European Polymer Journal 49 (2013) 2748–2758

Contents lists available at SciVerse ScienceDirect

European Polymer Journal

journal homepage: www.elsevier.com/locate/europolj

Effect of high molar mass species on linear viscoelastic

properties of polyethylene melts
Jon Otegui, Javier Ramos, Juan F. Vega ⇑, Javier Martínez-Salazar
BIOPHYM, Departamento de Física Macromolecular, Instituto de Estructura de la Materia-Consejo-Superior de Investigaciones Científicas (IEM-CSIC),
C/Serrano 113 bis, 28006 Madrid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The molecular features and the linear viscoelastic melt properties of a series of polydis-
Received 22 November 2012 perse high density polyethylene are investigated. The main molecular characteristic of
Received in revised form 29 May 2013 the materials studied is the presence of high molar mass species, which give rise to unusual
Accepted 11 June 2013
values of the Mz/Mw ratio. This molecular particularity strongly affects to the linear visco-
Available online 22 June 2013
elastic melt properties as the Newtonian viscosity and the steady-state shear recoverable
compliance. Both the third and the fourth moments of the molecular weight distribution
Keywords:
affect to the values of these properties, which follow the trend expected by the reptation
Linear polyethylene
Molecular weight distribution
model. The evidences clearly prove the effect not only of the polydispersity index, but more
High molecular weight tails interestingly of the shape of the molecular weight distribution on the dynamics of the sys-
Melt linear viscoelasticity tems. The behavior shown by the experimental samples studied in this work is interesting
Reptation as the linear viscoelastic properties have become widely used, not only to test the molec-
ular weight distribution dependence of the viscoelastic fingerprint and the processability,
but also to assess the possible presence of long chain branching.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction empirical modification of the average molecular weight

for the dependence of the rheological properties, later
The effect of the molecular weight distribution (MWD) validated for polydisperse polypropylene (PP) [6,7] and
on the viscoelastic response of linear polyethylene (PE) PE [7]. The equation of Malkin and co-workers [5],
melts has been widely reported in the literature since the obtained for polydisperse polybutadiene (PB), nicely
1960s, but the results are certainly contradictory. Bueche agrees with recent theoretical predictions [8,9], based in
[1] and Graessley [2] considered the existence of an aver- the determination of the linear viscoelastic response from
aged molecular weight of the distribution, Mt, located be- the MWD trace using the double reptation approach. The
tween the second and the third moments, Mw and Mz, approximate correlation found by Nobile and Cocchini
respectively, that best represents the average to explain takes the form [8]:
the dependence of the Newtonian viscosity, g0, in linear  b
polymers with polydispersity index higher than Mw/ Mz
g0 ¼ 0:51KMaw ð1Þ
Mn = 2. Later, Saeda and co-workers [3] found the experi- Mw
mental support to these ideas in polydisperse samples.
Locati and Gargani [4] empirically expressed Mt in terms of with K the corresponding pre-exponential constant for
the polydispersity index using Saeda’s experimental data. In monodisperse species at a given temperature, a = 3.4 and
the same direction, Malkin and co-workers [5] proposed an b = 0.8. This expression resembles the correlation found
by den Doelder, but in this case with b = 0.7 [9]. Other
authors did not observe these MWD effects on g0 for poly-
⇑ Corresponding author. Tel.: +34 915 616 800; fax: +34 915 615 413. disperse PE samples with Mz/Mw = 2.2 ± 0.7 [10]. A more
E-mail address: jf.vega@csic.es (J.F. Vega). recent work has revealed that the width of the MWD

0014-3057/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.eurpolymj.2013.06.015
 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758 2749

strongly affects to both g0 and steady-state shear recover- those reported until now in linear polydisperse PE), so
able compliance, J 0e , in linear PE samples with values of Mz/ the effect of the very high-Mw tail has to be especially
Mw between 4.0 and 11.2 [11,12]. It is widely accepted that noticeable. For this reason, we extend the study of the
J 0e does not relate to Mw, but it is very sensitive to the mea- physical properties of these materials and report: (i) basic
sure of the MWD. Many relations (empirical and theoreti- physical properties of the materials; (ii) the effect of the
cal) have been proposed to correlate J 0e with MWD. The one presence of high-Mw tails on the melt linear viscoelastic
suggested by Mieras and van Rijn [13] (Eq. (2)) with power properties, and more specifically on J 0e , and (iii) the model-
law exponent of c = 3.7, was successfully tested by Mills ing of the linear viscoelastic response within the reptation
[14] in polydimethylsiloxane (PDMS), polystyrene (PS) framework by using the experimental MWD together with
and PE. simulated values of the entanglement features.
 c
Mz
J 0e / ð2Þ 2. Experimental methods
Mw
The blending laws of Bogue et al. [15] and Kurata et al. The materials were kindly supplied by Centro Tecnológ-
[16] established values for c of 2.0 and 3.7, respectively. ico Repsol-YPF (Spain). The samples labeled from PE1 to
More recently, Nobile and Cocchini [8] have obtained a PE4 listed in Table 1 were obtained from a bis(imino)pyri-
similar correlation based in the double reptation approach, dyl iron catalysts [28]. The polymerizations were carried
but with an exponent of c = 5.5. Other correlations were out in a Buchi reactor (except the PE1 sample in Table 1,
proposed by Ferry et al. [17] based in the Rouse model, obtained in autoclave reactor in equivalent conditions
and Agarwal [18] from empirical observations, in which and a slightly higher concentration of catalyst). Isobutene
higher moments of the MWD are involved: solvent (1 dm3) and the required amount of MAO co-cata-
! lyst were introduced into the reactor under inert nitrogen
M zþ1 M z atmosphere. The total pressure was adjusted to 40 bars.
J 0e / ð3Þ
M 2w Reactions were carried out at a temperature of 90 °C and
kept constant throughout the polymerization. In the Buchi
  series hydrogen was introduced continuously in the poly-
Mzþ1 M z
J 0e / ð4Þ merization reactor to control the molecular weight of the
Mw Mn
final products. The total concentration of hydrogen is ad-
By the other hand, Ressia et al. [19] proposed an expres- justed to remain constant during polymerization and in-
sion based on the blending laws of Montfort et al. [20] and creased from 0 to 1 mol% in the subsequent
Zang et al. [21] for mixtures of monodisperse systems, polymerizations (from PE2 to PE4). Polymerizations were
which included not integer moments of the MWD (Qa = carried out along 1 h. Two samples obtained from a metal-
RwiMai , with ‘‘a’’ values between 3.4 and 4.4): locene single-site catalyst (mPE) have been included in this
  study, for comparison purposes [30].
Qa The samples were analyzed by 13C nuclear magnetic
J 0e / ð5Þ
Q a1 resonance (13C NMR) on a Bruker DRX500 spectrometer.
Finally, den Doelder has recently established a relation- The spectra were recorded at 100 °C on a Bruker DRX 500
ship based also on the double reptation model [9]: spectrometer operating at 125 MHz. The samples were dis-
solved in hot 1,2,4-trichlorobenzene (TCB) and d6-ben-
 2  4
Mz Mzþ1 zene. 13C NMR results corroborate the linear structure of
J 0e / ð6Þ all the materials studied in this work in agreement with
Mw Mz
previous results [28].
From all these expressions strong effects of the highest The MWD of the samples was determined by means of
moments of the MWD on the melt elastic character of high-temperature size exclusion chromatography (SEC)
polymeric materials are expected, as a slight increase in analyzer Waters 150C, equipped with refractive index
the moment ratios can cause values of J 0e several orders (RI) and viscometer (V) detectors. All measurements were
of magnitude higher than the characteristic value of mono- conducted at a flow rate of 1 mL min1 and at temperature
disperse species.
In previous works we have reported that tridentate
bis(imino)pyridyl iron complexes, when combined with a Table 1
co-catalyst, such as methylaluminoxane (MAO) and alumi- Molecular features of the samples studied.

num alkyls (AlEt3 and AliBu3), form very active catalytic Samples Mw Mw/ Mz Mz+1
systems that yields linear PE with very broad MWD [22– (kg mol1) Mn (kg mol1) (kg mol1)
24]. We have investigated the performance of this type of sd = ±5%
catalyst from the simulation [25–27] and experimental mPE1a
65.0 3.6 195.0 585.0
points of view [28]. We have also reported the effect of mPE2a 152.0 2.3 350.0 805.0
the MWD on g0 for these materials in a recent contribution PE1 35.0 6.1 1000 4730
[29], on the basis of the applicability of Eq. (1). We must PE2 49.4 7.8 1340 5770
PE3 71.7 9.7 1400 5300
point out here that the reported values of Mz/Mw in the
PE4 92.5 13.3 1730 5200
experimental PE samples studied in this work are extre-
a
mely high, between 20 and 30 (2–3 times higher than Single-site metallocene samples [30].
2750 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758

of 145 °C using 1,2,4-trichlorobenzene (TCB) as solvent. To tures were shifted along the angular frequency axis to a
accurately calibrate the column we have obtained the reference temperature of TR = 160 °C. Vertical shifts along
hydrodynamic volume, which is the product of the molec- the modulus axis were not necessary. The frequency shift,
ular weight, M; and the intrinsic viscosity, [g], and the aT, was expressed in terms of the flow activation energy, Ea,
retention volume using monodisperse polystyrene stan- using the Arrhenius relationship [35]:
dards (universal calibration). The viscometer detector  
Ea 1 1
makes this possible because it directly measures [g], which aT ¼ exp  ð7Þ
R T TR
is inversely proportional to the molecular density of the
polymer coil. The molecular weight of the fractions has with R the universal gas constant and T the absolute tem-
been estimated from the intrinsic viscosity values together perature. The values of Ea are given in Table 2.
using the hydrodynamic volumes obtained from of the uni- For determining g0 and J 0e , creep and creep-recovery
versal calibration. True molecular weights can be obtained experiments were performed at T = 160 °C. The rheological
in this way independently of the chemical and molecular tests were carried out using 25 mm parallel-plate and/or
structure of the polymers and the calibration standards cone-plate geometries. We have ensured that the measure-
[31]. The Mark–Houwink correlation of the samples stud- ments were conducted in the linear viscoelastic regime
ied has a power law exponent of av = 0.70 [30], in good and that the steady-state was reached in the preceding
agreement with literature results for linear PE [32–37], creep and the recovery tests by applying different stresses,
providing further evidence of the linear structure of the s0, and creep times, tc. It was proven that the stresses ap-
samples. The molecular features obtained by the SEC/RI/V plied were within the linear viscoelastic regime (between
system are listed in Table 1. The result obtained for each 0.78 and 12.5 Pa in the materials studied here). The time-
of the samples has been averaged from 10 independent dependent deformation c(t) was measured. The quotient
measurements. The standard deviation of the averaged val- of the deformation and the applied stress is the creep com-
ues obtained for the molecular features is s.d. = ±5%. pliance, J(t):
The thermal properties (melting point, Tm, and crystal-
t
line content, Xc) of the polymers were obtained on a Perkin JðtÞ ¼ J 0 þ WðtÞ þ ð8Þ
Elmer 7 series thermal analyzer by heating to 160 °C at a g0
rate of 10 °C min1 and cooling at the same rate to 20 °C. being J0 the instantaneous elastic compliance, W(t) the
A second heating cycle at 10 °C min1 was used for data creep function and t/g0 is the irreversible contribution to
analysis. The reported Tm is the peak value from the last the deformation. If tc is sufficiently long, W(t) will reach a
heating. Crystallinity was calculated using a value of constant value and g0 reads as follows:
288.4 J g1 as the reference melting enthalpy for 100% t
crystalline PE. The measured properties are listed in g0 ¼ t!1
lim ð9Þ
JðtÞ
Table 2.
Small-amplitude oscillatory viscoelastic measurements After a defined tc, the applied stress was set to zero
in the melt were carried out by means of a Bohlin CVO (s0 = 0 Pa) and the recoverable deformation, and then the
stress-controlled rheometer using a parallel disk (15 mm recoverable compliance, Jr, was measured as a function of
diameter) geometry, in the angular frequency range be- the recovery time, tr. If the creep and recovery time are
tween 102 and 102 rad s1. The measurements were per- long enough and the experiment is performed in the linear
formed well inside the linear viscoelastic region (shear range of deformation, J 0e is obtained as:
strain lower than 0.1) previously located by the aid of
J 0e ¼ lim J r ðtc ; t r Þ ð10Þ
amplitude sweeps. The following viscoelastic functions t c !1
t r !1
were measured as a function of the angular frequency, x,
at different temperatures (between the melting tempera- It was proven that the thermal stability of the materials
ture and 190 °C): storage shear modulus, G0 , and loss shear was sufficient for all measuring times at the applied tem-
modulus, G00 . The data obtained at the different tempera- perature. The values (averaged over at least 3 independent

Table 2
Physical properties of the samples studied: density (q), melting temperature (Tm), crystallinity (Xc), flow activation energy (Ea); Newtonian viscosity (g0) and
steady-state shear recoverable compliance (J0e ).

Samples q (g cm3)a Tm (°C) Xc (%) Ea (kJ mol1) g0 (kPa s)b g0 (kPa s)c J0e 103 (Pa1)c

mPE1 n.d. n.d. n.d. 23.8 1.09 n.d. n.d.

mPE2 0.9516 134.2 0.70 23.0 23.4 23.0 0.055
PE1 0.9737 130.8 87.0 27.1 0.985 1.03 55.0
PE2 0.9716 133.2 86.0 25.0 3.38 3.12 39.2
PE3 0.9687 133.6 81.0 26.7 13.4 12.5 13.1
PE4 0.9705 134.1 78.0 27.9 25.0 24.2 8.65

n.d. Means not determined.

a
Density measured on a gradient column.
b
Newtonian viscosity, g0, measured at T = 160 °C from dynamic measurements [29].
c
Newtonian viscosity, g0, and steady-state shear recoverable compliance, J 0e , measured at T = 160 °C from creep measurements.
 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758 2751

measurements) obtained for both g0 and J 0e are listed in dependence of g0 and J 0e on the molecular features. We have
Table 2. The standard deviation of the averaged values measured the values g0 of the materials under study from
obtained for the molecular features is s.d. = ±6.5%. dynamic [29] and creep measurements. The creep measure-
ments at 160 °C for the samples studied are shown in Fig. 1.
In order to guarantee the linearity of the viscoelastic re-
3. Results and discussion sponse in these experiments the results obtained at differ-
ent shear stresses, s0, are compared (between 1.56 and
The 13C NMR analysis reveals that all the products ob- 12.5 Pa). All the samples show linear behavior in the shear
tained from the bis(imino)pyridyl iron catalysts posses a stress range explored. At long creep times all the samples
linear structure. The linear structure of the samples was show a straight line of a slope 1 in the double logarithmic
theoretically predicted in our group by computer simula- plot, which is equal to a plateau in t/J(t) versus t plot. This
tions of the copolymerization of ethylene with 1-hexene demonstrates that for the samples investigated the steady-
[36], and it was also probed experimentally [28]. state of creep is obtained and that the determination of g0
SEC results show a variation in the Mw of the samples. is possible, by means of the application of Eq. (9). For the
The data listed in Table 1 reveal that both Mw and Mw/Mn polydisperse PE samples the creep compliance decreases
systematically increase as the presence of hydrogen during (viscosity increases) as the molecular mass increases, as ex-
polymerization does from PE2 to PE4. Additionally, the pected. The results of g0 obtained from these data are listed
materials are characterized by high values of Mz/Mw, indi- in Table 2, being in perfect agreement with those obtained
cating the presence of high-Mw tails. Huang and co-work- previously from the dynamic measurements [29].
ers have argued that the presence of molecular hydrogen The g0  Mw dependence is represented by the well-
results in a greatly decreased formation of low Mw polymer known relationship observed for many linear monodis-
in the case of catalysts with less hindered metal complexes perse polymers [39]:
[37]. Moreover, the reported data by these authors have
suggested not only a decrease of the amount of low-Mw g0 ¼ KMaw ð11Þ
fraction but also an increase in the Mw and the amount of
where, for PE, a takes values from 3.36 to 3.64 and K is a
the high-Mw fraction of the MWD. These authors have con-
constant that depends on temperature [30]. This relation-
cluded that the dominant effect of hydrogen in this less
ship is valid for Mw values above Mc, being Mc a critical
steric bulk system is the deactivation of active species that
molecular weight related to the molecular weight of an
polymerize low Mw material and the activation of those
entanglement strand, Me. Fig. 2A includes the results ob-
species that produce the high Mw fraction. Recent com-
tained for our polydisperse PE materials and other results
puter simulations made in our group using global and local
obtained from the literature for polydisperse PE samples
descriptors of chemical reactivity and selectivity based on
[3,7,10–12]. We have also included the reported correla-
the DFT theory, have shown that these catalysts produce
tion obtained for selected data of linear PE fractions, mono-
materials with broad MWD and high-Mw species [25–27].
disperse hydrogenated polybutadiene and mPE samples
Concerning the physical properties, the materials with
with narrow MWD [30]. Significant deviations from the
the lowest Mw values show increased density and crystal-
trend followed by monodisperse PE samples are observed
linity and decreased melting temperatures, similarly to
for the polydisperse PE series studied in this work and
analogous samples studied by other authors [37], and also
for the most of the data collected from the literature. The
to linear PE samples synthesized using single-site catalysts
presence of high-Mw tails could be the cause of this devia-
[38]. There are some differences among the samples stud-
tion, as it seems to be more pronounced as the value of Mz/
ied here, but these are not expected to be due to the pres-
Mw increases. For the materials with Mz/Mw  Mw/Mn = 2–3
ence of short chain branches, as judged by the 13C NMR
results and the computer simulations. The apparent molec-
ular weight dependence of the microstructural features is a 1
10
well-know effect caused by the non-equilibrium condi-
0
tions at which thermal properties are measured. The crys- 10
tallization time during the DSC scan is similar in all the -1
10
samples, but because of the different Mw, the time needed
Jc (Pa-1)

-2
10
to reach equilibrium crystallization conditions is also dif-
-3
ferent in each case. As the Mw increases, the chain move- 10
PE1
ments are more restrained as a consequence of the -4
10 PE2
increased average number of entanglements, and thus PE3
-5
more time is required to reach the same state than the cor- 10 mPE2
x0.5
responding to a sample with a lower Mw value. This obvi- -6
10
-2 -1 0 1 2 3 4
ously gives rise to lower values of crystallinity and 10 10 10 10 10 10 10
density in the samples with the highest Mw. t (s)
In addition to the considerations about the microstruc-
ture, the materials studied are of interest because of their Fig. 1. Creep compliance versus creep time for the materials studied at
T = 160 °C and different shear stresses: (squares) 1.56 Pa; (circles)
rheological properties. The relationship between molecular 3.12 Pa; (up triangles) 6.25 Pa and (down triangles) 12.5 Pa. Creep
features and rheological properties is a topic of long- compliance values for mPE2 sample have been shifted vertically (0.5)
standing discussion in polymer science, especially the to avoid crowding.
2752 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758

10
5
tial use as an indirect method to determine the amount of
4
(A) LCB in polyolefins and to differentiate its effect from that
10 induced by the MWD. In particular, low LCB contents (less
10
3 than 1 branch per 10,000 carbon atoms) lead to substantial
(kPa s)

increments in g0 values in comparison with that exhibited

2
10 by the linear counterpart in samples obtained from single-
site catalysts [40–42]. These catalyst systems cannot only
0

1
10
produce polyolefins with narrow MWD and LCB, but they
10
0 also show the capacity to incorporate high amounts of
comonomer. The most probable mechanism to produce
-1
10 LCB in this type of polymerization is terminal branching
1 2 3
10 10 10
[43]. In this mechanism, terminated chains formed during
Mw (kg/mol) the polymerization process by b-hydride elimination reac-
tion or by transfer to ethylene are inserted in another
5
10 growing chain, as occurs in a copolymerization step. These
10
4 (B) terminated chains containing vinyl-end groups act as
(Mz /Mw) (kPa s)

3
macro-monomers. Thus to form a LCB, the copolymeriza-
10 tion ability of the catalyst is a necessary condition. Not-
10
2
3.4 withstanding, it has been reported both experimentally
-

1 and theoretically the inefficiency of the catalyst used here

10
for the incorporation of either a-olefin as 1-hexene [28,36],
0
10 or even more bulky comonomers as styrene [44]. In addi-
0

tion to this, it should be noted that PE is one of the few

2

-1
10
-2
polymers for which the presence of LCB affects the temper-
10 ature dependence of rheological properties [45]. In fact, the
1 2 3
10 10 10
values of Ea increases as LCB content does in PE samples
Mw (kg/mol)
with branching content between 0.1 and 1 branches per
10,000 carbon atoms [46]. For these samples the observed
Fig. 2. (A) Newtonian viscosity versus weight average molecular weight
at T = 160 °C (h) polydisperse PE samples, (j) mPE with narrow MWD, increase of g0 of nearly one order of magnitude (same as
(O) polydisperse HDPE with Mz/Mw = 3.0–5.4 [3], (}) polydisperse HDPE the observed in the materials under study) with respect
with Mz/Mw = 2–9 [7]; (+) mHDPE with narrow MWD with Mz/Mw = 2.2– to the linear species, would corresponds to LCB contents
3.5 [10], (s) polydisperse mHDPE with Mz/Mw = 3–5 [11], (N) HDPE with
close to 1 branch per 10,000 carbon atoms and values of
narrow MWD [12], (M) polydisperse HDPE with Mz/Mw = 6.5–11.2 [12].
The solid line corresponds to the correlation found for monodisperse
Ea of close to 40 kJ mol1 [42]. In our samples an average
HDPE fractions and mPEs [30]. (B) Weight-average molecular weight value of Ea = 26.7 ± 1.3 kJ mol1 has been obtained, which
dependence of the normalized Newtonian viscosity for the materials is consistent with the reported value of linear PE, so the
studied at T = 160 °C. The symbols are the same than in Fig. 2A. presence of LCB should be completely ruled out.
Then the effect should definitively be due to the pres-
ence of very high-Mw tails in the MWD of the samples.
[10,12], including our mPE samples with narrow MWD
Interestingly, the samples studied show a strong positive
(crosses and solid symbols in Fig. 2A), the deviation from
asymmetric deviation in the MWD with respect to the
the correlation of monodisperse samples is only marginal.
mean value, as it can be observed in Fig. 3. The positive
This result suggests that in this type of materials (single-
deviation accounts for closely the 20% of the whole MWD
site catalyzed PE samples) the amount of high-Mw tails is
within the molecular weight range from 10 to
not high enough to cause an impact in the values of g0,
10,000 kg mol1. This deviation of the MWD causes extre-
with respect to that exhibited by linear monodisperse PE
mely high Mz/Mw ratio values of around 20–30, as it has
samples with the same Mw. In fact, the term (Mz/Mw)0.8
been pointed out in preceding lines. Additionally, the val-
cancels the pre-exponent 0.51 in Eq. (1) for values of Mz/
ues of Mz/Mw are higher than the polydispersity index,
Mw around 2.0 [29]. In contrast, the materials with values
Mw/Mn, in all the cases (the maximum value of Mw/Mn in
of Mz/Mw = 4.0–11.2 [3,7,11,12] clearly deviate from the
Table 1 is 13.3). The characteristic high Mz/Mw values in
correlation (open symbols in Fig. 2A), and also our polydis-
the samples are related to the existence of a large number
perse PE samples with the highest Mz/Mw ever measured in
of components with high relaxation times contributing to
this type of polymers to our knowledge. These results sug-
the linear viscoelastic response. In the framework of Eq.
gest that polydispersity index given as Mw/Mn is not en-
(1), using the pre-exponential factors given by Nobile and
ough to judge the effect of the MWD on the linear
Cocchini [8] used also in our previous work [29], it is pos-
viscoelastic behavior.
sible to fit all the data in Fig. 2A to a single correlation con-
However, additional interpretations further than high-
sidering variable values of c, depending of the sample
Mw tails, could lead us to argue on the possible presence
series considered. The result of the fitting can be observed
of molecules with long chain branches (LCB) in the samples
in Fig. 2B. Our polydisperse PE samples fit very well by con-
studied. In the last decade a significant number of studies
sidering a value of c = 0.80 [29]. For the results taken from
have been carried out in order to establish the exact g0–
the literature (shifted to T = 160 °C by using the character-
Mw relationship in PE. The reason behind that is its poten-
istic value of Ea for linear PE), the fitting requires values of
 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758 2753

0.8 obtain J 0e from G0 . For these reasons creep and creep recov-
(A) ery test is a more suitable technique for the determination
of this viscoelastic function. Examples of the creep and
0.6
creep-recovery tests performed to obtain J 0e are given in
dlogW/dlog M

Fig. 4A–C for three of the samples studied: PE3, PE4 and
0.4 mPE2. The applied stress values, s0 = 3.12 and 12.5 Pa,
respectively, are within the linear viscoelastic range as it
Mw>500 kg mol
-1 is demonstrated in Fig. 1: identical time dependence of Jc
0.2
are obtained in all the cases in this range of shear stress.
However, it has been reported in the literature that the
0.0 recoverable compliance also depends on the creep time,
0 1 2 3 4 tc [48]. In the case of the polydisperse PE samples studied
log M (M in kg/mol) the values of the Jr plateau at long times are highly depen-
dent on tc, as it can be observed in Fig. 4A for the PE3 sam-
0.10
ple. Constant values of J 0e are reached for tc values of 1250 s
(B) and higher for an applied shear stress of s0 = 3.12 Pa. An in-
0.08
crease of more than twice in tc does not significantly
dlogW/dlog M

change the final value of J 0e . In order to assure that J 0e has

0.06
really been achieved for the polydisperse PE samples we
have additionally performed creep-recovery measure-
0.04
ments at different shear stresses within the range 0.78–
3.12 Pa and during high enough creep times of
0.02
tc = 3,125 s. As an example these tests are shown in
Fig. 4B for PE4 sample. In this figure it is clearly seen that
0.00
2 3 4 both Jc and Jr coincide, being the standard deviation of
log M (M in kg/mol) the final value of J 0e measured of around s.d. = ±3.5%. Simi-
lar results about the linearity of the response were ob-
Fig. 3. SEC traces of some of the samples studied: (j) PE2; (d) PE 3 and tained for the other polydisperse PE products in the same
(N) PE4. range of shear stress and creep time. In the case of mPE2
sample a higher value of the shear stress (s0 = 12.5 Pa)
has been applied to achieve a higher resolution of the rhe-
c ranging from 1.0 [7,11] to 1.30 [8,10,12]. The average va- ometer. In this case, and in contrast to polydisperse PE
lue obtained for c for the whole set of materials is around samples, the steady state value is reached after quite short
1.0, which is in agreement with the empirical correlations recovery times (Fig. 4C), in agreement with previous re-
reported in the literature [5,7] and close to the theoretical sults obtained in materials with narrow MWD [49–51].
values of 0.7–0.8 [8,9]. The variability found in the values Also the value of J 0e obtained in this sample is in agreement
of c is higher than the experimental uncertainly expected with that reported for this type of materials (J 0e  0.06–
in g0 (±20–30%) due to the variability in the measurement 0.07 mPa1).
of MWD by SEC (±5–10%). A proper determination of both The most interesting feature observed in the polydis-
g0 and MWD is crucial, which is particularly problematic in perse PE samples is the very high J 0e values obtained, being
the case of very high Mw polymers, i.e. above 300 kg mol1 between 200 and 1000 times higher than that obtained for
and broad MWD. Notwithstanding, it should be pointed the mPE2 sample. Moreover, the increase observed in J 0e
out that the investigated extremely broad PE samples with values in the polydisperse series is not explained by the
Mz/Mw significantly larger than Mw/Mn are experimental different Mw of the samples. In fact an apparent decrease
specimens with little relevance to commercial products of J 0e as Mw increases is directly read from the results in Ta-
for which the independence of g0 on the MWD is a very ble 2 for this series. The value of J 0e of a polymeric material
good approximation [10,30]. depends on a number of molecular parameters. From the
The presence of these very high-Mw components should literature it is known that for linear monodisperse poly-
have an important impact not only in g0 values, but also in mers above the critical molar mass J 0e is molar mass inde-
the elastic character of the materials. The values of J 0e are pendent [52], but increases with the broadening of the
usually better determined in creep and creep-recovery MWD. As it has been pointed out in the introduction, dif-
experiments than in oscillatory shear tests. There are vari- ferent empirical and theoretical correlations between J 0e
ous reasons for the use of creep and creep-recovery tests in and the MWD can be found in the literature. Fig. 5A depicts
determining this viscoelastic function [47–51]: (i) the stea- the experimental values of J 0e for the materials studied in
dy-state of deformation is reached much faster than in the present work together with experimental data col-
oscillatory shear [47]; (ii) oscillatory shear measurements lected from the literature for linear PE, including model
are sometimes limited due to torque transducer accuracy hydrogenated polybutadienes (HPB), in which Mz/Mw is
in the low frequency range of the rheometers [48]. In this available [11,14,53–56]. In the case of monodisperse HPB
region for loss angles close to d = 90° the resolution affects samples we have considered that Mw/Mn = Mz/Mw. These
G0 more than G00 . This fact makes possible to achieve g0 data were obtained at different temperatures within
from G00 , but conversely it makes extremely difficult to the range 150–190 °C. This should not affect to the
2754 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758

Fig. 4. Creep, Jc (solid symbols), and recovery compliance, Jr (open symbols), for the materials studied at T = 160 °C: (A) PE3 (s0 = 3.12 Pa) and (C) mPE2
(s0 = 12.5 Pa) with variable creep time, tc (squares) 50 s, (circles) 250 s, (up triangles) 1250 s, (down triangles) 3125 s. (B) PE4 (tc = 3125 s) with variable
shear stress, s0 (squares) 0.78 Pa, (circles) 1.56 Pa and (triangles) 3.12 Pa.

0
10 (c = 3.0) and Nobile and Cocchini (c = 5.5). An increase of
-1
(A) =5.5
=3.7 the experimental values of J 0e with Mz/Mw is observed as a
10
general trend. However, each correlation seems to explain
-2
10 =3.0 the behavior of a particular series. It should be pointed out
Je (Pa )

that, except the experimental results recently reported by

-1

-3
10 Ansari et al. [11], which follow the squared power law re-
=2.0
0

-4
10
ported by Bogue et al. [15], the other results seem to deviate
towards the Kurata (c = 3.0) and/or Nobile and Cocchini
-5
10 (c = 5.5) correlations. It should be also noted that for the
-6
most of the samples from the literature, detailed informa-
10
0
10 10
1
10
2 tion on the molecular structure is missing. It is known that
the presence of small amounts of LCB causes a strong effect
Mz /Mw
in J 0e [49,50,57]. Another source for the data scattering could
be the presence of tiny amounts of high-Mw tails (<1 wt.%
0
10 with Mw > 1000 kg mol1), which will be hardly detected
-1
(B) [58–60]. As it is clear from Fig. 5B, it is the correlation given
10
by den Doelder (Eq. (6)) that best explains the results ob-
-2
10 tained in our materials and those recently reported by
Je (Pa )

Ansari et al. [11]. The functional dependence of J 0e is

-1

-3
10 2
then described by the MWD in the form of M 2zþ1 /MwMz with
0

-4
10 a power law exponent of 2, suggesting the strong effect of
the very high-Mw tail in this viscoelastic function. The veri-
-5
10 fication of this dependence asks for a wider experimental
-6 support in well characterized samples with broad MWD.
10
0
10
1
10 10
2 3
10 10
4 Unfortunately, we have not found additional results
2
in the literature than those of Ref. [11] reporting both J 0e
(Mz+1) /Mz Mw and Mz+1 in linear PE samples in order to further test this
correlation.
Fig. 5. (A) Steady-state shear recoverable compliance, J 0e , versus Mz/Mw
for the materials studied at T = 160 °C; (j) polydisperse PE samples and
It is of interest to additionally check the experimental
(j) mPE2, and literature results for linear polydisperse PE samples; (j) results obtained in both the creep-recovery and dynamic
Mills et al. [53]; (s) Plazek et al. [54]; (4) HPB from Carella et al. [55]; (O) modes. The values of the shear creep, Jc, and shear recov-
Yoshikawa et al. [56] and (}) Ansari et al. [11]. The lines correspond to Eq. ery, Jr, compliances can be converted to the dynamic mod-
(2) with different exponents: (solid line) c = 3.7 [13], (dashed-dotted line)
uli, G0 and G00 , following the procedure described by
c = 2.0 [15], (dashed line) c = 3.0 [16], and (dotted line) c = 5.5 [8]. (B)
Correlation between J0e and a measure of MWD, M2zþ1 /MzMw, for the Kaschta and Schwarzl [61]. This methodology has been
materials under study and for materials from the literature. The symbols widely used to obtain the complete linear viscoelastic fin-
are the same than in Fig. 5A. The slope is 2.0 as it is expected according to gerprint of polyolefins in the terminal region [48–51]. The
Eq. (6) [9]. discrete retardation spectra {sk, Jk} of the materials under
study have been obtained by fitting both the experimental
comparison, as J 0e is temperature independent in a thermo- values of Jc and Jr to the following equations [35]:
rheologically simple polymer as linear PE [56]. Also Eq. (2) X
n
1
is plotted considering the different exponents proposed by J c ðtÞ ¼ J 0 þ J k ð1  et=sk Þ þ ð12Þ
Mieras and van Rijn (c = 3.7), Bogue et al. (c = 2.0), Kurata K¼1
g0
 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758 2755

Fig. 6. Comparison of the experimental (symbols) and computed (lines) linear viscoelastic response at the T = 160 °C of the samples studied: (A) PE1, (B)
PE2, (C) PE3 and (D) PE4: squares for G0 and circles for G00 (open symbols for creep-recovery converted data and solid symbols for oscillatory measurements).
Solid lines correspond to the computed response using the experimental MWD obtained by SEC by the application of Eqs. (18)–(21). Dashed lines
correspond to the computed response using equivalent log-normal MWD (same Mw and Mw/Mn = Mz/Mn).

X
n X
n
1
J r ðtÞ ¼ J 0 þ J k ð1  et=sk Þ ð13Þ J 0 ðxÞ ¼ Jk ð14Þ
K¼1 K¼1
1 þ x2 s2k

The instantaneous compliance, J0, has been X

n
1 xs2k
considered negligible. The different sets of {sk, Jk} data J 00 ðxÞ ¼ þ Jk ð15Þ
have been found by using the Reptate software [62]. xg0 K¼1
1 þ x2 s2k
As examples, the solid lines in Fig. 4A–C are the
Finally, from the relation between the complex compli-
results obtained for the fitting procedure applied to
ance J*(x) and the complex modulus G*(x), the following
PE3, PE4 and mPE2 samples. Knowing the retardation
equations can be derived to extract the corresponding val-
spectra, the theory of linear viscoelasticity allows the
ues of the storage and loss moduli, G0 (x) and G00 (x) as:
conversion of the time-dependent compliances into the
dynamic moduli as a function of the angular frequency, J 0 ðxÞ
G0 ðxÞ ¼ 2
ð16Þ
x: ½jJ  ðxÞj
2756 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758

8G X1 2
10
5
etp =sd
F 1=2 ðt; MÞ ¼ lðt; MÞ ¼ ð19Þ
p p¼odd p
2 2
'bT /aT (Pa s)

10
4
where G is the renormalization contour length fluctuations
(CLF) factor for the modulus [66]:
 0:5
Me
10
3
G¼1j ð20Þ
M

and sd is the terminal relaxation time of a chain of molar

2
10 mass M that includes the CLF effect on the basis of the
-5 -3 -1 1 3
10 10 10 10 10 entanglement relaxation time, se [66,67]:
-1
a T (rad s )
 3 "  0:5 #2
M Me
Fig. 7. Comparison of the experimental values (symbols) of the dynamic s d ¼ 3s e 1j ð21Þ
Me M
viscosity, g0 , versus angular frequency, x, for the materials studied at
T = 160 °C with those expected by the model for symmetrical log-normal
MWD with the same Mw and Mw/Mn than the polydisperse PE samples being j is a constant of the order unity. We have used in
studied (dotted lines): (hj) PE1, (sd) PE2, (4N) PE3, (O.) PE4. The open our calculations the value of j = 1.69 obtained by Likhtman
symbols correspond to experimental dynamic measurements and the and McLeish [67]. The expression for sd provides the expla-
solid symbols to creep measurements.
nation for the decrease in the relaxation time as the molec-
ular weight decreases, deviating from the predicted
J 00 ðxÞ exponent 3.0 that holds for very long chains to the well-
G00 ðxÞ ¼ 2
ð17Þ known value of 3.4 for shorter chains [60,69]. The other
½jJ ðxÞj
parameters of the model are the entanglement features:
The calculation of the dynamic moduli has been re- the plateau modulus, G0N , the molecular weight between
stricted to the experimental window of the creep-recovery entanglement strands, Me, and the entanglement strand
measurement. The calculated values have been compared relaxation time, se. We have recently obtained all these
with the experimental dynamic shear moduli that were di- parameters for linear PE, by combining experiments and
rectly measured at the same temperature. The dynamic melt atomistic computer simulations [70,71]. These values
shear moduli from calculations and measurements show are: G0N = 2.17 ± 0.23 MPa (experimental average),
a good agreement, as it can be seen in Fig. 6A–D. Addition- Me = 0.790 ± 0.010 kg mol1 (Monte Carlo and Molecular
ally, the range for the dynamic shear moduli has been ex- Dynamics computer simulation) and se = 2.50 ± 0.12 ns
tended to lower frequencies, usually difficult to access in (Molecular Dynamics computer simulation) at T = 160 °C.
the oscillatory mode, as it has been pointed out in preced- This later value has been calculated from the obtained by
ing lines. computer simulations at T = 177 °C, by applying the corre-
It is also of interest to apply the general double repta- sponding temperature shift factor of linear PE. The calcu-
tion mixing rule to obtain the whole linear viscoelastic re- lated values of G0 and G00 are found by appropriate Fourier
sponse of the materials in the broad range of frequencies transform of the results obtained for G(t) from Eqs. (18)–
explored (close to six decades) by making use of the exper- (21).
imental MWD obtained by SEC. Within the double repta- Fig. 6A–D shows the comparison between the calcu-
tion theory framework the expression for the relaxation lated and the experimental results in the polydisperse PE
modulus, G(t), reads: samples listed in Table 1. The agreement between the
model and the experimental data is exceptional, taking
Z 1 2 into account the complex shape of the MWD and the differ-
GðtÞ ¼ G0N F 1=2 ðt; MÞwðMÞd log M ð18Þ ences among the samples. Additionally, it should be men-
log M e
tioned that no empirical or fitting parameters have been
In this equation F(t, M) is the kernel function describ- used to calculate the values of dynamic moduli. We have
ing the relaxation of a molecular weight fraction, w(M), compared in the same figures the experimental results
G0N is the plateau modulus and Me the entanglement with those obtained from the model with the consider-
molecular weight. The exponent in this mixing rule ation of ideal Gaussian MWD functions with the same
was derived theoretically; the Doi-Edwards’ reptation Mw and Mw/Mn than those obtained by SEC in the samples.
model yielded the linear mixing rule with an exponent The model has been widely reported to describe the linear
of 1 [63], and the quadratic mixing rule was obtained viscoelastic fingerprint in linear polymers with very differ-
by des Cloiseaux and Tsenoglou for double reptation ent shape in the MWD, including generalized exponential
[64,65]. Other approaches as dynamic dilution or the functions (GEX) [8], Gaussian or log-normal distributions
Rouse-like motion of the tube considered a slightly high- [69,70] and multimodal log-normal combinations [9,60].
er value of 7/3 [66–68]. Considering the double reptation As it can be observed in Fig. 6, the predicted rheological re-
approach, the kernel function in Eq. (18) can be obtained sponse of a log-normal distribution is shifted to higher fre-
from the tube survival probability l(t, M) of a chain with quencies, pointing towards to a minor contribution of
molecular weight M at time t established by the repta- high-Mw species in the linear viscoelastic fingerprint. This
tion theory [63,67]: shift to higher frequencies causes a difference in the
 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758 2757

viscosity, which is one order of magnitude lower than the cant, but more precisely the shape and predominantly
measured for the polydisperse PE samples with the same the third and the fourth moments of the distribution, Mz
Mw and Mw/Mn (but different Mz/Mw) as it can be appreci- and Mz+1, which seem to play a key role in the viscous
ated in Fig. 7. The results in Figs. 6 and 7 suggest then a and elastic properties of the melt.
strong dependence of the linear viscoelastic response not
only due to Mw/Mn, but also to the shape of the MWD. In
fact, as it has been pointed out in the introduction, this Acknowledgements
dependence has been widely discussed in the literature,
but in the case of linear PE only for samples with values Thanks are due to the CICYT (MAT2009-12364 and
of the Mz/Mn ratio in the range 1.5–11.2 always lower or MAT2012-36341) for financial support. J.R. acknowledges
similar to Mw/Mn [7–18,53–56]. It should be then pointed financial support through the Ramon y Cajal program:
out that the effects observed in the linear viscoelastic prop- Contract RYC-2011-09585. The supply of the samples is
erties of the polydisperse PE samples studied here are not acknowledged to Carlos Martín (Centro Tecnológico Rep-
merely due to a broadening of the MWD in terms of Mw/ sol-YPF, Spain).
Mn, but more precisely to the presence of a significant frac-
tion of very high-Mw species, as judged by the extremely
References
high values of Mz/Mw (2–5 times higher than Mw/Mn) and
to the contribution of the fourth moment of the distribu- [1] Bueche F. Melt viscosity of polymers: effect of polydispersity. J
tion, Mz+1. These high values of Mz and Mz+1 together with Polym Sci 1960;43:527–30.
the asymmetric character of the MWD represent a good [2] Graessley WW. Viscosity of entangling polydisperse polymers. J
Chem Phys 1967;47:1942–53.
opportunity to test the influence of the high-Mw tail of [3] Saeda S, Yotsuyan J, Yamaguchi K. The relation between melt flow
the distribution in the physical properties of these linear properties and molecular weight of polyethylene. J Appl Polym Sci
polymers. The high-Mw tail of the distribution in linear 1971;15:277–92.
[4] Locati G, Gargani L. Dependence of zero-shear viscosity on
PE, even for contents lower than 1% in weight, has a signif-
molecular-weight distribution. J Polym Sci – Polym Lett Ed
icant effect not only on the linear viscoelastic fingerprint 1973;11:95–101.
but also on the non-linear behavior, and more precisely [5] Malkin AY, Blinova NK, Vinogradov GV, Zabugina MP, Sabsai OY,
in the processing and extrudate instabilities regimes dur- Shalganova VC, et al. On the rheological properties of polydisperse
polymers. Eur Polym J 1974;10:445–51.
ing extrusion processes, as we have also probed in previous [6] Zeichner GR, Patel PD. A comprehensive evaluation of polypropylene
works [58–60]. melt rheology. In: Second World Conference Chemical Engineering,
Montreal, Canada 1981;6:333–7.
[7] Wasserman SH, Graessley WW. Prediction of linear viscoelastic
response for entangled polyolefin melts from molecular weight
4. Conclusions distribution. Polym Eng Sci 1996;36:852–61.
[8] Nobile MR, Cocchini F. Predictions of linear viscoelastic properties
for polydisperse entangled polymers. Rheol Acta 2000;39:152–62.
We have studied a series of experimental polyethylene [9] den Doelder J. Viscosity and compliance from molar mass
samples with broad molecular weight distributions and a distributions using double reptation models. Rheol Acta
significant fraction of high molecular weight species. The 2006;46:195–210.
[10] Stadler FJ, Piel C, Kaschta J, Rulhoff S, Kaminsky W, Münstedt H.
analysis of the linear viscoelastic response of the materials Dependence of the zero shear-rate viscosity and the viscosity
shows a conspicuous increase of both the Newtonian vis- function of linear high-density polyethylenes on the mass-average
cosity and the steady-state shear recoverable compliance molar mass and polydispersity. Rheol Acta 2006;45:755–64.
[11] Ansari M, Hatzikiriakos SG, Sukhadia AM, Rohlfing DC. Rheology of
when compared with model linear samples with the same
Ziegler–Natta and metallocene high-density polyethylenes: broad
weight average molecular weight but narrower molecular molecular weight distribution effects. Rheol Acta 2011;50:17–27.
weight distribution. In addition, the polymers exhibit the [12] Vittorias I, Lilge D, Baroso V, Wilhelm M. Linear and non-linear
characteristic temperature dependence of viscoelastic rheology of linear polydisperse polyethylene. Rheol Acta
2011;50:691–700.
properties of linear polyethylene, and previous experi- [13] Mieras HJMA, van Rijn CFH. Elastic behaviour of some polymer
ments and computer simulations also showed the ineffi- melts. Nature 1968;218:865–6.
ciency of the catalyst used in the polymerization to [14] Mills NJ. Elasticity of polydimethylsiloxane melts. Nature
1968;219:1249–50.
incorporate other species than ethylene, such as vinyl [15] Bogue DC, Masuda T, Einaga Y, Onogi S. A constitutive model for
macromonomers, to the growing polymer chains during molecular weight and concentration effects in polymer blends.
the polymerization. In this context, the presence of long Polym J 1970;5:563–72.
[16] Kurata M, Osaki K, Einaga Y, Sugie T. Effect of molecular weight
chain branching causing the variations observed should distribution on viscoelastic properties of polymers. J Polym Sci
be ruled out. In fact, the reptation mixing rule, together 1974;12:849–69.
with the Doi’s relaxation kernel applied to linear chains [17] Ferry JD, Williams ML, Stern DJ. Slow relaxation mechanisms in
concentrated polymer solutions. J Phys Chem 1954;58:987–92.
dynamics, is able to predict the dynamic shear moduli
[18] Agarwal PK. A relationship between steady state shear compliance
using the experimental Size Exclusion Chromatography and molecular weight distribution. Macromolecules 1979;12:342–4.
traces of the samples. This result confirms previous predic- [19] Ressia JA, Villar MA, Vallés EM. Influence of polydispersity on the
viscoelastic properties of linear polydimethylsiloxanes and their
tions about the contribution of the high molecular tail of
binary blends. Polymer 2000;41:6885–94.
the distribution to the melt linear viscoelastic response, [20] Montfort JP, Marin G, Arman J, Monge Ph. Blending law for binary
which in the materials studied here seems to be very blends of fractions of linear polystyrene. Polymer 1978;19:277–84.
important, due to their extremely high values of the Mz/ [21] Zang YH, Muller R, Froelich D. Influence of molecular weight
distribution on viscoelastic constants of polymer melts in the
Mw ratio. It is of particular interest to recall that not merely terminal zone. New blending law and comparison with
the width of the molecular weight distribution is signifi- experimental data. Polymer 1987;28:1577–82.
2758 J. Otegui et al. / European Polymer Journal 49 (2013) 2748–2758

[22] Small BL, Brookhart M, Bennett AMA. Highly active iron and cobalt [45] Carella JM, Gotro JT, Graessley WW. Thermorheological effects of
catalysts for the polymerization of ethylene. J Am Chem Soc long-chain branching in entangled polymer melts. Macromolecules
1998;120:4049–50. 1986;19:659–67.
[23] Britovsek GJP, Gibson VC, Kimberley BS, Maddox PJ, McTavish SJ, [46] Wood-Adams P, Costeux S. Thermorheological behavior of
Solan GA, et al. Novel olefin polymerization catalysts based on iron polyethylene: effects of microstructure and long chain branching.
and cobalt. Chem Commun 1998;7:849–50. Macromolecules 2001;34:6281–90.
[24] Mikenas TB, Zakharov VA, Echevskaya LG, Matsko MA. Kinetic [47] Gabriel C, Kaschta J. Comparison of different shear rheometers with
features of ethylene polymerization over supported catalysts [2,6- regard to creep and creep recovery measurements. Rheol Acta
Bis(imino)pyridyl iron dichloride/magnesium dichloride] with AlR3 1998;37:358–64.
as an activator. J Polym Sci Part A – Polym Chem 2007;45:5057–66. [48] Gabriel C, Kaschta K, Münstedt H. Influence of molecular structure
[25] Cruz VL, Ramos J, Gutiérrez-Oliva S, Toro-Labbé A, Martínez-Salazar on rheological properties of polyethylenes. I. Creep recovery
J. Theoretical study on a multicenter model based on different metal measurements in shear. Rheol Acta 1998;37:7–20.
oxidation states for the bis(imino)pyridine iron catalysts in ethylene [49] Gabriel C, Münstedt H. Creep recovery behavior of metallocene
polymerization. Organometallics 2009;28:5889–95. linear low-density polyethylenes. Rheol Acta 1999;38:393–403.
[26] Cruz VL, Martínez J, Martínez-Salazar J, Ramos J, Reyes ML, Toro- [50] Gabriel C, Münstedt H. Influence of long-chain branches in
Labbé A, et al. QSAR model for ethylene polymerisation catalysed by polyethylenes on linear viscoelastic flow properties in shear. Rheol
supported bis(imino)pyridine iron complexes. Polymer Acta 2002;41:232–44.
2007;48:7672–8. [51] Stadler F, Münstedt H. Terminal viscous and elastic properties of
[27] Martínez J, Cruz V, Ramos J, Gutiérrez-Oliva S, Martínez-Salazar J, linear ethane/a-olefin copolymers. J Rheol 2008;52:697–712.
Toro-Labbé A. On the nature of the active site in bis(imino)pyridyl [52] Dealy JM, Larson RG. Structure and rheology of molten
iron, a catalyst for olefin polymerization. J Phys Chem C polymers. Munich: Hanser Publishers; 2006.
2008;112:5023–8. [53] Mills NJ. The rheological properties and molecular weight
[28] Vega JF, Otegui J, Exposito MT, Lopez M, Martín C, Martínez-Salazar J. distribution of polydimethylsiloxane. Eur Polym J 1969;5:675–95.
Structure and physical properties of polyethylenes obtained from [54] Plazek DJ, Ragupathi N, Kratz RF, Miller WR. Recoverable compliance
dual catalysis process. Polym Bull 2008;60:331–42. behavior of high-density polyethylenes. J Appl Polym Sci
[29] Vega JF, Otegui J, Ramos J, Martínez-Salazar J. Effect of molecular 1979;24:1305–20.
weight distribution on Newtonian viscosity of linear polyethylene. [55] Carella JM, Graessley WW, Fetters LJ. Effects of chain microstructure
Rheol Acta 2012;51:81–7. on the viscoelastic properties of linear polymer melts:
[30] Aguilar M, Vega JF, Sanz E, Martínez-Salazar J. New aspects on the polybutadienes and hydrogenated polybutadienes. Macromolecules
rheological behavior of metallocene catalysed polyethylenes. 1984;17:2775–86.
Polymer 2001;42:9713–21. [56] Yoshikawa K, Toneaki N, Moteki Y, Takahashi M, Masuda T. Dynamic
[31] Grubisic Z, Rempp P, Benoit H. A universal calibration for gel viscoelastic, stress relaxation and elongational flow behavior of high
permeation chromatography. J Polym Sci Part B – Polym Lett density polyethylene melts. In: Collier I, Utracki LA, editors. Polymer
1967;5:753–9. rheology and processing. New York: Elsevier; 1990. p. 37–53.
[32] Drott EE, Mendelson RA. Determination of polymer branching with [57] Resch JA, Stadler FJ, Kaschta J, Munstedt H. Temperature dependence
gel-permeation chromatography. II. Experimental results for of the linear steady-state shear compliance of linear and long-chain
polyethylene. J Polym Sci Part A2 Polym Phys 1970;8:1373–85. branched polyethylenes. Macromolecules 2009;42:5676–83.
[33] Cote JA, Shida M. Long-chain branching in low-density polyethylene. [58] Aguilar M, Expósito MT, Vega JF, Muñoz-Escalona A, Martínez-
J Polym Sci Part A2 – Polym Phys 1971;9:421–30. Salazar J. Elimination of extrudate distortions in metallocene-
[34] Williamson GR, Cervenka A. Characterization of low-density catalyzed polyethylene. Macromolecules 2004;37:681–3.
polyethylene by gel permeation chromatography-I. The universal [59] Aguilar M, Martín S, Vega JF, Muñoz-Escalona A, Martínez-Salazar J.
calibration curve and Mark-Houwink equation. Eur Polym J Processability of a metallocene-catalyzed linear PE improved by
1972;8:1009–17. blending with a small amount of UHMWPE. J Polym Sci Part B –
[35] Ferry JD. Viscoelastic properties of polymers. 3rd ed. New York: Jon Polym Phys 2005;43:2963–71.
Wiley and Sons; 1980. [60] Vega JF, Expósito MT, Otegui J, Martínez-Salazar J. Eliminating
[36] Ramos J, Cruz V, Muñoz-Escalona A, Martínez-Salazar J. A sharkskin distortion in polyethylene extrusion via a molecular route.
computational study of iron-based Gibson–Brookhart catalysts for J Rheol 2011;55:855–74.
the copolymerisation of ethylene and 1-hexene. Polymer [61] Kaschta J, Schwarzl ER. Calculation of discrete retardation spectra
2002;43:3635–45. from creep data-I. Method Rheol Acta 1994;33:517–29.
[37] Huang R, Koning CE, Chadwick JC. Effects of hydrogen in ethylene [62] Ramírez J, Likhtman A. Rheology of Entangled Polymers: Toolkit for
polymerization and oligomerization with magnesium chloride- Analysis of Theory and Experiment. <http://www.reptate.com>.
supported bis(imino)pyridyl iron catalysts. J Polym Sci Part A – [63] Doi M, Edwards SF. The theory of polymer
Polym Chem 2007;45:4054–61. dynamics. Oxford: Clarendon; 1986.
[38] Piel C, Stadler FJ, Kaschta J, Rulhoff S, Münstedt H, Kaminsky W. [64] Tsenoglou C. Viscoelasticity of binary homopolymer blends. Am
Structure-property relationships of linear and long-chain branched Chem Soc Polym Prep – Div Polym Chem 1987;28:187–8.
metallocene high-density polyethylenes characterized by shear [65] des Cloizeaux J. Double reptation vs simple reptation in polymer
rheology and SEC-MALLS. Macromol Chem Phys 2006;207:26–38. melts. Europhys Lett 1988;5:437–42.
[39] Berry GC, Fox TG. The viscosity of polymers and their concentrated [66] Milner ST, McLeish TCB. Parameter-free theory for stress relaxation
solutions. Adv Polym Sci 1967;5:261–357. in star polymer melts. Macromolecules 1997;30:2159–66.
[40] Vega JF, Santamaría A, Muñoz-Escalona A, Lafuente P. Small- [67] Likhtman AE, McLeish TCB. Quantitative theory for linear dynamics
amplitude oscillatory shear flow measurements as a tool to detect of linear entangled polymers. Macromolecules 2002;35:6332–43.
very low amounts of long chain branching in polyethylenes. [68] van Ruymbeke E, Keunings R, Stephenne V, Hagenaars A, Bailly C.
Macromolecules 1998;31:3639–47. Evaluation of reptation models for predicting the linear viscoelastic
[41] Malmberg A, Liimatta J, Lehtinen A, Lofgren B. Characteristics of long properties of entangled linear polymers. Macromolecules
chain branching in ethene polymerization with single site catalysts. 2002;35:2689–99.
Macromolecules 1999;32:6687–96. [69] Vega JF, Rastogi S, Peters GWM, Meijer HEH. Rheology and
[42] Wood-Adams PM, Dealy JM, DeGroot AW, Redwine OD. Effect of reputation of linear polymers. Ultra high molecular weight chain
molecular structure on the linear viscoelastic behavior of dynamics in the melt. J Rheol 2004;48:63–678.
polyethylene. Macromolecules 2000;33:7489–99. [70] Ramos J, Vega JF, Theodorou DN, Martínez-Salazar J. Entanglement
[43] Soares JBP, Hamielec AE. Bivariate chain length and long chain relaxation time in polyethylene: simulation versus experimental
branching distribution for copolymerization of olefins and polyolefin data. Macromolecules 2008;41:2959–62.
chains containing terminal double-bonds. Macromol Theor Simul [71] Ramos J, Vega JF, Martínez-Salazar J. Assessment of entanglement
1996;5:547–72. features and dynamics from atomistic simulations and experiments
[44] Expósito MT. Ph.D. Thesis. Universidad Complutense de Madrid, in linear and short chain branched polyolefins. Soft Matter
Madrid (Spain); 2006. 2012;8:6256–63.