Synergistic toughening of nanocomposite double network hydrogels by physical adsorption and chemical bonding of polymer chains to inorganic nanospheres and nanorods: a comparative study

Abstract

Previously, we have reported nanocomposite double network (ncDN) hydrogels by introducing bare or reactive inorganic nanospheres or nanorods into double network hydrogels. The obtained ncDN gels showed very high compression strength and toughness. However, the toughening mechanisms remains yet to explore. In this work, a comparative study is presented to provide detailed investigations on the polymer–nanoparticle interactions for ncDN gels with bare or vinyl-grafted nanoparticles. First, the effects of physical adsorption and/or chemical bonding of polymer chains to nanoparticles on the mechanical properties of the parent single network hydrogels of poly(2-acrylamido-2-methyl-propane sulfonic acid) (PAMPS) and polyacrylamide (PAAm) are compared. The nanoparticles showed significant toughening to PAAm gel than to PAMPS gel, due to the strong adsorption of PAAm to nanoparticles. Second, by using PAMPS–nanoparticle hydrogel as a host for in situ polymerization of AAm monomers, the obtained ncDN gels showed outstanding compression strength and toughness, with vinyl-grafted nanoparticles toughening more than bare nanoparticles. Detailed comparative analysis on the initial and ultimate modulus of the ncDN gels suggests that, after PAMPS network fracturing upon compression at high strains, the strong polymer–nanoparticle adsorption/bonding plays a critical role in the mechanical properties of the gels, with silica nanospheres working more effectively than ATP nanorods. TEM images revealed that, ATP nanorods were fractured upon large strain compression, while the silica nanospheres served as energy dissipation center. This study provides toughening mechanisms of nanospheres and nanorods for nanocomposite double network hydrogels.

---

Introduction

Polymer hydrogels are known as soft and wet materials with highly hydrated three-dimensional networks. Conventional polymer hydrogels are limited in their practical applications due to poor mechanical properties, due to the lack of energy dissipation mechanisms of the highly swollen network. Recently, numerous strategies have been demonstrated to develop hydrogels with outstanding strength and toughness by introducing efficient energy dissipation mechanisms. For example, double network hydrogels,^1,2 nanocomposite hydrogels,^3,4 hybrid hydrogels,^5,6 and micelle crosslinked hydrogels^7,8 etc., show extraordinary strength, stretchability, or fatigue resistance. Therein, sacrificial bonds are utilized to dissipate energy through fracturing upon loading. Some sacrificial bonds are reversible and thus could entitle a recovery of the fractured network under proper conditions.

Nanoparticles have been widely used to reinforce polymer hydrogels. Different from the compositing effects for polymer/inorganic composites with weak polymer–filler interactions, the nanoparticles used usually show strong interactions with the hydrophilic polymer chains of hydrogels. As a result, both the strength and toughness are enhanced for most nanoparticle/polymer composite hydrogels. For example, silica nanoparticles show very strong reversible adhesion with poly(dimethylacrylamide) (PDMA) chains,⁹ and silica suspension could be used as ‘glue’ to firmly adhere two pieces of PDMA hydrogels. Silica nanoparticles have also been introduced into chemically crosslinked PDMA hydrogels to form dual crosslinked hybrid hydrogels.^10,11 The modulus, strength, and flexibility of these hydrogels increased with nanoparticle content. It is suggested that transient crosslinked networks formed in these hydrogels through silica nanoparticles/PDMA junctions, which could release accumulated mechanical force within polymer chains through detachment under large strains. Yang et al.^12,13 introduced cellulose nanocrystals into chemically crosslinked hydrophilic polymer networks to form nanocomposite hydrogels. The strength, toughness and effective network chain density increased with nanoparticles content, and effective hysteresis loops were observed in tensile loading–unloading cycles. Therein, nanoparticles were observed to bridge cracks during crack propagation. These results suggest that the strong physical adsorption between polymer chains to nanoparticles plays critical roles in enhancing both the strength and toughness of hydrogels.

In previous studies, we have demonstrated the use of inorganic nanoparticles, e.g., silica nanospheres¹⁴ and clay nanorods,¹⁵ to reinforce double network (DN) hydrogels. DN gels comprised of two (semi-) interpenetrating contrasting polymer networks exploit the rigid network as sacrificial bonds to dissipate energy upon loading, while the ductile network acts as hidden length to sustain large deformation.¹⁶ Thus, double network hydrogels show high strength and toughness. We used vinyl-grafted nanoparticles to copolymerize with the rigid network, which was subsequently interpenetrated with loosely cross-linked ductile network to form nanocomposite double network (ncDN) hydrogels. Compared to DN gels, these ncDN gels showed enhanced compressive strength, modulus, and toughness.^14,15 With the polymer chains bonded to nanoparticles, embedded micro-network porous structures are observed for ncDN gels, presumably due to the microphase separation between the nanoparticles and the polymer network.¹⁴ This phenomenological explanation, however, did not consider the polymer–chain interactions. Upon cyclic loading–unloading tests, the ncDN gels show remarkable hysteresis during the second and subsequent cycles, indicating residual energy dissipation after the fracturing of the rigid network. This energy dissipation decays with more cycles but could be recovered at elevated temperatures. Interestingly, this residual energy dissipation is observed for both bare and vinyl-grafted nanoparticle composited gels, and increases monotonically with the nanoparticle content.¹⁵ Both physical adsorption and chemical bonding are attributed to this phenomenon and, further, to the significant reinforcement to the gels. However, these results did not provide details on how these nanoparticles interact with each polymer network. Besides, it is not clear how the geometric difference of nanospheres and nanorods influence the toughening mechanisms of polymer hydrogels.

In order to investigate the toughening mechanisms of nanoparticles to double network hydrogels, this work presents a comparative study on the mechanical behaviors of nanoparticle-composited single and double network hydrogels. Hydrogels composited with bare or modified nanoparticles showed remarkable differences in compression stress–strain curves, as well as the modulus and fracture properties. Besides, the moduli of hydrogels composited with silica nanospheres and clay nanorods at high strains are compared in order to study the polymer–nanoparticle interactions after the fracture of the rigid PAMPS/nanoparticle composite network. TEM images provide direct evidence to the different behaviors of nanospheres and nanorods under load, suggesting a critical role played by nanoparticle geometry on the toughening mechanisms.

Experimental

Materials

Tetraethoxysilane (TEOS), vinyltriethoxysilane (VTEOS), acetic acid (HAc), ammonium hydroxide (NH₄OH, 25 wt% aqueous solution), ethanol, acrylamide (AAm), 2-acrylamido-2-methyl-propane sulfonic acid (AMPS), N,N′-methylene bisacrylamide (MBAA) and potassium persulfate (KPS) were purchased from the Sinopharm Chemical Reagent Co. Ltd., Attapulgite (ATP) was provided by Jiuchuan Nano-Material Technology Co. Ltd. De-ionized water was used for all experiments, with oxygen removed by bubbling nitrogen gas for more than 2 hours before use.

Synthesis of vinyl-grafted silica (G-silica) and bare (B-silica) silica nanospheres

Bare and vinyl modified silica nanoparticles were synthesized by in situ hydrolysis and precipitation of precursors. Briefly, TEOS (8.5 mL) in ethanol (10 mL) was quickly added to a solution of NH₄OH (7 mL), water (23 mL) and ethanol (120 mL) with stirring at room temperature. As the mixture turned turbid, VTEOS was added until the TEOS/VTEOS molar ratio reached 85/15. The solution was dialyzed in a dialysis bag (cut-off molecular weight: 8000–14 000) against water to remove unreacted chemicals. The product, G-silica, was collected by centrifugation and freeze dryed.¹⁷ B-silica were also prepared following this procedure without using VTEOS.

Synthesis of vinyl grafted attapulgite (G-ATP) nanorods

ATP (2 g) was sonicated in an ethanol/water mixture (75 mL/75 mL) for 2 h. Then the VTEOS (5 mL) in water (45 mL) solution was added under vigorous stirring, followed by acidification to pH 3.0 with HAc. The mixture was then transferred into a flask and refluxed at 70 °C for 30 min. Subsequently, the suspension was centrifuged and washed with de-ionized water to pH 7.0, followed by washing with ethanol three times. The obtained G-ATP powders were lyophilized before use.¹⁸

Synthesis of nanocomposite single network (ncSN) hydrogels

The ncSN gels were synthesized by free radical polymerization of mixed solution of monomers (AMPS or AAm), crosslinker (MBAA), and nanoparticles. The feed ratio of MBAA to AMPS was fixed at 4%, and that of MBAA to AAm was 0.01%. The water content of resulted gels was 90 wt%. For example, silica (0.885 g), MBAA (0.265 g), initiator KPS (0.012 g) and AMPS (8.85 g) were mixed in water (90 mL) under stirring. After stirring for 30 min at 25 °C, the solution was transferred into glass moulds for free radical polymerization at 60 °C for 10 h to form hydrogels.

By using bare or grafted nanoparticles for the synthesis, a series of ncSN gels, namely, PAMPS-BS1, PAMPS-GS1, PAMPS-BA1, PAMPS-GA1, PAAm-BS1, and PAAm-BA1 were obtained, with BS for bare silica, GS for G-silica, BA for bare ATP, and GA for grafted ATP. The nanoparticle content was 1 wt% with respect to the monomers. For comparison, single network hydrogels of PAMPS (PAMPS-SN) and PAAm (PAAm-SN), free of nanoparticles, were synthesized by using MBAA as crosslinker.

Synthesis of nanocomposite double network (ncDN) hydrogels

The nanocomposite single network hydrogels of PAMPS were used to host the AAm monomers for in situ free radical synthesis of the second network. Briefly, the nanocomposite PAMPS gel was swollen in a aqueous solution of 3 mol L⁻¹ AAm, 0.01 mol% MBAA, and 0.01 mol% KPS for 24 h. The swollen gel was sandwiched by two glass plates and sealed, followed by polymerization at 60 °C for 10 h. Finally, the gels were immersed in water for 7 days, with water been changed twice a day.

The synthesized ncDN gels were designated as ncDN-BSm, ncDN-GSm, ncDN-BAm and ncDN-GAm gels, where m is the weight percentage (wt%) of nanoparticles to AMPS. On the other hand, the ncDN gels with bare nanoparticles were denoted as ncDN-B gels, the ncDN gels with vinyl grafted nanoparticles as ncDN-G gels, the ncDN gels with silica nanospheres (grafted or not) as ncDN-silica gels, and the ncDN gels with ATP nanorods (grafted or not) as ncDN-ATP gels. For comparison, the PMAPS/PAAm double network (DN) gels with the formulations mentioned above, free of nanoparticles, were prepared.

Scanning electron microscopy (SEM)

Silica or G-silica dispersions were cast on silicon wafers and then dried for imaging by using a Hitachi S4800 scanning electron microscope (Hitachi Inc., Japan) at 15 kV. On the other hand, gel samples were freeze-fractured in liquid nitrogen, freeze-dried and then sputter coated with platinum for SEM imaging.

Transmission electron microscopy (TEM)

The suspensions of bare ATP or G-ATP were cast on carbon supported copper grids, and then dried for TEM imaging by using a Tecnai F20 transmission electron microscope (FEI Inc., Oregon) at 200 kV. On the other hand, lyophilized gels were embedded in epoxy resin for cryo-microtoming at −40 °C into ca. 70 nm sections, which were collected onto copper grids for imaging operating.

Mechanical tests

Cylindrical gel samples (9 mm diameter and 5–7 mm height, n = 5 each) were compression tested by using an Instron 5567 (Instron Inc., MA) instrument equipped with a 3 kN load cell. All the samples were loaded at a crosshead speed of 10% min⁻¹ to up to 98% strain to protect the load cell.

The engineering compression stress (σ) was calculated as σ = F/πR², where F is the loading force and R is the original radius of specimen. The engineering compression strain (ε) was defined as the change in height (h) relative to initial height (h₀) of the specimen, ε = (h₀ − h)/h₀ × 100%. The initial modulus (E_i) was calculated as the slope of stress–strain curve within the range ε = 5–10%. The fracture energy (U), is calculated by integrating the area under the stress–strain curve:

ΔU = ∫σdε (1)

Cyclic compression tests with a maximum strain of 90% were conducted on specimens with the same size and at the same crosshead speed. The dissipated energy for each cycle, ΔU, is defined as the area encompassed by the loading–unloading curve:

Results and discussion

Preparation of nanoparticles modified with vinyl groups

Silica nanoparticles synthesized by in situ precipitation showed well controlled particle size with narrow distributions and smooth surface (Fig. 1a). By controlling the stirring speed and reaction time, silica nanoparticles with an average diameter of 257.9 ± 7.4 nm were used for this study. To graft vinyl groups to silica nanoparticles, VTEOS was used after the pre-hydrolysis of TEOS, leading to nanoparticles with raspberry like surface (Fig. 1b), likely due to the in situ hydrolysis and polymerization of VTEOS at the nanoparticle surface. On the other hand, vinyl groups were grafted onto attapulgite (ATP) nanorods dispersed in aqueous solutions. TEM results showed that bare ATP nanorods were about 600–700 nm long and 20–24 nm in diameter (Fig. 1c). After vinyl modification, the nanorods became hairy, and the average length and diameter were not significantly different from the inact ones (Fig. 1d). The presence of vinyl groups in G-silica and G-ATP were identified through FTIR measurements in previous work.^14,15

Fig. 1 SEM images of (a) bare and (b) vinyl-grafted silica nanospheres, and TEM images of (c) pristine, and (d) vinyl-grafted attapulgite (ATP) nanorods.

Compression properties of nanocomposite single network (ncSN) hydrogels

Previously, nanoparticle-reinforced double network hydrogels showed very high compression strength and toughness.

Therein, the outstanding reinforcing effect was presumably attributed to the physical adsorption and/or covalent bonding of polymer chains to nanoparticles. Herein, we provide evidences to the synergistic contribution of physical adsorption and chemical bonding to nanoparticles to the toughening effect of single network and double network hydrogels. For this purpose, the compression properties of single network (SN) hydrogels of PAMPS and PAAm with the presence of bare or modified nanoparticles have been systematically and comparatively investigated.

Fig. 2 compares representative compression stress–strain curves of these ncSN gels. The PAMPS-SN gel showed a fracture strength (σ_b) of 59.7 ± 9.6 kPa, fracture strain (ε_b) of 31.9 ± 1.5%, initial modulus (E_i) of 180 ± 10.4 kPa, and fracture energy (U) of 6.6 ± 0.9 kJ m⁻³. With the presence of 1 wt% bare silica nanospheres, the σ_b was increased to 87.7 ± 7.1 kPa, ε_b to 35.9 ± 1.2%, E_i to 191.5 ± 12.1 kPa, and U to 10.5 ± 1.8 kJ m⁻³. These results show a simultaneous enhancement in both compression strength and toughness. This behavior is much different from the conventional reinforcement of polymer materials with nanoparticles, where there are usually no or weak polymer–particle interactions. Herein, this toughening effect, although relatively low, may be related to the adsorption of polymer chains to silica nanoparticles. As the silica nanoparticles were modified with vinyl groups, the compression properties were further enhanced. The PAMPS-GS1 gels with 1 wt% G-silica nanoparticles showed a σ_b of 121.2 ± 2.9 kPa, ε_b of 33.9 ± 0.9%, E_i of 277.7 ± 30.2 kPa, and U of 14.1 ± 1.6 kJ m⁻³, which are significantly higher than those for the PAMPS-BS1 hydrogels except that both ε_b values are not significantly different (p > 0.05). The chemical bonding between polymer chains to silica nanoparticles further improved the strength and toughness of the PAMPS-SN gel.

Fig. 2 Representative stress–strain curves of (a) PAMPS-SN, PAMPS-BS1 and PAMPS-GS1 gel, and (b) PAAm-SN and PAAm-BS1 gel. (c) E_i and U of silica composited SN gels. Representative stress–strain curves of (d) PAMPS-SN, PAMPS-BA1 and PAMPS-GA1 gel, and (e) PAAm-SN and PAAm-BA1 gel. (f) E_i and U of ATP composited SN gels.

PAAm chains are well known for their strong adsorption to inorganic particles (e.g., clay¹⁹ or silica nanoparticles^20,21). In order to investigate the effect of PAAm–nanoparticle adsorption to the mechanical properties of SN gels, nanocomposite PAAm gel was synthesized with the presence of 1 wt% bare silica nanoparticles. Fig. 2b compares representative compression stress–strain curves of the PAAm-BS1 gel and PAAm gel. In comparison to the PAMPS gel, the PAAm gel showed higher compression properties (Fig. 2c). The σ_b of PAAm-SN gels was 703.6 ± 32.8 kPa, and ε_b was 80.2 ± 3.3%. With 1 wt% bare silica nanoparticles, the σ_b was largely enhanced to 6.7 ± 0.4 MPa with ε_b of about 90.7 ± 4.1%. As a result, the U was increased from about 108.3 ± 10.2 kJ m⁻³ to about 521.7 ± 30 kJ m⁻³.

As the silica nanoparticles were replaced by attapulgite (ATP) nanorods, similar toughening of PAMPS and PAAm SN gels were also investigated. Bare ATP nanorods slightly improved the compressive strength and toughness of the PAMPS gel, while the gel containing G-ATP nanorods showed a bit higher strength and toughness (Fig. 2d). In contrast, the compression strength and fracture strain of the PAAm-BA1 gel were 8.7 ± 0.7 MPa and 92.1 ± 3.5%, much higher than those for the PAAm gel (Fig. 2e). As a result, the compression toughness of PAAm-BA1 was 583.9 ± 47.2 kJ m⁻³, which is almost 6 times that of PAAm-SN gel (Fig. 2f).

Such a big increase in toughness is attributed to the strong adsorption of PAAm chains to silica/ATP nanoparticles. In contrast to the small difference in the strength and toughness between PAMPS-BS1 (or PAMPS-BA1) and PAMPS-SN gels, the remarkable toughening effect of PAAm gel indicates a stronger physical adsorption of PAAm chains to silica/ATP nanoparticles than those for PAMPS chains. The bare silica nanoparticles and bare ATP nanorods contain a lot of hydroxyl groups on surface, which may form hydrogen bonding to the AAm monomers prior to polymerization. Similar adsorption of monomers to inorganic particles has been presumed in nanocomposite hydrogels by Haraguchi et al.,²² who used synthetic hectorite clay nanosheets to adsorb monomers for synthesis of hydrogels with very high strength and toughness. Besides, the surface modification enables chemical bonding, in addition to physical adsorption, of polymer chains to nanoparticles. This is important, as to be shown and discussed below, for the outstanding toughening of double network hydrogels.

Compression properties of nanocomposite double network (ncDN) hydrogels

The nanocomposite single network PAMPS hydrogels were used to host free radical polymerization of AAm monomers into a second network interpenetrating to the rigid first one, as reported previously,^14,15 resulting in ncDN hydrogels. Previous studies have demonstrated outstanding compression strength and toughness of these hydrogels. Herein, a series of ncDN hydrogels with different nanoparticle contents were synthesized in order to further comparatively study the effect of bare and functional nanoparticles on the mechanical properties of the gels.

Fig. 3 compares representative compression stress–strain curves of ncDN gels with 1 wt% silica nanospheres or ATP nanorods. In comparison to DN gels, the ncDN gels showed much higher compression strength and fracture strain. For example, the ncDN-GA1 and ncND-BA1 gels did not fracture at 98% strain, while the σ_b was 47.5 ± 6 MPa for ncDN-BS1 gel and 73.5 ± 2.6 MPa for ncDN-GS1 gel, which are much higher than 18.6 ± 2.1 MPa for the DN gel. Apparently, the compression toughness is much higher for the ncDN gels. On the other hand, the vinyl grafted nanoparticles showed a higher reinforcement effect on the ncDN gels than the bare ones did.

Fig. 3 Representative compression stress–strain curves of DN, ncDN-GS1, ncDN-BS1, ncDN-GA1 and ncDN-BA1 gels.

The mechanical properties of ncDN hydrogels with different nanoparticle contents are summarized and compared in Fig. 4 and 5. First of all, all the ncDN gels showed significantly higher strength, moduli, and toughness than those for DN gel as control. With high silica nanosphere content, the fracture strain were not necessarily higher than that of DN gel (Fig. 4b). Other than this, the strength, modulus, and toughness of ncDN gels with vinyl-functionalized silica nanospheres are significantly higher than those for ncDN gels with bare silica nanospheres (Fig. 4a, c and d). As m was increased from 0.5 to 1, 2, 3, and 4, the σ_b was varied from 32.2 ± 5.3 to 47.5 ± 6, 47.9 ± 10.2, 46.7 ± 3.8, and 42.8 ± 6.5 MPa for ncDN-BSm gels, and from 59 ± 6.6 to 73.5 ± 2.6, 62.0 ± 5.2, 53.1 ± 5.8, and 46.1 ± 8.4 MPa for ncDN-GSm. Moreover, the U were enhanced from 1.1 ± 0.1 MJ m⁻³ for DN gel to a maximum of 2.9 ± 0.7 MJ m⁻³ for the ncDN-BSm gels, and further to a maximum of 3.9 ± 0.5 MJ m⁻³ for the ncDN-GSm gels. These results suggest that, in addition to physical adsorption, chemical bonding of polymer chains to nanospheres provides further reinforcement to the network.

Fig. 4 Mechanical properties of ncDN-silica gels with various silica nanosphere contents. (a) Fracture strength (σ_b), (b) fracture strain (ε_b), (c) initial modulus (E_i), and (d) fracture energy (U).

Fig. 5 Mechanical properties of ncDN-ATP gels with different ATP contents. (a) Fracture strength (σ_b), (b) fracture strain (ε_b), (c) initial modulus (E_i), and (d) fracture energy (U).

Similar high strength and toughness of the ncDN gels with bare or vinyl-modified ATP nanorods are showed in Fig. 5. With 0.1, 0.5, and 1.0 wt% nanorods (bare or grafted), the gels did not fracture at 98% strain, and the σ_b (or σ_0.98) was higher than 18.6 ± 2.1 MPa for DN gel, showing a maximum of 65.7 ± 3.7 MPa with 1 wt% nanorods (Fig. 5a). The ε_b was increased with the presence of nanorods, but decreased at high ATP contents (1.5 and 2 wt%, Fig. 5b). The E_i was increased monotonically with increasing nanorod content (Fig. 5c). The U of these gels were higher than that of DN gel, showing a maximum of 2.6 ± 0.2 MJ m⁻³ with 1 wt% G-ATP (Fig. 5d). Gels with vinyl-grafted ATP nanorods showed slightly higher strength, modulus, and toughness than those with bare ATP nanorods (p > 0.05 with in nanorod content of 1 wt% or lower). With relatively high bare nanorod content (1.5 and 2 wt%), the strength and toughness was decreased, probably due to the slight precipitation and aggregation of nanorods in the hydrogel matrix. However, the loss mechanical properties became less for gels with vinyl-grafted ATP nanorods.

Previously, we demonstrated the porous network of such ncDN gels by SEM. With increasing nanoparticle contents, the network meshes become denser, and the average pore diameter was decreased. Besides, the gels with vinyl-grafted nanoparticles showed smaller pore size. Measurements on the equilibrium swelling ratio (ESR) of these ncDN-B gels showed a monotonic decrease in ESR with increasing nanoparticle content. These results suggested that the polymer–nanoparticle interactions led to increases in crosslink density. Such a description on the network structures, however, does not explain how the polymer chain–nanoparticle interactions affect the toughness of the gels.

According to Fig. 2, the toughening extent of PAMPS hydrogel by nanoparticles is limited, and much lower than that for the ncDN gels (Fig. 4 and 5). Moreover, the results that most ncDN gels did not fracture at high strains imply additional mechanism to maintain the integrity of the network after the fracture of rigid PAMPS/nanoparticle network. Fig. 2 indicates that the PAAm/nanoparticle interactions are much stronger than those between PAMPS and nanoparticles, presumably due to the hydrogen bonding of amide groups in PAAm with hydroxyl groups on nanoparticles.^21,23

Previous cyclic loading–unloading tests on ncDN-ATP hydrogels at high strain showed a decay of hysteresis energy and recovery after restoration to the level of the second cycle.¹⁵ Since the PAMPS network has been fractured upon the first loading, such a recovery suggests a critical contribution from reversible polymer–nanoparticle adsorption. Interestingly, the ncDN-ATP gels containing vinyl-functionalized nanoparticles showed less recovery ratio than those with bare nanoparticles. The presence of grafted functional groups reduces the sites for physical adsorption.

Herein, the polymer–nanoparticle interactions have been further investigated by cyclic compression loading–unloading tests on ncDN-silica gels and ncDN-ATP gels. Fig. 6 shows two successive loading–unloading curves of ncDN-GS1 (Fig. 6a), ncDN-BS1 (Fig. 6b), ncDN-GA1 (Fig. 6c), and ncDN-BA1 gels (Fig. 6d). All hydrogels show obvious energy dissipation in the first cycle, mainly due to the fracture of PAMPS network.¹⁶ Meanwhile, these hydrogels show smaller energy dissipation in the second run (Fig. 6).

Fig. 6 Representative two successive loading–unloading curves of (a) ncDN-GS1, (b) ncDN-BS1, (c) ncDN-GA1, and (d) ncDN-BA1 gels, with dissipated energy (ΔU), or loop area, labelled in the panels.

The dissipated energies (ΔU) of ncDN-silica and ncDN-ATP gels in the first and second compression cycles were compared in Fig. 7. With the same particle contents, ncDN-silica gels always exhibit bigger ΔU than ncDN-ATP gels, no matter in the first or second loading–unloading cycles (Fig. 7a and b). These results indicate a greater energy dissipation with silica nanospheres than that with ATP nanorods. Futhermore, gels with vinyl-grafted nanoparticles show higher ΔU than those with bare nanoparticles.

Fig. 7 Comparison of the energy dissipation (ΔU) of (a) the first and (b) the second loading cycle of ncDN gels with various formulations.

These results suggest a synergistic effect of covalent bonding and adsorption of polymer chains to nanoparticles on mechanical properties of hydrogels. In ncDN-G gels, the nanoparticles are wrapped by a layer of covalent networks (Fig. 8a and b). The adsorbed PAAm chains are embedded, entangled, and enwrapped in the surface layer bonded to nanoparticles (Fig. 8c and d). Upon large strains, the adsorbed polymer chains detach from nanoparticle surface. However, different from the desorption from bare nanoparticles, the desorption may be hindered by surrounding covalent networks grafted on the nanoparticles (Fig. 8d), which may cause higher energy dissipation than those from bare nanoparticles.

Fig. 8 Schematic illustration to the network structures of (a) ncDN-GSm and (b) ncDN-GAm hydrogels. (c and d) The PAAm chains are adsorbed to nanoparticle surface and intertwined with PAMPS chains bonded to nanoparticle surfaces.

In order to further understand how the PAAm–nanoparticle interactions affect the mechanical behavior of the gels, the moduli at the initial stage and those at high strain (90–95%) were compared, together with the overall compression work at high strain (Fig. 9). The initial modulus (E_i) is defined as the slope of stress–strain curve between ε of 5–10%, while the modulus above 90% strain is defined as the slope of stress–strain curve between ε of 90–95% (E_90–95%). At low strain, the modulus is highly related to the rigidity of the network. The stress is exerted on the rigid PAMPS/nanoparticle composite network. The E_i values monotonically increased with nanoparticle content, with those for gels with functionalized nanoparticles higher than those with bare nanoparticles (Fig. 9a). Interestingly, there are no significant differences in E_i between gels with silica or ATP nanoparticles (Fig. 9a).

Fig. 9 Comparison of the (a) initial modulus (E_i), (b) large strain modulus (E_90–95%), and (c) work of compression at 95% strain (U_95%) of ncDN gels with various formulations.

At high strain (above 90%), the ncDN gels showed increased difference in the capability to resist deformation. The E_90–95% values for hydrogels with grafted nanoparticles are much higher than those with bare nanoparticles, indicating that both PAMPS and PAAm chains may have bonded to the vinyl-functionalized nanoparticles (Fig. 9b). On the other hand, the gels containing silica nanoparticles showed higher modulus than those with ATP nanorods, no matter the nanoparticles are bare or functionalized with vinyl groups. For example, the E_90–95% of ncDN-GS1 is 3.8 times as that of ncDN-GA1. These results suggest that, with high strain, the ncDN-silica gels showed a higher resistance against deformation, and more energy was needed to deform ncDN-silica gels than ncDN-ATP gels.

Moreover, the compression work of these ncDN gels at high strain is compared in Fig. 9c. The work to deform these ncDN gels to 95% strain (U_95%) is defined as the area under the compression stress–strain curve until 95% strain. This value was increased with increasing nanoparticle content, and the vinyl-functionalized nanoparticles rendered higher U_95% values than those with bare nanoparticles. Again, the U_95% values for ncDN gels with silica nanospheres are significantly higher than those with ATP nanorods. With more nanoparticle contents, the gap became larger, indicating that the effect from silica nanospheres was amplified at higher contents.

The comparative study on the mechanical properties of ncDN gels with silica nanospheres and ATP nanorods indicates a possible geometry influence on the toughening effect, in addition to the synergistic polymer–nanoparticle interactions from both physical adsorption and chemical bonding. This issue will be investigated below.

Effect of nanoparticle geometries on mechanical properties of ncDN gels

The above results indicate an effect of nanoparticle geometry on the toughening of the DN gels. In order to investigate the morphologies of nanoparticles in the gels before and after high strain compression, the ncDN gels were embedded in epoxy resin and microtomed for TEM imaging.

Fig. 10 shows representative TEM images of ncDN-BA1 gel and ncDN-BS1 gel before and after compression. For the as-prepared ncDN-BA1 gels, the ATP nanorods embedded in the gel matrix maintained were about 600 nm long (Fig. 10a), which is close to that of pristine ATP nanorods (Fig. 1c). In contrast, the ATP nanorods in ncDN-BA1 gels were fractured into short segments (about 100 nm long, Fig. 10b) after compression tests up to 95% strain. On the other hand, silica nanospheres in ncDN-BS1 gels maintained the original size before and after compression tests (Fig. 10c and d). During compression testing, the silica nanoparticle acted as an energy dissipation center, with the surrounding polymer chains/network are radially stretched (Fig. 10d). These observations explained the outstanding toughening effect of silica nanospheres in comparison to the ATP nanorods.

Fig. 10 TEM micrographs of ncDN-BA1 gels (a) before and (b) after 95% compression (b), and ncDN-BS1 gels (c) before and (d) after 95% compression.

Conclusion

This comparative study demonstrates that covalent bonding and non-covalent adsorption of polymer chains to inorganic nanospheres and nanorods synergistically and effectively toughened double network hydrogels. Both the physical adsorption and covalent bonding of polymer chains to nanoparticles improved the strength and toughness of the parent PAMPS and PAAm single network hydrogels, while the PAAm chains showed much stronger adsorption to nanoparticles and thus provided more toughening effect to single network hydrogels. Moreover, by using the rigid PAMPS/nanoparticle hydrogel to host the polymerization of AAm monomers, the obtained nanocomposite double network (ncDN) hydrogels showed very high strength and toughness, in comparison to the parent DN hydrogel. Detailed comparison of the modulus of these hydrogels at very high strain (90–95%) indicated that the strong adsorption of PAAm network to nanoparticles may account for the outstanding toughening effect. TEM images revealed that silica nanospheres were more efficient in energy dissipation than ATP nanorods, as the latter fractured upon high loadings.

Acknowledgements

We are grateful to financial support from the Natural Science Foundation of China (21574145), the Hundred Talents Program of the Chinese Academy of Sciences (JF), the Zhejiang Natural Science Foundation of China (LR13B040001), and the Ningbo Natural Science Foundation (2015A610025).

Notes and references

1. Q. Chen, L. Zhu, H. Chen, H. Yan, L. Huang, J. Yang and J. Zheng, Adv. Funct. Mater., 2015, 25, 1598–1607.
2. J. P. Gong, Y. Katsuyama, T. Kurokawa and Y. Osada, Adv. Mater., 2003, 15, 1155–1158.
3. K. Haraguchi and T. Takehisa, Adv. Mater., 2002, 14, 1120–1124.
4. T. Wang, S. Zheng, W. Sun, X. Liu, S. Fu and Z. Tong, Soft Matter, 2014, 10, 3506–3512.
5. J.-Y. Sun, X. Zhao, W. R. K. Illeperuma, O. Chaudhuri, K. H. Oh, D. J. Mooney, J. J. Vlassak and Z. Suo, Nature, 2012, 489, 133–136.
6. C. H. Yang, M. X. Wang, H. Haider, J. H. Yang, J.-Y. Sun, Y. M. Chen, J. Zhou and Z. Suo, ACS Appl. Mater. Interfaces, 2013, 5, 10418–10422.
7. Y.-N. Sun, G.-R. Gao, G.-L. Du, Y.-J. Cheng and J. Fu, ACS Macro Lett., 2014, 3, 496–500.
8. Y. Sun, S. Liu, G. Du, G. Gao and J. Fu, Chem. Commun., 2015, 51, 8512–8515.
9. S. Rose, A. Prevoteau, P. Elziere, D. Hourdet, A. Marcellan and L. Leibler, Nature, 2014, 505, 382–385.
10. W.-C. Lin, W. Fan, A. Marcellan, D. Hourdet and C. Creton, Macromolecules, 2010, 43, 2554–2563.
11. S. Rose, A. Dizeux, T. Narita, D. Hourdet and A. Marcellan, Macromolecules, 2013, 46, 4095–4104.
12. J. Yang, C. Han, J. Duan, F. Xu and R. Sun, ACS Appl. Mater. Interfaces, 2013, 5, 3199–3207.
13. J. Yang, C. Han, X. Zhang, F. Xu and R. Sun, Macromolecules, 2014, 47, 4077–4086.
14. Q. Wang, R. X. Hou, Y. J. Cheng and J. Fu, Soft Matter, 2012, 8, 6048–6056.
15. G. Gao, G. Du, Y. Cheng and J. Fu, J. Mater. Chem. B, 2014, 2, 1539–1548.
16. J. P. Gong, Soft Matter, 2010, 6, 2583–2590.
17. M. Marini, B. Pourabbas, F. Pilati and P. Fabbri, Colloids Surf., A, 2008, 317, 473–481.
18. L. H. Wang and J. Sheng, Polymer, 2005, 46, 6243–6249.
19. O. Okay and W. Oppermann, Macromolecules, 2007, 40, 3378–3387.
20. J. Li, Z. Li, H. Chen, L. Yang, H. Zheng, Y. Shang, D. Yu, J. deClaville Christiansen and S. Jiang, J. Mater. Sci., 2016, 51, 3080–3096.
21. J. Yang, C.-R. Han, J.-F. Duan, F. Xu and R.-C. Sun, J. Phys. Chem. C, 2013, 117, 8223–8230.
22. K. Haraguchi, H.-J. Li, K. Matsuda, T. Takehisa and E. Elliott, Macromolecules, 2005, 38, 3482–3490.
23. G. Gao, G. Du, Y. Sun and J. Fu, ACS Appl. Mater. Interfaces, 2015, 7, 5029–5037.

---