Accepted Manuscript

Rheology and porosity control of poly(2-hydroxyethyl methacrylate) hydrogels

Evgeny Karpushkin, Miroslava Dušková-Smrčková, Miroslav Šlouf, Karel Dušek

PII: S0032-3861(12)01020-8
DOI: 10.1016/j.polymer.2012.11.055
Reference: JPOL 15831

To appear in: Polymer

Received Date: 17 August 2012

Revised Date: 26 October 2012
Accepted Date: 20 November 2012

Please cite this article as: Karpushkin E, Dušková-Smrčková M, Šlouf M, Dušek K, Rheology and
porosity control of poly(2-hydroxyethyl methacrylate) hydrogels, Polymer (2012), doi: 10.1016/
j.polymer.2012.11.055.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

PT
RI
SC
tan δ

U
AN
M
D
TE
EP
C
AC

frequency
 ACCEPTED MANUSCRIPT

Rheology and porosity control of poly(2-hydroxyethyl methacrylate) hydrogels

Evgeny Karpushkin, Miroslava Dušková-Smrčková, Miroslav Šlouf, Karel Dušek

Institute of Macromolecular Chemistry AS CR, v. v. i. Heyrovského nám. 2, 162 06 Prague 6,

Czech Republic

PT
Phone +420 776543980
Fax + 420 296809410

RI
e-mail: [email protected]

SC
Abstract
Swelling and dynamic mechanical behavior in a broad frequency range of homogeneous and

U
porous poly(2-hydroxyethyl methacrylate) gels were studied in relation to their matrix
structure, gel morphology and nature of the swelling medium. Very lightly cross-linked gels
AN
showed up relatively large mechanical losses caused by slow relaxation of hydrophobic
physical associates still proceeding at frequency 10-3 Hz. The macroporous gels of fused-
M
spheres type morphology contained many dangling aggregates not bearing the stress which
caused concentration of the stresses in the load bearing paths through the matrix. This is the
D

reason for the sudden drop of storage modulus. In the α-relaxation region, the loss factors for
homogeneous and porous gels merged into a single master curve. Reswelling of gels in a poor
TE

solvent caused a decrease of the swelling degree and shifted their α- relaxation to lower
frequencies; reswelling in good solvent shifted the maximum of α- relaxation to higher
EP

frequencies.
C

Keywords:
AC

hydrogel; mechanical properties; swelling.

1
 ACCEPTED MANUSCRIPT

1. Introduction
Cross-linked poly(2-hydroxyethyl methacrylate) (PHEMA) hydrogels of various
morphologies rank among most frequently applied synthetic hydrogels in bio-related areas
[1]. Macroporous hydrogels including PHEMA are prepared in various ways [2-4] but mainly
by phase separation occurring during polymerization, when the amount of added diluent

PT
exceeds the critical concentration [5-7]. The effective diluents active in phase separation are
in fact monomer (HEMA)-diluent mixtures which often show up cosolvency towards

RI
PHEMA in a certain composition region [5]. Therefore, various morphologies can be
achieved by varying the nature and amount of diluent. The properties of the PHEMA matrix

SC
have the special feature in that they are controlled to a large extent by transient (physical)
interactions. These hydrophobic interactions are viewed as associations of hydrophobic parts
[8] coupled with changes in the iceberg structure of water [9]. They play an important role in

U
dynamic and static properties of the PHEMA hydrogels: they are strong enough not to allow
chemically uncross-linked PHEMA to dissolve in water but only to swell. Covalent
AN
permanent interactions (cross-links), initially having a negligible effect on swelling in water,
play an increasing role with increasing concentration of the cross-linker or when the gel is
M
reswollen in a good solvent. Both the matrix properties and morphology control the
mechanical and swelling properties of hydrogels important in applications.
D

Whether the physical interactions are sensed by rheometry, depends on the experimental
window of frequencies and relaxation times of physical bonds. Discrimination between
TE

permanent and transient networks was attempted by examining the storage and loss moduli in
the low-frequency region [10]. The interplay between covalent bonds and attractive physical
EP

interactions as seen in the low-frequency region is not enough clear – the chemical cross-links
can suppress the formation of physical bonds hindering the alignment of chains (e. g.,
formation of helices). Alternatively, increasing number of chemical cross-links can enhance
C

the weaker, non-specific interactions by increasing the probability of pairing of interacting

AC

sites [11-14]. This is probably the case of PHEMA-water system. Decreased degree of
swelling is manifested by the increase of storage G’, and loss G’’ moduli, and of the loss
factor tan δ [15], [16].
When interpreting mechanical properties, the situation for the phase-separated macroporous
PHEMA gels which are in the center of our interest is more complicated than for
homogeneous gels. Both the deformation geometries of pore walls and the topological
structure of the matrix gel are not known. A working hypothesis was put forward assuming
that the gel matrix is the same as that formed at dilution at which the gel is still homogeneous
2
 ACCEPTED MANUSCRIPT

(non-porous and transparent) [16]. This may be questionable especially for highly porous gels
and needs verification or rejection.
In our previous study [16], we showed that the frequency dependence of the loss factor, tan δ,
served as a good diagnostic tool for the morphology of the material. At low frequencies, the
loss factors of macroporous phase-separated hydrogels were higher than those of samples
without well-developed pores. There were indications that for some macroporous samples the

PT
loss factor increased with decreasing frequency (down to 0.05 Hz) pointing to reorganization
of structures held by physical bonds. The rapid increase of the loss factor with the cross-link

RI
density at 10 Hz was likely due to the decreased segmental mobility resulting in the onset of
swollen gel vitrification. Very important was the finding that the dynamic rheology was able

SC
to detect existence of communicating pores.
In this contribution, the porosity of PHEMA hydrogels was controlled by addition of various
amounts of diluents and the resulting morphology was characterized by the frequency

U
dependences of G’ and tan δ in a broad range of frequencies (from10-4 to103 Hz). While the

AN
G’ dependence at low frequencies indicates whether the structure has fully relaxed or not, the
loss factor tan δ dependence characterizes the relaxation processes at higher frequencies
(vitrification) as well as at low frequencies (relaxation of substructures – clusters,
M
entanglements). To broaden the sensing range, we have chosen swelling media which are
better or poorer solvents for PHEMA than water. In this way, the intersegmental interactions
D

could be made weaker or stronger and their nature can be better understood. These swelling
media (solutions of salting-in or salting-out salts) were also used as diluents for HEMA and
TE

the change of their solvation power resulted in a dramatic change of morphology. The effect
of additives on the morphology of the formed gels has been demonstrated in [17].
EP

The physical interactions of homogeneous systems and their weakening or strengthening is

well seen in the low-frequency range when the contribution by covalent cross-links is kept at
minimum. Alternatively, the effect of covalent cross-links can be better characterized when
C

the physical interactions are removed by reswelling in a good solvent. The knowledge
AC

accumulated by the study of homogeneous gels serves as a prerequisite for interpreting the
behavior of macroporous systems.
The study of phase-separated macroporous gels is focused on their swelling and dynamic
mechanical behavior in relation to matrix structure and morphology: on the segmental
mobility limitations when entering the glass-transition region and to the quasi-equilibrium
state at low frequencies when the rubber-elasticity of the network structure of the matrix and
the gel morphology control the mechanical and swelling responses.

3
 ACCEPTED MANUSCRIPT

2. Experimental
2.1. Materials
2-hydroxyethyl methacrylate HEMA (Rohm), di(ethylene glycol) dimethacrylate DEGDMA
(Aldrich), ammonium persulfate (Fluka), N,N,N’,N’-tetramethylethylenediamine (Helicon),
dimethylsulfoxide DMSO (Aldrich), and sodium chloride (Aldrich) were used as received.
HEMA contained 0.07% ethylene glycol dimethacrylate and 0.44% of di(ethylene glycol)

PT
monomethacrylate.
2.2. Preparation of gels

RI
2-hydroxyethyl methacrylate (HEMA) and di(ethylene glycol) dimethacrylate (DEGDMA)
were copolymerized in the presence of various amounts of water with ammonium persulfate

SC
(APS) as initiator and N,N,N’,N’-tetramethylethylenediamine (TEMED) as promoter. APS
was first mixed with water, HEMA and DEGDMA were added, and the solution was stirred
for 5 min (the solutions were always clear at this point, proving full miscibility of

U
components). Then, the sample was bubbled with nitrogen for 5 min, TEMED was added, and
the mixture was quickly poured between two glass plates separated with 3 mm silicone rubber
AN
sealing. Typically, a sample lost its fluidity after 5 to 20 minutes indicating the cross-linked
structure formation. The reaction was carried out overnight at room temperature to ensure
M
complete conversion. This was confirmed by FTIR measurements of C=C bonds conversion.
Corresponding data is presented in the Supplementary data part.
D

After the polymerization, the sample was washed several times with an excess of distilled
water (at least 5 washes, 24 hours each). Sol fraction for all samples was determined and was
TE

found to be less than 3% except for samples prepared with 80% of water. Sol fraction data is
presented in the Supplementary data part. Then, a disk (20 mm in diameter) was cut to be
EP

used in the rheology tests. The rest of the sample was used for swelling and morphology
studies. The following parameters were varied at preparation: diluent concentration (40 to 80
wt%) and DEGDMA amount with respect to HEMA (0 to 2.0-6.0 mol%, upper limit
C

depended of miscibility). APS to monomers ratio was 0.75 wt% for all samples.
AC

Sample coding: The samples prepared with water as diluent and then reswollen in water were
coded as follows: 70/1.0, where the first number (70) is the initial water concentration (IWC)
in wt%, and the second (1.0) stands for the concentration of cross-linker added to monomer
(CC) in mol% with respect to monomers. If a sample was reswollen in solvent other than
water before the measurement, the type of swelling medium was indicated in brackets after
the sample code: 70/1.0(DMSO). See examples in Scheme 1.

4
 ACCEPTED MANUSCRIPT

PT
RI
SC
Scheme 1 Coding of gel samples.

2.3. Rheological measurements

U
All rheological measurements were run with parallel plates geometry using the Bohlin Gemini
AN
HR Nano rotational rheometer (Malvern). Swollen samples immersed in water during the test
were used; the presence of water did not influence the measured values (additional friction of
rotating part was negligible). The swollen sample diameter was 20 mm, sample thickness
M
varied between 2.2-2.5 mm (samples swollen in 3 M NaCl), 2.8-3.0 mm (samples swollen in
water) and 4.0-4.4 mm (samples swollen in DMSO). During the measurement the sample was
D

conditioned at a selected temperature in the range of 5-90 °C using Peltier-type temperature

TE

control. Prior to measurement, a sample was equilibrated at a given temperature until the
recorded mechanical properties were constant for at least 10 minutes.
The amplitude sweep tests were performed in the stress-control mode at frequency 1 Hz,
EP

typical strain range was 10-5 to 10-2. The frequency sweep tests were performed in the stress-
control mode, within limits of linear viscoelasticity. Frequency range was typically between
C

0.05 and 15-40 Hz. The measuring gap (distance between bottom and top plates of the
rheometer) for correct rheological studies was selected as described previously [16]. A special
AC

Piezoelectric Rotary Vibrator (PRV) [18] attachment was used to directly measure hydrogels
behavior in high-frequency oscillations (10-2000 Hz), in these tests sample thickness was
around 0.5 mm, and diameter was 40 mm.
2.4. Time-temperature superposition
Frequency sweep curves measured at individual temperatures were superimposed to get a
single curve using the algorithm built in the rheometer software (version 6.51.0.3). The
solution is obtained by taking pairs of curves in ascending value of temperature, and applying

5
 ACCEPTED MANUSCRIPT

a weighted polynomial fit of suitable order to obtain the best pairwise shift factor, then
accumulating all pairwise shift factors, relative to a chosen reference temperature, to obtain
the overall superimposed curve. In this work, the reference temperature was always set to 25
°C. The experimental variable to calculate shift factors was the loss factor, while storage and
loss moduli curves were obtained from the raw data using the shift factors calculated from
loss factor superposition.

PT
2.5. Morphology studies
The morphology of the hydrogels was decided from the consolidated observation by means of

RI
LV SEM, ESEM and light microscopy.
SEM studies were made with the Quanta 200 FEG microscope (FEI, Czech Republic). With

SC
Cryo Low Vacuum Scanning Electron Microscopy method, LV SEM, samples were observed
in the frozen state. The hydrogel sample was immersed into liquid nitrogen (-196 ºC) and
transferred to a table conditioned at -10 ºC. In the frozen state, the sample was cut with a thin

U
blade to remove the upper layer and ice crystals. In the course of scanning the pressure in the
microscope chamber was kept at low value of 90-110 Pa and the temperature was maintained
AN
below 0 °C. With Environmental Scanning Electron Microscopy method, ESEM, samples
were observed in hydrated state. The hydrogel was trimmed with a blade, transferred to a
M
table conditioned at 2 ºC, and scanned at this temperature. The water vapor pressure in the
chamber was adjusted to about 74% relative humidity. Light microscopy study was performed
D

with the microscope Zetopan Pol. (Reichert) equipped with a digital camera using a phase
contrast objective (Ph 40x). Swollen hydrogel sample was cut with blade to prepare a wedge
TE

shaped specimen that was observed at room temperature.

2.6. Swelling studies
EP

The equilibrium weight content of water EWC (denoted as w1 in equations), was determined
gravimetrically by weighing the samples in their equilibrium swollen state (msw) and after
drying to constant weight under vacuum at 100 °C (md). The swollen sample was lightly
C

blotted with filter paper to remove excess liquid, and immediately weighted. The EWC was
AC

expressed as weight of water in the swollen sample per 1 g of dry polymer:

msw − md
w1 = (1)
md

while the initial water content was expressed either in wt% of IWC (denoted as w10 )
mw
w10 = × 100% (2a)
m w + mp

or as volume fraction of polymerizable material during polymerization ( φ 20 )

6
 ACCEPTED MANUSCRIPT

mp / ρ p
φ 20 = (2b)
mp / ρ p + mw / ρ w

where mp is the mass of the polymer or monomer(s) and mw is the mass of water. The density
of dry polymer ρ p was used in calculations instead of monomer density to account for

contraction of the monomer during polymerization; ρ w is water density.

PT
The equilibrium swelling of networks in medium other than water was expressed similar to
w1 , as weight of swelling medium in the swollen sample per 1 g of dry polymer.
The cross-linker concentration CC, is expressed in molar percent of the cross-linker

RI
DEGDMA in its mixture with HEMA
n(DEGDMA)
CC = × 100%

SC
(3)
n(DEGDMA) + n(HEMA)

U
3. Results and Discussion
Several series of gels were prepared from HEMA with varying amounts of diluents and cross-
AN
linker. Water and aqueous solutions of sodium chloride (poorer solvent) or magnesium
perchlorate (better solvent) [19] were used as diluents. When during polymerization the initial
M
diluent concentration and cross-linker concentration exceeded the critical value, the system
phase separated. As described in more detail in ref. [16], phase separation occurred by
microsyneresis or combination of micro- and macrosyneresis [20] and was caused by
D

deterioration of solvation power of HEMA-water mixtures and build-up of molecular

TE

weight/cross-link density of PHEMA [6-7, 21]. The microphase separation resulted in gels of
different morphologies.
EP

3.1. Classification of studied hydrogels

The hydrogels described in this contribution differ in appearance, morphology, and
C

mechanical properties. Those called “homogeneous” are visually clear with no heterogeneities
detectable by electron microscopy. The group of non-transparent gels with large
AC

communicating pores between fused-spheres like structure of the matrix constitutes another
category called here “macroporous”. The gels called here “macroporous” are distinguished by
low storage modulus G’ and its sudden drop when passing from the “intermediate” category
[16]. By using the term “macroporous” throughout this communication we have in mind this
part of macroporous gels as defined by IUPAC (pore size > 50 nm). The gels which are
classified as neither homogeneous nor macroporous are grouped in the “intermediate”
category. It is a wide category of gels that fulfill the criterion of not being optically clear and

7
 ACCEPTED MANUSCRIPT

not to belong to the category of macroporous gels. The intermediate group includes gels of
very different turbid gels of fine droplet like and interlocking bicontinuous morphologies at
their borders, but their shear moduli change only moderately.
The gels prepared with water as diluent having different IWC and CC are assigned to these
classes in Fig. 1. Typical morphological differences are characterized by ESEM micrographs
attached (more micrographs are to be found in the Supporting info). The main reason for the

PT
morphology change with variation of initial water concentration is the deterioration of
solvation power of the HEMA-water mixture as polymerization proceeds and the diluent gets

RI
richer in water component.

U SC
AN
M
D

Figure 1 Appearance and morphology of gels swollen in water to equilibrium, prepared with
TE

water as diluent. Each point corresponds to a hydrogel prepared at certain IWC and CC. Open
circles: homogeneous gels, grey circles: intermediate gels, black circles: macroporous gels.
EP

The demarcation lines delimit the respective categories.

When diluents other than water are used, the demarcation lines are dramatically shifted as
C

shown in Figs. 2, and 3.

AC

8
 ACCEPTED MANUSCRIPT

a/

PT
RI
U SC
AN
M

b/
D
TE
C EP
AC

Figure 2 Plots of gel morphologies as revealed by SEM. The gels were prepared in varied
diluent. The schematics is self-explanatory.

The effect increasing concentration of Mg(ClO4)2 seen in Fig. 2a is illustrated by the series
70/2, i.e., 70 wt% Mg(ClO4)2 solution of varying molarity. The originally white gel of fused-
spheres type morphology obtained with water as diluent, remains white when 0.05M

9
 ACCEPTED MANUSCRIPT

Mg(ClO4)2 is used instead of water, but the morphology changes to a what we call an
interlocking bicontinuous one, in which both pores and swollen matrix form continuous
phases. Further increase of molarity of Mg(ClO4)2 to 0.1-0.2 M offers an opalescent (blue) gel
characterized by dispersed droplet morphology and above 0.3 M Mg(ClO4)2 the gels are clear.
For 60/1 series, 0.05 M Mg(ClO4)2 is sufficient to get blue gel and above 0.2 M Mg(ClO4)2 all
gels are clear.

PT
The effect of increasing concentration of the salting-out additive NaCl on morphology is even
more dramatic (Figs. 2b and 3).

RI
The chart of Fig. 2b is sufficiently self-explanatory but let us highlight the changes by the
series 60/1. The morphology of the 60/1 prepared with water as diluent is practically

SC
featureless. It turns to interlocking bicontinuous morphology when 0.4 M NaCl is used. A
further increase of molarity of NaCl to 0.475 M offers mixed structure where liquid phase of
the interlocking system start being filled with agglomerates of spheres and the whole story

U
ends above 0.55-0.6 M with exclusive formation of macroporous gels of fused-spheres type
morphology. Such structures do not develop when clear or intermediate turbid gels are
reswollen in aqueous NaCl solutions.
AN
When the content of diluent is low (series 40/1), addition of NaCl stimulates evolution of the
M
droplet morphology.
D
TE
EP

prepared in prepared at prepared at prepared at

pure water 0.40 M NaCl 0.45 M NaCl 0.525 M NaCl

Figure 3 The effect of diluent quality at gel preparation on hydrogel morphology as revealed
C

by SEM of water-swollen samples. Macroscopic appearance of all these samples in their

AC

swollen state is white non-transparent gel. Gels were polymerized in aqueous NaCl solutions
of varying concentrations and then reswollen in water while salt was thoroughly washed
away.

3.2. Characterization of rheological and swelling properties of gels

The connection between the microscopic structure of the sample and its mechanical response
in linear viscoelastic shear region is the main focus of this paper. In this section, we will

10
 ACCEPTED MANUSCRIPT

describe and briefly comment the rheological and equilibrium swelling results of PHEMA
gels prepared across the whole range of preparation conditions as depicted in Fig.1-3. As a
prerequisite for data evaluation, the applicability of the time-temperature superposition to
swollen PHEMA gels has been tested.

3.2.1. Time-temperature superposition for PHEMA gels

PT
The expansion of the frequency range of measurement towards low frequencies necessary for
this analysis is time consuming and for viscoelastic solids can be overcome by time-

RI
temperature superposition introduced by Ferry [22]. In principle, it is not applicable to
equilibrium swollen systems due to temperature dependent volume changes possibly

SC
combined with swelling transitions. Janáček and Ferry [23, 24] applied this superposition to
PHEMA gels swollen in di(ethylene glycol), and discussed the results in conjunction with
secondary relaxations possibly due to entanglements. For temperature-independent swelling

U
systems or nearly so (e.g., the PHEMA-water system [25]), frequency-temperature
superposition can be successfully accomplished, but one has to be careful when interpreting
AN
the results. Two characteristic samples – 40/0 homogeneous and 80/1 highly porous – were
selected for comparison of the superimposed and directly measured loss factors. The behavior
M
the 40/0 sample is characteristic of low tan δ below 1 Hz and well developed increase of tan δ
in the high-frequency region towards the glass transition. The 80/1 sample is characterized by
D

larger losses in the low frequency region, with a pronounced increase at high frequency,
similar to 40/0, and a weaker increase below 10-2 Hz (secondary relaxation).
TE

1.5
time-temperature superposition time-temperature superposition
direct conventional measurement 0.6 direct conventional measurement
EP

direct PRV measurement

1.0
0.4
tan δ

tan δ
C

0.5
0.2
AC

0.0 0.0
1E-4 1E-3 0.01 0.1 1 10 100 1000 1E-4 1E-3 0.01 0.1 1 10 100
a) Frequency (Hz) b) Frequency (Hz)
Figure 4 Frequency dependences of loss factor for the gels 40/0 (a) and 80/1 (b) swollen in
water to equilibrium. Open symbols are result of time-temperature superpositions
(measurements at 5-70 °C were shifted to a reference temperature of 25 °C), full symbols are
result of direct measurements at 25 °C in conventional mode (down to 10-4 Hz), or in PRV
mode (up to 103 Hz, 40/0 sample only).

11
 ACCEPTED MANUSCRIPT

In both cases (Fig. 4), the superimposed and directly measured dependences cannot be
distinguished from one another within the experimental accuracy. Thus, in this special case of
swollen PHEMA network, when the swelling changes are not of large magnitude – less than
5% of EWC in the range of 5 °C to 85 °C – the time-temperature superposition can be applied
to achieve the enhanced range of shear oscillation frequencies within reasonable testing time.

PT
3.2.2. Rheological and swelling properties
The model series XX/1 spanning from homogenous to highly porous samples comprises

RI
hydrogels prepared with 1 mol% of cross-linker and varied initial water concentration. The
shear storage moduli depended strongly on IWC (Fig. 5a). This can be explained by the two

SC
effects operative at the same time: 1) different amount of polymer material per unit volume of
the sample swollen to equilibrium, and 2) inhomogeneous stress field in highly porous gels as
will be discussed in Section 3.4.2. All curves showed significant upturn above 10 Hz and a

U
rubbery plateau at lower frequencies (or at least convergence to a plateau).
The loss factor reflects the dynamics of the structure of the hydrogels. Because tan δ =
AN
G’’/G’, the dependence of G’ and G’’ on the amount of the polymeric material in unit volume
is eliminated. The loss factors (Fig. 5b) showed the trend towards α-relaxation with onset
M
around 5 Hz for all samples in this series. The difference between the loss factors diminished
above 10 Hz, and they all followed one master curve up to at least 102 Hz. At low frequencies
D

(10-3 to 1 Hz), the loss factor significantly increased with increasing IWC.
TE

6
10
0
40/1 10

5 50/1 60/1
10
EP

80/1
G' (Pa)

70/1 10
-1
tan δ

50/1,
4
10 60/1,
70/1
-2
C

10
3
10 80/1 40/1
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
AC

10 10 10 10 10 10 10 10 10 10 10 10 10 10
a) Frequency (Hz) b) Frequency (Hz)
Figure 5 Frequency dependences of a) elastic shear modulus and b) loss factor for gels
prepared with 1 mol% of cross-linker and variable initial concentration of water, swollen in
water to equilibrium. Curves are result of time-temperature superpositions (measurements at
5-70 °C were shifted to a reference temperature of 25 °C). Corresponding sample codes are
shown near the curves.

12
 ACCEPTED MANUSCRIPT

A somewhat different trend was observed when the concentration of the added cross-linker
increased from 0 to 5 mol% while the water content was kept constant: the shift of α-
relaxation to lower frequencies was obvious. All curves showed the rubbery plateau at low
frequencies. Figures 6a,b illustrate the 40/X series, the trends for other IWC-constant series
are the same. The sample prepared without added cross-linker DEGDMA (40/0) was also
covalently cross-linked due to some minor amount (0.07 wt%) of ethylene glycol

PT
dimethacrylate present in the commercial 2-hydroxyethyl methacrylate.

RI
0
7
10 10 40/5
40/3
G' (Pa)

SC
-1
10 40/0, 40/0.1

tan δ
6
10 40/1
40/5
40/3

U
10
-2 40/1
40/0.1, 40/0
5

a)
10 -4 -3 -2 -1 0 1 2
10 10 10 10 10 10 10 10 10
Frequency (Hz)
3
AN4

b)
-3
10
-2
10
-1
10 10
0

Frequency (Hz)
10
1 2
10
3
10

Figure 6 Frequency dependences of a) elastic shear modulus and b) loss factor for gels
M
prepared with 40 wt% of water as diluent and variable concentration of cross-linker, swollen
in water to equilibrium. Curves are result of time-temperature superpositions (measurements
D

at 5-70 °C were shifted to a reference temperature of 25 °C). Corresponding sample codes are
shown near the curves.
TE

When the samples were tested after reswelling in solvent better than water, dimethylsulfoxide
EP

(DMSO) (Table 1), the loss factors were much lower than those for gels swollen in water.
Also, reswelling in DMSO lead to a considerable weakening of the frequency dependence of
G’, and the α-relaxation maximum is shifted far way outside the accessible frequency range.
C

Reswelling of samples in 3M NaCl aqueous solution, which is a poorer solvent for PHEMA
AC

than water (Tab. 1), has an opposite effect: the frequency of the α-relaxation maximum was
shifted from values >103 Hz for water-swollen gels to ~1 Hz. Example of the swelling media
effect on shear properties for the sample 40/1 is presented in Fig. 7, the effect of swelling
media (DMSO or NaCl solution) on the behavior of samples prepared at different IWC/CC
combinations was similar.

13
 ACCEPTED MANUSCRIPT

0
10 40/1
7
10 40/1(3M NaCl)
40/1 (3M NaCl) 10
-1

G' (Pa)

tan δ
6
10
40/1 10
-2 40/1(DMSO)

5
40/1(DMSO) -3
10 -4 10

PT
-3 -2 -1 0 1 2 3 -4 -3 -2 -1 0 1 2 3
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
a) Frequency (Hz) b) Frequency (Hz)
Figure 7 Frequency dependences of a) elastic shear modulus and b) loss factor for gels

RI
prepared at 1 mol% of cross-linker and 40 wt% of water as diluent, and reswollen to
equilibrium in water, DMSO or 3 M aqueous NaCl. Curves for samples swollen in water and

SC
aqueous NaCl are result of time-temperature superpositions (measurements at 5-70 °C were
shifted to a reference temperature of 25 °C). Corresponding sample codes are shown near the
curves.

U
AN
Table 1 Swelling of selected samples prepared with water as diluent and re-swollen to
equilibrium in water, DMSO and 3M aqueous NaCl at room temperature.
Equilibrium solvent content
M
[g swelling medium/g dry gel], 25 °C
Sample
3 M NaCl
water DMSO
solution
D

40/1 0.59 4.95 0.28

50/1 0.71 5.85 n/a
TE

60/1 1.14 7.6 0.64

70/1 2.18 12.0 n/a
80/1 3.11 17.8 2.41
EP

3.3. High-frequency dynamic shear rheometry and approach to the main transition
region
C

At high oscillation frequencies, the loss factor is mainly governed by decreasing segmental
mobility: experimentally, it depends on the cross-link density (practically, for CC ≥ 2 mol%,
AC

Fig. 6b) while the initial dilution with water at this cross-linker level has little effect (Fig. 5b).
With increasing cross-link density, the onset of α-relaxation is shifted to lower frequencies.
The decrease of segmental mobility is caused by segmental association which is assisted by
the covalent cross-links. The fact that the glass transition process as sensed by shear
rheometry of macroporous (80/1), intermediate (50/1, 60/1, 70/1) and homogeneous (40/1)
water-swollen samples occurred in the same frequency range and merged in a single master
curve (Fig. 5b) is a strong indications that the matrix in phase-separated hydrogels has the

14
 ACCEPTED MANUSCRIPT

same swelling properties in water and poorer solvents as for the homogeneous hydrogels of
corresponding concentration of cross-linker. Such conclusion is corroborated by the closeness
of α-relaxation maxima for gels of various morphologies reswollen in 3 M NaCl solution
(Fig. 8).

1.5 60/1

PT
1.0
tan δ

RI
0.5
80/1 40/1

SC
0.0
1E-4 1E-3 0.01 0.1 1 10 100 1000
Frequency (Hz)

U
Figure 8 Frequency dependences of loss factor for hydrogels prepared at 1 mol% cross-linker
and variable concentration of water, reswollen in 3M NaCl to equilibrium. Curves are result
AN
of time-temperature superposition (measurements at 5-55 °C were shifted to a reference
temperature of 25 °C).
M
The assumption of the same swelling properties in water and poorer solvents, now proved,
was used previously [16] for calculation of pore volume in swollen porous gels. The equality
D

of the degrees of swelling in water of the matrix in porous gels and respective homogeneous
TE

gels may imply that also the network structure is the same. However, it is not necessarily so.
The equality of degrees of swelling in water is caused by the notoriously known insensitivity
of the degree swelling of the lightly cross-linked PHEMA gels in water to changes of network
EP

parameters (state of coiling of network chains and cross-link density at its low level). It is no
longer true when water is changed for a better swelling agent.
C

Such test has been performed with the XX/1 series (diluent was water) swollen in DMSO.
The impact of viscous losses was significantly decreased as compared with the same water-
AC

swollen sample (Fig. 9), this trend was similar for all IWC/CC compositions examined. When
swollen in DMSO, for the samples prepared at IWC 40-70 wt%, the losses were very low and
not distinguishable. At IWC = 80 wt%, the losses were significantly higher pointing to a
more complicated topological structure of the matrix (Fig. 10).

15
 ACCEPTED MANUSCRIPT

0.15

80/1 (water)
0.10
tan δ

0.05 80/1 (DMSO)

PT
0.00
1E-3 0.01 0.1 1
Frequency (Hz)

RI
Figure 9 Frequency dependences of loss factor for 40/1 gels, re-swollen in water or DMSO.
Curve for water-swollen sample is a result of time-temperature superposition (measurements

SC
at 15-55 °C were shifted to a reference temperature of 25 °C).

U
40/1
50/1
0.06 AN 60/1
tan δ

70/1
80/1
M
0.03
D

0.00
0.1 1 10
TE

Frequency (Hz)
Figure 10 Frequency dependences of loss factor for gels prepared with 1 mol% of cross-
linker and variable initial concentration of water, and reswollen in DMSO to equilibrium, at
EP

25 °C.
C

3.4. Near-to-equilibrium state and slow relaxations

The low-frequency range of measurement is suitable for characterization of the effect of
AC

permanent bonds – cross-links, and possible slow relaxations related to dissociation and
reformation of arrays of physical bonds of intermediate life-time. We will first consider the
series of homogeneous gels and examine the dependence of modulus and concentration of
elastically active network chains (EANC) obtained from low-frequency storage shear
modulus, and then concentrate on the dilution (morphology) dependence.
3.4.1. Homogeneous gels

16
 ACCEPTED MANUSCRIPT

Figure 4 shows that for the majority of gels in the range ≤ 0.1 Hz, the storage modulus does
not change and it can be considered as the equilibrium value of shear modulus, Ge (Tab. 2).
This is, however, not true for gels of the lowest cross-linker contents 40/0 and 40/0.1 for
which a constant value of G’ has not been reached even at frequency 10-3 Hz. For the G’
values at frequency 10-3 Hz, it was assumed that G’ = Ge, and the concentration of elastically
active network chains (EANC), ν e , in moles per dry unit volume of the network, was

PT
calculated using eq. (4)
G e = RT ν e (φ 20 ) 2 / 3 (φ 2 )1 / 3 (4)

RI
where φ2 is volume fraction of polymer in the gel swollen to equilibrium and φ 20 is volume
fraction of polymerizable substances at preparation, R and T are the gas constant and

SC
temperature in K, respectively. Since for this homogeneous gels series, φ20 = φ2 (see more
discussion on that in ref. [16]) perhaps except of the sample 40/5 where macrosyneresis was

U
observed,
Ge = RTν eφ 20
AN
The values of G’ were also determined for the gels swollen in DMSO, better solvent than
(4a)

water (Table 1, Fig. 11). The tan δ values were very low (Fig. 7, 8) proving no interference of
M
physical interactions with the effect of covalent cross-links. For PHEMA networks swollen in
DMSO, already at 0.1 Hz, G’ reached its equilibrium value.
D

Results for water- and DMSO-swollen gels are summarized in Tab. 2 and compared with the
ideal concentration of the elastically active network chains, νchem in Fig. 12. The quantity νchem
TE

was calculated from the composition of polymerization mixture assuming that each consumed
double bond of the cross-linker contributed to one EANC. The contribution of ethylene glycol
EP

dimethacrylate present in commercial HEMA (0.07 wt% determined by gas chromatography)

to νchem was also considered.
C

Table 2 Equilibrium shear moduli and EANC concentration of EANCs for gels prepared with
AC

40 wt% water and varied cross-linker concentration, reswollen in water or DMSO.

ideal water DMSO
[EANC] G’(10-3 Hz) [EANC] G’(0.1Hz) [EANC]
Sample
νchem [MPa] νe [MPa] νe
[mol/L] [mol/L] [mol/L]
40/0 0.014 (0.110)a (0.083)a 0.0425 0.049
40/0.1 0.034 (0.115)a (0.090)a 0.0440 0.057
40/0.5 0.113 0.135 0.100 n/a n/a
40/1 0.212 0.180 0.130 0.110 0.135

17
 ACCEPTED MANUSCRIPT

40/2 0.406 0.210 0.147 n/a n/a

40/3 0.596 0.300 0.205 0.250 0.271
40/4 0.786 0.390 0.269 n/a n/a
40/5 0.971 0.430 0.289 0.565 0.579
a
at 10-3 Hz, equilibrium modulus not yet reached: [EANC] may be overestimated

The dependence of νe on νchem for water-swollen samples is close to linear with slope of 0.22
(incomplete utilization of cross-linker in forming EANCs) with a positive intercept at νchem =

PT
0 which can be explained by still ongoing relaxations and by permanent constraints
considered usually in the form of trapped entanglements.

RI
SC
6
swelling (g/g)

40/X (DMSO)
4

U
2
40/X
AN
0
0 1 2 3 4 5
M
CC (mol%)
Figure 11 Equilibrium solvent uptake at 25 °C by PHEMA gels prepared with 40 wt% water
D

and variable concentration of cross-linker, CC, reswollen in water (
) or in DMSO ().
TE

0.6
40/X νe=νchem
40/X (DMSO)
EP
νe (mol/L)

0.4
C

0.2
AC

0.0
0.0 0.2 0.4 0.6 0.8 1.0
νchem (mol/L)
Figure 12 Concentration of elastically active network chains determined experimentally and
ideal concentration of elastically active network chains.

18
 ACCEPTED MANUSCRIPT

That slow relaxations caused likely by regrouping of hydrophobic clusters are still operative is
documented by a relatively high value of tan δ for gels 40/0 and 40/0.1 (Fig. 6b). In DMSO,
these clusters do not exist and tan δ falls down to a very small value.
However, another special feature for DMSO-swollen samples can be observed: Compared to
water-swollen gels, the values of G’ and νe calculated from them are considerably higher at
higher cross-link densities and the νe vs. νchem dependence is curved upward. This feature is

PT
caused by a large stretching of the EANCs by swelling, over the Gaussian limit, and thus
inapplicability of eqs. (4) or (4a). For the 40/5 gel, the EANCs are stretched in DMSO to

RI
about 50% of their contour length which is well above the Gaussian limit. The finite
extensibility models of rubber elasticity are available [26], but at this stage we do not analyze

SC
the G’ – CC dependence for DMSO swollen samples any further.
Both dependences show a positive intercept for νchem → 0 of similar magnitude; that
corresponding to DMSO-swollen samples seems to be due to trapped entanglements. The

U
somewhat larger value for water-swollen gels is due to physical associates still relaxing in this
AN
frequency region. The downfall to νchem = 0 at the gel point is not seen because gelation occurs
at νchem < 0.014 mol/L which is the amount contained in the used HEMA monomer.
M
3.4.2. Macroporous gels
As the initial dilution of the polymerizing mixture increases up to IWC = 70 wt%, the G’
D

values (Fig. 6a) in the range of frequencies 10-1 to 10-3 Hz are independent of frequency and
can be taken as equilibrium values. The 80/1 gel shows up a mild decrease of G’ even at 10-3
TE

Hz. Also, the tan δ curve for the gel 80/1 is markedly different from the others: not only in
that the mechanical losses are significantly higher but also in that they increase with
EP

decreasing frequency. Such behavior points to a large nonuniformity in structures of varying

mobilities (involving large dangling structures and even sol). The low-frequency tan δ
C

behavior discussed above for 80/1 (Fig. 6b) is characteristic of all highly porous hydrogels
down to low frequencies. Selected examples are presented in Fig. 13. The intensity of these
AC

slow relaxations increases with increasing initial dilution of the polymerized mixture.

19
 ACCEPTED MANUSCRIPT

0.2
80/1
80/2
0.15 70/3
60/6

0.1
tan δ

PT
0.05
1E-6 1E-5 1E-4 1E-3 0.01 0.1 1
Frequency (Hz)

RI
Figure 13 Frequency dependences of loss factor for macroporous hydrogels swollen in water
to equilibrium. Curves are result of time-temperature superposition (measurements at 25-90

SC
°C were shifted to a reference temperature of 25 °C).

U
To understand the marked change when passing from 60/1-70/1 gels to 80/1 gels (compare
Fig. 5b and Fig. 13), this result should be combined with relaxation behavior in the main
AN
transition region (Section 3.3) and with our knowledge about the mechanism of phase
separation during cross-linking polymerization of HEMA. At IWC = 80 wt%, the composition
M
of the monomer-water mixture has passed its cosolvency maximum [5]: phase separation
starts early and the polymer phase volume is small compared with the liquid phase. During
D

further polymerization, monomers and short growing chains are transported from the liquid
phase to the polymer phase and the structure grows more by forming new layers rather than
TE

by absorption and polymerization of monomer in the already separated polymer phase. At the
same time, the liquid phase gets poorer in the cross-linker molecules because they are
EP

incorporated faster (they have two double bonds). As a result, the spherical formations and the
connecting bridges have a gradient structure.
Tab. 3 summarizes the G’ values across the whole dilution range. Unlike the “homogeneous”
C

series 40/X, the porous samples contain voids. When considering the elastic response of the
AC

matrix, it should be related to the polymeric material. Using the proved assumption that the
swelling degree of the matrix is the same as that of the homogeneous hydrogel of the same
cross-link density, a trivial correction for the modulus should be applied in the form
Ge = G 'φ 2(40) /φ 2 (XX) where φ 2(40) is the volume fraction of polymer in the reference

homogeneous sample and φ 2 (XX) is the volume fraction of the polymer counting also the
water in the pores. This corrected value of the modulus has been used for calculating the

concentration of EANCs using eq. (4) and (4a) with φ2 = φ20 = φ2 (40) (Table 3).

20
 ACCEPTED MANUSCRIPT

Table 3 Equilibrium shear modulus and calculated EANC concentration for gels prepared
with 1 mol% of cross-linker and varied amount of water, and reswollen in water or DMSO.
Equilibrium shear modulus Ge was calculated from the measured G’: Ge = G 'φ 2(40) /φ 2 (XX)
(correction for pores fraction).

PT
ideal water-swollen DMSO-swollen
[EANC] G’ Ge [EANC] G’ Ge [EANC]
Sample
νchem (10-3 Hz) [MPa] νe (0.1 Hz) [MPa] νe
[mol/L] [MPa] [mol/L] [MPa] [mol/L]

RI
40/1 0.212 0.18 0.18 0.130 0.11 0.11 0.135
50/1 0.212 0.13 0.14 (0.102) 0.074 0.088 (0.105)
60/1 0.212 0.074 0.10 (0.071) 0.038 0.055 (0.066)

SC
70/1 0.212 0.037 0.074 0.020 0.043
80/1 0.212 0.0010 0.025 0.0012 0.038

U
However, the values of νe for higher initial dilutions (that is, for macroporous gels) would be
fictive because the trivial correction of modulus is not sufficient for calculating the
AN
concentration of EANCs of the matrix in porous gels and they are not displayed. For
intermediate gel the correction may well approximate the existence of closed pores.
M
Inhomogeneous stresses, and possibly inhomogeneous structure across the matrix profile as
well as participation of deformation modes other than shear count among the other factors.
D

Inhomogeneous stresses are the most important reason for the sudden drop of modulus or
“corrected” modulus when passing to fused-sphere morphology. Inhomogeneous stresses
TE

arise from the fact that the force applied to the outer surface concentrates to a small fraction of
the cross-section of the porous material. This situation is visualized by Fig. 14. The necks
connecting the fused spheres are the effective load bearing parts of the macroporous material.
EP

The stress concentration is strongly enhanced by existing dangling structures (shown in Fig.
14), which means more material per the same cross-section of the neck. The stress
C

concentration results in larger deformation and, therefore, in lower modulus or its corrected
AC

value of the matrix compared to the true modulus of the matrix.

21
 ACCEPTED MANUSCRIPT

PT
RI
Figure 14 SEM image of fused–sphere like macroporous gel (sample 80/0.1). Sketch

SC
visualizes necks connecting particles as loci of higher stress causing larger deformation, and
large dangling structures.

U
The magnitude of the stress concentration and, probably, different deformation geometries
AN
(bending, etc.) is difficult to estimate. The correct answer about the mechanical behavior of
such heterogeneous swollen systems can be obtained by applying the Finite Element Method
M
to a model of such network and finding the value of νe matching best the experimental value
of G’.
D

3.5 Properties of gels prepared using diluents better or worse than water
TE

It was shown in Section 3.1 that changing the nature of diluent can have a profound impact on
gel morphology. Using diluents better than water (solutions of Mg(ClO4)2) can result in
EP

homogeneous gels. Very remarkable is the situation when the solvent power of the diluent
gets worse. Small changes in the concentration of NaCl cause transition from bicontinuous
interlocking to fused spheres morphologies. How are these changes reflected in the gel
C

properties when the gels are reswollen in water? The following set of data (Fig. 15) gives the
AC

answer.

22
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Figure 15 Shear dynamic modulus at 0.1 Hz, loss factor tand, and equilibrium water content,
AN
EWC as functions of quality of diluent used for gel preparation: molar concentrations of NaCl
and Mg(ClO4)2. Dependencies shown for gel series 40/1, 60/1, and 70/2.
M
Making the solvation power better by increasing concentration of Mg(ClO4)2 results in a
decrease of the degree of swelling in water (EWC), because the porous structure obtained for
D

c(Mg(ClO4)2) = 0 renders non-porous and EWC is equal to the value for homogeneous gel.
TE

The EWC for all three gels of is practically the same which supports our previous statement
that swelling of PHEMA gels in water is insensitive to changes in network parameters and ν e

and ϕ 20 . The storage modulus G’ somewhat increases with increasing concentration of

EP

Mg(ClO4)2. This is due to selective binding of the perchlorate anion to the monomer segment
which makes the chains somewhat more extended. The loss factor tan δ is low and somewhat
C

decreases with increasing concentration of Mg(ClO4)2, which means that some remaining
AC

physical associates still relaxing at 0.1 Hz are removed.

Stronger are changes caused by changing the morphology by using NaCl solutions as diluents.
Transition region from bicontinuous interlocking to fused-spheres morphology occurring in
the narrow concentration range of 0.4-0.5 M NaCl for the 60/1series is manifested by a fall of
G’ by an order of magnitude, while the development of bicontinuous interlocking from
assumedly microdroplet like morphology (<0.4 M NaCl) is not reflected in the value of G’.
The degree of swelling of gels in water (EWC) increases with increasing or developing
porosity (60/1, 40/1), the fall of EWC for the 60/1 series at higher concentrations of NaCl is
23
 ACCEPTED MANUSCRIPT

ascribed to macrosyneresis (shrinkage of the gel volume) already observed for some other
gels of high porosity [16].

4. Conclusions
Homogeneous and porous PHEMA gels were prepared by cross-linking polymerization of
HEMA in the presence of water at various dilutions and cross-linker contents. Large changes

PT
in morphology were achieved when water as diluent was replaced by salt solutions
(Mg(Cl04)2 and NaCl) which improve and deteriorate, respectively, the solvation power of

RI
the diluent. Varying of salt concentrations made possible to fine tune the morphology and to
control the passage from bicontinuous interlocking to fused-spheres like gel-liquid

SC
morphology.
The swelling and dynamic mechanical behavior of gels was investigated in relation to their
matrix structure and morphology. The associates existing in water-swollen gels are caused by

U
hydrophobic interactions which limit the solubility of PHEMA and swelling of PHEMA gels
and are mobile enough to enable reaching mechanical equilibrium within the available
AN
frequency range (down to 10-3 Hz). This is valid except of gels of the lowest cross-link
density where ongoing relaxations can be assigned to slow movement of larger physical
M
associates. Reswelling of the gels in a good solvent for PHEMA dissolves the associates and
removes the slow relaxations, which is documented by vanishing magnitude of the loss factor.
D

The low-frequency (equilibrium) mechanical behavior of porous gels depends on the gel
morphology. In the equilibrium mechanical response of fused-spheres type morphologies,
TE

inhomogeneous stress fields cause stress concentration in engaged necks connecting a

network of (elastically active) fused spheres. This network bears elastically inactive dangling
EP

aggregates of spheres (cf. scheme in Fig. 16).

C
AC

24
 ACCEPTED MANUSCRIPT

Figure 16 Network of fused spheres; dark structure – network of elastically active necked-
sphere chains

This feature must be taken into account when the concentration of EANCs of the matrix is
calculated from the modulus of the macroporous material. If this is not respected, the
calculated concentration of EANCs is incorrect. The borderline between porous gels

PT
composed of fused small particles without dangling chains and those macroporous gels
composed of large particles with abundant dangling structures and thinner necks is that

RI
demarcation line shown in Fig 1. The increased losses observed at low frequencies for these
fused-spheres-type gels can be ascribed to slow movement of dangling associates of spheres.
From the dynamic mechanical properties in the α-relaxation region, the frequency

SC
dependences of the loss factor for homogeneous gels and samples of different porosity merge
into a single master curve which indicates that the swelling properties of porous hydrogel

U
matrix are the same as those of homogeneous hydrogels of the same cross-link density

AN
because the content of diluent in the gel is the main factor determining the magnitude of
losses.
M
Acknowledgement
The authors wish to acknowledge the financial support of this work by the Czech Science
D

Foundation (grant P101/12/1306) and by the Ministry of Education, Youths and Sports of the
Czech Republic (grant SVV-2012-265305).
TE

5. References
EP

1. Montheard JP, Chatzopoulos M, Chappard D. J Macromol Sci-Pol R 1992;32:1-34.

2. Wu D, Xu F, Sun B, Fu R, He H, Matyjaszewski K. Chem Rev 2012;112:3959-4015.
3. Svec F. J Chromatogr A 2010;1217:902-924.
C

4. Hentze HP, Antonietti M. Rev Mol Biotechnol 2002;90:27-53.

AC

5. Dušek K, Sedláček B. Eur Polym J 1971;7:1275-1285.

6. Šprincl L, Kopeček J, Lím D. Calcif Tissue Res 1973;13:63-72.
7. Chirila TV, Constable IJ, Crawford GJ, Vijayasekaran S, Thompson DE, Chen YC,
Fletcher WA, Griffin BJ. Biomaterials 1993;14:26-38.
8. Rubinstein M, Dobrynin V. Trends Polym Sci 1997;5: 181-186.
9. Tanford C. The Hydrophobic Effect, 2nd ed. New York: Wiley, 1980.
10. Kavanagh GM, Ross-Murphy SB. Prog Polym Sci 1998; 23:533-562.

25
 ACCEPTED MANUSCRIPT

11. Bode F, Alves da Silva M, Drake AF, Ross-Murphy SB, Dreiss AC. Biomacromolecules
2011;12:1341-1357.
12. AI-Ruqaie IM, Kasapis S, Richardson RK. Polymer 1997;38(22):5685-5694.
13. Sworn G, Kasapis S. Carbohydr Res 1998;309:353-361.
14. Kesavan S, Prud'homme RK. Macromolecules 1992;25:2026-2032.
15. Meakin JR, Hukins DWL, Apsden RM, Imrie CT. J Mater Sci-Mater M 2003;14:783-787.

PT
16. Karpushkin E, Dušková-Smrčková M, Remmler T, Lapčíková M, Dušek K. Polym Int
2012;61(2):328-336.

RI
17. Liu Q, Hedberg EL, Liu Z, Bahulekar R, Meszlenui RK, Mikos AG Biomaterials
2000;21:2163-2169.

SC
18. Kirschenmann L, Pechhold W. Rheol Acta 2002;41:362–368.
19. Dušek K, Bohdanecký M, Vošický V. Collect Czech Chem Commun 1977;42:1599-1607.
20. Dusek K, Duskova-Smrckova M. Prog Polym Sci. 2000;25:1215–1260.

U
21. Wichterle O, Chromeček R. J Polym Sci 1969;16:4677-4686.
22. Williams ML, Landel RF, Ferry JD. J Am Chem Soc 1955;77:3701–3707.
AN
23. Janáček J, Ferry JD. Macromolecules 1969;2(4):370-378.
24. Janáček J, Ferry JD. Macromolecules 1969;2(4):379-385.
M
25. Dušek K, Sedláček B. Collect Czech Chem Commun 1971;35:1569-1577.
26. Treloar LRG. The physics of rubber elasticity, New York: Oxford University Press, 1975.
D
TE
C EP
AC

26