+ All Categories
Home > Documents > Nano-hybrid self-crosslinked PDMA/silica hydrogels

Nano-hybrid self-crosslinked PDMA/silica hydrogels

Date post: 17-Dec-2016
Category:
Upload: alba
View: 225 times
Download: 4 times
Share this document with a friend
13

Click here to load reader

Transcript
Page 1: Nano-hybrid self-crosslinked PDMA/silica hydrogels

PAPER www.rsc.org/softmatter | Soft Matter

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online / Journal Homepage / Table of Contents for this issue

Nano-hybrid self-crosslinked PDMA/silica hydrogels

Linn Carlsson, S�everine Rose, Dominique Hourdet* and Alba Marcellan*

Received 19th February 2010, Accepted 8th April 2010

DOI: 10.1039/c0sm00009d

We discovered that the free radical polymerization of N,N-dimethylacrylamide in water can lead, above

a certain concentration, to gels without any added difunctional crosslinker. These so called ‘‘self-

crosslinked’’ hydrogels were prepared and their weak mechanical properties were improved by

introducing silica nanoparticles.

From swelling experiments performed at equilibrium in aqueous media, it was shown that silica

particles behave as adhesive fillers and strongly interact with PDMA chains. These interactions are

responsible for the reinforcement of mechanical properties. From initial elastic moduli, determined in

the preparation state, we show that the elastic behaviour of these hydrogels mainly originates from

entanglements and from physical crosslinks that can be controlled by the polymer concentration and

the ratio between silica particles and polymer chains, respectively. The mechanical behaviour was

characterized using: monotonic tensile tests, loading-unloading cycles at large strains and stress

relaxation experiments in order to investigate long time behaviour. The introduction of silica highly

increases the stiffness of the network without greatly reducing its extensibility, implying that strong

interactions take place between PDMA chains and silica surfaces.

Non-linear behavior was pointed out: softening at small deformations and hardening at high

deformations which is related to finite chain extensibility. All these effects have been shown to strongly

depend on the silica content. The analysis of hysteresis and residual strains induced by cycles, clearly

indicate that contrary to chemical crosslinkers, hybrid interactions increase the dissipative process.

1. Introduction

Hydrogels are crosslinked polymers forming a three-dimensional

network and are highly swollen in aqueous environments. From

a macroscopic point of view, if the lifetime of polymer crosslinks is

long enough compared to the experimental time, hydrogels behave

like solids; they have a defined geometry and they do not flow like

liquids in the corresponding timescale. Simultaneously, these

macromolecular reservoirs are similar to real solutions or dialysis

bags wherein the soluble molecules can diffuse and be outsourced

with diffusion constants depending on the mesh size of the network

and on the level of interactions. Since the beginning of the 80’s

hydrogels have known a continuous growth with important appli-

cations like molecular sieves, super absorbent polymers or contact

lenses.1–4 More recently, with the development of stimuli-responsive

hydrogels, they continue to attract a very important research activity

with potential applications in biomedical areas as biotechnological

devices, tissue engineering, and drug delivery systems.5–10

We can consider today that the chemist is able to design very

sophisticated macromolecular architectures capable of swelling or

collapsing under various stimuli like temperature, pH, ionic

strength, light, electric field, magnetic field, etc.,11–19 but one of the

main issues remaining is the mechanical integrity of these soft

materials. The mechanical behaviour of soft hydrogels is actually

an important research area and different groups are intensively

working in this field trying to solve the paradox: how to get tough

Physico-chimie des Polym�eres et des Milieux Dispers�es, UMR 7615,UPMC-CNRS-ESPCI, 10 rue Vauquelin, 75005 Paris, France. E-mail:[email protected]; [email protected]; [email protected]; [email protected]; Fax: +33 (0)1 40 79 46 86; Tel: +33 (0)1 40 79 46 82

This journal is ª The Royal Society of Chemistry 2010

hydrogels? During this decade, several groups, mainly in Japan

have proposed some new strategies to improve the mechanical

properties of soft networks. Two main paths have been investi-

gated. The first strategy consists in reducing as much as possible

the heterogeneities in the network, implying that defect free

materials will not easily nucleate cracks. We can mention the

pioneering works of Okumura et al. on slide-ring polyrotaxane

hydrogels20,21 and more recently the Tetra-PEG gels of Shibayama

and coworkers.22 These gels are highly extensible and elastic but

are not very resistant to the propagation of an existing crack. A

second path has consisted of toughening the material by enhancing

the energy dissipation during crack propagation. Gong et al.,23–25

with so-called double networks have designed partially inter-

connected interpenetrated networks (IPN), that greatly enhance

the fracture toughness by increasing the dissipative volume ahead

of the crack tip. These gels are however permanently damaged

upon deformation and are softer when loaded a second time.

Both of these strategies involve a rather sophisticated chemistry

which is not off the shelf. A third and much easier strategy has been

pioneered by Haraguchi et al. with nanocomposite hydrogels

(NC).26–31 Most NC gels were synthesized by free radical polymeri-

zation of N-isopropylacrylamide (NIPA) or N,N-dimethylacryla-

mide (DMA) in the presence of exfoliated clays (Laponite: diameter

20–30 nm and thickness 1 nm). Since no crosslinking agent was

added, they were classified as physical gels. The structure of the gels is

still not clear, but the authors assume that the polymerization starts

from the clay platelet surface, leading to a percolating network where

clay particles act as multi-functional crosslinks.30 Accordingly, the

elasticity of the hybrid network would originate from polymer

chains making bridges between the clay platelets. Haraguchi et al.31

attribute the high deformability of NC gels (3max ¼ 800–1000%),

Soft Matter, 2010, 6, 3619–3631 | 3619

Page 2: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

compared to chemically cross-linked gels (3max ¼ 20–50%), by the

larger average distance between cross-links. In that case the average

distance between particles (physical crosslinks) is directly controlled

by the clay/monomer ratio. NC gels combine a high modulus with

a high extensibility, suggesting a large density of elastic chains

(contributing to the modulus) but a large distance between cross-

link points (high extensibility). The key would then be in the mul-

tifunctionality of the crosslinks. The weakness of such nanoclay

gels is however their highly viscoelastic character and residual

deformation due to the lack of chemical crosslinks. The concept of

nanocomposite gels with reinforced mechanical properties has been

successfully extended to other polymers, like poly(acrylamide)

using modified clays32 but conversely the replacement of clay by

other inorganic particles like silica or titanium oxide has not

provided comparable properties.33

Quite recently a similar concept of hybrid networks applied to

semi-dilute solutions was developed in our group34,35 by using

specific interactions between silica nanoparticles and polymer

side-chains (PNIPA or PDMA) grafted onto a non-adsorbing

poly(acrylamide) backbone. In a semi-dilute solution, physical

gels having a long lifetime, can be easily formed by simply mixing

silica nanoparticles with the copolymer. The main problem with

this procedure is that homogeneous hybrid gels become rather

difficult to prepare at polymer concentration above 2 wt% due to

the strong interactions taking place between the graft copolymers

and the silica particles.

This limitation can be overcome by polymerizing the monomer

in the presence of the inorganic filler. Following this idea, the

simplest way was to replace graft copolymers by homopolymers

using only DMA as monomer and basing our approach on the

strong interaction that will be developed during polymerization

between PDMA chains and silica nanoparticles. This idea has

been applied successfully using some additional chemical cross-

linker N,N0-methylenebisacrylamide (MBA) in the preparation

of the hybrid network.36 From a detailed comparison of the

hybrids with unmodified PDMA gels, we show in a first

approach that the incorporation of only 7 vol% silica nano-

particles in the hydrogel increases the initial stiffness by a factor

of 6 and the fracture toughness up to an order of magnitude.36

With the idea of increasing the extensibility of these silica-filled

gels, it was clearly interesting to reduce or completely eliminate

the chemical crosslinker (MBA) from the formulation.

In this paper, we report the facile synthesis of new hybrid

hydrogels, what we have called ‘‘self-crosslinked’’ hydrogels,

prepared only from water, monomer, initiator and silica nano-

particles. Then we describe the swelling behaviour and the

mechanical properties of these hybrid networks as a function of

DMA and silica concentration, comparing also our results with

those obtained in the presence of some MBA crosslinker. Finally,

from mechanical analyses performed at large strains we will

discuss the role of physical crosslinks and dissipation processes

on the reinforcement of organic networks.

Experimental part

Materials

N,N-Dimethylacrylamide (DMA, 99%, Aldrich), potassium persul-

fate (KPS, Acros Organics), N,N,N0,N0-tetramethylethylenediamine

3620 | Soft Matter, 2010, 6, 3619–3631

(TEMED, 99,5%+ redistilled, Sigma Aldrich) and N,N0-methyl-

enebisacrylamide (MBA, Fluka) were used as received without

further purification.

The silica particles (Ludox TM-50 from Dupont) were kindly

supplied by IMCD France SAS. The silica suspension was

characterized by dynamic light scattering using a ALV/CGS-3

compact goniometer (ALV, Langen, Germany). The measure-

ments performed at different scattering angles (30�, 90� and 150�)

in dilute conditions give an average radius R ¼ 17 nm for the

silica particles and a specific area Sspe ¼ (3/rSiR) ¼ 77 m2 g�1,

considering rSi ¼ 2.3.106 g m�3 for the density of pure silica. The

silica suspension (52 wt%; pH¼ 8.5–9.5) was used as received for

the synthesis of hybrid hydrogels.

Preparation of hydrogels

Hydrogels were prepared at 25 �C by free-radical polymerisation

of DMA in an aqueous suspension of silica nanoparticles using

KPS and TEMED as a redox initiator. For all the syntheses, the

molar ratio (DMA)/(KPS)/(TEMED) was set equal to 100/1/1

while the relative amounts of DMA, silica and water were varied

keeping constant either the weight ratio polymer/water (PW) or

silica/polymer (SP). In some cases, a very small amount of

chemical crosslinker (MBA) was introduced (R ¼ (MBA)/

(DMA) ¼ 0.1–0.5 mol%) in order to investigate the impact of

covalent crosslinks on the mechanical properties.

The nomenclature of hydrogels is SPX_PWY-RZ, with S for

silica, P for polymer, W for water, R for crosslinker (when used).

X is the weight ratio between silica and DMA, Y is the weight

ratio between DMA and water and Z is the molar percentage

between MBA and DMA. The composition of the hydrogels we

tested is summarized in Table 1.

A typical procedure for the synthesis of hydrogels can be

described as follows in the case of the sample SP2_PW0.14.

First, 1.485 g of DMA and 0.0405 g of KPS were dissolved at

25 �C in an aqueous suspension of silica particles initially

prepared by mixing of 7.699 g of water and 5.712 g of Ludox

TM-50. The homogeneous suspension was purged with nitrogen

during 30 min under magnetic stirring and then 22.5 mL of

TEMED was rapidly added under vigorous stirring. The mixture

was then transferred under nitrogen atmosphere into a mold,

made with two hydrophobized glass plates (chemically treated

with octadecyltrichlorosilane) separated by a spacer. The poly-

merization was left to proceed during 1 night. In these condi-

tions, it was shown from size exclusion chromatography

experiments performed on hydrogel samples swollen in a large

excess of solvent that the conversion of monomer is quantitative

and that the amount of extractable (polymer chains) was less

than 5 wt%.

At the end of the reaction, the gels were cut with a die-cutter of

rectangular shape and the samples were stored into paraffin oil

until mechanical testing in order to avoid any change in hydrogel

composition induced by swelling or drying.

Unfilled DMA networks were also synthesized in plastic

containers of cylindrical shape (diameter ¼ 33 mm) in order to

produce sample shapes suitable for rheological experiments.

For each gel (see Table 1), the initial swelling in the prepara-

tion state (Q0) was calculated from their initial composition,

This journal is ª The Royal Society of Chemistry 2010

Page 3: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Table 1 Composition and nomenclature of hydrogels classified by increasing silica/polymer ratio and then by polymer/water ratio

Sample Name SPX_PWY-RZ DMA/g H2O/g Silica/g KPS/g MBA/g TEMED/g SwellingaQ0

SP0_PW0.041 0.594 14.38 0.016 0.007 26.5SP0_PW0.047 0.668 14.31 0.018 0.008 23.5SP0_PW 0.052 0.7425 14.23 0.02 0.009 21.2SP0_PW0.11 1.485 13.50 0.041 0.017 10.6SP0_PW0.14 1.485 10.44 0.041 0.017 8.4SP0.5_PW0.14 1.485 10.44 0.74 0.041 0.017 8.4SP0.5_PW0.12 1.485 12.67 0.74 0.041 0.017 10.0SP1_PW0.12 1.485 11.92 1.49 0.041 0.017 9.5SP1_PW0.14 1.485 10.44 1.49 0.041 0.017 8.4SP2_PW0.14 1.485 10.44 2.97 0.041 0.017 8.4SP2_PW0.24 2.093 8.6 4.19 0.057 0.025 5.1SP2_PW0.44 2.79 6.31 5.58 0.076 0.033 3.3SP2.2_PW0.47 2.79 5.89 6.18 0.076 0.033 3.2SP3.1_PW0.34 2.093 6.21 6.57 0.057 0.025 4.1SP3.5_PW0.14 1.485 10.44 5.20 0.041 0.017 8.4SP5_PW0.1 1.395 14.14 6.97 0.038 0.016 11.7SP5_PW0.14 1.395 9.81 6.97 0.038 0.016 8.4SP5_PW0.21 1.395 6.53 6.97 0.038 0.016 5.9SP2_PW0.14_R0.1 1.485 10.44 2.97 0.041 0.002 0.017 8.4SP2_PW0.14_R0.2 1.485 10.44 2.97 0.041 0.005 0.017 8.4SP2_PW0.14_R0.5 1.485 10.44 2.97 0.041 0.012 0.017 8.4

a Q0, swelling at the preparation state.

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

assuming full conversion of monomers into the polymer network

as mentioned previously and the additivity of volumes:

Q0¼1þvw

spe

vpspe

m0

p;w

m0p

� 1

!(1)

with vwspe the specific volume of water, vp

spe the specific volume of

the dry polymer network (vpspe ¼ 0.95 mL g�1), m0

p the mass of the

polymer network (m0p ¼ mDMA + mMBA) and m0

p,w ¼ m0p + m0

w

the total mass of the swollen gel in the preparation state with m0w

the total mass of water excluding the silica contribution as the

water does not swell the inorganic phase.

Drying

Drying experiments were carried out on various hydrogel

samples under ambient conditions by measuring the loss of water

as a function of time from a rectangular piece of swollen gel

(100 mm � 5 mm � 2 mm) exposed to air but without any

specific convective flow. For a gel sample of approximately 1 g,

the drying rate is equal to 0.1 mg min�1 cm�2.

Swelling measurements

For each sample, small pieces of hydrogel in the preparation state

were initially cut, weighed and then placed at room temperature

into a glass container with a large excess of sodium chloride 0.5

mol L�1. The solution was changed every day during 10 days and

the gels, swollen at equilibrium, were finally weighed. The equi-

librium swelling ratio (Qe) of the hydrogels in sodium chloride

0.5 mol L�1 were calculated according to eqn ({1}) taking into

account the mass of polymer network and the total mass of

absorbed solution.

Rheology

Rheological experiments were carried out at 25 �C on

a controlled stress rheometer (Haake RS600) equipped with

This journal is ª The Royal Society of Chemistry 2010

parallel roughened plates, a solvent trap filled with water for

minimizing evaporation of hydrogel sample and a Peltier device

for the temperature control. The gap between the plates was set

at 3 mm before the onset of the measurements and the sample

was coated with a thin layer of paraffin oil to prevent water

evaporation. The frequency sweep was applied at low deforma-

tion, in the linear regime, over the frequency range 0.1–10 Hz.

Mechanical testing

Tensile tests were performed on a standard tensile Instron

machine, model 5565. The device used a 10 N load cell (with

a relative uncertainty of 0.16% in the range from 0 to 0.1 N) and

a video extensometer which follows the local displacement up to

120 mm (with a relative uncertainty of 0.11% at full scale).

The gel samples (L ¼ 100 mm, w ¼ 5 mm, t ¼ 2 mm) were

prepared in triplicate and stored as previously described. The

gauge length was taken constant for all the tests (L0 � 30 mm)

and the gel strip was marked with two dots with a white marker,

for their recognition by the video extensometer. Three different

mechanical tests were carried out at room temperature: mono-

tonic tensile tests, loading-unloading cycles and stress relaxation

tests.

Tensile tests were carried out at two different strain rates: at

100 mm min�1 (nominal strain rate _3 � 0.06 s�1) for standard

conditions and at higher rate 1000 mm min�1 (nominal strain rate

_3 � 0.60 s�1).

Loading–unloading experiments were performed in order to

characterize the dissipated strain energy (hysteresis). Measure-

ments were carried out on hydrogels of similar size, using the

same conditions as reported for the tensile measurements in

standard conditions (_3 � 0.06 s�1). The loading was applied at

a main constant crosshead velocity, from 0 to the maximal strain

value, 3max and then the sample was unloaded. This loading–

unloading cycle was repeated 3 times. Two levels of maximal

Soft Matter, 2010, 6, 3619–3631 | 3621

Page 4: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

strains were investigated (3max ¼ 0.5 and 3max ¼ 5) and all the

experiments were duplicated.

Stress relaxation tests consist in maintaining a fixed strain level

and recording the stress as a function of time. The same proce-

dure was carried out for all the samples using a constant strain

level of 30 ¼ 0.5 and a fixed strain rate _3 � 0.06 s�1 to reach the

applied strain.

Results and discussion

Polymerization of DMA and self-crosslinking process

Alkyl-acrylamides like DMA or NIPA are very common water-

soluble monomers often used in the preparation of hydrophilic

macromolecular structures. One of the main advantages of DMA

compared to the well-known unsubstituted acrylamide is its high

solubility in both water and organic solvents which opens the

route to homogeneous copolymerization with hydrophobic

comonomers and the preparation of amphiphilic polymers or

networks.37–39 In the case of NIPA or more hydrophobically

substituted monomers, the use of organic solvent is also an

interesting way to prepare homogeneous networks and to avoid

any phase separation that generally occurs in aqueous media,

even at low temperature, just above the lower critical solution

temperature (LCST) of the polymer.

Polymerization in aqueous media is of course more friendly for

the environment but in that case radical initiators can induce

transfer reactions leading to nonlinear polymer structures.

Potassium persulfate can react with substituted amines and the

extent of the reaction strongly depends on the level of substitu-

tion of nitrogen.40 For instance, in the case of acetamide deriv-

atives, the reaction of KPS carried out in equimolar conditions

(0.4 mol L�1) at 85 �C in oxygen-free conditions leads to a full

transformation of N,N-dimethylacetamide, whereas only 56% of

N-methylacetamide is modified and less than 10% of acetamide.

According to the authors, KPS dealkylates N-substituted and

N,N-disubstituted amides to amides and N-substituted amides

respectively, while the dealkylated group appears in the reaction

mixture as the corresponding aldehyde or ketone. As shown in

Fig. 1A, the primary radicals issued from the thermal decom-

position of persulfate can react with substituted amides via

radical attack on the a-carbon to the amide nitrogen. If we

Fig. 1 (A) First steps of the radical attack of persufate on N,N-dimethylal

polymerization of N,N-dimethylacrylamide induced by persulfate initiator.

3622 | Soft Matter, 2010, 6, 3619–3631

assume a similar mechanism between KPS and the methyl groups

of DMA, these transfer reactions, which could occur either with

the monomer or the polymer, will give rise to the formation of

branched structures and/or three-dimensional networks with

a functionality equal to 3 (see Fig. 1B).

This very important feature of the polymerization of

N-substituted acrylamides is a major drawback if one aims to

prepare model linear polymers but could be useful if the main

goal is to develop soft materials with a solid rather than a liquid

behaviour. Self-crosslinking has been reported for instance in the

preparation of N-isopropylacrylamide gel nanospheres by Gao

and Frisken.41 On the basis of the reaction mechanism given in

Fig. 1, it appears that the viscoelastic properties of macromo-

lecular formulations would strongly depend on both monomer

and persulfate concentrations. In the framework of the present

study we have investigated the role of the monomer concentra-

tion between 0.1 and 1 mol L�1 (1 to 10 wt%) keeping the molar

ratios (DMA)/(KPS)/(TEMED) equal to 100/1/1.

At low concentrations, typically below 4 wt%, the formula-

tions obtained after polymerization remain rather inhomoge-

neous (suspension of polymer clusters) and therefore it was

difficult to get reproducible results with rheological measure-

ments. In the following, we will simply consider the data

obtained at higher concentrations. As plotted in Fig. 2, the elastic

character of the formulations obtained after polymerization

clearly increases with DMA concentration with a progressive

shift from liquid-like behaviour (G0 < G0 0 : solution) to solid-like

behaviour (G0 > G0 0 : gel).

At the gel point, both elastic and loss moduli should follow the

same frequency dependence (G0 �G0 0 � un).42 And this situation

is clearly observed in Fig. 2 at a concentration Cgel close to

4.5 wt%. Below Cgel, the formulation consists of a distribution of

finite clusters as already mentioned previously. Beyond the gel

point, the materials consists of a percolating and infinitely large

macromolecule which can swell but cannot dissolve in a solvent,

and lower molecular weight molecules (sol fraction) which are

still extractable from the gel. In the following, we will take

advantage of this self-crosslinking reaction of DMA to prepare

slightly crosslinked hybrid hydrogels.

In the work of Haraguchi et al.27 on NC gels, it was shown that

nanocomposite gels prepared at low polymer concentration

(typically below CDMA¼ 0.5 mol L�1, i.e. about 5 wt%) were very

kylamide40 and (B) extrapolation of this mechanism to self-crosslinking

This journal is ª The Royal Society of Chemistry 2010

Page 5: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Fig. 2 Elastic (filled symbols) and loss (hollow symbols) moduli of

PDMA formulations prepared at different DMA concentrations (without

MBA crosslinker): 4 wt%: SP0_PW0.041 (C); 4.5 wt%: SP0_PW0.047

(-); 5 wt%: SP0_PW0.052 (:) and 10 wt%: SP0_PW0.14 (;).

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

brittle and often neither uniform nor transparent. On the other

hand, a significant increase of both modulus and strength was

observed above this concentration. This behaviour was corre-

lated to the progressive formation of a real organic network with

increasing DMA concentration (sol/gel transition) in the pres-

ence of clay nanoparticles. In the present work, this idea is

confirmed in the same concentration range without added

particles and we think that self cross linking of PDMA is an

important requirement in the formation of NC gels as it will

restrict the flow properties of polymer chains and will emphasize

their elastic behaviour.

Formulation of nano-hybrid hydrogels

Several criteria have to be considered for the design of hybrid

hydrogels.

(1) As previously reported, a critical DMA concentration is

needed to reach the percolation threshold and to prepare

a homogeneous macroscopic network. This aspect is directly

related to the weight ratio ‘‘polymer/water’’ (PWY in the

nomenclature) which is inversely proportional to the swelling Q0.

(2) In a previous study34 with the same Ludox particles, it has

been shown that small polymer chains of PNIPA and PDMA

interacted strongly and adsorbed onto silica surfaces. The

adsorption isotherms were almost the same for the two

polymers and the maximum amounts of adsorbed polymers

were Gmax y 1 mg m�2. Moreover, from calorimetric experi-

ments performed on PNIPA/silica mixtures, it was shown that

at low coverage (G < 0.5 mg m�2) the polymer chains strongly

adsorbed in a flat conformation at the surface of the particles

although at higher coverage (0.5 < G # 1 mg m�2) mainly loops

and tails were formed in the outer shell with swelling and

responsive properties.

On the basis of Ludox TM-50 particles characterized by an

average specific surface Sspe ¼ 77 m2 g�1, the maximum polymer

adsorption (Gmax ¼ 1 mg m�2) is expected for a weight ratio

‘‘silica/polymer’’ of 13, corresponding to SP13 according to the

chosen nomenclature.

This journal is ª The Royal Society of Chemistry 2010

Taking into account these two key aspects, we have tried to

explore a wide range of compositions but as shown in Table 1, we

still have worked with DMA concentrations above 10 wt%

(PWY¼ 0.10 to 0.44) in order to get hybrid hydrogels with solid-

like properties. Moreover the preparation of homogeneous

samples, starting from a well-dispersed suspension of particles

(concentration of 52 wt%), leads us to work well below Gmax

(SPX � 13), i.e. with a large excess of non-adsorbed polymer

chains.

At the end of the synthesis, a very simple comparison between

organic hydrogels of DMA and hybrid ones clearly shows that

the introduction of silica particles into the formulations gives rise

to much stiffer materials that can be easily die-cut and submitted

to tensile tests.

While hybrid hydrogels made from DMA are generally very

sticky at low concentration of silica particles (viscoelastic

behaviour), either the addition of a small amount of MBA (even

at 0.1 mol%) or silica particles strongly changes the mechanical

properties of the material that becomes less sticky but much more

brittle.

Swelling behaviour of hybrid hydrogels.

When a polymer network is immersed in a solvent, the swelling

equilibrium is reached when the osmotic pressure inside the gel

(Pgel) becomes equal to that of the surrounding medium (Pout),

as described by the following relation:

Pgel ¼ Pm + Pel + Pion ¼ Pout (2)

where Pm, Pel and Pion are, respectively, the mixing, the elastic

and the ionic contributions to the osmotic pressure.

In the case of hybrid hydrogels, the silica particles embedded

inside the network bring counter ions coming from partial

dissociation of silanol groups and the latter contribute positively

to the osmotic pressure (Pion > 0), like in polyelectrolyte gels.

With the aim of studying specifically the role of silica beads as

potential crosslinkers, all the swelling experiments have been

performed at high ionic strength (NaCl 0.5 mol L�1), in order to

screen the ionic contribution (Pion y Pout ¼ P0ion).

The general comparison between equilibrium swelling values

reported in Fig. 3 and 4, points out the influence of the polymer

concentration used in the preparation state, which determines the

level of entanglements inside the network, as well as the impact of

MBA and silica particles as crosslinkers. By comparison with

MBA, the silica particles are much less effective in limiting the

equilibrium swelling (see Fig. 4), but we have to consider that: (1)

the PDMA chains only interact with the silica surface and (2) as

opposed to organic networks which are generally considered

homogeneous, hybrid ones are heterogeneous in nature alter-

nating dense zones, where PDMA strongly interacts with silica

beads, with swollen ones which correspond to the pure PDMA

matrix. The swollen zones can therefore absorb water in large

quantities.

From a general point of view, the addition of fillers in elas-

tomer networks and the consequences on mechanical and

swelling properties is a very important issue which has been

widely studied experimentally and theoretically.43–48

Soft Matter, 2010, 6, 3619–3631 | 3623

Page 6: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Fig. 3 Swelling equilibrium of hybrid hydrogels in NaCl 0.5 mol L�1 as

a function of the molar percentage between MBA and DMA (Z param-

eter). The weight ratios between polymer and water at the preparation

state (Y parameter) are respectively: 0.14 (black), 0.24 (grey) and 0.44

(white).

Fig. 4 Swelling equilibrium, Qe of hybrid hydrogels SPX_PW0.14 in

NaCl 0.5 mol L�1 versus silica volume fraction (dots). Comparison with

the theoretical model of Lequeux without correction for adsorbed poly-

mer layer (solid line) and with correction (dashed line : r0 ¼ 17 nm and

t ¼ 2 nm, Qlayer ¼ 2).

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

Swelling measurements of filled elastomers is for instance

a classical test to probe the level of interactions inside the

network. If there is no adhesion at all between the polymer

matrix and the particles, the filled polymer is expected to swell

slightly more at the equilibrium (Qe) than for the unfilled matrix

itself (Qref) due to the formation of cavities filled with solvent

around the particles (Fig. 5 (a)). Conversely, if strong interac-

tions exist between the matrix and the particles, the swelling Qe

will be lower than Qref and heterogeneous due to the constraints

at the polymer–filler interface (Fig. 5 (b)).

Using these assumptions and a continuous media mechanics

approach, Lequeux and coworkers48 have derived an analytical

expression relating the swelling properties of the filled elastomer

(Qe), the swelling of the polymer matrix alone (Qref) and the

volume fraction of particles in the initial state (f):

3624 | Soft Matter, 2010, 6, 3619–3631

Qe¼

24 Q

1=3ref þ

3ð1�Q1=3ref Þð1� nÞ

2ð1� 2nÞ þ ð1þ nÞð0:64=fð1þ t=r0Þ3Þ

!2

�f

35,ð1� fÞ

(3)

with n the Poisson coefficient, taken equal to 0.5, r0 the radius of

neat particles and t the thickness of the adsorbed polymer layer

(Fig. 5c).

Using the previous model, simply taking into account the

volume fraction of neat particles (t¼ 0), we get theoretical values

which slightly overestimate the experimental swelling of filled

network (see Fig. 4). As discussed by Lequeux and co-workers48

this general behaviour could be explained by the existence of

unswollen polymer layer strongly adsorbed at the particle

interface. The main effect is an increase of the effective volume of

hard particles which can be taken into account in the model using

an additional polymer layer of thickness, t. In the framework of

the present study, assuming that 1 mg of PDMA is adsorbed on

1 m2 of silica particles (r0 ¼ 17nm), we can estimate the thickness

of the PDMA layer t ¼ 1 – 2 nm, considering that this layer is

fully dry (Qlayer ¼ 1) or slightly swollen (Qlayer ¼ 2), respectively.

In Fig. 4, we can see that the model can be slightly improved

assuming a polymer layer of 2 nm but clearly a full agreement

cannot be obtained on the whole range of silica concentration.

One possible reason of the relatively slow decrease of the

equilibrium swelling at high silica fraction could be due to the

heterogeneous polymerization taking place in a dispersed media.

Based on strong interactions between silica and PDMA, we can

reasonably assume that during the polymerization process, there

is a competition between the kinetic chains propagating at the

interfaces and in the solution. Assuming that, at the end of the

polymerization, the total amount of adsorbed PDMA is given by

Gmax ¼ 1 mg m�2, the concentration of PDMA chains forming

the bulk matrix is obviously less concentrated than the one

prepared without silica. For example, in the case of SP5_PW0.14

with 38 wt% of DMA, is expected to polymerize and self-adsorb

on silica surfaces. The remaining 62 wt% of DMA will poly-

merize homogeneously in the solution, leading to an average

ratio polymer/water in the matrix of 0.08 instead of 0.14.

Therefore, although a constant polymer/water ratio was used for

this series, the local swelling ratio of the matrix may increase with

increasing silica content.

Mechanical properties

Tensile behaviour: initial modulus and non-linear behaviour. For

each sample, various tests have been performed in order to

ensure the reproducibility of the experiments. Typical stress–

strain curves are shown Fig. 6 for the standard nominal strain

rate _3 �0.06 s�1. Results and particularly the initial part of the

curve (i.e. initial stiffness) are highly reproducible for a given

formulation.

Self-crosslinked hybrid hydrogels are seen to be highly

deformable, with average strains above 1000%. However, the

unfilled samples SP0 samples are too soft and sticky to be

handled and only the modulus can be measured by rheometry (at

1 Hz: E0 ¼ E0 ¼ 3G0 ¼ 1.05 kPa).

The tensile properties of nanosilica PDMA gels strongly

depend on silica content, but the general tensile behaviour is

This journal is ª The Royal Society of Chemistry 2010

Page 7: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Fig. 5 Schematic representation of the swelling behaviour of a polymer layer (initial radius R0) surrounding a hard sphere (r0) : (a.) swelling without

adhesion between the particle and the polymer matrix, swelling with adhesion without (b) or with (c) an unswollen polymer layer.49

Fig. 6 Typical tensile stress-strain curves at _3 � 0.06 s�1: effect of filler content for SP0.5_PW0.14, SP1_PW0.14, SP2_PW0.14 and SP5_PW0.14.

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

similar for all the samples. The behaviour is highly non-linear

and shows high levels of strain at break compared to the

conventional PNIPA29 or PDMA27 crosslinked gels. In the first

part of the curve, at low strains (below 2%), the initial modulus Ei

is defined and identified as the Young’s modulus. At intermediate

and large strains a marked softening followed by a distinctive

hardening relative to standard Gaussian elasticity, is observed.

The deformation process is homogeneous during the test: no

necking process is seen in the strain range investigated.

As shown in Fig. 7, a significant increase of the initial stiffness

is observed as the volume fraction of nanoparticles increases.

According to the well-known Guth and Gold model,50 the

modulus of the reinforced material, E is given as a function of the

unfilled matrix modulus, E0 and the volume fraction of filler, f by:

E ¼ E0(1 + 2.5f + 14.1f2) (4)

This model, typically used for f <0.2, implies an incompress-

ible matrix (Poisson’s ratio n ¼ 0.5), hard spherical particles and

no interaction between the fillers and the matrix. The particle

interface is considered to be inert. As shown in Fig. 7, the

addition of silica nanoparticles leads to a high increase in the

initial modulus, much more significantly than Guth and Gold’s

prediction (represented for comparison). This reinforcement

clearly points out that the introduction of silica does not only

lead to a so-called ‘‘hydrodynamic’’ reinforcement, but obvi-

ously, interactions between PDMA and silica surface should be

taken into account.

This journal is ª The Royal Society of Chemistry 2010

In principle the average molar mass between crosslinks can be

derived from experimental swelling and elastic modulus experi-

ments. Although, due to the heterogeneity of hybrid networks,

this calculation of the average molar mass value between cross-

links remains only qualitative, it remains useful for comparative

purposes. Assuming an infinite average functionality, both affine

and phantom models converge giving rise to the following

equations:

Mc;Q ¼V1

vspe

"�Q

�2=30 Q�1=3

e

Ln�1�Q�1

e

�þQ�1

e þ c12Q�2e

#(5)

Mc;E ¼3RT

vspeQ0Ei

(6)

where V1 ¼ 18 mL g�1 is the molar volume of water, vspe ¼ 0.95

mL g�1 the specific volume of PDMA, c12 ¼ 0.48 the Flory

interaction parameter for PDMA in water,51 Q0 the swelling ratio

of the hydrogels in their preparation state, Qe their equilibrium

swelling, Ei the initial modulus, R the ideal gas constant and T

the absolute temperature.

The molar masses obtained from swelling experiments and

tensile tests are compared in Table 2.

As expected from the previous relations, the molar mass Mc

decreases with increasing modulus or decreasing swelling at

equilibrium. The agreement between Mc values obtained from

tensile tests and swelling can be considered as rather good

Soft Matter, 2010, 6, 3619–3631 | 3625

Page 8: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Fig. 7 Effect of silica volume fraction on initial modulus ( _3 � 0.06 s�1).

Guth and Gold’s model is represented as dash line with E0 ¼ 1.05 kPa,

obtained from rheology.

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

(Mc,E/Mc,Q y 1 � 1.5) for most of the samples except those

which are physically crosslinked with a large amount of silica

(Mc,E/Mc,Q < 0.5). As previously mentioned, the main difficulty

for the calculation of Mc arises from the heterogeneity of hybrid

networks. The discrepancy observed for SP2_PW0.14 and

SP5_PW0.14 demonstrates that the introduction of large

amounts of silica particles, which effectively behave as physical

crosslinkers, has a stronger impact on mechanical properties

than on the swelling equilibrium. The close correlation between

mechanical and swelling properties, obtained for hybrid

networks additionally crosslinked with MBA, could be explained

by a better homogeneity of network with chemical crosslinker.

In Fig. 8a, the initial modulus, Ei versus Q0 for two series of

hydrogels (SP2 and SP5) shows the same scaling behaviour :

E � Q�2.60 . This result is in good agreement with the theoretical

prediction of Obukhov et al. (E � Q�2.250 ) for lightly crosslinked

hydrogels in good solvent conditions, where modulus and

swelling are dominated by trapped entanglements.52

The modulus obtained for the SP5 series are systematically

higher than those of SP2 gels at the same Q0 reflecting a higher

content of silica particles and higher degree of physical cross-

linking. Interestingly we can show in Fig. 8b that a master curve

can be obtained by plotting the normalized modulus [(Ei/(Q0)�2.6]

versus the weight ratio silica/PDAM (X). This plot, which gathers

all the results obtained with hybrid hydrogels (without MBA),

clearly shows the linear dependence of the reduced modulus with

the concentration of silica particles. Taking into account entan-

glements and physical crosslinks, the Young modulus can be

written as:

Table 2 Comparison between Mc values calculated from (a) initial modulusperformed in NaCl 0.5 M (index Q)

Hydrogel SPX_PWY-RZ Ei/kPa Mc,E/kg mol�1

SP0_PW0.14 1.05 705SP0.5_PW0.14 1.7 540SP1_PW0.14 2.9 317SP2_PW0.14 6.5 143SP5_PW0.14 22 42SP2_PW0.14-R0.1 15 62SP2_PW0.14-R0.2 21 44SP2_PW0.14-R0.5 40 23

3626 | Soft Matter, 2010, 6, 3619–3631

E � XQ�2.60 � X.Y2.6 (7)

According to this relation we can see that for a given silica/

polymer ratio, X, the modulus only depends on trapped entan-

glements (E � Q�2.250 � Y2.25), while for a given concentration of

trapped entanglements (Q0 y Y�1), the modulus will scale with

the number of physical crosslinkers and X is the right variable in

that case. This result differs from classical theory of chemically

crosslinked entangled networks where the contribution of

chemical crosslinks and entanglements are additive as:53

E � CRT

�1

Mc

þ 1

Me

where Mc and Me are the average molar mass strands between

crosslinks and entanglements, respectively, and C is the polymer

concentration.

Contrary to model networks, where physical and chemical

crosslinks are expected to be randomly distributed, it is not the

case with hybrid gels where the silica nanoparticles could intro-

duce a high level of heterogeneity in the distribution of cross-

links. From the results plotted in Fig. 8b we can infer that the

elastic modulus of hybrid networks is originally controlled by the

level of entanglements between PDMA chains and that silica

nanoparticles, which strongly interact with the polymer matrix,

dramatically increase or amplify the connectivity of the network.

A schematic picture is given in Fig. 9, where the adsorption of

PDMA chains on silica surfaces create domains of high polymer

density which are connected to each other through the entangled

PDMA matrix.

The important stiffening parameter in hybrid networks is not

the absolute concentration of silica particles but the weight ratio

silica/polymer that is proportional to the fraction of PDMA

which effectively interacts with silica surfaces. This result can be

also closely compared with the data obtained by Haraguchi and

coworkers on NC PDMA and PNIPA gels.27,30 In the case of NC

PDMA hydrogels the modulus and the ultimate tensile strength

was shown to increase almost proportionally to the clay content,

for a given polymer concentration. From a more detailed study,

performed by neutron scattering and mechanical testing on NC

PNIPA gels, they conclude that the number of PNIPA chains

bridging clay platelets per unit volume (Nchain), increases with the

concentration of inorganic particles as: Nchain � [clay]4/3. In that

case, the additional [clay]1/3 dependence of Nchain is ascribed to

the fact that crosslinks are not dispersed in a three-dimensional

space in the gel but are localized on a two-dimensional space

(planar crosslinking), i.e., on the surface of clay platelets.

Ei obtained at standard strain rate (index E) and (b) swelling experiments

Qe Mc,Q/kg mol�1 Mc,E/Mc,Q

122 605 1.293 370 1.584 306 1.182 296 0.555 141 0.337 65 1.027 34 1.317 13 1.7

This journal is ª The Royal Society of Chemistry 2010

Page 9: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Fig. 8 (a) Variation of the Young modulus versus the initial swelling of hybrid hydrogels SP2 and SP5. (b) Master curve of the normalized Young

modulus [(Ei/Q0)�2.6] versus the weight ratio silica/polymer (X).

Fig. 9 Schematic representation of the structure of hybrid hydrogels

with silica nanoparticles covered by PDMA layer embedded into an

entangled PDMA matrix with additional chemical crosslinks.

Fig. 10 Reduced stress plotted as a function of reciprocal l for PW0.14

hydrogels.

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

While many publications note that the ‘‘mechanical proper-

ties’’ can be greatly improved by a proper design of the gel

network, most publications simply report the elastic modulus

and the nominal strain at break. The non-linearities in the

behaviour (i.e. non-linear viscoelasticity, hysteretic behaviour,

residual strain) are rarely investigated but can provide mean-

ingful understanding on micro-mechanisms of deformation or on

damage processes. For weakly crosslinked systems, the nonlinear

behaviour in uniaxial extension is very sensitive to subtle changes

in structure and can be used to gain insight on the organization of

the gel.54 It is the purpose of this section to investigate in more

detail the large strain behaviour of the hybrid hydrogels.

The stress-strain curves shown in Fig. 6 are markedly non

linear and significantly deviate from the classical rubber elasticity

prediction. These deviations from the rubber elasticity model are

generally represented in a Mooney diagram that considers the

variation of the reduced stress versus the reciprocal elongation, l

� 1.55,56 In this representation the reduced stress (s*) is given by:

s* ¼ F/A0(l � 1/l2) (8)

with F the force, A0 the cross-sectional area of the undeformed

sample (at Q0).

The advantage of this type of representation for soft non linear

materials is that the softening or hardening relative to the

This journal is ª The Royal Society of Chemistry 2010

Gaussian elasticity prediction appears simply as a downwards or

an upwards slope as a function of 1/l. As shown in Fig. 10,

a softening is observed for all our samples but the amplitude of

the softening increases with the fraction of silica particles. From

a molecular point of view, this softening behaviour can be caused

by different mechanisms: the reversible orientation of entangle-

ments in the tensile direction,57 the viscoelastic relaxation of the

non trapped entanglements or the desorption of polymer chains

from the silica particles.

At high extension (low 1/l) the upturn emphasizes a hardening

behaviour, usually attributed in rubbers either to finite chain

extensibility or to other processes such as strain-induced crys-

tallisation commonly observed in natural rubber.58,59 Looking

at the minimum of the curves, there is a clear correlation between

the amount of added chemical crosslinker (R ¼ 0.1 mol%) and

the reduction of chain extensibility, shorter values of Mc leading

to lower chain extensibility, since lmaxf M1/2c , see Table 2.

If we compare the influence of added silica for the self-cross-

linked samples, the minimum in the reduced stress curves occurs

Soft Matter, 2010, 6, 3619–3631 | 3627

Page 10: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

for higher values of l for higher amounts of added particles

which it is quite surprising if we consider that the interactions

between silica and PDMA are responsible for the formation of

‘‘transient crosslinks’’ which should decrease the extensibility of

the chains. At very high strains, the upturn appears sharper with

increasing amounts of added particles. The silica particles

essentially add a significant concentration of elastically active

chains with a relatively narrow distribution of distances between

particles.

Fig. 12 First loading-unloading cycle for SP0.5_PW0.14, SP1_PW0.14,

Dissipative behaviour: strain rate effect and hysteresis

Qualitatively, the impact of the strain rate on the mechanical

behaviour is shown in Fig. 11, with a significant increase of the

strain rate dependence with increasing amounts of added parti-

cles. This suggests that the fraction of the softening due to

viscoelastic relaxation processes is greatly increased as the

particles are added. Furthermore the characteristic relaxation

times introduced by the presence of particles are obviously in the

range of the applied strain rates.

Three loading–unloading cycles have been repeated with two

levels of maximal strains investigated, respectively 3max¼ 0.5 and

3max ¼ 5. The nominal strain rate has been mainly kept constant

during the test (i.e. _3� 0.06 s�1). After a loading–unloading cycle,

the dissipated energy has been calculated integrating the area

under the loop. The dissipated energy, Ed calculated for each first

cycle has been normalized by Ea, the applied strain energy

(i.e. the area under the loading curve). The instantaneous

residual deformation, 3res has been determined as soon as the

sample has been unloaded. Moreover, in order to look at the

possible damage in the structure during the cycle, the initial

moduli have been determined for the different repeated loading.

Fig. 12 compares the first cycle for 3max ¼ 5.

These relative values, plotted in Fig. 13, clearly indicate that

dissipative mechanisms increase with increasing amounts of silica

nanoparticles. The normalized dissipated energy is seen to

depend on the silica volume fraction, independently of the strain

level, both at intermediate strains (below 50%) and at large

Fig. 11 Stress–strain curves for SP1_PW0.14 (sparse marker, B), and

SP5_PW0.14 (sparse marker, ,) at nominal _3 ¼ 0.06 s�1 (solid line) and

nominal _3 ¼ 0.6 s�1 (broken line).

3628 | Soft Matter, 2010, 6, 3619–3631

strains (up to 500%). Interestingly, the extrapolation of the

results at f/0, seems in contradiction with the unfilled hydrogel

SP0_PW0.14 formulation which is very soft and too sticky to be

handled for tensile tests. Addition of chemical crosslinks leads to

a significant decrease in relative disspated mechanisms.

We can see that the addition of particles also implies a higher

part of irreversible processes in mechanisms of deformation,

increasing the instantaneous residual strain, 3res. A similar

behaviour has been reported by Haraguchi et al.60 with NC

PNIPA gels prepared with a high clay content. Nevertheless, the

process here is seen to be very fast since the areas obtained

remain very low regarding the range of silica concentrations

investigated (silica/PDMA ¼ X ¼ 0.5 to 5) compared to weight

ratios clay/PNIPA ¼ 0.3 to 1.5.60

The observed increase in residual strain due to the presence of

silica is consistent with the hypothesis that PDMA and silica

nanoparticles form non-permanent and reversible interactions.

SP2_PW0.14, SP5_PW0.14, respectively for 3max ¼ 5 at nominal

_3 ¼ 0.06s�1.

Fig. 13 Normalized dissipated strain energy and residual strain for the

first cycle for two maximal applied strain 50% and 500%.

This journal is ª The Royal Society of Chemistry 2010

Page 11: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

By comparison, both dissipated energy and residual strain

decrease with the introduction of a small proportion of chemical

crosslinker (comparing SP2_PW0.14 vs. SP2_PW0.14-R0.1).

This implies that the dissipative mechanisms are not only due to

the presence of particles, as an analogy with Mullins effect, but

they should be also due to the ‘bulk’ material properties and to its

very low crosslinking density.

Moreover, looking at the second and third applied cycles,

attention has been paid on the evolution of the initial moduli

during consecutive loadings, respectively noted E2ndi and E3rd

i . By

comparison with NC gels prepared with anisotropic clay plate-

lets, the effect of residual orientation that has been reported60 is

not expected in the case of silica nano-particles.

The loss in initial stiffness can be simply interpreted as a global

estimate of irreversible processes. A damage parameter, Dnth is

defined as follows:

Dnth ¼ 1� Enth

i

E1st

i

(9)

with Enth

i the initial modulus during the nth loading cycle and E1st

i

the initial modulus during the first loading cycle.

This parameter has been calculated for cycles at intermediate

strains (below 50%) and for large strains cycles (up to 500%). As

shown in Table 3 for intermediate strains, there is no significant

Table 3 Damage parameter for the second and third cycles at 50% and500% deformation

Sample

3max ¼ 0.5 3max ¼ 5

D2nd D3rd D2nd D3rd

SP0.5_PW0.14 0.07 � 0.01 0.07 � 0.01 �0.2 � 0.19 0.04 � 0.08SP1_PW0.14 0.01 � 0.03 0.03 � 0.02 0.27 � 0.03 0.24 � 0.09SP2_PW0.14 0.02 � 0.02 0.02 � 0.02 0.40 � 0.01 0.37 � 0.00SP5_PW0.14 0.06 � 0.02 0.05 � 0.02 0.58 � 0.05 0.66 � 0.07SP2_PW0.14-R0.1 0.07 � 0.00 0.07 � 0.00 – –

Fig. 14 Stress relaxation behavio

This journal is ª The Royal Society of Chemistry 2010

change in the initial stiffness with a quantitative recovery inde-

pendent of the particle content.

At large strains, damage appears more markedly with

increasing amount of silica particles. No significant increase in

damage has been observed between the second cycle and the

third one, since D2nd are nearly equivalent to D3rd.

Long time behaviour: stress relaxation. The stress relaxation

behaviour is plotted in Fig. 14, using the normalized stress

s(t)/smax, with smax equal to the maximum stress value. The time

reference is taken to be equal to zero when the imposed strain

level, 30 ¼ 0.5 is attained (i.e. for s ¼ smax).

For all the samples, a viscoelastic effect is pointed out, since

stress relaxation is observed as a function of time. For a suffi-

ciently long time, above 3 min, the stress is stabilized at a given

level and it is noticeable that the amount of silica strongly

influences the relaxation process, clearly showing that the life-

time of silica–PDMA interactions is a key-aspect of their

properties.

Moreover, as it has been shown by Adler and coworkers,32,61

the relaxation amplitude is also dependant on inherent properties

of polymer chains. In our case, the crosslinker addition,

comparing SP2_PW0.14 vs. SP2_PW0.14-R0.1 relaxation stress

behaviours, leads to a reduction of the viscoeslastic component.

As discussed by Haraguchi et al.,27–29,33,60,62 the exceptional

properties of nanocomposite hydrogels mainly originate from

the very large size of elastically active chains which are assumed

to be connected through inorganic particles. The situation of

silica/PDMA hybrid networks is very close even if we assume

some additional chemical crosslinks issued from the synthesis.

Taking into account the strong adsorption behaviour of PDMA

on silica surfaces, we can schematically picture the structure of

hybrid hydrogels as spherical particles covered by an adsorbed

polymer layer; the connectivity of the network being provided by

polymer chains bridging particles through an entangled polymer

matrix (Fig. 9).

ur of hybrid gels for 30 ¼ 0.5.

Soft Matter, 2010, 6, 3619–3631 | 3629

Page 12: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

Conclusions

In this paper, we have developed a simple synthetic strategy to

obtain highly extensible and yet mostly elastic nano-composite

‘self-crosslinked’ hydrogels. Taking advantage from side reac-

tions during homopolymerization of DMA in water with clas-

sical monomer/initiators, the ‘self-crosslinked’ gels reported here

are made from relatively high amounts of silica nano-particles

(silica/polymer weight ratio from 0.5 to 5, i.e. volume fractions

from 0.03 to 0.21).

The improvement in properties relies on the synergy between

a lightly swollen crosslinked polymer matrix and strongly inter-

acting silica nanoparticles creating a hydrogel with combination

of physical and chemical crosslinks. Several important conclu-

sions can be drawn from our results.

The samples obtained from the polymerization of DMA

without any added crosslinker were lightly crosslinked and had

a solid-like character above 5 wt% polymer in water as shown by

rheological measurements. This behaviour, unreported so far, is

consistent with the existence of transfer reaction to the monomer

and justifies the use of the term ‘‘self-crosslinked’’ to refer to the

gels. At those polymer concentrations, the self-crosslinked gels

made from PDMA only were however too soft and sticky to be

mechanically tested as self-standing films. The addition of a small

amount of silica particles led to a dramatic change in macro-

scopic mechanical properties and in particular to a great increase

in the elastic character. This improvement of mechanical prop-

erties did not result in any loss in transparency.

From mechanical tests performed with gels in the preparation

state, experiments showed that small strain modulus is controlled

by the product of the density of entanglements by the weight

ratio of silica nanoparticles, a remarkable result suggesting that

each silica particle contributes an additional density of elastic

chains which is proportional to the existing entanglement density

of the polymer in water.

The strong adsorption of the PDMA polymer on each particle

led up to a factor of 20 in elastic modulus increase while retaining

an extensibility of 1000%, a remarkable result. Instantaneous

recovery in strain is seen to be quite high for loading/unloading

cycles at 0.5 strain, inferior to 0.08 strain for a silica volume

fraction of 0.10. However, after higher strain levels (loading–

unloading cycle at 5 in strain) beyond about 0.05 volume fraction

of silica, the gel starts to show a significant level of hysteresis and

a non negligible residual deformation.

Tensile, hysteresis and relaxation experiments all point to the

following physical picture: at a few percent of silica volume

fraction, the viscoelastic poorly crosslinked matrix become

physically crosslinked by the particles and behaves as an elastic

gel with very high extensibility and a modulus of the order of

2–3 kPa. When additional silica is added the resulting interac-

tions between silica particles become comparatively more

important, resulting in a modulus increasing up to 20–40 kPa at

20–25 vol% silica but the gel demonstrates also an increasingly

significant strain rate dependence, hysteresis and residual

deformation.

In summary our results demonstrate that hybrid hydrogels,

with mechanical properties similar to the clay-filled hydrogels,

can be obtained with silica particles provided that the polymer

strongly adsorbs on the surface of the particles. These results are

3630 | Soft Matter, 2010, 6, 3619–3631

very promising and further optimization of the mechanical

properties can most likely be obtained by tuning the ratio of

chemical vs. physical crosslinks as well as the level of particle–

matrix interaction.

Acknowledgements

The authors would like to acknowledge Wei Fan, Freddy

Martin, Ludovic Olanier, Dr Wei-Chun Lin, Dr Tetsuharu

Narita and Dr Guylaine Ducouret for their help and advice in

synthesis and mechanical characterization of hydrogels. We

would like to specifically thank Prof Costantino Creton for many

helpful discussions.

References

1 F. L. Buchholz and A. T. Graham, Modern Superabsorbent PolymerTechnology, Wiley-VCH, 1997.

2 W. Oppermann, B. Lindemann and B. Vogerl, in Polymer Gels:Fundamentals and Applications, ed. H. B. Bohidar, P. Dubin andY. Osada, ACS Symposium Series 833, Oxford University Press,2003, vol. 833, pp. 37–50.

3 N. A. Peppas, Hydrogels in Medicine and Pharmacy, CRC Press, BocaRaton, FL, 1986.

4 O. Wichterle, Soft Contact Lenses, Wiley, New York, 1978.5 M. Haider, Z. Megeed and H. Ghandehari, J. Controlled Release,

2004, 95, 1–26.6 B. Jeong and A. Gutowska, Trends Biotechnol., 2002, 20, 305–360.7 J. Kopecek, Eur. J. Pharm. Sci., 2003, 20, 1–16.8 R. Langer and D. A. Tirrell, Nature, 2004, 428, 487–492.9 Z. Megeed, J. Cappello and H. Ghandehari, Adv. Drug Delivery Rev.,

2002, 54, 1075–1091.10 S. Varghese and J. Elisseeff, Adv. Polym. Sci., 2006, 203, 95–144.11 T. Ikeda, M. Nakano, Y. L. Yu, O. Tsutsumi and A. Kanazawa, Adv.

Mater., 2003, 15, 201–210.12 I. C. Kwon, Y. H. Bae and S. W. Kim, Nature, 1991, 354, 291–293.13 Y. Osada, H. Okuzaki and H. Hori, Nature, 1992, 355, 242–244.14 M. Shibayama, F. Ikkai, S. Inamoto, S. Nomura and C. C. Han,

J. Chem. Phys., 1996, 105, 4358–4366.15 A. Suzuki and T. Tanaka, Nature, 1990, 346, 345–347.16 D. Szabo, G. Szeghy and M. Zrinyi, Macromolecules, 1998, 31, 6541–

6548.17 T. Tanaka, Phys. Rev. Lett., 1978, 40, 820–823.18 R. Yoshida, K. Uchida, Y. Kaneko, K. Sakai, A. Kikuchi, Y. Sakurai

and T. Okano, Nature, 1995, 374, 240–242.19 A. S. Huffman, A. Afrassiabi and L. C. Dong, J. Controlled Release,

1986, 4, 213–222.20 Y. Okumura and K. Ito, Adv. Mater., 2001, 13, 485.21 C. M. Zhao, Y. Domon, Y. Okumura, S. Okabe, M. Shibayama and

K. Ito, J. Phys.: Condens. Matter, 2005, 17, S2841–S2846.22 T. Matsunaga, T. Sakai, Y. Akagi, U. I. Chung and M. Shibayama,

Macromolecules, 2009, 42, 6245–6252.23 J. P. Gong, Y. Katsuyama, T. Kurokawa and Y. Osada, Adv. Mater.,

2003, 15, 1155.24 Y. Tanaka, R. Kuwabara, Y. H. Na, T. Kurokawa, J. P. Gong and

Y. Osada, J. Phys. Chem. B, 2005, 109, 11559–11562.25 R. E. Webber, C. Creton, H. R. Brown and J. P. Gong,

Macromolecules, 2007, 40, 2919–2927.26 K. Haraguchi, Curr. Opin. Solid State Mater. Sci., 2007, 11, 47–54.27 K. Haraguchi, R. Farnworth, A. Ohbayashi and T. Takehisa,

Macromolecules, 2003, 36, 5732–5741.28 K. Haraguchi and T. Takehisa, Adv. Mater., 2002, 14, 1120–1124.29 K. Haraguchi, T. Takehisa and S. Fan, Macromolecules, 2002, 35,

10162–10171.30 S. Miyazaki, T. Karino, H. Endo, K. Haraguchi and M. Shibayama,

Macromolecules, 2006, 39, 8112–8120.31 M. Shibayama, T. Karino, S. Miyazaki, S. Okabe, T. Takehisa and

K. Haraguchi, Macromolecules, 2005, 38, 10772–10781.32 M. F. Zhu, Y. Liu, B. Sun, W. Zhang, X. L. Liu, H. Yu, Y. Zhang,

D. Kuckling and H. J. P. Adler, Macromol. Rapid Commun., 2006,27, 1023–1028.

This journal is ª The Royal Society of Chemistry 2010

Page 13: Nano-hybrid self-crosslinked PDMA/silica hydrogels

Publ

ishe

d on

07

June

201

0. D

ownl

oade

d by

Por

tland

Sta

te U

nive

rsity

on

06/1

0/20

13 1

8:46

:08.

View Article Online

33 K. Haraguchi, H. J. Li, K. Matsuda, T. Takehisa and E. Elliott,Macromolecules, 2005, 38, 3482–3490.

34 L. Petit, L. Bouteiller, A. Brulet, F. Lafuma and D. Hourdet,Langmuir, 2007, 23, 147–158.

35 D. Portehault, L. Petit, N. Pantoustier, G. Ducouret, F. Lafuma andD. Hourdet, Colloids Surf., A, 2006, 278, 26–32.

36 W.-C. Lin, W. Fan, A. Marcellan, D. Hourdet and C. Creton,Macromolecules, 2010, 43, 2554–2563.

37 X. Y. Xie and T. E. HogenEsch, Macromolecules, 1996, 29, 1734–1745.

38 L. Guillaumont, G. Bokias and I. Iliopoulos, Macromol. Chem. Phys.,2000, 201, 251–260.

39 S. L. Cram, H. R. Brown, G. M. Spinks, D. Hourdet and C. Creton,Macromolecules, 2005, 38, 2981–2989.

40 H. L. Needles and R. E. Whitfield, J. Org. Chem., 1964, 29, 3632.41 J. Gao and B. J. Frisken, Langmuir, 2003, 19, 5212–5216.42 H. H. Winters and M. Mours, Adv. Polym. Sci., 1997, 134, 165.43 A. R. Payne, J. Appl. Polym. Sci., 1962, 6, 57–63.44 G. Kraus, J. Appl. Polym. Sci., 1963, 7, 861–871.45 J. Rehner, Reinforcement of Elastomers Interscience, New York, 1965.46 S. S. Sternstein and A. J. Zhu, Macromolecules, 2002, 35, 7262–7273.47 G. Huber and T. A. Vilgis, Macromolecules, 2002, 35, 9204–9210.48 J. Berriot, H. Montes, F. Lequeux, D. Long and P. Sotta,

Macromolecules, 2002, 35, 9756–9762.

This journal is ª The Royal Society of Chemistry 2010

49 J. Berriot, F. Lequeux, H. Montes and H. Pernot, Polymer, 2002, 43,6131–6138.

50 E. Guth, J. Appl. Phys., 1945, 16, 20–25.51 N. Gundogan, O. Okay and W. Oppermann, Macromol. Chem. Phys.,

2004, 205, 814–823.52 S. P. Obukhov, M. Rubinstein and R. H. Colby, Macromolecules,

1994, 27, 3191–3198.53 S. F. Edwards, Proc. Phys. Soc., 1967, 92, 9.54 F. Deplace, C. Carelli, S. Mariot, H. Retsos, A. Chateauminois,

K. Ouzineb and C. Creton, J. Adhes., 2009, 85, 18–54.55 M. Mooney, J. Appl. Phys., 1940, 11, 582.56 R. S. Rivlin, Philos. Trans. R. Soc. London, Ser. A, 1948, 241, 379–

397.57 M. Rubinstein and S. Panyukov, Macromolecules, 2002, 35, 6670–

6686.58 J. E. Mark, Rubber Chemestry and Technology, 1975, 48, 495–512.59 L. R. G. Treloar, in The physics of rubber elasticity, Clarendon Press,

Oxford, 1975, pp. 210–310.60 K. Haraguchi and H. J. Li, Macromolecules, 2006, 39, 1898–1905.61 Y. Liu, M. F. Zhu, X. L. Liu, Y. M. J. Ang, Y. Ma, Z. Y. Qin,

D. Kuckling and H. J. P. Adler, Macromolecular Symposia, 2007,353–360.

62 K. Haraguchi, H.-J. Li and L. Song, J. Colloid Interface Sci., 2008,326, 41–50.

Soft Matter, 2010, 6, 3619–3631 | 3631


Recommended