the system to becloser to its equilibrium state, and therefore polymer crystals approaching equilibrium are expected to form as inthe thermal fractionation methodl'4. These highly ordered crystals gave a higher melting termperature and showedno tendency to rearrange during heating, prior to melting. The thermal fractionation results provide avisualization of differing branch content fractions, and the distribution can be analyzed from the discretepeakshs. The discrete peaks provide a means of observing if rearrangement of crystals takes place, because ifrearrangement occurs the peaks tend to merge. The loss of peak resolution can be readily observed. The meltingcurves after slow cooling provided a smooth distribution curve that can be analyzed in the same way as a molarmass distribution curel6.17.C6-BPE-ZNgas2- C6-BPE-ZNsoln0t0叶840 6080100 120 14010 100 120 140T(C)14s 12- C-BPE-M> 12- C.-BPE-S108tE 8+营54+24050 80 100 1201402040 6080 100120 140T('C)Fig.1 DSC specific heat curves for the polyethylenes after the three crystallization methodsThe thinner lines: cooling at a rate of 10 K/min; the normal lines: stepwise isothermalcooling; the thicker lines: cooling at a rate of 0.08 K/minThe degree of branching (B) and crystallinity (X) were calculated according to the equations, derived fromTREF results'", and the SCB distributions obtained after中国煤化工are presented in Fig. 2.YHCNMHGBranching Distribution in PE235The average SCB values are also obtained from the curves, and listed in Table 2. Results for single site catalyzedpolyethylenes having lower melting range are not shown here. Over the lower melting range, the calculationsprovide negative crystallinity data because of the lower starting and equilibration temperature of the DSCcompared with TREF. This is a problem arising from the calibration, not from the DSC results. Another singlesite catalyzed polyethylene that we have studied has a peak melting temperature of 60C and very lowcrytallinity of about 12%. When slowly cooled to -50C the polyethylene continues to crystallize giving a finalcrytallinity of about 25%. Upon heating the polyethylene started to melt at about -20C. Such polymers are notsuited to the present branching distribution analysis by our slow continuous cooling method. A broadertemperature range calibration is required where Eqs. (4), (5), and (6) are modifed to allow for crystallinity ofhighly branched polyethylenes that crystallize or continue to crystallize at sub-ambient temperatures.12C6-BPE-ZNgas1.1.0Cr-BPE-ZCNsoln0.0.8-0.6司0.4.2-:0.:0120300506010 2030405060SCBFig. 2 SCB distributions for the Zeigler-Natta catalyzed gas phase and solution polyethylenesafter slow continuous coolingTable 2. The SCB, L, MSL and their distribution for the ethylene copolymersobtained by slow continuous coolingProperties VLDPE2Cg-BPE-ZNgasCq-BPE-ZNsolnTc, 0.o8('C)121.5116.7Tmo.ogP(C)126.7112.7, 124.1 .SCB.(/1000C)18.014.3Osca'14.17.7CVscB"0.780.540.9640.971T"(C)139.3140.7Lw (nm)12.99.8L (nm)6.63.9CVr°0.510.40[D (nm)20.69.3, 15.7MSLw118.991.5OMSL88.454.5CVMsL0.740.60A σ= standard deviation; b CV = cofficient of variationBranching density is expressed as the number of short chain branches per 1000 backbone carbon atoms.However, another way of expressing the data is into convert it into the methylene sequence length betweenbranches. Crystallinity arises when the unbranched chain segments joi^中国煤化itstaliationTYHCNMHG236R Shanks et al.process, the branches are excluded from the crystals and the methylene sequence length will limit the lamellarthickness. Thus, MSL provides a number that reflects the ability of segments to crystallize. The compositiondistributions of the ethylene copolymers expressed in MSL are shown in Fig. 3.0.4 (0.C6-BPE-ZNgasCg-BPE-ZNsoln” 0.0.2. 0.20.1 t立0.10.010020030000400MSLFig.3 MSL distributions for the Zeigler-Natta catalyzed gas phase andsolution polyethylenes after slow continuous coolingThe lamellar thickness of each LLDPE was calculated using the Thomson-Gibbs equation (Eq. 7), and theaverage lamellar thickness for each LLDPE was obtained from integration of the curves. Other researchers haveused DSC data for calculation of lamella thicknessll, and the use of Tm° has been recommendedl2, 13. Thecurves were required to be integrated because the linear scale of temperature became not-linear after conversionto lamella thickness. Integration provided a linear scale that could be used to calculate averages and standarddeviations. In the analysis, the thermodynamic copolymer melting temperature, Tm, was chosen as the limitingtemperature for each of the polymers. This temperature is preferred to the equilibrium melting temperature (Tm)for branched copolymers since the branches provide an absolute limitation on the melting temperature that can bereached by equilibration. The corresponding lamellar thickmess ditributions are shown in Fig. 4.0.6 |0.6会民0.40.4自0.250.2-0.0 5101520253055202530Lamellar thicknessLamellar thickness (nm)Fig.4 Lamella thickness distributions for the Zeigler Natta catalyzed gas phase andFigure 5 shows the SCB distribution curves obtained after thermal fractionation. The distributions arecomposed of discrete fractions where each peak in the DSC curve has been converted into a bar of the bar graphby converting the peak temperature into SCB content and converting the peak area into total polyethylene bycorrection for the crystallinity for that level of branching. If a smooth curve were drawn through the tops of thebars, a graph the same as Fig. 2 would be obtained. Similarly, if the steps in the thermal fractionation procedurewere decreased the discrete distribution graph would approach the continuous distribution. Figure 6 shows theanalogous discrete distribution of the ethylene copolymers for MSL. Figure 7 shows the discrete distribution ofthe ethylene copolymers for lamellar thickness. These dii1 be analyzed to provideaverage parameters, though with lower resolution than thos(中国煤化工; curves.TYHCNMHGBranching Distribution in PE2373030 rC6-BPE-ZNgasCg-BPE-ZNsoln20善102台10个46101315182022252729314810121415171921 23SCBFig. 5 SCB discrete distributions for the Zzigler-Natta catalyzed gas phase andsolution polyethylenes after thermal fractionationC6-DPE-ZNgasCq-BPE-ZNsoln夏叉i 2(出台1(37.8 44.5 54.1 68.2 92.9 1704246515663718298120202MSLFig.6 MSL discrete distributions for the Zeigler-Natta catalyzed gas phase and solution polyethylenesafter thermal fractionation40 r10 rCG-BPE-ZNgasC-BPE-ZNsoln30 tn20t0t)1115211116Lamellar thickness (nm)Larmellar thickness (nm)Fig.7 Lamella discrete thickness distributions for the Zeigler-Natta catalyzed gas phase andTable 2 shows the analyses of the continuous distribution curves. The peak crysallization and meltingtemperatures from the DSC scans after slow continuous cooling are listed first. The SCB weight average (SCB.)is 18.0 with a dispersity of 14.1 for C6-BPE-ZNgas and 14.3 and中国煤化工average shortYHCNMHG238R Shanks et al.branching distribution of the Cg-BPE-ZNsoln was found to be similar to value of 13.8土2.1 as reported byStarcks9. Tm were 139.3 and 140.7C for C-BPE-2Ngas and Cg-BPE-Znsoln, respectively. The weight averagelamella thickness in each case is similar to values of 13.0 and 11.6 nm for two LLDPEs by TEM studiest1 and12.9 and 8.6 nm from AFM[19!. Table 3 shows the analysis from the discrete distribution curves. The results aresimilar to those from the continuous curves, but the averages are based on a smaller number of data thanavailable from the continuous curves.Table 3. The SCB, L, MSL and their distribution for the ethylene copolymersobtained by thermal fractionationPProperties VLDPE2C6-BPE-ZNgasCg-BPE-ZNsolnSCBw(1/1000C)12.611.9Osca8.7CVscB'0.690.50Lw (nm)13.610.5o6.14.3CVb0.450.40MSLw138.5105.9OysL'82.469.4CVMsu"0.590.65# σ= standard deviation; b CV = cofficient of variationCONCLUSIONSDSC melting curves have been used to obtain short chain branching distributions using data from TREF forcalibration of the branch content and crystallinity with melting temperature. The results from continuous slowcooling and thermal fractionation techniques are similar. The distributions were also represented as meanmethylene sequence length and lamellar thickness distributions. Each of the distributions was used to providetypical averages, analogous to those obtained from size exclusion chromatography for molar mass. The weightaverage short chain branching, mean methylene sequence length between branches and lamellar thickness wereobtained, with the standard deviation of each used as a dispersity index. The assumptions made for each of thecalculations were discussed. The thermal fractionation data has been tested to ensure that recrystallization orrearrangment of crystals does not occur after the crystallization conditions are applied. A limitation was that themethod was found to be suitable only for polyethylenes that start to melt sufficiently above ambient temperaturethat the DSC data can be correlated with TREF data.ACKNOWLEDGEMENTS Fei Chen would like to acknowledge support of a Postgraduate Research Scholarship fromRMIT University.REFERENCES1 Adams, JL, Foster, G.N, Rastogi, S.R., Vogel, R.H. and Wasserman, S.H, Polym. Preprints, 1998, 39: 1902 Zhou, x. and Hay, J.N, Eur. Polym. J, 1993, 29: 2913 Wild, L, Adv. Polym. Sci, 1991, 98: 1Shanks, R.A. and Amarasinghe, G., J. Thermal Anal. Cal, 2000, 59: 4715 Starck, P, Polym. Intern, 1996, 40: 111Amal, M.L, Sanchez, JJ. and Maller, A.J, ANTEC Proceed,中国煤化工MHCNMHGBranching Distribution in PE2397 Hosoda, S., Kojima, K. and Furuta, M, Macromol Chem, 1986, 187: 15018 Psarski, M., Piorkowska, E. and Galeski, A, Macromolecules, 2000, 33: 9169 Zhou, H. and Wilkes, G.L, Polymer, 1997, 38: 573510 Mathot, V.B.F,“The crytallization and melting region", in“Calorimetry and Thermal Analysis of Polymers", ed. byMathot, V.B.F, Hanser, Munich, 1993, p.23111 Hosoda, S, Polym. J,1998, 20: 38312 Crist, B. and Claudio, E.S, Macromolecules, 1999, 32: 894513 Crist, B. and Howard, P.R., Macromolecules, 1999, 32: 305714 Keating, M.Y, Lee, I.h. and Wong, C.S, Thermochim Acta, 1996, 284: 4715 Zhang, F, Fu, Q, Lu, T, Huang, H and He, T, Polymer, 2002, 42: 103116 Zhang, F,, Liu, J., Fu, Q., Huang, H, Hu, z, Yao, s, Cai, X. and He, T, J. Polym Sci: B: Polym Phys, 2002, 40: 81317 Zhang, F, Liu, J, Xie, F, Fu, Q. and He, T, J. Polym Sci: B: Polym Phys, 2002, 40; 82218 Xu, J, Xu, X. and Feng, L, Eur. Polym. J, 1999, 36: 68519 Liu, J, Zhang, F., Xie, F, Du, B., Fu, Q. and He, T., Polymer, 2001, 42: 5449中国煤化工MYHCNMHG