crystal structure, melting behaviour and equilibrium melting point of star polyesters with...
TRANSCRIPT
Crystal structure, melting behaviour and equilibrium melting point of star
polyesters with crystallisable poly(1-caprolactone) arms
E. Nuneza, C. Ferrandoa, E. Malmstroma, H. Claessona, P.-E. Wernerb, U.W. Geddea,*
aDepartment of Fibre and Polymer Technology, Royal Institute of Technology, SE-100 44 Stockholm, SwedenbDepartment of Inorganic, Physical and Structural Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden
Received 12 June 2003; received in revised form 1 March 2004; accepted 3 May 2004
Available online 11 June 2004
Abstract
Star polymers consisting of poly(1-caprolactone), PCL, attached to third generation dendrimer, hyperbranched and dendron cores have
been studied by differential scanning calorimetry and wide-angle X-ray scattering. The degree of polymerisation of the PCL arms of the star
polymers ranged from 14 to 81. The crystal unit cell was the same for the star polymers as for their linear PCL analogues. The star polymers
showed a lower degree of crystallinity than the linear PCL, suggesting that the dendritic cores imposed restriction on PCL crystallisation.
Slow heating of rapidly cooled samples led to crystal rearrangement—a gradual increase in melting point with decreasing heating rate and
recrystallisation followed by additional high temperature melting. The tendency for crystal rearrangement was less pronounced in star
polymers based on dendrimer or hyperbranched cores, suggesting that the dendritic cores constitute an obstacle to crystal rearrangement. The
star polymers showed higher equilibrium melting points than the linear PCL analogues. It is suggested that covalent attachment of the PCL
arms to the dendritic core reduced the positional freedom and the entropy of the melt with respect to that of linear PCL.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Star polymer; Poly(1-caprolactone); Equilibrium melting point
1. Introduction
A star polymer consists of a central core linking together
three or more linear polymer chains. The core, as in the
present study, can be a dendritic molecule based on an ABx
ðx $ 2Þ monomer to which linear polymer chains are
attached. The core may, depending on the method of
synthesis, be a dendrimer (monodisperse) or hyperbranched
polymer with vacancies in the branching. The high degree of
branching prohibits crystallisation of the core moieties. The
attachment of regular linear units, such as poly(1-capro-
lactone) (PCL), to the fully amorphous, dendritic core
molecule makes the polymer semicrystalline [1]. The
attachment of the crystallisable units to the rigid, non-
crystallisable core is expected to have profound effects on
their crystallisation and on the crystalline morphology. This
paper, which is the first of a series, reports the effect of the
molecular architecture on the crystal unit cell structure, the
melting behaviour and the equilibrium melting point. Most
studies [2–5] on semicrystalline star polymers have
reported only ‘kinetic’ melting points without any attempt
to assess the equilibrium melting point. Risch et al. [6],
however, reported that the equilibrium melting point of star-
branched polyamide-6 of finite molar mass decreased with
increasing branch-point functionality, and attributed this to
a lower enthalpy of fusion of the star-branched polymers.
Linear PCL is a highly crystalline polymer. Thermodyn-
amic data relating to the fusion of linear PCL were reported
by Crescenzi et al. [7]: the heat of fusion at the equilibrium
melting points was determined to 135 J g21 and the
equilibrium melting point of high molar mass linear PCL
by the diluent method to 336 K. The most extensive studies
on the crystallization and melting behaviour of linear PCL
are due to Phillips et al. [8,9]. They reported equilibrium
melting point data obtained by the Hoffman–Weeks method
in the range 341–343.5 K depending on molar mass (7000–
40,000 g mol21), spherulite growth rate data obeying a
single crystallization regime (Regime II) according to the
Lauritzen–Hoffman theory and yielding a fold surface free
energy of ,90 mJ m22 and isothermal data on crystal
thickening. Strobl et al. [10] recorded the small-angle X-ray
0032-3861/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2004.05.047
Polymer 45 (2004) 5251–5263
www.elsevier.com/locate/polymer
* Corresponding author. Tel.: þ46-8-790-7640; fax: þ46-8-208-856.
E-mail address: [email protected] (U.W. Gedde).
scattering pattern during isothermal crystallization and at a
later stage during melting of the isothermally crystallized
samples. They found for a linear PCL with a molar mass of
42,000 g mol21, as they did for a few other polymers
(syndiotactic PP, PE), that the intersection of melting point
vs. (crystal thickness)21 and crystallization temperature vs.
(crystal thickness)21 appeared at a finite crystal thickness
and not as expected at near-infinite crystal thickness.
Extrapolation of the melting point data presented by Strobl
et al. [10] according to the Thomson–Gibbs equation
yielded an equilibrium melting point near 370 K. It should
be pointed out that the extrapolation range was very long
and that the data used for extrapolation covered only a
narrow crystal thickness range. The slope coefficient of the
melting point-reciprocal crystal thickness data presented by
the Strobl group yielded a very low fold surface free energy
ðsÞ; ,60 mJ m22. Melting point-reciprocal crystal thick-
ness data for polyethylene were presented in the same paper
and surprisingly even these corresponded to
s < 60 mJ m22, which is significantly less than the
established value, ,90 mJ m2 [11]. The Strobl group
reported in their paper [10] that no isothermal crystal
thickening occurred during crystallization and that crystal
thickening occurred only during the final stages of melting.
The crystal thickening occurring during the slow heating
used was particularly strong effect for crystals grown at
temperatures below 313 K. Nojima et al. [12] reported for a
linear PCL (degree of polymerisation < 60) a maximum
30% increase in the long period during 2 K min21 heating
from the crystallization temperature (310–319 K) to the
melting point. Phillips et al. [9] found that the melting point
increased with the logarithm of the crystallization time
according to an S-shaped curve. Crystal thickness data
calculated from melting point data using the Thomson–
Gibbs equation taken for samples crystallized at different
temperatures could be shifted along the logarithmic time
axis to obtain a master curve; the temperature dependence of
the shift factor followed the Arrhenius equation. The
heating rate used by Phillips et al. was 10 K min21 which
was higher than that used by Strobl et al.
The crystal unit cell of linear PCL resembles to some
extent that of polyethylene: the lateral packing of the chains
is very similar and the sub-cell is orthorhombic with the
space group P212121 [13,14]. Chatani et al. [13] was the first
to determine the crystal structure by X-ray diffraction of
stretched films after annealing: a ¼ 0:747 nm,
b ¼ 0:498 nm, and c ¼ 1:705 nm (orthorhombic sub-cell).
These values refer to the cell dimension at room
temperature. The crystalline stem was assumed to be
twisted slightly to account for the c spacing obtained.
Bittiger et al. [14] reported for linear PCL an orthorhombic
sub-cell with the dimensions a ¼ 0:7496 nm,
b ¼ 0:4974 nm, and c ¼ 1:7297 nm (room temperature).
They proposed that the chain conformation of the crystalline
stem was all-trans with two repeating units included in the
sub-cell along c: Later work by Hu and Dorset [15] using
electron diffraction showed that the crystalline stem had to
be twisted slightly along c: The dimensions of the
orthorhombic cell at room temperature was according to
this study: a ¼ 0:748 nm, b ¼ 0:498 nm, and c ¼ 1:726 nm.
The equilibrium melting point of a polymer is seldom
measured directly, because chain folding limits the crystal
thickness and the experimental melting points are much
lower than that of the equilibrium crystal. The equilibrium
melting point can however be determined, although often
with considerable uncertainty, by suitable extrapolation of
experimental data. The Thomson–Gibbs equation can be
used to extrapolate accurately taken melting point data of
polymers with known crystal thicknesses to infinite crystal
thickness in order to obtain the equilibrium melting point
[16]. The Hoffman–Weeks method [17] determines the
equilibrium melting point by linear extrapolation of Tm ¼
f ðTcÞ data to Tm ¼ Tc: The method is strictly applicable only
to a series of samples with crystals of a certain constant
crystal-thickening factor. Strobl et al. [10] raised serious
doubts about the validity of the Hoffman–Weeks method by
showing that the extrapolated equality Tm ¼ Tc occurred at
a finite crystal thickness of several semicrystalline poly-
mers. A third class of methods uses equilibrium melting
point data of oligomers with n repeating units with the same
crystal unit cell structure as the polymer by extrapolation to
n!1 [18–20]. The crystallisable PCL chains included in
this study are all of finite size, with maximum 117 repeating
units corresponding to a stem length of ,100 nm. The large
size of the dendritic core sets the limit for maximum crystal
thickness of the star polymers to the extended chain length
of single PCL arms. This puts the maximum crystal
thickness of the studied star polymers in the range 12–
70 nm. Different strategies were used to extrapolate thermal
analysis data to obtain the equilibrium melting point of the
finite-sized system. A theoretical discussion concerned with
this problem is presented in Section 3.
This paper reports equilibrium melting point data
obtained by analysis of melting point–crystallization
temperature data with a series of linear and star polymers
based on dendrimer, hyperbranched and dendron cores. X-
ray scattering was used to characterize the crystal structures
of the polymers and it was confirmed that the crystal
structures of the different polymers were the same. The
melting behaviour was studied at different heating rates and
it was found that the crystals in the star-branched polymers
rearranged at a lower rate than the crystals in the linear PCL
analogues. The equilibrium melting point data of the
polymers with crystallisable PCL units of finite molar
mass were finally extrapolated to infinite PCL molar mass
using the Flory–Vrij method and the Broadhurst equation.
The experimental melting point data recorded for the star
polymers were typically 3–5 K higher than those of the
linear PCL’s with corresponding PCL chain length crystal-
lized at the same temperature. The equilibrium melting
temperatures calculated for the star polymers were also
found to be higher than those of linear PCL.
E. Nunez et al. / Polymer 45 (2004) 5251–52635252
2. Experimental
Star polymers, consisting of PCL grafted to dendritic
hydroxyl-functional cores, were studied. The cores were
third-generation hyperbranched polyester with approxi-
mately 32 terminal hydroxyl groups (Boltorn, Perstorp
AB, Sweden), a third-generation dendrimer with 24
hydroxyl groups and a third-generation dendron with 8
hydroxyl groups. The dendrimer and dendron core polymers
were synthesized according to Ihre et al. [21]. The synthesis
of star polymers, i.e. the attachment of the PCL onto the
dendritic cores, was achieved according to Claesson et al.
[22]. Linear PCL’s - Tone P300, P1270 and P1241—
purchased from Union Carbide, USA were studied as
received. The structures of the star polymers studied are
displayed in Fig. 1. Molecular structure data of the polymers
studied are presented in Table 1.
The crystal unit cell structure was determined by wide-
angle X-ray scattering using a Huber Imaging Plate Guinier
Camera G670 operating at 40 kV and 30 mA using strictly
monochromatic CuKa1 radiation ðl ¼ 0:15406 nmÞ: Data
in the transmission mode were collected at 0.0058 incre-
ments in the range 4–1008 in 2u: The indexing program
TREOR [23] was used to assess the unit cell parameters.
Samples weighing 5 ^ 2 mg were encapsulated in 40 ml
aluminium pans and analysed in a temperature- and energy-
calibrated Mettler-Toledo DSC 820 purged with dry
nitrogen. Temperature calibration was performed by
recording the melting of highly pure indium at the actual
heating rates used. The samples were crystallized at
constant temperature in the range 303–321 K after 2 min
storage at 348 K and 10 K min21 cooling to the selected
crystallisation temperature. This was the maximum cooling
rate possible to use to avoid ‘over-shooting’. Calorimetric
data were recorded during both the cooling phase and during
the isothermal stage to make certain that all crystallization
occurred during the isothermal stage. The samples were
crystallized under isothermal conditions to 10% of the final
attainable degree of crystallinity at the particular tempera-
ture in an attempt to keep the crystal-thickening factor
constant at the different crystallisation temperatures. The
melting peak temperature was finally recorded at
10 K min21. The recorded melting point data were cor-
rected for the thermal lag between the sample pan and the
sample holder in accordance with Gedde and Jansson [24].
Finally, the melting curves of samples of each material were
recorded at different heating rates, after a specific crystal-
lisation procedure: cooling from 348 to 293 K at
30 K min21. The mass crystallinity of samples cooled at
10 K min21 from 353 to 250 K was assessed by the total
Fig. 1. Chemical structures of the star polymers studied: Dn—dendrimer core HBn—hyperbranched core; Donn—dendron core. The number of repeating units
in a single PCL arm is indicated by n:
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5253
enthalpy method [16] using 135 J g21 [7] as the heat of
fusion of 100% crystalline PCL.
3. Theory: assessment of equilibrium melting point
The equilibrium melting point ðT0mÞ of a homopolymer is
the melting point of an infinitely thick crystal based on a
specific crystal phase with equilibrium density of internal
defects. Crystals of thicknesses in the range of micrometer
are made by high-pressure crystallisation of a few polymers,
e.g. polyethylene [25]. The melting point depression
originating from the finite crystal thickness of crystals in
the micrometer thickness range is less than 0.1 K according
to calculations based on the Thomson–Gibbs equation. In
many cases, the equilibrium melting point is obtained by
extrapolation of melting point data ðTmÞ of polymer crystals
of finite crystal thickness ðLcÞ: The negligible contribution
from the lateral surfaces to the stability of polymer crystals
is the basic assumption underlying the Thomson–Gibbs
equation:
Tm ¼ T0m 1 2
2s
Dh0Lc
� �ð1Þ
where s is the specific surface free energy of the fold
surface and Dh0 is the heat of fusion at the equilibrium
melting point. Extrapolation of Tm 2 L21c data to L21
c ¼ 0
yields the equilibrium melting point provided that the
following requirements are fulfilled: (i) the same crystal
phase must be present in the crystals of finite thickness as in
the infinitely thick, equilibrium crystal. (ii) Crystal
rearrangement (e.g., crystal thickening) has to be inhibited
during the heating and the recording of the melting point.
(iii) Superheating effects must be inhibited. (iv) The specific
fold surface free energy must be the same for all samples
used for extrapolation. (v) The concentration of defects in
the crystals must be the same for all the samples studied and
also the same as in the equilibrium crystal. In addition,
many polymer samples contain crystals with a distribution
in crystal thickness and it is difficult to select a ‘typical’
crystal thickness in a given sample.
Hoffman and Weeks [17] proposed a method based on
the assessment of melting points of crystals grown at
different crystallization temperatures ðTcÞ :
Tm ¼Tc
bþ T0
m 1 21
b
� �ð2Þ
where b is the crystal thickening factor ¼ Lc=Lpc ; where Lpc is
the thickness of the virgin crystal with the melting point
equal to the crystallization temperature. Eq. (2) is
approximately valid provided that the stability of the first
formed crystal is just slightly greater than the minimum
requirement, i.e. Lpc ¼ Lc;min þ dLc < Lc;min; where Lc;min is
the minimum crystal thickness corresponding to a melting
point equal to the crystallization temperature. The equili-
brium melting point is according to Eq. (2) obtained by
extrapolation of Tm 2 Tc data to Tm ¼ Tc; provided b is
constant between the different samples used for extrapol-
ation. More recent data on linear polyethylene [26] and
poly(ethylene oxide) [27] showed that the thickness of the
virgin crystals exceeds the minimum thickness by a more
significant dLc value, 4–5 nm. This gives the Tm 2 Tc
relationship a curvature, which is often seen in practice. The
equation that relates melting point and crystallization
temperature is:
Tm ¼ T0m 1 2
2s
Dh0b2sT0
m
Dh0ðT0m 2 TcÞ
þ dLc
" #266664
377775 ð3Þ
Table 1
Nomenclature and molecular structure of samples studied
Group Core Mcore (g mol21)a Number of PCL arms Sample code nb
Linear PCL – – 1 PCL17 17 (1.2)
PCL39 39 (1.5)
PCL117 117 (1.4)
Dendrimer-PCL 3rd-Generation dendrimer 3006 24 D14 14 (1.3)
D24 24 (1.3)
D42 42 (1.3)
D51 51 (1.3)
Hyperbranched-PCL 3rd-Generation Boltorn 3607 32 HB51 51 (1.3)
HB79 79 (1.4)
Dendron-PCL 3rd-Generation dendron 1008 8 Don15 15 (1.3)
Don46 46 (1.3)
Don81 81 (1.4)
a Molar mass of core calculated theoretically from the chemical formula.b Number average of the degree of polymerisation of the PCL arms ðnÞ; obtained by H NMR [22]. The values within parentheses show the polydispersity
ð �Mw= �MnÞ obtained by size exclusion chromatography applying the universal calibration procedure. The values presented for the star polymers are estimated
from data for linear PCL polymerized under similar conditions.
E. Nunez et al. / Polymer 45 (2004) 5251–52635254
The equilibrium melting point can be obtained by implicit
solution of Eq. (3). The importance of a significant dLc term
on the Tm 2 Tc curvature and the tendency for under-
estimation of the equilibrium melting point by rectilinear
extrapolation of Tm 2 Tc curvature was recognized in the
comprehensive study of Marand et al. [28]. It was also
mentioned briefly in an earlier publication by one of the
authors of this paper [16].
The equilibrium melting point of oligomers and some
low molar mass polymers can in some cases by determined
directly by melting of extended-chain crystals grown at high
temperatures. Extrapolation of Tm 2 Tc data to Tm ¼ Tc in
order to obtain the equilibrium melting point is not strictly
valid because the relation dLc p Lc;max (Lc;max is the length
of the extended chain, i.e. maximum crystal thickness) does
not hold. However, the equilibrium melting point of the
finite sized system can be obtained by assuming that the
relationship Lc ¼ Lc;min þ dLc holds with a constant dLc
independent of crystallisation temperature. The difference
ðDTÞ between melting point and crystallisation temperature
for the equilibrium crystal with the thickness Lc;max can be
estimated from the Thomson–Gibbs equation according to:
DT ¼ T0mðnÞ2 Tc
¼ T0mðn!1Þ
2s
Dh0
1
Lc;max 2 dLc
� � 2 1
Lc;max
" #ð4Þ
where T0mðn!1Þ is the equilibrium melting point of a
polymer with infinite molar mass. The equilibrium melting
point is thus obtained at the point where the fitted curve
according to Eq. (3) deviates from Tm 2 Tc line by DT (see
the example shown in Fig. 10). The temperature at which
the curve obtained by fitting of Eq. (3) intersects the Tm ¼
Tc line is referred to as the virtual equilibrium melting point.
One of the established methods to determine the
equilibrium melting point of a polymer of infinite molar
mass is to extrapolate melting point data of extended-chain
crystals of oligomers showing the same crystal packing as
the polymer to infinite molar mass. A prerequisite of the
method is that no intermediate phases exist. Extrapolation
methods have been proposed by Broadhurst [18,19], Eq. (5),
and Flory and Vrij [20], Eq. (6):
T0mðnÞ ¼ T0
mðn!1Þðnþ aÞ
ðnþ bÞð5Þ
where a and b are constants to be adjusted in the fit of the
equation to the experimental data.
nDh0
R
!DT 2
nDCp
2R
� �DT2 2 T0
mðnÞT0mðn!1Þln n
¼T0
mðn!1Þ
R
!ðT0
mðnÞDse 2 DheÞ ð6Þ
where DCp is the difference in specific heat between the
amorphous and crystalline PCL, Dse and Dhe are the end-
group contributions to entropy and enthalpy changes at the
melting point. Hay [29] showed that Eq. (6) could be
simplified to:
Tm ¼ T0m 1 2 2RTm
ln n
nDh0
� �� �ð7Þ
where Dh0 is the heat of fusion per mole repeating units. Eq.
(7) predicts that the equilibrium melting point is obtained as
the intercept in a plot of Tm vs. ln n=n:
The equilibrium melting point of linear polyethylene has
been determined by the three methods and also by direct
measurement of the melting point of extended-chain
crystals grown at elevated pressure. Extrapolation based
on the Thomson–Gibbs equation gave the values 414.7 K
[30] and 415.2 K [31]. Extrapolation of Tm 2 Tc data
according to the Hoffman–Weeks method yielded 418.2 K
[17] and 418.4 using the non-linear method (significant dLc)
[28]. Extrapolation of melting point data of extended-chain
crystals of oligomers yielded 414.3 K according to the
Broadhurst equation [18,19], 418.7 K according to Flory
and Vrij [20] and 419 K according to Hay [29]. The melting
point of crystals grown at elevated pressure gave 414.6 K
[32]. Thus, the different estimates of the equilibrium
melting point ranges from 414.3 to 419 K. Strobl et al.
[10] estimated from the their melting point–crystal thick-
ness data an equilibrium melting point data of 418–
419 K. The intersection between Tc 2 L21c and Tm 2 L21
c
occurred at finite crystal thickness at 411 K, which was
considerably lower than earlier estimates of the equilibrium
melting point obtained by the Hoffman–Weeks method.
The following data for linear PCL have been used in the
calculations: Dh0 ¼ 163 MJ m23 ¼ 135 kJ ðkgÞ21 ¼ 2:2 kJ
(mol backbone atoms)21 [7]; s ¼ 90 mJ m22 [8]. Crystal
stem length per PCL repeating unit ¼ 0.863 nm [15].
4. Results and discussion
4.1. Crystal unit cell
The X-ray diffraction patterns of the star polymers and
the linear polymers were almost identical. The recorded
diffraction patterns were in accordance with the earlier
reported data of Hu and Dorset [15] and the assessment of
the unit cell by the indexing program TREOR was
consistent with the suggested orthorhombic cell. The
positions of the two intense reflections originating from
(110) and (200) of the orthorhombic unit cell were used to
calculate the corresponding d-spacings (Table 2).
The star polymers showed the following d-spacings:
0.41615 ^ 0.00119 nm (110) and 0.37364 ^ 0.00055 nm
(200). These values are very similar to the values obtained
for the linear PCL, 0.41617 nm (110) and 0.37392 nm
(200). The percentage difference between the mean values
of the star polymers and the values obtained for the linear
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5255
PCL were: 20.4% (110) and 20.1% (200); where the
minus signs indicate that the d-spacings were larger for the
linear PCL than for the star polymers. Thus, the X-ray
diffraction data showed that the dendritic cores had no
impact on the unit cell structure of the PCL crystals.
4.2. Degree of crystallinity
The PCL crystallinity (i.e. the mass of PCL crystal phase
divided by the total PCL mass) of samples crystallized
during a 10 K min21 cooling from 353 to 273 K was larger
in linear PCL than in the star polymers (Fig. 2). The degree
of crystallinity of these samples sensitively reflects
differences between different samples with regard to their
ability to crystallize. A common feature of linear polymers
is that the degree of crystallinity decreases with increasing
molar mass. This was demonstrated by Ergoz et al. [33] for
narrow fractions of linear polyethylene covering three
orders of magnitude in molar mass. The linear PCL samples,
not fully ranging one order of magnitude in molar mass,
showed a 10% change in mass crystallinity, which is five
times greater than the standard deviation of the crystallinity
data taken (Fig. 2).
The star polymers based on dendrimer and dendron cores
showed a similar depression in PCL crystallinity with
respect to linear PCL, whereas the star polymers based on
the hyperbranched cores showed an even larger depression
in PCL crystallinity. The depression in PCL crystallinity
ðDwcðnÞÞ of the star polymers with reference to linear PCL
with a corresponding degree of polymerisation of the PCL
units is given by:
DwcðnÞ ¼ wc;PCLðnÞ2 wc;star-PCLðnÞ ð8Þ
where wc;PCLðnÞ is the mass crystallinity of linear PCL with
n repeating units and wc;star-PCLðnÞ is the PCL mass
crystallinity of the star polymer with PCL arms with n
repeating units. The mass crystallinity values for linear PCL
used in the calculations of Dwc were based on a linear
function fitted to the experimental data (Fig. 2). The number
of additional, amorphous PCL repeating ðDnÞ units per PCL
arm in the star polymers with reference to linear PCL was
calculated from:
Dn ¼ Dwcn ð9Þ
Fig. 3 shows the depression in PCL crystallinity in the star
polymers with respect to linear PCL expressed in Dn; as a
function of degree of polymerisation of the PCL arms ðnÞ:
The data obtained for the star polymers with dendrimer and
dendron cores followed a linear trend according to:
Dn ¼ Dn0 þ Cn ð10Þ
where Dn0 is the intercept, i.e., a term independent of the
PCL chain length. The data obtained for the star polymers
with dendrimer or dendron cores suggest that only one
repeating unit ðDn0 < 1Þ of PCL nearest to the core is
prohibited from crystallizing. The linear increase in Dn with
n expressed by the slope coefficient ðCÞ—equal to 0.08 for
the star polymers with dendrimer or dendron cores—
suggests that the chain folding occurring in the star
polymers with longer PCL arms was influenced by the
presence of the dendritic cores near the fold surfaces, as that
this leads to less tight folds.
The star polymers with hyperbranched core showed
an even greater crystallinity depression, corresponding to
Table 2
Crystal structure data obtained by X-ray diffraction
Sample d110 (nm) d200 (nm)
PCL117 0.41617 0.37392
D42 0.41572 0.37397
HB79 0.41523 0.37301
Don81 0.41750 0.37394
Fig. 2. Mass PCL crystallinity ðwcÞ as a function of degree of
polymerisation of a single PCL arm ðnÞ for linear PCL (B), dendrimer-
PCL (A), hyperbranched-PCL (W) and dendron-PCL (X). The crystallinity
assessments were made on samples kept at 353 K in the molten state for
1 min and then cooled at 10 K min21 to 273 K.
Fig. 3. Crystallinity depression in star polymers with reference to linear
PCL expressed as the number of PCL repeating units ðDnÞ per PCL arm
plotted against the degree of polymerisation ðnÞ of a single PCL arm:
dendrimer-PCL (A), hyperbranched-PCL (W) and dendron-PCL (X). The
standard deviation of single data was ,0.02. The continuous line is a linear
fit of the data for dendrimer-PCL and dendron-PCL. The broken line is a
linear fit of the data for hyperbranched PCL assuming the same slope
coefficient as that obtained for dendrimer-PCL and dendron-PCL.
E. Nunez et al. / Polymer 45 (2004) 5251–52635256
Dn=n ¼ 0:14–0:20: Adjustment of Eq. (10) to the exper-
imental data obtained for the polymers with hyperbranched
core would lead to a very high Dn0; between 8 and 9. This in
combination with the low C-value obtained, 0.02–0.03,
indicates that the two data points available for the
hyperbranched star polymers are not sufficient for reliable
determination of Dn0 and C: It seems reasonable to assume,
however, that the irregular branching in the hyperbranched
cores makes it difficult for the ‘inner’ PCL units to
crystallize, which leads to a high Dn0: It can be argued
that the slope coefficient ðCÞ should be the same for the
hyperbranched star polymers as for dendrimer and dendron-
based star polymers. In that instance, the intercept ðDn0Þ
becomes ,5 for the hyperbranched polymers.
4.3. PCL crystal melting at different heating rates
The melting of linear PCL first crystallized during a
30 K min21 cooling from the melt may serve is a good
example showing the principal effects of varying heating
rate (Fig. 4(a)). The melting trace obtained on slow heating
(1 K min21) contained two well-resolved melting peaks.
The two peaks were shifted to lower temperatures with
increasing heating rate. The temperature difference between
the two peaks was approximately constant, 3.2 ^ 0.1 K, in
the heating rate range 1–5 K min21. The two peaks
gradually merged at the higher heating rates; the tempera-
ture difference was only 2.3 K at 10 K min21. The relative
size of the high temperature peak decreased with increasing
heating rate (Fig. 4(b)). These findings suggest that the
bimodal melting was due to the following sequence of
events: melting of the initial crystals, recrystallisation and
final melting of the crystals grown during the heating. The
same basic scheme applied to both the linear and the star
polymers. The differences between the different polymers
were only in the rate of the recrystallisation process.
The other possible cause for bimodal melting—the initial
presence of two crystal populations different in crystal phase
or crystal thickness—seems unlikely in the present case.
The X-ray diffraction patterns of the polymers studied
showed only the presence of crystals with the orthorhombic
unit cell. The relatively low polydispersity of the samples
studied and the fact that all the samples crystallized during a
constant rate cooling makes it unlikely that two crystal
populations with distinctly different crystal thickness would
be present immediately after the constant rate cooling.
The star polymers with dendrimer or hyperbranched
corers showed a significantly smaller high temperature peak
than the linear PCL at all corresponding heating rates (Fig.
4(a)–(d)). The melting merged into one melting peak at
heating rates $10 K min21 for the star polymers with
dendrimer or hyperbranched corers (Fig. 4(c) and (d)).
Linear PCL and star PCL with a dendron core given the
same thermal treatment showed bimodal melting even up to
a heating rate of 30 K min21 (Figs. 4(a), (e) and 5(a)). The
lower tendency of the star polymers with dendrimer or
hyperbranched cores to rearrange on heating with reference
to that of linear PCL is consonant with earlier data reported
by Nojima et al. [12]. They reported that the long period
associated with the PCL crystals in block copolymers based
on PCL and polybutadiene remained essentially constant
during heating from the crystallisation temperature to the
melting point. Linear PCL showed on the other hand some
increase in the long period on heating to the melting point.
The low temperature peak was due to the melting of the
crystals formed during the initial cooling phase. The heating
rate dependence of the low temperature peak for both linear
PCL and star PCL with dendrimer core suggests, however,
that this material undergoes gradual rearrangement before
final melting during the slow heating experiments (Figs.
5(a) and (b)). Heating at a rate of 10 K min21 or greater had
no further implication on the melting point. This is very
similar to the behaviour of thin polyethylene crystals that
show perfectioning and crystal thickening on slow heating
and, above a certain critical heating rate, melting occurs
essentially without previous crystal rearrangement [34].
The commonly used heating rate, 10 K min21, lead to no
significant crystal rearrangement for the rapidly cooled star
polymers with dendrimer or hyperbranched cores, but in
linear PCL and star PCL with dendron core, bimodal
melting, i.e. melting of initial crystals followed by crystal-
lisation and final melting, occurred even during faster
heating (#30 K min21). The melting point of the initial
crystals can be obtained, however, by considering only the
low temperature peak of these polymers.
4.4. Assessment of equilibrium melting point
The equilibrium melting temperatures of the different
samples were determined by extrapolation of Tm 2 Tc data.
The melting point was taken as the melting peak
temperature on samples crystallized to 10% of the final
attainable degree of crystallinity at the particular tempera-
ture. Correction was made for the thermal lag between
sample and sample holder according to Gedde and Jansson
[24]. Melting points recorded for samples crystallized at
even lower temperatures during constant rate cooling at
different heating rates approached a constant value at 5–
10 K min21 (Figs. 5(a) and (b)). No correction was thus
made for possible crystal rearrangement during heating
from the crystallisation temperature to the melting point.
The melting peaks recorded on the direct heating from the
crystallisation temperature were strictly unimodal and
narrow, observations that indicate that the orthorhombic
crystal phase is directly transformed into the molten state
without significant crystal rearrangement. Fig. 6 shows
Tm 2 Tc data for a series of samples of linear PCL. Data
reported by Phillips et al. [8] for a linear PCL with a molar
mass of 40,000 g mol21 (corresponding to DP < 350) are
included. It is evident that a linear function like the
Hoffman–Weeks equation, Eq. (2), is not in accordance
with the pronounced curvature shown by the experimental
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5257
data (Fig. 6). It was also noted that the melting point at a
given crystallisation temperature, e.g. at 316 K, increased
with increasing molar mass.
The effect of molecular architecture on the melting point
is demonstrated by the data presented in Fig. 7. The degrees
of polymerisation of the PCL units in these polymers are
approximately the same, ranging from 39 to 51. The Tm 2
Tc curves of the different samples are essentially parallel-
shifted along the melting point axis; the melting points of
the star polymers are ,2 K higher than those of linear PCL
at any given Tc within the studied temperature range. The
star polymers showed a progressive increase in melting
point with increasing PCL degree of polymerisation (Figs. 8
and 9). The data obtained for the star polymers showed a
pronounced curvature in the Tm 2 Tc diagrams (Figs. 8 and
9).
Eq. (3), which is a development of the Hoffman–Weeks
equation, considers that the crystals are more stable
(expressed by dLc) than the minimum requirement at the
particular crystallisation temperature. This equation pre-
dicts a positive second derivative of Tm with respect to Tc:
The crystal-thickening factor ðbÞ was set unity. This
supposition was made because of the short crystallisation
times and the small-angle X-ray data presented by Strobl
et al. [10] and Nojima et al. [12]. Furthermore, the goodness
of fit decreased when b set to values greater than 1. The two
adjustable parameters, the temperature at which the fitted
curve intersects Tm ¼ Tc (the virtual T0m) and dLc; were
adjusted in order to minimise the sum of square differences.
Optimum fit was achieved in most cases with dLc ¼ 5 nm:
A few of the low molar mass samples showed higher
goodness of fit with a slightly lower value, dLc ¼ 4 nm:
However, dLc was set to 5 nm as a universal value for all the
different samples.
Fig. 10 and Table 3 show the fit of Eq. (3) to the
experimental data obtained for Don81. The curvature of the
experimental data in the Tm 2 Tc diagram is stronger than
predicted by Eq. (3). This was a general feature for all the
samples studied. The positive deviation of the experimental
data with respect to the fitted line typically occurring at the
two highest crystallisation temperatures is most probably
due to a relatively small increase of the crystal thickening
factor. Crystallisation to 10% of the final degree of
crystallinity means that more time was available for crystal
Fig. 4. Melting traces recorded at different heating rates (shown adjacent to each curve; in K min21) of samples initially crystallized during a 30 K min21
cooling from the melt: (a) linear PCL (PCL39); (b) relative size of the high temperature melting peak ðAh=AtotÞ as a function of heating rate ðQÞ; star polymers
based on (c) dendrimer (D42); (d) hyperbranched (HB51); (e) dendron (Don46).
E. Nunez et al. / Polymer 45 (2004) 5251–52635258
thickening at the higher crystallisation temperatures. For
Don81 the following crystallisation times were used to reach
the 10% relative level of crystallinity: 313.5 K—0.04 h;
317.5 K—0.17 h; 321.5 K—0.83 h; 325.5 K—4.03 h;
327.5 K—8.88 h.
The implication of single data points on the fitted
equation and on the ‘virtual’ T0m value obtained is however
limited (Table 3). The range in the virtual T0m obtained from
different choices of data for extrapolation was from 356.5 to
358.9 K with an average value of 357.3 K and a standard
deviation of 0.7 K. The virtual Tm0 value by fitting Eq. (3) to
the all the data taken was 357.3 K. The equilibrium melting
point of Don81 was calculated by considering the finite
thickness of the equilibrium crystal (see Section 3). The DT
value used in the assessment, see Fig. 10, was calculated
from Eq. (4) using s ¼ 40 mJ m22. This relatively low
value was used because the equilibrium crystal is lacking
Fig. 5. Effect of heating rate on melting temperature of samples crystallized
during 30 K min21 cooling from the melt of (a) PCL39 and (b) D51. The
values plotted correspond to the averages based on two individual
measurements, and the error bars represent the standard deviations. Filled
symbols indicate high-temperature peak and unfilled symbols indicate low
temperature peak.
Fig. 6. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for PCL17 (B), PCL39 (A), PCL117 (X) and a linear PCL with average
DP < 350 (W). The data for this polymer was reported by Phillips et al. [8].
The data shown for PCL17, PCL39 and PCL117 are averages based on three
separate measurements and they refer to samples crystallised to 10% of the
final crystallinity at each temperature. The standard deviation of the single
data points was smaller than 0.2 K. The lines are second-degree polynomial
fits to the experimental data.
Fig. 7. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for PCL39 (W), D42 (X), Don46 (A) and HB51 (B). The samples
crystallised to 10% of the final, attainable crystallinity at each temperature.
The data are averages based on three separate measurements. The standard
deviation of the single data points was smaller than 0.2 K. The lines are
second-degree polynomial fits to the experimental data.
Fig. 8. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for D14 (B), D24 (A), D42 (X) and D51 (W).The samples crystallised to
10% of the final, attainable crystallinity at each temperature. The data are
averages based on three separate measurements. The standard deviation of
the single data points was smaller than 0.2 K. The lines are second-degree
polynomial fits to the experimental data.
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5259
chain folds. It is important to point out, however, that
s ¼ 90 mJ m22 was used as input in Eq. (3) for the
extrapolation of the melting point data of the folded-chain
crystals grown at the crystallisation temperatures studied.
The value used for s is important since DT is strictly
proportional to s: The main emphasis, however, is to
compare equilibrium melting point data for linear and star-
branched PCL. A change in s alters the calculated
equilibrium melting point data but the qualitative con-
clusions are not influenced by any realistic alteration of s:
Fig. 11 shows data for a relatively high molar mass linear
PCL earlier reported by Phillips et al. [8]. They used a linear
extrapolation procedure in accordance with the original
Hoffman–Weeks method and obtained an equilibrium
melting point of 343.5 K [8]. The extrapolation of the
experimental data according to Eq. (3) yielded a signifi-
cantly higher value, T0m ¼ 356:7 K: The curvature of the
data in the Tm 2 Tc plot is obvious and also the difficulty to
perform a linear extrapolation of the data to determine the
equilibrium melting point.
Fig. 12 presents the results from extrapolation of the
fitted Eq. (3) using Eq. (4) to calculate DT to obtain Tm0 (n)
for the different samples. The following universal parameter
values were used: Eq. (3): s ¼ 90 mJ m22;
Dh0 ¼ 163 MJ m23; b ¼ 1; dLc ¼ 5 nm; Eq. (4):
s ¼ 40 mJ m22; Dh0 ¼ 163 MJ m23. The extrapolation
method using these parameter values produced, as it
Fig. 9. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for Don15 (A), Don46 (X) and Don81 (W).The samples crystallised to
10% of the final, attainable crystallinity at each temperature. The data are
averages based on three separate measurements. The standard deviation of
the single data points was smaller than 0.2 K. The lines are second-degree
polynomial fits to the experimental data.
Fig. 10. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for Don81: (a) fitted Eq. (3) using the following parameter values:
Dh0 ¼ 163 MJ m23; s ¼ 90 mJ m22; b ¼ 1; dLc ¼ 5 nm. (b) Tm ¼ Tc: (c)
the intersection between curve (a) and line (b) denoting the virtual
equilibrium melting point: (d) equilibrium melting point assuming constant
dLc ¼ 5 nm; DT denotes the difference in melting and crystallisation
temperatures calculated from the Thomson – Gibbs equation with
s ¼ 40 mJ m22.
Table 3
Virtual equilibrium melting point data for Don81 calculated from melting
point data obtained at different crystallisation temperatures
Tc (min) (K)a Tc (max) (K)a T0m (K)b
313.5 327.5 357.3
313.5 325.5 357.1
313.5 323.5 356.8
313.5 321.5 356.7
313.5 319.5 356.6
313.5 317.5 356.8
313.5 315.5 357.1
315.5 327.5 357.2
317.5 327.5 357.4
319.5 327.5 357.6
321.5 327.5 358.1
323.5 327.5 358.5
325.5 327.5 358.9
315.5 325.5 357.9
317.5 323.5 356.7
319.5 321.5 356.5
a Minimum and maximum crystallisation temperatures used for
extrapolation.b Virtual equilibrium melting point obtained by fitting of Eq. (3) to the
experimental data and using the following parameter values: Dh0 ¼ 163
MJ m23; s ¼ 90 mJ m22; b ¼ 1; dLc ¼ 5 nm:
Fig. 11. The melting point ðTmÞ as a function of crystallisation temperature
ðTcÞ for PCL350. The experimental data are from Phillips et al. [8]. The
continuous line shows the best fit of Eq. (3) using the parameter values:
Dh0 ¼ 163 MJ m23; s ¼ 90 mJ m22; b ¼ 1; dLc ¼ 5 nm. The cross in the
right-hand upper corner indicate the estimated equilibrium melting point
using DT calculated from Eq. (4) with s ¼ 40 mJ m22.
E. Nunez et al. / Polymer 45 (2004) 5251–52635260
appeared accurate results for the majority of the samples.
The results obtained by the extrapolation were not sensitive
to the choice of the data. Removal of the data points taken at
the highest crystallization temperatures (where possible
crystal thickening may have occurred) only altered the
result by a few tenths of a Kelvin. Extrapolation of the data
obtained for the lowest molar mass samples (PCL17, D14 and
Don15) gave, however, results with significant uncertainty.
This was primarily due to the large DT values shown by
these samples. The values presented in Fig. 12—marked
with arrows—should be considered as highly approximate.
The equilibrium melting point data are presented in Fig.
12 in which the data are plotted according to the Hay [29].
The equilibrium melting point data follow a linear trend
with respect to ln n=n: The following values for T0m !1
were obtained by extrapolation to ln n=n ¼ 0 : linear-PCL:
359.2 K; dendrimer-PCL: 362.0 K; dendron-PCL: 363.2 K;
hyperbranched-PCL: 363.6 K. Hence, the difference in
equilibrium melting point between the linear and the star-
branched polymers is 3–4 K. The use of a lower s-value
(10 mJ m22 was tested) in the calculation led only to small
changes in the extrapolated values of T0mðn!1Þ : 358.7 K
(linear-PCL); 363.0 K (dendrimer-PCL); 363.0 K (dendron-
PCL); 362.7 K (hyperbranched-PCL). The goodness of fit
was however lower for the data based on calculation of DT
with s ¼ 10 mJ m22.
Fig. 13 presents equilibrium melting point data plotted as
a function of n together with best fits of the Broadhurst
equation (Eq. (5)) to the experimental data. The following
values were obtained for T0mðn!1Þ : linear PCL: 358.5 K;
dendrimer PCL: 366.2 K; dendron PCL: 363.7 K; hyper-
branched PCL: 365.0 K. The determinations of the equili-
brium melting points for the low molar mass samples
PCL17, D14 and Don15 were associated with considerable
uncertainty and it was noticed from the fittings according to
the Broadhurst equation that a change of these equilibrium
melting point values by 1 K changed the extrapolated
T0mðn!1Þ-value by ,2 K. The extrapolation of the
equilibrium melting point data according the Hay equation
was less influenced by the uncertain equilibrium melting
points of the low molar mass samples. The equilibrium
melting point data obtained by using s ¼ 10 mJ m22 in the
calculation of DT were also described by the Broadhurst
equation and although the goodness of fit was not as high as
in the case of using s ¼ 40 mJ m22 essentially the same
differences in T0mðn!1Þ between linear and star polymers
were obtained: 360.6 K (linear PCL) and 363–364 K (star
polymers).
Fig. 14 shows the melting point ranges for the linear and
star polymers and the calculated crystal thickness using the
Fig. 12. The equilibrium melting point ðT0mðnÞÞ as a function of ln n=n (n is
the degree of polymerisation of PCL of a single PCL arm) for: linear PCL
(B), dendrimer-PCL (A), hyperbranched-PCL (W), and dendron-PCL (X).
The lines are best fits of Eq. (7) to the experimentally based equilibrium
melting point data. The arrows point towards data points associated with
considerable uncertainty (^2–3 K).
Fig. 13. The equilibrium melting point ðT0mðnÞÞ as a function of the degree of
polymerisation of PCL of single PCL arms ðnÞ for: linear PCL (B),
dendrimer-PCL (A), hyperbranched-PCL (W), and dendron-PCL (X). The
curves are best fits of Eq. (5) to the experimentally based equilibrium
melting point data. The arrows point towards data points associated with
considerable uncertainty (^2–3 K).
Fig. 14. Crystal thickness plotted as a function of melting point calculated
from the Thomson–Gibbs equation using the following input data:
Dh0 ¼ 163 MJ m23; s ¼ 90 mJ m22 for (curve a) linear PCL:
T0m ¼ 359 K; (curve b) star PCL: T0
m ¼ 363 K. The broken lines shows
the melting point and crystal thickness ranges included in this study. Data
for the samples with very low molar mass (PCL17, D14 and Don15) are not
included.
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5261
Thomson–Gibbs equation with the following parameter
values: Dh0 ¼ 163 MJ m23, s ¼ 90 mJ m22, Tm0 ¼ 359 K
(linear PCL), 363 K (star PCL). The crystal thickness ranges
for the higher molar mass samples (not including PCL17,
D14 and Don15) were 12–15.5 nm for linear PCL and 11–
15 nm for star-branched PCL. The extended-chain length
was between 31 and 302 nm for the linear PCL’s and 21 and
70 nm for the star polymers. The PCL chains must then be
folded at least once and in most cases twice or more times.
Hence, the presence of great many chain folds in the fold
surface of the crystals is consonant with the s value used
(90 mJ m22) in the calculation of the equilibrium melting
point according to Eq. (3) and in the calculation of the
crystal thickness according to the Thomson – Gibbs
equation. Strobl et al. [10] reported by small-angle X-ray
scattering crystal thicknesses between 7 and 9 nm corre-
sponding to melting points in the range 333–338 K. This is
considerably less than that shown in Fig. 14; 15–17 nm for
linear PCL crystals melting between 333 and 338 K. The
melting point–crystal thickness data presented by Strobl
et al. [10] yielded a unexpected low s-value, ,60 mJ m22.
A simple, qualitative analysis of these findings can be
made using the equation:
T0mðnÞ ¼
Dh0ðnÞ
Ds0ðnÞð11Þ
where Dh0 and Ds0 are the enthalpy and entropy of fusion.
The similarity in the unit cell parameters of the polymers
with different molecular architectures suggests that Dh0 is
independent of molecular architecture. However, the
depression in crystallinity of the star polymers with respect
to linear PCL indicates that a portion of the PCL arms was
hindered from crystallizing. Only the repeating PCL units
close to core should be influenced in the case of the
equilibrium extended-chain crystals. It seems reasonable, on
the basis of the data presented in Fig. 3, that only a few (,1)
repeating PCL units of each PCL arm of the dendrimer/
dendron star polymers are prohibited from crystallizing. In
hyperbranched star polymers, a larger number of PCL
repeating units (,5) are unable to crystallize. Hence, the
dendrimer/dendron star polymers would in this instance
experience a reduction in enthalpy of fusion to DH0ðn2 1Þ
and hence a lowering of the equilibrium melting point. This
effect should be more important in polymers with the short
PCL arms. More important is the reduction in Ds0ðnÞ that is
responsible for the increase in equilibrium melting point
with respect to linear PCL for the star polymers. The
attachment of the crystallisable PCL arms to the core
reduces the positional freedom of the crystallisable units in
the molten state leading to a lower Ds0ðnÞ for these polymers
than for linear PCL.
5. Conclusions
PCL star polymers based on dendritic cores showed the
same unit cell structure as linear PCL. The crystallinity of
the star polymers was lower than that of the linear PCL
analogues suggesting that a few PCL repeating units nearest
to the dendritic core were unable to crystallize and that
chain folds were less tight in the star polymers. The
crystallinity depression was particularly strong in the star
polymers with hyperbranched core. Crystal rearrangement
during the slow heating of polymers previously crystallized
during fast cooling occurred to the same extent in star
polymers based on dendron core as in the linear polymers,
but it was retarded in the star polymers based on dendrimer
or hyperbranched cores. The equilibrium melting points of
the star polymers were higher than those of the linear
analogues. This effect can be attributed to positional
restriction imposed on the PCL chains by their covalent
attachment to the dendritic cores, which lowers the melt
entropy of the star polymers with respect to their linear
analogues.
Acknowledgements
The financial support from the Swedish Research
Council (grant # 5104-20005764/20) is gratefully
acknowledged.
References
[1] Malmstrom E, Johansson M, Hult A. Macromol Chem Phys 1996;197:
3199.
[2] van Caeter P, Goethals EJ. Macromol Rapid Commun 1997;18:393.
[3] Chen E-Q, Lee S-W, Zhang A, Moon B-S, Mann I, Harris FW, Cheng
SZD, Hsiao BS, Yeh F, von Merrewell E, Grubb DT. Macromolecules
1999;32:4784.
[4] Wei HY, Shi WF, Nie KM, Shen XF. Polymer 2002;43:1969.
[5] Zhao YL, Cai Q, Jiang J, Shuai XT, Bai JZ, Chen CF, Xi F. Polymer
2002;43:5819.
[6] Risch BG, Wilkes GL, Warakomski JM. Polymer 1993;34:2330.
[7] Crescenzi V, Manzini G, Calzolari G, Borri C. Eur Polym J 1972;8:
449.
[8] Phillips PJ, Rensch GJ, Taylor KD. J Polym Sci, Polym Phys 1987;25:
1725.
[9] Phillips PJ, Rensch GJ. J Polym Sci, Polym Phys 1989;27:155.
[10] Heck B, Hugel T, Iijima M, Sadiku E, Strobl G. New J Phys 1999;1:
171.
[11] Gedde UW, Mattozzi A. Adv Polym Sci 2004;169:29.
[12] Nojima S, Kikuchi N, Rohadi A, Tanimoto S, Sasaki S. Macromol-
ecules 1999;32:3727.
[13] Chatani Y, Okita Y, Tadokoro H, Yamashita Y. Polym J 1970;1:555.
[14] Bittiger H, Marchessault RH, Niegisch WD. Acta Crystallogr 1970;
B26:1923.
[15] Hu H, Dorset DL. Macromolecules 1990;23:4604.
[16] Gedde UW. Polymer physics. London: Chapman & Hall; 1995.
[17] Hoffman JD, Weeks JJ. J Res Natl Bur Std-Phys Chem 1962;A66:13.
[18] Broadhurst MG. J Chem Phys 1962;36:2578.
[19] Broadhurst MG. J Res Natl Bur Std-Phys Chem 1966;70A:481.
E. Nunez et al. / Polymer 45 (2004) 5251–52635262
[20] Flory PJ, Vrij A. J Am Chem Soc 1963;85:3548.
[21] Ihre H, Hult A, Soderlind E. J Am Chem Soc 1996;118:6388.
[22] Claesson H, Malmstrom E, Johansson M, Hult A. Polymer 2002;43:
3511.
[23] Werner P-E, Eriksson L, Westdahl M. J Appl Crystallogr 1985;18:
367.
[24] Gedde UW, Jansson J-F. Polymer 1983;24:1521.
[25] Wunderlich B. Crystal nucleation, growth, annealing. Macromolecu-
lar physics, vol. 2. New York: Academic Press; 1978.
[26] Barham PJ, Chivers RA, Keller A, Martinez-Salazar J, Organ SJ.
J Mater Sci 1985;20:1625.
[27] Cheng SZD, Chen J, Barley JS, Zhang A, Habenschuss A, Zschack
PR. Macromolecules 1992;25:1453.
[28] Marand H, Xu J, Srinivas S. Macromolecules 1998;31:8219.
[29] Hay JN. J Polym Sci, Polym Chem 1976;14:2845.
[30] Illers K-H, Hendus H. Makromol Chem 1968;113:1.
[31] Bassett DC, Hodge AM, Olley RM. Proc Roy Soc London 1981;
A377:39.
[32] Arakawa T, Wunderlich B. J Polym Sci, Polym Symp 1967;C16:653.
[33] Ergoz E, Fatou JG, Mandelkern L. Macromolecules 1972;5:127.
[34] Wunderlich B. Crystal melting. Macromolecular physics, vol. 3. New
York: Academic Press; 1980.
E. Nunez et al. / Polymer 45 (2004) 5251–5263 5263