1-octene copolymers synthesized from living coordination polymerization

European Polymer Journal 54 (2014) 160–171

Contents lists available at ScienceDirect

European Polymer Journal journal homepage: www.elsevier.com/locate/europolj

Structure analysis of ethylene/1-octene copolymers synthesized from living coordination polymerization Weifeng Liu a, Wen-Jun Wang a, Hong Fan a, Luqiang Yu b, Bo-Geng Li a,⇑, Shiping Zhu c,⇑ a

State Key Laboratory of Chemical Engineering, Department of Chemical & Biological Engineering, Zhejiang University, Hangzhou 310027, PR China SINOPEC Beijing Research Institute of Chemical Industry, Beijing 100013, PR China c Department of Chemical Engineering, McMaster University, Hamilton, Ontario L8S 4L7, Canada b

a r t i c l e

i n f o

Article history: Received 28 January 2014 Received in revised form 4 March 2014 Accepted 11 March 2014 Available online 17 March 2014 Keywords: Ethylene/1-octene copolymer Living coordination polymerization Composition distribution Olefin block copolymers

a b s t r a c t A series of ethylene/1-octene random and block copolymer samples were synthesized via a living coordination polymerization catalyzed by bis[N-(3-methylsalicylidene)-2,3,4,5,6pentafluoroanilinato] titanium(IV) dichloride/dMAO. The chain microstructures of the copolymers were elucidated by analytic TREF and DSC thermal fractionation techniques in this work. It was found that the living random copolymers possessed narrower intrachain composition distributions than those prepared from conventional metallocene catalysts. With the same short chain branching content, the melting temperature of living random copolymer was about 16 °C lower than metallocene-catalyzed random copolymer. The living block copolymers were also studied in comparison with the olefin multiblock copolymer (OBC) produced from chain shuttling polymerization. The interchain composition distribution of OBC was found broader than living block copolymers. However, the crystal thickness distribution of OBC was narrower. The diblock copolymer had a bimodal distribution in the crystal thickness and the step triblock copolymer showed a trimodal distribution, in contrast to OBC’s single modal distribution. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Polyethylene is the most widely used commodity polymer. Depending on the short chain branching (SCB) distribution, its property varies from plastic to elastomer [1]. It is well known that the chain microstructure of polyolefin is determined by the catalyst [2]. Commercial ethylene/ a-olefin copolymers are mainly produced by Ziegler–Natta or metallocene catalysts. Most of them have a random chain microstructure. Copolymers by traditional Z–N catalysts (e.g. LLDPE) exhibit broad molecular weight and composition distributions [3]. Metallocene products (e.g. ⇑ Corresponding authors. Tel.: +86 571 8795 2483; fax: +86 571 8795 1612 (B.-G. Li). E-mail addresses: [email protected] (B.-G. Li), [email protected] (S. Zhu). http://dx.doi.org/10.1016/j.eurpolymj.2014.03.010 0014-3057/Ó 2014 Elsevier Ltd. All rights reserved.

m-LLDPE, POE) have much narrower SCB distribution than Z–N’s [4], due to single-site nature of the former. Recently, ethylene/1-octene multiblock copolymer (OBC) has been commercialized using a chain shuttling polymerization technique [5,6]. Chain shuttling polymerization employs two post-metallocene catalysts. One catalyst, good for ethylene/a-olefin copolymerization produces soft amorphous segments. The other, poor in copolymerization, produces hard semi-crystalline segments. The hard and soft segments are alternatively connected through the function of an appropriate chain shuttling agent. Both soft and hard segments are random copolymers. The number and length of soft and hard segments are statistical. This type of multiblock chain structures gives the materials better elasticity at high temperature than random copolymers [7]. Nowadays, ethylene/a-olefin copolymers become increasingly important in high-valued

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

applications. For example, POE and OBC have become major materials for polyolefin-based thermoplastic elastomers [8]. Remarkable progress has been made in the areas of living coordination polymerization over the past decades [9,10]. As chain termination and transfer reactions are negligible, the chain microstructure of polyolefin could be precisely designed and controlled via living polymerization [11–14]. In living ethylene/a-olefin copolymerization, the comonomer units are incorporated into chains in a living manner, polyethylenes with novel and complex architectures could thus been synthesized [15–17]. This level of richness and controllability in chain microstructure was not possible with traditional Z–N or metallocene catalysts. There is no doubt that living polymerization technique provides a powerful tool for tailor-making polyolefin materials. Structure analysis of the ethylene/a-olefin copolymers thus becomes most desirable. It is of practical importance for designing new polyolefin materials, though challenging due to structural complexity. The composition distribution of ethylene/a-olefin copolymers can be characterized by two types of established methods. One is based on chromatographic separation of polymer chains with different crystallizabilities from a dilute solution, such as temperature rising elution fraction (TREF) technique, crystallization analysis fraction (CRYSTAF) technique [18,19] and the newly developed crystallization elution fraction (CEF) technique [20,21]. All these techniques could provide comparable results [18–21]. TREF is more time-consuming due to two fractionation steps, crystallization and elution. CRYSTAF was found to be more sensitive to crystallization kinetics and cocrystallization effects, although it takes only one crystallization step. CEF process is similar to TREF but time saving. It bears less cocrystallization effect comparing with CRYSTAF, because of the superior dynamic crystallization. All these techniques need toxic solvents and expensive instruments. As polymer chains are physically separated according to the comonomer incorporation, polymer interchain heterogeneity can be fully analyzed. The other type of methods is based on DSC thermal fractionation, including step crystallization (SC) and successive self-nucleation and annealing (SSA) techniques [22]. In the DSC thermal fractionation, the polymer sample is fractionated from melt by a carefully designed temperature program. Polymer chains having different SCB densities form crystals or lamellas of different sizes. Subsequent heating of the fractionated sample gives a multimodal DSC melting curve, which can be correlated to its composition distribution. Because crystallization is sensitive to crystalline methylene sequence length (CMSL) and polymer chains are not physically separated, polymer intrachain sequence heterogeneity can also result in molecular segregation. The DSC thermal fractionation is reflective of both interchain and intrachain heterogeneities. In comparison, SSA has better resolution and takes shorter time than SC [23,24]. In general, the interchain and intrachain heterogeneities of olefin copolymers can be characterized by TREF/ SC and/or TREF/SSA cross-fractionation. These methods are widely used in the characterization of commercial eth-

161

ylene/a-olefin copolymer products [25–28]. However, it has not been applied to olefin copolymers produced from living coordination polymerization. The structural differences between the copolymers prepared by living and nonliving polymerization are yet to be elucidated. This work reports a study on the composition distribution of ethylene/1-octene random and block copolymers synthesized by living coordination polymerization. The interchain heterogeneities of all samples analyzed in this work were determined by the analytic TREF analysis. The SSA thermal fractionation technique was adopted for characterization of the intrachain heterogeneity. The intrachain heterogeneities of the living random and block copolymers were respectively compared to their commercial ethylene/ 1-octene random and block counterparts.

2. Materials and methods In this work, the ethylene/1-octene copolymers were synthesized from bis[N-(3-methylsalicylidene)-2,3,4,5,6pentafluoroanilinato] titanium(IV) dichloride/dMAO (fluorinated FI-Ti) [29], which is a famous living catalyst system. The synthesis procedures for the catalyst and the copolymers were described in our previous work [30]. In the synthesis of copolymers used in this work, the 1-octene conversion was controlled below 3% to avoid the copolymer composition drifting. The basic structure characteristics of the samples are listed in Table 1. LBC represents living block copolymer. In particular, LBC1 is a living diblock copolymer and LBC2 is a living step-triblock copolymer. The detailed specifications of LBC1 and LBC2 could be found in our previous work [31]. The commercial available products, Engage 8150 and Infuse 9000, are designated as MRC (metallocene-catalyzed random copolymer) and OBC (olefin multiblock copolymer), respectively. Engage 8150 is an ethylene/1-octene random copolymer produced by Dow Chemical’s constrained geometry catalyst (CGC) technique. Infuse 9000 is an ethylene/1-octene multiblock copolymer with statistical block length distribution, produced by Dow Chemical’s wellknown chain shuttling polymerization technique [5,6]. High temperature gel permeation chromatography (PLGPC 220 system) was used for the determination of molecular weights (Mn and Mw) and polydispersity index (PDI). The measurement was done at 150 °C, using 1,2,4-trichlorobenzene as solvent flowing at the rate of 1.0 ml/min. Universal calibration was performed on the monodisperse polystyrene (PS) standards. The Mark–Houwink constants of PS are K = 5.91  104 and a = 0.69, and those of PE are K = 1.21  104 and a = 0.707. The comonomer incorporation was determined by the high temperature 13C NMR spectra with deuterated odichlorobenzene (o-DCB) as solvent. Polymer solutions (10 wt%) were scanned at 125 °C using a Bruker AC 400 pulsed NMR spectrometer with a spectral width of 8000 Hz, pulse delay time of 8 s, acquisition time of 1.3 s and pulse angle of 90°. At least 5000 scans were required for a good signal to noise ratio. The ASTM D5017-96 method was employed for the carbon assignments and composition calculation [32].

162

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

Table 1 Characteristics of the studied copolymer samples. Sample namea

Mol%b Comon.

Mwc (104)

PDIc

Tgd (°C)

Tcd (°C)

Tmd (°C)

D Hm d (J/g)

XC,DSCe (wt%)

t1/2f (min)

MRC LRC1 LRC2i

13.3 11.0 13.2

12.13 38.71 28.51

2.22 1.19 1.15

55.2 54.6 55.6

35.2 31.6 22.4

50.9 49.8 41.3

40 39 28

14 13 10

3.0 2.2 2.1

– – –

– – –

– – –

OBC LBC1j LBC2j

8.7 8.8 8.3

13.20 54.45 64.28

2.20 1.25 1.29

62.5 53.7 53.1

97.3 19.6, 99.2 19.8, 61.6, 99.3

119.8 39.9, 113.9 39.3, 76.1, 113.6

50 73 87

17 25 30

1.1 5.5 4.4

16.4 14.0 13.0

12.2 9.3 8.9

26 17 24

L(110)g (nm)

L(200)g (nm)

XC,WAXDh (%)

a MRC: metallocene-catalyzed random copolymer, LRC: living random copolymer, OBC: olefin multiblock copolymer from chain shuttling polymerization technique, LBC: living block copolymer. b Molar fraction of 1-octene in copolymer, determined by NMR. c Determined by GPC. d Determined by DSC. e Crystallinity from heat of melting, X C;DSC ¼ DHm =DH0m , taking DH0m of 293 J/g for the polyethylene crystal. f The time of 50% crystallinity. g The average crystal sizes perpendicular to the crystallographic planes of (1 1 0) and (2 0 0), estimated from WAXD. h Relative crystallinity index from WAXD. i Data cited from Ref. [30]. j Data cited from Ref. [31].

the minimum Ts temperature, the sample was heated from 50 to 160 °C at 10 °C/min. The procedures applied are shown in Fig. 1. The corresponding last melting curve from SSA method was recorded. Flat samples with 1.5 mm thickness and 20 mm diameter for Wide-angle X-ray diffractograms (WXRD) were injected by Haake MiniJet system. Samples were scanned at room temperature using Empyrean 200895 diffractometer system with a scanning increment of 0.02°. The anode material is Cu and the Ka wavelength is 1.54 Å. 3. Results and discussion 3.1. Random copolymers As shown in Table 1, the comonomer incorporation of LRC1 was about 2 mol% smaller than that of MRC (11.0 vs. 13.3%), but the melting temperatures were almost the same (49.8 vs. 50.9 °C). For LRC2 and MRC, the comonomer incorporations were nearly the same (13.2 vs. 13.3%).

200 Temperature

Temperature (°C)

All the samples were submitted to the analytic TREF analysis by Polymer Char TREF 300. The sample was dissolved in o-DCB as 20 wt% polymer solution at 150 °C for 60 min. The solution was stabilized at 95 °C for 45 min. Crystallization was achieved by cooling the solution to room temperature at a rate of 0.1 °C/min. The precipitated polymer was finally eluted at a heating rate of 1 °C/min. The pump flow rate was 0.5 ml/min. DSC analysis was carried out using a TA Q200 thermal analyzer at a heating/cooling rate of 10 °C/min. Sample between 5.0 and 7.0 mg was first heated to 160 °C, maintained at 160 °C for 5 min to remove the thermal history. Recrystallization was then accomplished by cooling to 90 °C. The sample was reheated to 160 °C after isotherm at 90 °C for 3 min. The cooling curve was recorded for the crystallization temperature (Tc) and the second heating curve was recorded for the glass transition temperature (Tg) and peak melting temperature (Tm). DSC thermal fractionation was also done by the TA Q200 thermal analyzer. In this work, SSA was employed as the DSC thermal fractionation technique. Samples for SSA were weighed in the range of 3–5 mg. SSA temperature program was set as follows: the sample was first heated to 160 °C, maintained at 160 °C for 5 min, then cooled to 50 °C at 10 °C/min to create the initial ‘‘standard’’ state. The sample was then heated to the first self-seeding temperature (Ts) at 10 °C/min. Self-nucleation and annealing was achieved by holding isothermal at Ts for 5 min. The sample was subsequently cooled to 50 °C after the selfnucleation. Successive self-nucleation and annealing was achieved by repeatedly melting and crystallization on decreasing Ts. For MRC, LRC1 and LRC2, the first Ts was set at 87.5 °C with the fractionation window of 7.5 °C. For OBC, LBC1 and LBC2, the first Ts was set at 130 °C; the fractionation window was first chosen 5 °C when Ts decreased from 130 to 80 °C, then changed to 7.5 °C when Ts was lower than 80 °C. For all samples, the minimum Ts temperature was 5 °C and the annealing time was 5 min, with the same scanning rate of 10 °C/min. After final crystallization from

150 100 50 0 -50 0

100

200

300

400

500

600

700

Time (min) Fig. 1. Scheme of analysis procedure for the SSA method used for block copolymers.

163

W. Liu et al. / European Polymer Journal 54 (2014) 160–171 + +

4B6,δ δ +

ααγ αα

γδ

CH CHβ

(a) +

+

5B6,βδ 2B6

ββ

MRC

1B6

Endo Up

6B6,αδ 3B6

LRC2

CHββ

LRC1 LRC2

LRC1

MRC -50 45

40

35

30

25

20

15

0

10

50

100

150

Temperature (°C)

ppm Fig. 2. 13C NMR spectra of random copolymers measured at 125 °C using deuterated o-dichlorobenzene as solvent, from 10 to 45 ppm.

(b)

LRC1

LRC2

-50

0

50

100

150

Temperature (°C) 100

Crystallinity (%)

However, the melting temperature of LRC2 was about 10 °C lower than that of MRC (41.3 vs. 50.9 °C). This could be more clearly demonstrated in Figs. 2 and 3a. Fig. 3b shows that the crystallization temperature of LRC2 was much lower than that of MRC. A broad shoulder peak was found in the cooling curve of MRC, but it was not evident in the cooling curves of LRC1 and LRC2. Interestingly, under the same cooling rate, the crystallization rate of LRC1 and LRC2 seemed a little faster than that of MRC, as shown in Fig. 3c. Note that MRC had the highest crystallinity (14%). All the three samples (MRC, LRC1, LRC2) are random ethylene/1-octene copolymers. The distinctive thermal features must be induced by their microstructure differences. Table 2 provides the sequence fractions of the ethylene/ 1-octene copolymers obtained from the 13C NMR spectra. There were no obvious sequence differences detected from this result. The NMR method only gave the information about their average compositions and the measurable sequence length was limited. To disclose the molecular structure, analytic TREF is another important tool. Fig. 4 shows the TREF curves of MRC, LRC1 and LRC2. Only one well-resolved peak was observed in all curves. All the peaks located in the same temperature range (<30 °C) and had the same peak temperature (27.5 °C), indicating the same interchain composition distribution for these random copolymers. The structure differences of these random copolymers could not be distinguished by TREF method. The DSC thermal fractionation technique was then applied. Fig. 5 shows DSC heating curves of the ethylene/1octene random copolymers after SSA fractionation. There were nine melting peaks in the curve of MRC. In LRC1, the melting peak at around 70 °C disappeared. In LRC2, the melting peaks at around 63 and 70 °C disappeared. The difference in structure was qualitatively indicated by this variation in the number and intensity of melting peaks. Note that MRC and LRC2 had the same average comonomer incorporation, and all these random copolymers were produced from single-site catalyst system and had the same interchain composition distribution as shown in Fig. 4. It has been demonstrated that the

Endo Up

MRC

(c)

80

LRC1 LRC2 MRC

60 40 20 t1/2

0 0

2

4

6

8

10

Time (min) Fig. 3. DSC curves of ethylene/1-octene random copolymers. (a) Melting curves; (b) cooling curves; and (c) crystallinity vs. time curves.

molecular weight and molecular weight distribution had little influence on the number of peaks in the endothermic DSC curves [33,34]. Therefore, the variation in the number and intensity of melting peaks must be induced by the difference of the crystalline sequence length distribution (intrachain heterogeneity). The DSC thermal fractionation method revealed that the intrachain composition distribution of LRC was narrower than MRC. Hosoda [3] proposed a linear relationship to correlate the melting temperature of each peak to the degree of SCB for ethylene/1-octene copolymers. However, the copolymer samples used by the authors were produced

164

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

Table 2 The triad and diad distributions of ethylene/1-octene copolymers obtained by

13

C NMR.

Sample

EEE (%)

EEO + OEE (%)

OEO (%)

EOE (%)

EOO + OOE (%)

OOO (%)

EE (%)

EO + OE (%)

OO (%)

MRC LRC1 LRC2 OBC LBC1 LBC2

67.5 67.9 65.8 73.6 78.9 81.1

16.4 19.6 19.1 16.7 13.0 10.8

1.24 1.22 1.58 0.889 0.771 1.66

9.88 8.44 10.2 8.65 5.69 4.81

4.94 2.78 3.30 0.162 1.69 1.61

0 0 0 0 0 0

75.7 77.8 75.4 82.0 85.4 86.5

21.8 20.9 23.0 17.9 13.8 12.7

2.47 1.39 1.65 0.081 0.844 0.806

140

MRC LRC1 LRC2

Living Random Copolymer Linear Fit Metallocene Random Copolymer Linear Fit

120

Tm (°C)

100 Tm = -1.79SCB + 144

80 Tm = -1.78SCB + 128

60 40

20

40

60

80

100

20

120

Temperature (ºC) Fig. 4. Analytical TREF curves of ethylene/1-octene random copolymers, (a) MRC: 1-octene = 13.3 mol%; (b) LRC1: 1-octene = 11.0 mol%; and (c) LRC2: 1-octene = 13.2 mol%.

0

10

20

30

SCB

40

50

60

Fig. 6. Relationships between melting temperature and short chain branching content for ethylene/1-octene random copolymers produced by FI catalyst and CGC catalyst, respectively. Data for living random copolymers were cited from our previous work (Ref. [30]), data for metallocene-catalyzed random copolymers were from the product information of DOW ENGAGE™ Polyolefin Elastomers (POE), both data were measured in the same way.

Endo Up

LRC2

T m ¼ 1:78SCB þ 128

ð1Þ

For metallocene-catalyzed random copolymer:

T m ¼ 1:79SCB þ 144

LRC1

MRC

-25

0

25

50

75

100

Temperature (°C) Fig. 5. DSC heating curves of ethylene/1-octene random copolymers after SSA.

by the supported TiCl4 and triethyl-aluminum. The comonomer incorporation was low and the melting temperature was in the range of 90–130 °C. In comparison, the DSC melting peaks in Fig. 5 are below 75 °C. The equation obtained by Hosoda is not suitable for extrapolation. In this work, the melting temperature of ethylene/1-octene copolymer was linearly correlated to the SCB content in a broader temperature range, as shown in Fig. 6. The data for the living random copolymers and the metallocene-catalyzed random copolymers were fitted, respectively. The fitted equations are as follows: For living random copolymers:

ð2Þ

The SCB content of each fraction in Fig. 4 was calculated using these two equations. It was interesting that the slope for both types of ethylene/1-octene random copolymers was the same, while the intercepts were different. With the same SCB content, LRC exhibited lower melting temperature (about 16 °C lower) than MRC. In other words, LRC required a lower SCB content (about 9 branches/ 1000 C less) to obtain the same melting temperature. This phenomenon could suggest that, in the living coordination copolymerization, the comonomer units incorporated into the copolymer chains more uniformly distributed than in metallocene-catalyzed copolymerization. Scheme 1 illustrates this structure difference. As the melting temperature of each fraction corresponds to a special group of crystallizable chain segments, the crystal thickness and crystalline methylene sequence length (CMSL) could also be revealed from the melting temperature. The crystal thickness was evaluated by Thomson–Gibbs equation [3]:

l¼

2rT 0m

DHm ðT 0m  T m Þ

ð3Þ

165

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

MRC

LRC

1-Octene

Ethylene

Scheme 1. Illustration of structure difference between MRC and LRC. MRC: metallocene catalyzed random copolymers, LRC: living random copolymers.

where Tm is the melting temperature of each fraction in the SSA heating curve, T 0m is the equilibrium melting temperature for infinitely thick polyethylene lamella (418.5 K), r is the lateral surface free energy of polyethylene lamellar (90  103 J/m2) [35], DHm is the heat of fusion for infinitely thick polyethylene lamella (293  106 J/m3). The CMSL was calculated using Keating’s method [4,36]:

lnðXÞ ¼ 0:331 þ 135:5=T m

ð4Þ

CMSL ¼ 0:2534X=ð1  XÞ

ð5Þ

where X is the molar fraction of CH2. MRC LRC1 LRC2 MRC LRC1 LRC2

60

50

40

30

50

20

40 30 25

30

35

2

Fit y = A0 + A1*x + A2*x + A3*x 3 + A4*x 4 + A5*x 5 A0 96.24404 A1 -4.83606 A2 0.10406 A3 -0.00113 A4 6.1616E-6 A5 -1.33832E-8

9

10 20

293/ Hm Poly fit of 293/ Hm

12

293/ Hm

70

CMSL (Angstrom)

SCB (CH3/1000C)

80

Fig. 7 shows these calculation results. The dependence of crystal thickness on the SCB content was similar for both types of random copolymers. However, with the same lamellar thickness, LRC had fewer SCB content than MRC. As mentioned before, this was attributed to the more homogeneous distribution of SCB in the LRC chain than that in the MRC chain. The dependence of crystal thickness with CMSL exhibited a linear relationship for both types of copolymers. This was because, under such high SCB density, the crystalline methylene sequence was not long enough to fold and only formed fringed micelles or bundled crystals [37]. Wide-angle X-ray diffractograms (WXRD) also shows there was no obvious highly-ordered orthorhombic crystalline phase in these ethylene/1-octene random copolymers, as shown in Fig. 8. Experimental comparison of the SCB distributions between LRC and MRC has not been reported before. In order

40

Crystal thickness (Angstrom) Fig. 7. Variation of short chain branching (SCB) and crystalline methylene sequence length (CMSL) with crystal thickness for the ethylene/1-octene random copolymers.

6

3

0 40

60

80

100

120

140

Tm (°C)

LRC2 LRC1 MRC

Intensity (a.u.)

Fig. 9. The temperature dependence of 293/DHm. Data used were cited from Ref. [30].

Table 3 Polydispersity parameters of the ethylene/1-octene copolymers.

5

10

15

20

25

30

35

2θ (degree) Fig. 8. Wide-angle X-ray diffractograms of ethylene/1-octene random copolymers.

Sample name

Lamellar thickness Ln (nm)

Lw (nm)

I

Ln (nm)

CMSL Lw (nm)

I

MRC LRC1 LRC2 OBC LBC1 LBC2

2.3 2.3 2.2 9.3 4.2 4.2

2.4 2.3 2.2 9.9 5.7 5.5

1.03 1.00 1.00 1.06 1.36 1.31

2.3 2.2 2.1 29.3 7.1 6.9

2.5 2.3 2.2 47.0 13.4 12.3

1.09 1.05 1.05 1.39 1.88 1.78

166

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

(a)

MRC LRC1

Fraction

0.20

Lw ¼

0.15

I¼

0.10

0.05

0.00 15

20

25

30

35

40

Crystal thickness (Angstrom) 0.30

MRC LRC2

(b)

Fraction

0.24 0.18 0.12 0.06 0.00 15

20

25

30

35

40

Crystal thickness (Angstrom) Fig. 10. Comparison of crystal thickness distribution for (a) MRC and LRC1, (b) MRC and LRC2.

to further elucidate the differences of intrachain heterogeneities between LRC and MRC, it becomes essential to estimate the crystal thickness distribution. Similar mathematical expressions used for the molecular weight distribution were introduced to describe the sequence length distribution [4]:

Ln ¼

n1 L1 þ n2 L2 þ    þ ni Li X ¼ fi Li n1 þ n2 þ    þ ni

ð6Þ

P 2 n1 L21 þ n2 L22 þ    þ ni L2i fi L ¼P i fi Li n1 L1 þ n2 L2 þ    þ ni Li

Lw Ln

ð7Þ

ð8Þ

where ni is the normalized partial area of fraction i in the final SSA melting curve, Li is the crystal thickness or CMSL. Some authors have pointed out that the peak area of each fraction cannot be directly converted into the weight percent, due to the temperature dependence of specific heat fusion [27]. We used the ratio of the heat fusion of perfect crystal (DHf = 293 J/g) to the fusion of each fraction (DHm(T)) as the correction factor [28]. Fig. 9 shows the temperature dependence of 293/DHm(T). The weight percent of each fraction was evaluated by multiplying the corresponding peak area with 293/DHm(T). The calculation results for the three random copolymers were included in Table 3. For all the three random copolymers, the values of crystal thickness were nearly the same to those of CMSL, as there was no chain folding. This also confirmed that only fringed micelles or bundled crystals existed in these random copolymers. The values of Lw and polydispersity index for LRC were smaller than MRC, as there was more short crystalline methylene sequence in LRC. Fig. 10 clearly shows that LRC had narrower distribution of crystal thickness than MRC, demonstrating the more homogeneous distribution of SCB in LRC than in MRC. This might also account for the faster crystallization rate of LRC than MRC, as shown in Fig. 3c. It would be beneficial to LRC in the area of fast processing applications with less product warp distortion. 4. Block copolymers The thermal properties of our living block copolymer samples were compared with a commercial OBC product, which was produced from chain shuttling polymerization. Their structure differences are illustrated in Scheme 2. The average comonomer incorporations were almost the same.

OBC Total Mw = 132 kg/mol Total Average 1-Octene Content = 8.7 mol% LBC1 Mw1 = 196 kg/mol Mw2 = 349 kg/mol 1-Octene Content = 0.83 mol% 1-Octene Content = 16.4 mol% Total Average 1-Octene Content = 8.8 mol% LBC2 Mw1 = 196 kg/mol Mw3 = 226 kg/mol Mw2 = 221 kg/mol 1-Octene Content = 0.83 mol% 1-Octene Content = 6.0 mol% 1-Octene Content = 20.5 mol% Total Average 1-Octene Content = 8.3 mol% Scheme 2. Structure illustration for the block copolymers. OBC: olefin multiblock copolymer, LBC: living block copolymer.

167

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

However, the thermal behaviors were very different, as shown in Fig. 11. LBC1 and LBC2 showed bimodal and trimodal melting behaviors respectively, as each block could crystallize [31]. OBC had a single melting peak, as the comonomer incorporation in the soft blocks was higher than 20 mol% and thus only the hard blocks could crystallize [38–40]. The crystallinity of OBC was the smallest among the three block copolymers, while the order of crystallization rate was: OBC > LBC2 > LBC1. This was probably because that the molecular weight of each hard block in OBC was much lower than that in the LBC’s, and thus the effect of chain entanglement on crystallization was less for OBC.

OBC LBC1 LBC2

20

40

60

80

100

120

Temperature (ºC)

(a)

Fig. 12. Analytical TREF curves of ethylene/1-octene block copolymers. OBC: 1-octene = 8.7 mol%; LBC1: 1-octene = 8.8 mol%; LBC2: 1-octene = 8.3 mol%.

Endo Up

LBC2 LBC1

OBC -50

0

50

100

150

Temperature (°C)

LBC2

Endo Up

LBC1 OBC

(b) -50

0

50

100

150

Temperature (°C)

Crystallinity (%)

100

(c) OBC LBC1 LBC2

80 60 40 20 t1/2

0 0

3

6

9

12

15

18

time (min) Fig. 11. DSC curves of ethylene/1-octene block copolymers. (a) Melting curves; (b) cooling curves; and (c) crystallinity vs. time curves.

Fig. 12 shows the analytical TREF curves of OBC, LBC1 and LBC2. Each curve shows two peaks. The low temperature peak corresponds ethylene–octene random copolymer. The OBC sample contained about 21 wt% random copolymer, estimated from the weight fraction of low temperature peak in the TREF curve, which was much higher than LBCs (2–3 wt%), as shown in Table 4. For the high temperature peak, the peak distribution of OBC was much broader than LBCs. Considering the mechanism of chain shuttling polymerization [5,6], the commercial OBC sample must also contain some LLDPE or ‘‘free’’ hard blocks as impurities [41], resulting in broad distribution in the TREF curve at the temperature range of 80–100 °C. The LBC samples exhibited sharp narrow peak distribution with only a small fraction of random copolymer contamination. The analytical TREF results indicated that the interchain composition distribution of commercial OBC sample was much broader than LBCs. SSA analysis was also performed on these block copolymers to study their crystalline sequence length differences. Fig. 13 shows the DSC heating curves of the ethylene/ 1-octene block copolymers after SSA fractionation. Interestingly, the commercial OBC sample showed narrower crystal thickness distribution than LBCs. The calculation results of average crystal thickness and CMSL were given in Table 3. The Ln and Lw values of OBC were larger than those of LBCs, due to the longer average CMSL in the OBC sample. The OBC possessed much smaller polydispersity index value of the crystal thickness distribution. This was because the comonomer branches concentrated in the soft blocks in OBC (illustrated in Scheme 2) and only the hard blocks could crystallize and form crystals with similar thermal stability. Therefore, the crystal thickness of OBC showed a single modal distribution, as shown in Fig. 14a. However, the crystal thickness distribution of LBC1 was bimodal, having an inversed parabolic shape, shown in Fig. 14b. This was because not only the hard blocks formed lamellar crystals but also the soft blocks with about 16 mol% 1-octene incorporation crystallized into fringed micelles or bundled crystals. The crystal thickness distribution of

168

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

Table 4 Comparison of ethylene/1-octene block copolymers obtained from analytic TREF. Samples

OBC

LBC1

LBC2

Peak temperature range in TREF Peak temperatures in TREF Content of low temperature fraction from TREF (wt%)

T < 30 °C, 71 °C < T < 98.5 °C 27.5 °C, 91.4 °C 21.2%

T < 30 °C, 83 °C < T < 94 °C 27.5 °C, 88.7 °C 3.1%

T < 30 °C, 81.5 °C < T < 94.5 °C 27.5 °C, 89.4 °C 1.9%

0.35

OBC

0.28

Fraction

LBC2

Endo Up

(a)

LBC1

0.21 0.14 0.07

OBC -25

0

25

50

75

100

125

0.00

150

Temperature (°C)

20

40

60

80

100

120

140

Crystal thickness (Angstrom)

Fig. 13. DSC heating curves of ethylene/1-octene block copolymers after SSA.

0.18

(b)

LBC1

Fraction

0.15 0.12 0.09 0.06 0.03 0.00

15

30

45

60

75

90

Crystal thickness (Angstrom) 0.15

(c)

LBC2

0.12

Fraction

LBC2 was x-shaped trimodal (Fig. 14c), which is consistent with its triblock structure. This was due to that the middle transition block of about 6 mol% 1-octene incorporation increased the fraction of the middle sized crystals. Comparing with the chain shuttling polymerization process for OBC, the living coordination polymerization for LBCs only uses one main catalyst, the product intrachain microstructure was clear and could be tailor-designed. The accurate and facilely tunable blocky structures in living block copolymers would benefit its applications. Fig. 15 shows the dependence of crystal thickness on the SCB content and CMSL for the block copolymers. The variation trend of SCB vs. crystal thickness was similar for LBC and OBC. However, the SCB content in LBC was less than that in OBC at the same lamellar thickness. This difference in the crystal thickness tended to disappear as the SCB content increased. The variation of crystal thickness with CMSL was the same for the three block copolymers. It deviated from the linear relationship as CMSL increased, due to the chain folding in lamellar crystals at the high CMSL values. Figs. 14 and 15 indicated that three phases existed in the living block copolymers: lamellar crystals with a thickness larger than 5 nm, fringed micelles or bundles smaller than 4 nm, and amorphous phase. It is interesting that the three phase microstructure has been found in a binary blend system made of OBC and MRC with identical comonomer incorporation [42].

0.09 0.06 0.03 0.00

15

30

45

60

75

90

Crystal thickness (Angstrom) Fig. 14. Comparison of crystal thickness distribution. (a) OBC; (b) LBC1; and (c) LBC2.

5. Solid-state morphology of block copolymers Study on the solid-state morphology is of great interest and can provide further insight into the chain microstructure. The morphology of block copolymers is determined

by both block crystallization and block incompatibility [43]. The block incompatibility is usually expressed by the value of vN, where v represents the Flory–Huggins

169

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

OBC LBC1 LBC2 OBC LBC1 LBC2

50 40

600

400

30 200

20

(110)

LBC2 LBC1 OBC

Intensity (a.u.)

SCB (CH3/1000C)

60

CMSL (Angstrom)

70

(200) (210)

10 0

20

40

60

80

100

120

0 140

5

10

Crystal thickness (Angstrom) Fig. 15. Variation of short chain branching (SCB) and crystalline methylene sequence length (CMSL) with crystal thickness for the ethylene/1octene block copolymers.

interaction parameter and N is the polymerization degree. According to the mean-field theory [44], the order–disorder transition (ODT) occurs at vN = 10.5 for a monodisperse diblock copolymer of symmetric composition. When vN is smaller than the ODT value, the melt is disordered or weakly segregated [45]. The solid-state morphology tends to be alternative crystalline/amorphous lamellae after crystallization. When vN is larger than the ODT value, the melt phase separation may confine crystallization and the melt phase morphology may be preserved in the solid after crystallization [46]. When vN is comparable to the ODT value, the melt phase separation competes with crystallization and the solid-state morphology is significantly determined by the history of crystallization [47]. The value of vN was about 24.8 for LBC1, estimated using the literature method [41]. It was much higher than the ODT value 10.5, and the crystallization process of LBC1 was thus assumed to be strongly affected by the melt phase separation. LBC2 is a step triblock copolymer and the method of vN estimation is limited for such type of block copolymers. The vN value of OBC was estimated about 3, in the vicinity of the ODT value for polydisperse OBC multiblock copolymers [43]. This explains the significant effect of the mesophase separation on the crystallization behavior of OBC as reported by other authors [38,39]. The WAXD patterns revealed good phase separation in the solid state of living block copolymers, as shown in Fig. 16. The sharp diffraction peaks at 2h of 21.7° and 24.0° were assigned to the reflections of (110) and (200) crystallographic planes of the orthorhombic polyethylene crystal unit. The weak reflection of (210) plane at 2h of 30.2° could also be observed in OBC and LBC2. The average crystal sizes perpendicular to the crystallographic planes of (110) and (200) were estimated by Scherrer equation [48]. The results were listed in Table 1. The crystal size calculated from WAXD was larger than from SSA. This should be resulted from the difference between two methods. WAXD analysis showed that the crystal size of OBC was larger than LBCs, the same tendency as estimated from the SSA method, further confirming the longer CMSL in the OBC.

15

20

25

30

35

2θ (degree) Fig. 16. Wide-angle X-ray diffractograms of ethylene/1-octene block copolymers.

The relative crystallinity index was also estimated by resolving the WAXD patterns into the crystalline peaks and an amorphous halo, using the following equation [7]:

X C;WAXD ðwt%Þ ¼

Ac  100 Ac þ Aa

ð9Þ

where Ac is the area of crystalline peaks and Aa is the area of amorphous halo. The results were also given in Table 1. In comparison with the DSC determinations, the crystallinity of OBC estimated from WAXD, XC,WAXD, was larger than that from DSC, XC,DSC. This discrepancy phenomenon for OBC type sample has been demonstrated to be induced by the intermediate phase [40]. However, for the LBC samples, the crystallinity XC,DSC was larger than XC,WAXD. This was because in the LBC samples, the soft blocks could also form fringed micelles as mentioned before, the melting enthalpy of soft blocks was included in the total melting heat of crystalline phase (DHm) when calculating XC,DSC, which must result in overestimation of the ordered phase. 6. Conclusion The microstructural properties of ethylene/1-octene copolymers prepared from the living coordination copolymerization with bis[N-(3-methylsalicylidene)-2,3,4,5,6pentafluoroanilinato] titanium(IV) dichloride/dMAO as catalyst were thoroughly studied by analytic TREF and DSC thermal fractionation technique. The structure differences in the living random copolymer (LRC) and metallocene-catalyzed random copolymer (MRC) and those in the living block copolymer (LBC) and olefin multiblock copolymer (OBC) produced from chain shuttling polymerization were fully analyzed. The SCB contents were estimated using the calibration equations obtained in this work. The information of crystal thickness distribution was obtained by Thomson–Gibbs equation and Keating’s method. For the random copolymers, the interchain composition distributions were the same as shown by the analytic TREF results. However, the SSA method revealed that the crystal thickness distribution of LRC was narrower than that of MRC, indicating that LRC had narrower intrachain

170

W. Liu et al. / European Polymer Journal 54 (2014) 160–171

composition distribution. With the same SCB content, the melting temperature of LRC was about 16 °C lower than MRC. Equivalently, LRC needed about 9 branches/1000 C less than MRC in the SCB content to have the same melting temperature. For the block copolymers, LBCs had very different intrachain microstructures from OBC’s. Analytic TREF revealed that the interchain composition distribution of the commercial OBC sample was broader than LBC samples, and the commercial OBC sample had much higher content of random copolymer impurities. However, SSA analysis revealed the crystal thickness distribution of OBC was narrower than LBCs. The crystal thickness of OBC showed single modal distribution, different from the bimodal distribution of the diblock copolymer and the trimodal distribution of the step triblock. The SSA analysis indicated a three phase microstructure in the LBCs. Both WAXD and SSA methods showed the average crystal size of OBC was larger than LBCs, implying longer average crystalline methylene sequence length in the OBC polymer chain. Acknowledgements Thanks to the supports from the National Natural Science Foundation of China, China (No. 20936006), the National Basic Research Program of China, China (2011CB606001) and the Fundamental Research Funds for the Central Universities. The authors would also thank to Mr. Peng He & Dr. Wei Yu at Shanghai Jiao Tong University for their stimulating discussions and kind assistance in this work. References [1] Chum PS, Swogger KW. Olefin polymer technologies – history and recent progress at The Dow Chemical Company. Prog Polym Sci 2008;33:797–819. [2] Busico V. Metal-catalysed olefin polymerisation into the new millennium: a perspective outlook. Dalton Trans 2009:8794–802. [3] Hosoda S. Structural distribution of linear low-density polyethylenes. Polym J 1988;20:383–97. [4] Keating M, Lee IH, Wong CS. Thermal fractionation of ethylene polymers in packaging applications. Thermochim Acta 1996;284:47–56. [5] Arriola DJ, Carnahan EM, Cheung YW, Devore DV, Graf DD, Hustad PD, et al. Catalyst composition comprising shuttling agent for ethylene multi-block copolymer formation. PCT Int Appl; 2005 [WO2005/090427]. [6] Arriola DJ, Carnahan EM, Hustad PD, Kuhlman RL, Wenzel TT. Catalytic production of olefin block copolymers via chain shuttling polymerization. Science 2006;312:714–9. [7] Wang HP, Khariwala DU, Cheung W, Chum SP, Hiltner A, Baer E. Characterization of some new olefinic block copolymers. Macromolecules 2007;40:2852–62. [8] Drobny JG. Handbook of thermoplastic elastomers. William Andrew Inc.; 2007 [chapters 7 and 15]. [9] Domski GJ, Rose JM, Coates GW, Bolig AD, Brookhart M. Living alkene polymerization: new methods for the precision synthesis of polyolefins. Prog Polym Sci 2007;32:30–92. [10] Makio H, Terao H, Iwashita A, Fujita T. FI catalysts for olefin polymerizations – a comprehensive treatment. Chem Rev 2011;111:2363–449. [11] Ye Z, Xu L, Dong Z, Xiang P. Designing polyethylenes of complex chain architectures via Pd-diimine-catalyzed ‘‘living’’ ethylene polymerization. Chem Commun 2013;49:6235–55. [12] Landry E, Ye Z. Convenient Pd-catalyzed synthesis of large unimolecular star polyethylene nanoparticles. Macromol Rapid Commun 2013;34:1493–8.

[13] Zhang K, Ye Z, Subramanian R. Synthesis of block copolymers of ethylene with styrene and n-butyl acrylate via a tandem strategy combining ethylene ‘‘Living’’ polymerization catalyzed by a functionalized Pd-diimine catalyst with atom transfer radical polymerization. Macromolecules 2008;41:640–9. [14] Wang J, Zhang K, Ye Z. One-pot synthesis of hyperbranched polyethylenes tethered with polymerizable methacryloyl groups via selective ethylene copolymerization with heterobifunctional comonomers by chain walking Pd-diimine catalysis. Macromolecules 2008;41:2290–3. [15] Xiang P, Ye Z. Alternating, gradient, block, and block-gradient copolymers of ethylene and norbornene by Pd-diimine-catalyzed ‘‘Living’’ copolymerization. J Polym Sci A Polym Chem 2013;51: 672–86. [16] Edson JB, Wang Z, Kramer EJ, Coates GW. Fluorinated Bis(phenoxyketimine) titanium complexes for the living, isoselective polymerization of propylene: multiblock isotactic polypropylene copolymers via sequential monomer addition. J Am Chem Soc 2008;130:4968–77. [17] Hotta A, Cochran E, Ruokolainen J, Khanna V, Fredrickson GH, Kramer EJ, et al. Semicrystalline thermoplastic elastomeric polyolefins: advances through catalyst development and macromolecular design. PANS 2006;103:15327–32. [18] Anantawaraskul S, Soares JBP. Fractionation of semicrystalline polymers by crystallization analysis fractionation and temperature rising elution fractionation. Adv Polym Sci 2005;182:1–54. [19] Soares JBP, Anantawaraskul S. Crystallization analysis fractionation. J Polym Sci B Polym Phys 2005;43:1557–70. [20] Monrabal B, Romero L, Mayo N, Sancho-Tello J. Advances in crystallization elution fractionation. Macromol Symp 2009;282: 14–24. [21] Suriya K, Anantawaraskul S, Soares JBP. Cocrystallization of ethylene/1-octene copolymer blends during crystallization analysis fractionation and crystallization elution fractionation. J Polym Sci B Polym Phys 2011;49:678–84. [22] Müller AJ, Arnal ML. Thermal fractionation of polymers. Prog Polym Sci 2005;30:559–603. [23] Müller AJ, Hernández ZH, Arnal ML, Sánchez JJ. Successive selfnucleation/annealing (SSA): a novel technique to study molecular segregation during crystallization. Polym Bull 1997;39:465–72. [24] Arnal ML, Balsamo V, Ronca G, Sánchez A, Müller AJ, Cañizales E. Applications of successive self-nucleation and annealing (SSA) to polymer characterization. J Therm Anal Calorim 2000;59:451–70. [25] Kim YM, Kim CH, Park JK, Kim JW, Kim TI. Short chain branching distribution and thermal behavior of high density polyethylene. J Appl Polym Sci 1996;60:2469–79. [26] Zhang M, Lynch DT, Wanke SE. Characterization of commercial linear low-density polyethylene by TREF-DSC and TREF-SEC crossfractionation. J Appl Polym Sci 2000;75:960–7. [27] Zhang M, Wanke SE. Quantitative determination of short-chain branching content and distribution in commercial polyethylenes by thermally fractionated differential scanning calorimetry. Polym Eng Sci 2003;43:1878–88. [28] Zhang F, Fu Q, Lü T, Huang H, He T. Improved thermal fractionation technique for chain structure analysis of ethylene/a-olefin copolymers. Polymer 2002;43:1031–4. [29] Mitani M, Furuyama R, Mohri JI, Saito J, Ishii S, Terao H, et al. Syndiospecific living propylene polymerization catalyzed by titanium complexes having fluorine-containing phenoxy–imine chelate ligands. J Am Chem Soc 2003;125:4293–305. [30] Liu W, Zhang K, Fan H, Wang WJ, Li BG, Zhu S. Living copolymerization of ethylene/1-octene with fluorinated FI-Ti catalyst. J Polym Sci A Polym Chem 2013;51:405–14. [31] Liu W, Guo S, Fan H, Wang WJ, Li BG, Zhu S. Synthesis of ethylene/1octene copolymers with controlled block structures by semibatch living copolymerization. AICHE J 2013;59:4686–95. [32] ASTM D5017-96(2009)e1. Determination of linear low density polyethylene (LLDPE) composition by carbon-13 nuclear magnetic resonance, vol. 08.02. ASTM: West Conshohocken, PA; 2009. [33] Simanke AG, Galland GB, Freitas L, De Jornada JAH, Quijada R, Mauler S. Influence of the comonomer content on the thermal and dynamic mechanical properties of metallocene ethylene/1-octene copolymers. Polymer 1999;40:5489–95. [34] Czaja K, Sacher B, Białek M. Studies of intermolecular heterogeneity distribution in ethylene/1-hexene copolymers using DSC method. J Therm Anal Calorim 2002;67:547–54. [35] Hoffman JD, Miller RL, Marand H, Roitman DB. Relationship between the lateral surface free energy r and the chain structure of meltcrystallized polymers. Macromolecules 1992;25:2221–9.

W. Liu et al. / European Polymer Journal 54 (2014) 160–171 [36] Tian Z, Gu XP, Wu GL, Feng LF, Fan ZQ, Hu GH. Periodic switching of monomer additions for controlling the compositions and microstructures of segmented and random ethylene–propylene copolymers in polypropylene in-reactor alloys. Ind Eng Chem Res 2011;50:5992–9. [37] Bensason S, Minick J, Moet A, Chum S, Hiltner A, Baer E. Classification of homogeneous ethylene–octene copolymers based on comonomer content. J Polym Sci B Polym Phys 1996;34:1301–15. [38] Jin J, Du J, Xia Q, Liang Y, Han CC. Effect of mesophase separation on the crystallization behavior of olefin block copolymers. Macromolecules 2010;43:10554–9. [39] Liu G, Guan Y, Wen T, Wang X, Zhang X, Wang D, et al. Effect of mesophase separation and crystallization on the elastomeric behavior of olefin multi-block copolymers. Polymer 2011;52: 5221–30. [40] Tong Z, Huang J, Zhou B, Xu JT, Fan ZQ. Chain microstructure, crystallization, and morphology of olefinic blocky copolymers. Macromol Chem Phys 2013;214:605–16. [41] Li S, Register RA, Landes BG, Hustad PD, Weinhold JD. Crystallization in ordered polydisperse polyolefin diblock copolymers. Macromolecules 2010;43:4761–70.

171

[42] Hölzer S, Menzel M, Zia Q, Schubert US, Beiner M, Weidisch R. Blends of ethylene–octene copolymers with different chain architectures – morphology, thermal and mechanical behavior. Polymer 2013;54:5207–13. [43] Li S, Register RA, Weinhold JD, Landes BG. Melt and solid-state structures of polydisperse polyolefin multiblock copolymers. Macromolecules 2012;45:5773–81. [44] Leibler L. Theory of microphase separation in block copolymers. Macromolecules 1980;13:1602–17. [45] Lee LBW, Register RA. Equilibrium control of crystal thickness and melting point through block copolymerization. Macromolecules 2004;37:7278–84. [46] Xu JT, Turner SC, Fairclough JPA, Mai SM, Ryan AJ, Chaibundit C, et al. Morphological confinement on crystallization in blends of poly(oxyethylene-block-oxybutylene) and poly(oxybutylene). Macromolecules 2002;35:3614–21. [47] Loo YL, Register RA, Ryan AJ. Modes of crystallization in block copolymer microdomains: breakout, templated, and confined. Macromolecules 2002;35:2365–74. [48] Cullity BD. Elements of X-ray diffraction. 2nd ed. Reading, MA: Addison-Wesley; 1978.