Effect of water state and polymer chain motion on the mechanical properties of a bacterial cellulose and polyvinyl alcohol (BC/PVA) hydrogel

Abstract

In this study, a bacterial cellulose and polyvinyl alcohol (BC/PVA) hydrogel was developed. The compressive modulus of the composite hydrogel increased by enhancing the content of BC and/or PVA and increased significantly by increasing the PVA content to 20 wt%. However, the strain point of the initial compressive failure decreased as the BC content increased. Using a simple and novel method that compares the compressive moduli of dehydrated and original samples, the compressive modulus decreased unexpectedly in dehydrated samples with PVA contents exceeding 20 wt%; in dehydrate samples with PVA contents of less than 20 wt%, the results increased. According to differential scanning calorimetry (DSC) results, free water was transformed to freezable bound water within the composite hydrogels when the PVA content was increased to greater than 20 wt%. Thus, freezable bound water has a greater effect on the compressive modulus of the composite hydrogel than free water and plays a significant role in maintaining the flexibility and stiffness of the hydrogel. Differences in the polymer chain motion and elastic modulus between the composite and pure PVA hydrogels were also studied using dynamic mechanical analysis (DMA). The results demonstrated that the molecular chain relaxation and viscosity were lower and that the elastic modulus was higher for BC/PVA than for pure PVA, indicating that the PVA molecular chain of the BC/PVA composite tends to be rigid, possibly contributing to the marked improvement in the mechanical properties.

---

Introduction

Polyvinyl alcohol (PVA) is a promising cartilage replacement biomaterial because of its viscoelastic behaviour, hydrophilicity, chemical stability and biocompatibility.¹ However, one of the primary disadvantages of employing hydrogels in medical applications is their insufficient mechanical strength.^2,3 Bacterial cellulose (BC) which was produced by Acetobacter xylinum has been widely used for numerous biomedical applications.^4,5 Reinforcing a PVA hydrogel with BC (which has unique 3D nano-fiber structures) by mimicking the collagen structure and proteoglycan networks in natural cartilage, would be an ideal choice. Besides, enhancing the PVA content in BC/PVA hydrogel can also increase its mechanical strength. Since PVA solution (20–25 wt%) is difficult to handle because of its high viscosity,⁶ high temperature could be applied to produce PVA hydrogels with PVA concentration up to 40 wt%. In this study, we investigated the combinative effects of introducing BC into PVA hydrogel and increasing PVA concentration on the mechanical properties of PVA.

Many studies have investigated the effect of water on the mechanical properties of cartilage, bone tissue^7–9 and polymers;^10,11 however, a lack of studies on the direct relationship between the water state and the mechanical strength of hydrogels exists. Generally, the state of water in hydrogels is categorised as three types: free water, freezable bound water, and non-freezing water.¹² For hydrophilic polymers, the thermal properties of polymers and water were remarkably influenced by the interaction between water and polymer chain. Free water is not bound to the polymer chain and does not form hydrogen bond. Besides, its melting/crystallization temperature and enthalpy is analogous to bulk water (at 0 °C). Freezable bound water is weakly bound to the polymer chain or to non-freezing water, and its melting/crystallization temperature is lower than 0 °C. Non-freezing water is tightly bound to the polymer chain by hydrogen bond with undetectable phase transition.^13,14 In this paper, we studied the effects of BC and PVA content and the water state effects on the mechanical properties of hydrogels. In addition, we examined the tan δ curve peak and storage modulus differences between BC/PVA and pure PVA hydrogels using dynamic mechanical analysis (DMA).

Results and discussion

Effect of BC content on the compression properties of BC/PVA hydrogels

Fig. 1A exhibits the stress–strain curves from static compression testing. The compressive stress of all samples in this group increased with increasing strain, and initial compressive failure appeared at a strain point in all samples, with the exception of sample B0.5P15. The samples exhibited apparent cracks after compression (Fig. 1C). The strain point at which initial compressive failure occurred decreased from 0.7 to 0.5 mm/mm as the BC content increased, with a compressive strength of approximately 5–6 MPa.

Fig. 1 Effect of BC content on the compression properties of BC/PVA hydrogels. (A) Stress–strain curve; (B) compressive modulus (strain range of 0–0.1 mm/mm) (the contents of each component of the samples are listed in Table SI1†); (C) image of the samples before and after compression.

The compressive moduli of samples B0.5P15, B0.7P15, B1.0P15, and B1.5P15 were 0.716, 0.585, 2.320, and 3.934 MPa, respectively, and these values increased with increasing BC content (Fig. 1B).

Effect of PVA content on the compression properties of BC/PVA hydrogels

The stress–strain curves (Fig. 2A) from static compression testing indicated that the initial compressive failure of sample B0.5P10 occurred at 0.7 mm/mm strain. However, the remaining samples, B0.5P15, B0.5P20, and B0.5P25 exhibited no compressive failure during compression testing (over a strain range of 0.0 to 0.8 mm/mm), and the samples demonstrated no cracking after compression testing (Fig. 2C).

Fig. 2 Effect of PVA content on the compression properties of BC/PVA hydrogels. (A) Stress–strain curve; (B) compressive modulus (strain range of 0–0.1 mm/mm) (the contents of each component of the samples are listed in Table SI2†); (C) image of the samples before and after compression.

The compressive moduli of samples B0.5P10, B0.5P15, B0.5P20, and B0.5P25 were 0.051, 0.716, 4.141, and 4.086 MPa, respectively; these values increased with increasing PVA content (Fig. 2B). A maximum compressive modulus of 4.141 MPa was reached at a PVA content of 20 wt%; this modulus is 23 times that of pure PVA (the compressive moduli of pure PVA hydrogels are shown in Fig. SI8†) with equal PVA content.

Effect of the dehydration on the compression properties of BC/PVA hydrogels

As shown in Fig. 3B, the compressive moduli of dehydrated samples B0.5P10 and B0.5P15 were higher than those of the original samples. However, the moduli of dehydrated samples B0.5P20 and B0.5P25 were lower than those of the original samples.

Fig. 3 Effect of the dehydration on the compression properties of BC/PVA hydrogels. (A) Stress–strain curve (strain range of 0–0.1 mm/mm) of original samples (BB0.5P10, B0.5P15, B0.5P20, and B0.5P25) and dehydrated samples (B0.5P10(d), B0.5P15(d), B0.5P20(d), B0.5P25(d)); (B) compressive modulus (strain range of 0–0.1 mm/mm) of original samples and dehydrated samples.

In general, the compressive moduli of hydrogels increased with the increase of dry matter content (and with decreasing water content). Fig. 3 shows that for samples with PVA contents of less than 20 wt%, the compressive moduli of the dehydrated samples increased compared to the original samples as their dry matter content increased. However, for hydrogels with PVA contents that exceeded 20 wt%, the compressive moduli of the dehydrated samples were unexpectedly lower than those of the original samples. Thus, in samples with PVA contents exceeding 20 wt%, the loss water can decrease the compressive modulus and negate the effect of the increased dry matter content.

We repeated this experiment with PVA samples with different molecular weights, and similar results were obtained (Fig. SI4†).

Water state of BC/PVA hydrogels with different PVA content

As shown in Fig. 4, the melting/crystallization peak of pure water appeared at 0 °C as free water. The melting/crystallization peak of water in pure PVA hydrogels appeared at approximately 0 °C and −1 °C; freezable bound water was the main component. As the PVA content increased, the melting/crystallization peak move to higher temperature slightly. The melting/crystallization peak of BC hydrogels also appeared at 0 °C and −1 °C; however, free water represented the main component. For BC/PVA composite hydrogels, the melting/crystallization peak appeared at 0 °C and −1 °C, and most of the water was transformed from free water into freezable bound water as the PVA content exceeded 20 wt%.

Fig. 4 DSC curve of water in pure water, and BC, PVA, BC/PVA hydrogels. (A) DSC curve of water in pure water, and BC, PVA hydrogels; (B) DSC curve of water in BC/PVA hydrogels.

The DSC results obtained for samples B1.0P10, B1.5P15, B2.0P20, and B3.0P30 demonstrated similar results but to a greater extent (Fig. SI6B†).

Effect of BC and PVA content on the compression properties of BC/PVA hydrogels

The stress–strain curves (Fig. 5A) from static compression testing showed that all samples in this group exhibited compressive failure. The strain points of initial compressive failure decreased significantly from 0.7 to 0.2 mm/mm.

Fig. 5 Effect of BC and PVA content on the compression properties of BC/PVA hydrogels. (A) Stress–strain curve; (B) compressive modulus (strain range of 0–0.1 mm/mm) (the contents of each component of the samples are listed in Table SI3†).

The compressive moduli of samples B1.0P10, B1.5P15, B2.0P20, and B3.0P30 were 0.118, 3.933, 25.416, and 24.630 MPa, respectively, and these values increased as the BC and PVA content increased (Fig. 5B). The maximum compressive modulus of the composite hydrogel with 2.0 wt% BC and 20 wt% PVA was 141 times that of the corresponding PVA hydrogel (20 wt%).

Storage modulus and tan δ peak of pure PVA and BC/PVA hydrogels

As shown in Fig. 6A, the storage modulus of the pure PVA hydrogel was approximately 0.2 MPa in the 0.1–200 Hz range. The storage modulus of the BC/PVA hydrogel was higher by almost one order of magnitude than that of pure PVA (up to 7–11 Mpa in sample B2.0P20 in the 200.1–200 Hz range (Fig. 6B)).

Fig. 6 Storage modulus curve of pure PVA and BC/PVA hydrogels. (A) Storage modulus curve of pure PVA hydrogels; (B) storage modulus curve of BC/PVA hydrogels.

As shown in Fig. 7, in the 0.1–200 Hz range, multiple peaks are observed in the tan δ curve for pure PVA (Fig. 7A and C); however, only two peaks are observed in the tan δ curve for BC/PVA (Fig. 7B) (with the exception of B1.0P10; several weak peaks are observed).

Fig. 7 tan δ curve of pure PVA and BC/PVA hydrogels. (A) tan δ curve of pure PVA hydrogels; (B) tan δ curve of BC/PVA hydrogels; (C) tan δ of pure PVA hydrogels in 0.1–125 Hz.

The tan δ peaks for pure PVA with various PVA contents were located at approximately the same positions in the 0.1–200 Hz range; however, the peak intensities are different.

Two peaks are present in the tan δ curve for a BC/PVA hydrogel in the ranges from 50–70 Hz and 130–140 Hz; the intensities of these two peaks were lower than those of the two corresponding peaks for a pure PVA hydrogel, especially that of the peak at 140 Hz. A decrease in tan δ indicates a decrease in the phase angle δ; i.e., the viscosity of the BC/PVA hydrogel is lower than that of the pure PVA hydrogel.

Discussion

Effect of BC pretreatment on the strain point of the initial compressive failure of BC/PVA hydrogels.

As the results of the static compression test shown in Fig. 1 and 5 showed that all of the BC/PVA hydrogel samples with pretreated BC failed during compression testing (strain range of 0–0.8 mm/mm). The strain points of the initial compressive failure decreased as the BC content increased. Cellulose is a rigid molecule and polymers containing this molecule tend to break if its content increased. Otherwise, the BC ribbons aggregated non-uniformly (as shown for B2.5 in Fig. SI1-1A†) during the pretreatment of BC, causing the final BC/PVA composite hydrogel product to possess some defects, which led to failure during compression testing.

As the results shown in Fig. 2, with the exception of B0.5P10, the BC/PVA hydrogel samples containing BC without pretreatment did not fail during compression testing (strain range of 0–0.8 mm/mm). The BC/PVA hydrogels containing BC without pretreatment (see B0.5 in Fig. SI1-1A†) exhibited uniform matrix structures (see B0.5P15, B0.5P20, and B0.5P25 in Fig. SI1-1C†), and the tiny defects that were present in these samples did not cause initial compressive failure during compression testing (strain range of 0–0.8 mm/mm). The sample B0.5P10 exhibited an initial compressive failure at 0.7 mm/mm due to the presence of a large hole that was produced from low PVA content (see B0.5P10 in Fig. SI1-1C†); in addition, PVA is subjected to tension in samples containing low PVA contents.

Thus, the presence of a homogeneous structure can increased the strain point at which initial compressive failure occurs.

Effect of BC and/or PVA content on the compression properties of BC/PVA hydrogels

From the results of the static compression test presented in Fig. 1, 2, and 5, the compressive moduli of the BC/PVA composite hydrogels increased with increasing BC and/or PVA content. Compared to pure PVA (Fig. SI8†), the mechanical strength of BC/PVA significantly increased.

The mechanical strength of BC/PVA hydrogel that have been reported with PVA content about 10 wt% (ref. 15 and 16) is limited. We prepared BC/PVA hydrogel with PVA content up to about 30 wt% to enhance its mechanical strength furthermore.

In particular, the compressive modulus of the BC/PVA composite hydrogel increased significantly, and reached a maximum at PVA contents of up to 20 wt% (the compressive modulus of B0.5P20 is 82 fold high than that of B0.5P10, and the compressive modulus of B2.0P20 is 214 fold high than that of B1.0P10). However, when the PVA content exceeded 20 wt%, the compressive modulus decreased slightly. As Nakaoki et al. reported that the pore size of PVA hydrogel decreased with increasing PVA content.¹⁷ The imperfect crystallisation of PVA might be caused by the small interspaces between the PVA domains, which would limit movement of the PVA chains during the freeze–thaw cycling, thus leading to the compressive modulus decrease. X-ray diffraction (XRD) results (Fig. SI3C†) support this perspective: the degree of pure PVA crystallinity decreased with increasing PVA content, and the degree of B2.0P20 crystallinity was higher than that of B3.0P30.

Effect of water state on the compression properties of BC/PVA hydrogels

The large number of hydroxyl groups in BC and PVA can form intramolecular hydrogen bonds or competitively form hydrogen bonds with water.¹⁸ In BC samples, most of the water is free (as the results of DSC in Fig. 4). In addition to the fact that the content range of BC in this study is only 0.5–3.0 wt%, the large number of hydroxyl groups result in assembly of the three level structure.¹⁹ In the BC production process, there are a limited number of hydroxyl groups in the BC that can form bonds with water molecules.

The DSC results showed that most of the water in the pure PVA hydrogels is freezable bound water, the melting/crystallization peak move to higher temperature slightly as PVA content increasing. We supposed that after crosslinking, the molecular chain of PVA is dispersed randomly in water; the hydrogen bonds between PVA and water are present in high amounts, and thus, most of the water is in the freezable bound state in the PVA hydrogels. As PVA content increases, the molecular interchain distance decreases, resulting in hydrogen bond formation in PVA molecular or between PVA molecular. The interaction between freezable bound water and PVA molecular weakened, which result in the melting/crystallization peak move to higher temperature slightly. For BC/PVA composite hydrogels, PVA molecules adhering to BC ribbons (see B0.5P0.5 in Fig. SI1 and 2†) formed hydrogen bonds first, and the dominant form of water was free water in the BC/PVA hydrogels with PVA contents of less than 20 wt%. At higher PVA content, the number of hydroxyl groups in the PVA molecules increased, increasing the possibility of PVA–water hydrogen bond formation. Finally, when the PVA content exceeded 20 wt%, the dominant form of water in the BC/PVA hydrogels was freezable bound water.

As shown by the results of the static compression test presented in Fig. 3, the compressive moduli of dehydrated BC/PVA hydrogels with PVA contents of lower than 20 wt% increased compared to the original samples (in which the main form of water was free water). However, the compressive moduli of dehydrated BC/PVA hydrogels with PVA contents exceeding 20 wt% were unexpectedly lower than those of the original sample, negating the effects of the increased dry matter content (in which the main form of water was freezable bound water). Thus, we can conclude that freezable bound water is more effective at improving the compressive modulus than free water. This phenomenon might occur because water molecules that are bound more tightly to polymer chains in hydrogels containing mostly freezable bound water than in hydrogels containing mostly free water, and freezable bound water molecules that are present in a semi-immobile state might be more difficult to squeeze out of the gels, improving their compressive moduli. Additionally, squeezing water out during compression testing was more difficult in samples with higher compressive moduli (Fig. SI7†).

We conducted similar experiments using samples B0.5P15, B0.7P15, B1.0P15, and B1.5P15. The dominant form of water in all of these samples was free water (Fig. SI6A†), and the compressive moduli increased in the dehydrated samples compared to the original samples in all cases (Fig. SI5†), supporting our conclusions.

So the large amount of freezable bound water in BC/PVA hydrogel with PVA content up to 20 wt% caused the dramatic increase in mechanical strength.

In addition, to the plasticizing effect of the non-freezing water,²⁰ the limited mobility of freezable bound water provided flexibility to the BC/PVA hydrogels based on its ability to act as a reversible and sacrificial component; hydrogen bonds could act as sacrificial bonds in robust hydrogels, as reported in several recent works.^21,22

Effect of BC composition on the chain motion and elastic moduli of PVA hydrogels

DMA is one of the most effective and sensitive methods for studying the motion of polymer chains; the location and shape of tan δ curve peaks are closely related to the mechanical relaxations of polymers, acting as a fingerprint.

BC, PVA, and BC/PVA composite hydrogels are crystalline polymers. Because crystalline and amorphous domains coexist, chain relaxation is more complex in crystalline polymers than in amorphous polymers. These behaviours include the glass transition of amorphous domains, crystal melting, crystal form transformation of the crystalline domain, and the transformation caused by the crystalline–amorphous domain interaction. In DMA frequency scans, the relaxation of the crystal domain is normally observed at lower frequencies.

As shown in Fig. 7A and C, multiple peaks are present in the tan δ curve of a pure PVA hydrogel, indicating that multiple motor units exist. For BC/PVA composite hydrogels, only two peaks (Fig. 7B) are observed, indicating that when associated with BC, the motion of the units of PVA are limited by BC and that the molecular chain tends to be rigid.

As shown by scanning electron microscopy (SEM) (Fig. SI1-1B†), the PVA hydrogel contains microcellular void structures, whereas BC comprises a nanofibre network, resulting in an orderly sheet texture (B0.5 and B2.5, see Fig. S1-1A†). PVA molecules were dispersed randomly in water during preparation and were well distributed in PVA hydrogels. However, in the BC/PVA hydrogel, PVA molecules first adhered to BC ribbons during crosslinking, and the BC ribbons oriented the PVA molecular chains to form a well-ordered structure (Fig. SI1 and 2†). As PVA content increased, a uniform, filled structure containing a microcellular BC/PVA matrix developed (Fig. SI1-1C†). Infrared (IR) spectroscopy showed BC/PVA hydrogen bonding (Fig. SI2†).

Stroescu et al. reported the similar result that the molecular mobility of PVA chains decreased in the interfacial zone because of the presence of the BC network,²³ and they think the main contributions to the exceptional mechanical performance of the nanocomposites are hydrogen bonding and chemical crosslinking between BC and PVA matrix.

As shown by the DMA results presented in Fig. 6, BC/PVA composite hydrogels increased by approximately one order of magnitude compared to pure PVA (Fig. 6) hydrogels. The results obtained in static compression testing also demonstrated that the mechanical strength of BC/PVA significantly increased compared to pure PVA. Based on the analysis presented above and considering that the lower crystallinity of BC/PVA compared to PVA (Fig. SI3B†) worsens the compressive modulus of the BC/PVA hydrogel, we believe that the orientation of the PVA molecule, which is oriented by BC, limited the motor units of PVA, rigidifying the molecular chain and improving the mechanical strength of the BC/PVA hydrogel.

Experimental

Materials

PVA (M_w = 89–98 kg mol⁻¹; degree of hydrolysis of >99%) was purchased from Sigma.

Pretreatment of BC

Laboratory-prepared BC (0.5 wt%) and its dehydrated (at 37 °C) samples containing 0.7, 1.0, 1.5, 2.0, and 2.5 wt% BC content were labelled B0.5, B0.7, B1.0, B1.5, B2.0, and B2.5, respectively.

Preparation of PVA solutions

Aqueous PVA solutions at various concentrations were prepared by swelling in deionised (DI) water sufficiently at room temperature, followed by heating at 115 °C for 15 min, then standing over night at 80 °C to remove the air bubble.

Preparation of PVA hydrogels

PVA solutions (10, 20, and 30 wt%) were placed into cylindrical moulds and then crosslinked by repeated freeze–thaw cycles at −20/+25 °C (6 times). The generated hydrogels were labelled P10, P20, and P30, respectively.

Preparation of BC/PVA hydrogels

BC samples were submerged in various PVA solutions at 80 °C for 7 days. The BC/PVA samples were then crosslinked by repeated freezing and thawing at −20/+25 °C (6 times). The sample names and estimated BC and PVA matrix contents in the BC/PVA hydrogels from each sample group are listed in the ESI (Tables SI1–3†).

Dehydration of BC/PVA hydrogels

Samples B0.5P10, B0.5P15, B0.5P20, and B0.5P25 were dehydrated by evaporating water at room temperature to 80 wt% of the original weight and labelled B0.5P10(d), B0.5P15(d), B0.5P20(d), and B0.5P25(d), respectively.

Each original sample was weighted and recorded its mass as m₁ (g), and was put into a 24-well plate without cover at room temperature. After that, it was weighted in every other hour until the mass is close to 0.8m₁ (g), and its mass was recorded as m₂, (s = |m₂ − 0.8m₁|/m₂, s should less than or equal to 0.05) Then the sample was operated with the static compression test immediately.

Sample characterisation

Static compression test. Static compression test specimens were prepared as cylinders (diameter of 8 mm; thickness of 3 mm) and tested under ambient conditions using a material testing machine (INSTRON 5967, Instron, High Wycombe, UK) at a crosshead speed of 0.18 mm min⁻¹ using a 500 N static load cell. All measurements were performed on at least three samples, and the average values were recorded.

DSC. Samples (5–15 mg) were hermetically sealed in an aluminium pan to analyse the water state using DSC (TA, DSC-Q200, USA). The pan was cooled to −20 °C, isothermally equilibrated for 5 min, and finally heated to 5.00 °C at a rate of 0.5 °C min⁻¹.

DMA. Cylindrical samples were prepared (diameter of 8 mm; thickness of 3 mm) and tested at room temperature using DMA (TA, Q800, USA) and multi-frequency scans (0.1–200 Hz).

Conclusions

Normal hydrogels that contain a large ratio of free water exhibit good flexibility but insufficient stiffness because water is easily extruded from the hydrogel. However, both flexibility and stiffness are necessary qualities of hydrogels used as cartilage replacements. Hydrogels with rigid polymer backbones and large freezable bound water ratios would provide both satisfactory flexibility and stiffness.

In this study, we developed a BC/PVA composite hydrogel, the compressive modulus increased by increasing the BC and/or PVA content. However, excessive BC content greatly decreased the flexibility, and the compressive modulus did not further increased when PVA contents exceeded 20 wt%. Using a simple and novel method that compares the compressive modulus variation between dehydrated and original samples and analyses the water state using DSC, we conclude that freezable bound water is more effective in improving the compressive modulus than free water and serves a significant role in maintaining the flexibility and stiffness of hydrogels.

In addition, by comparing the differences in the storage moduli and tan δ curve peaks between the BC/PVA and PVA hydrogels using DMA, we found that PVA molecules tend to be rigid in composites containing BC, with decreased viscosity and increased storage modulus.

Acknowledgements

The authors would like to thank Xuetao Shi for extensive help during the revision of this manuscript, greatly improving the overall quality of this paper. The authors gratefully acknowledge the financial support provided by the National Basic Research Program of China (Grant no. 2012CB619100), National Natural Science Foundation of China (Grant no. 50803018, 51273072, and 51232002), 111 project (B13039), and Guangdong Provincial Science and Technology Department Project (Grant no. 2012A080203010 and 2012A080800015).

Notes and references

1. V. M. Sardinha, L. L. Lima, W. D. Belangero, C. A. Zavaglia, V. P. Bavaresco and J. R. Gomes, Wear, 2013, 301, 218–225.
2. A. S. Hoffman, Ann. N. Y. Acad. Sci., 2001, 944, 62–73.
3. T. Huang, H. Xu, K. Jiao, L. Zhu, H. R. Brown and H. Wang, Adv. Mater., 2007, 19, 1622–1626.
4. N. Shah, M. Ul-Islam, M. Khattak and J. K. Park, Carbohydr. Polym., 2013, 98, 1585–1598.
5. N. Petersen and P. Gatenholm, Appl. Microbiol. Biotechnol., 2011, 91, 1277–1286.
6. L. E. Millon, C. J. Oates and W. Wan, J. Biomed. Mater. Res., Part B, 2009, 90, 922–929.
7. P. J. Basser, R. Schneiderman, R. A. Bank, E. Wachtel and A. Maroudas, Arch. Biochem. Biophys., 1998, 351, 207–219.
8. J. E. Berberat, M. J. Nissi, J. S. Jurvelin and M. T. Nieminen, Magn. Reson. Imaging, 2009, 27, 727–732.
9. J. Samuel, D. Sinha, J. C.-G. Zhao and X. Wang, Bone, 2014, 59, 199–206.
10. E. Otsuka, S. Komiya, S. Sasaki, J. Xing, Y. Bando, Y. Hirashima, M. Sugiyama and A. Suzuki, Soft Matter, 2012, 8, 8129–8136.
11. M. Abu-Abdeen and I. Elamer, Mater. Des., 2010, 31, 808–815.
12. R. Yudianti, M. Karina, M. Sakamoto and J. I. Azuma, Macromol. Res., 2009, 17, 1015–1020.
13. H. Hatakeyama and T. Hatakeyama, Thermochim. Acta, 1998, 308, 3–22.
14. J. Ostrowska-Czubenko, M. Pieróg and M. Gierszewska-Drużyńska, Prog. Chem. Appl. Chitin Its Deriv., 2011, 16, 147–156.
15. W.-D. Li, J.-H. Rong, Z.-D. Lin and X.-J. Zhang, Acta Polym. Sin., 2012, 4, 357–364.
16. A. F. Leitão, J. P. Silva, F. Dourado and M. Gama, Materials, 2013, 6, 1956–1966.
17. T. Nakaoki and H. Yamashita, J. Mol. Struct., 2008, 875, 282–287.
18. S. G. Kelmanovich, R. Parke-Houben and C. W. Frank, Soft Matter, 2012, 8, 8137–8148.
19. D. Klemm, D. Schumann, U. Udhardt and S. Marsch, Prog. Polym. Sci., 2001, 26, 1561–1603.
20. P. Blasi, S. S. D'Souza, F. Selmin and P. P. DeLuca, J. Controlled Release, 2005, 108, 1–9.
21. J.-Y. Sun, X. Zhao, W. R. Illeperuma, O. Chaudhuri, K. H. Oh, D. J. Mooney, J. J. Vlassak and Z. Suo, Nature, 2012, 489, 133–136.
22. J. P. Gong, Soft Matter, 2010, 6, 2583–2590.
23. C. Castro, A. Vesterinen, R. Zuluaga, G. Caro, I. Filpponen, O. J. Rojas, G. Kortaberria and P. Gañán, Cellulose, 2014, 21, 1745–1756.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra11594e

---