Dispersion stability, magnetivity and wettability of cellulose nanocrystal (CNC)-dispersed superparamagnetic Fe₃O₄ nanoparticles: impact of CNC concentration

Abstract

This study investigates the effects of cellulose nanocrystals (CNCs) on the dispersion and colloidal stability of Fe₃O₄–cellulose nanocrystal nanocomposites (MCNCs). The hybrid composites were prepared using an ultrasound assisted in situ co-precipitation technique in the presence of cellulose nanocrystals (CNCs) as the dispersant. The microscopy analysis showed that the dispersion of MNPs improved greatly with CNC as dispersant. STEM images showed that the mean particle size of Fe₃O₄ nanoparticle (MNPs) on all CNC samples was found to be less than 20 nm. However, the MNPs aggregated in the sample with 0.01 wt% of CNC. The colloidal stability improved substantially in the presence of 0.05 wt% CNC, and increasing the CNC concentration any more made no difference to the dispersive properties. The surface charges of MCNCs decreased drastically from −14.6 to −59.7 mV as CNC concentration increase from 0.00 to 1.00 wt%. The amount of MNPs deposited on CNC template decreased considerably as CNC concentration increased. Higher amount of MNPs deposited on the CNC surface gave rise to a higher surface wettability and magnetivity. All MCNC samples exhibited superparamagnetic properties and the saturation magnetization (M_s) of MCNC composites was reduced from 30.798 to 1.625 emu g⁻¹ with increasing CNC content from 0.01 to 1.00 wt%. Overall, the results of the study showed that the incorporation CNC led to an improvement of the MNP dispersion and colloidal stability. The as-prepared MCNCs can be used to stabilize palm olein-based emulsions, suggesting their potential usefulness as nanocarriers in food and drug delivery applications.

---

1. Introduction

Fe₃O₄ nanoparticles (MNPs) are excellent superparamagnetic materials that have been drawing increasing attention in various applications such as magnetic resonance imaging,¹ drug targeting,² hyperthermia,³ and ferrofluid⁴ since decades ago due to its low cost, negligible toxicity, and good biocompatibility properties.^5,6 However, MNPs in the suspension state are often thermodynamically unstable and have a great tendency for aggregation as a result of their high surface energy and magnetic dipole interaction.⁷ Their physical stability is a crucial problem that limits the industrial applications of MNPs and considerable research effort has been devoted to the preparation of stable MNP suspensions with good dispersibility. Controlling the dispersion and aggregation of the nanoparticles is crucial to exploit the advantages of the MNPs in diagnosis and therapy. Furthermore, a review done by Kharisov et al.⁸ reported that a highly aqueous solution of soluble MNPs has a stronger remediation effect compared with insoluble ones.

To tackle this problem, in order to prepare dispersible MNPs with good colloidal stability, some dispersants are often used to prevent agglomeration of nanoparticles during MNP synthesis.⁹ To date, natural dispersants including gelatin, dextran, starch, chitosan, and ethyl cellulose have also been extensively studied as an alternative dispersants to overcome the particle aggregation issue.^10–15 In literature, cellulose nanocrystals (CNCs) has been reported to exhibit remarkable performance to prevent the agglomeration of the MNPs. The improved dispersion and colloidal stability of MNPs in aqueous solution was attributed to the electrostatic repulsion imparted by the negatively charges surface sulfate groups, obtained after the acid hydrolysis of native cellulose.^16–20 Therefore, CNCs have been suggested as an ideal green, non-toxic, effective protectant or dispersant matrix in the development of magnetic nanocellulosic structures in which they have been suggested to be applicable as antibacterial agent,²¹ magnetically retrievable oil absorbent,²² recyclable catalyst,²¹ and drug delivery.²³

Using in situ chemical co-precipitation approach, the so-called Fe₃O₄–cellulose nanocrystal (MCNC) nanocomposites can be prepared by dispersing the inorganic MNPs in the cellulose matrix.^24,25 The incorporation of MNPs into CNC not only prevent the agglomeration of the MNPs in the suspension state, but also preserve the redispersibility of MNPs after in situ process. The surface functional groups of CNCs are able to act as nucleation sites for the MNPs to growth uniformly.^9,21 Recent study by Liu et al.⁹ demonstrated that the MNPs synthesized in situ in CNC matrix showed supreme resistant to agglomeration at both low and high MNPs loading. In addition, the resulted MNP suspension exhibited excellent colloidal stability due to the great dispersive properties of CNC.⁹ Besides MNPs, stable dispersion and colloidal stability of single-walled carbon nanotubes²⁶ and atactic polypropylene²⁷ have also been achieved using CNC as a dispersant.

Despite the numerous literature studies of CNC as renewable dispersant, the effect of CNC concentration on the dispersion and colloidal stability of MNPs has, however, not been adequately studied. Hence, the objective of this study was to investigate the impact of CNC concentration affects the dispersion stability, magnetivity, as well as the wetting behaviors of the resulting MCNC nanocomposites. As part of the present study, the as-prepared magnetic cellulose composites was used to stabilize palm oil-based emulsion and the morphology of the resulting Pickering emulsion was observed using optical microscopy.

2. Experimental section

2.1. Materials and instruments

Iron(II) chloride tetrahydrate (FeCl₂·4H₂O, ≥99%), iron(III) chloride hexahydrate (FeCl₃·6H₂O, 99%), and ammonium hydroxide (28% NH₃ in H₂O) were purchased from Sigma-Aldrich. CNC (freeze dried, 0.96 wt% sulfur content) was procured from University of Maine. β-Carotene rich-palm oil (275 mg β-carotene per kg, melting point 19 °C) was purchased from Sime Darby Jomalina Sdn Bhd (Malaysia). All water used in this experiment are ultrapure water obtained from Milli-Q® Plus apparatus (Millipore, Billerica, USA). Ethanol (AR standard) was acquired from R & M Chemical (Syarikat Saintifik Jaya, Malaysia). All experiments were conducted using an ultrasonic horn (20 kHz, 100 W system, NexTgen ultrasonic platform, Sinaptec, France) under pulse mode (15 s pulse on, 10 s pulse off). All chemicals in this study were of analytical grade.

2.2. In situ synthesis of MCNC composite

MCNC composites were prepared by ultrasound assisted in situ co-precipitation method. In particular, CNC was first dispersed in water at various concentrations (0.01, 0.05, 0.1, 0.5, and 1 wt%) under ultrasound cavitation for 2 minutes. Next, fixed amount of iron(III) and iron(II) chloride (1.5/1 Fe³⁺/Fe²⁺ mol ratio) were added into the CNC dispersion. Subsequently, the mixtures were stirred and heated to 45 °C. Then, the mixtures were sonicated in the presence of ammonium hydroxide (2.2 ml) for 5 minutes.

After sonication, MCNC composites were precipitated using ethanol. The MCNC composite residual obtained were magnetically separated and washes 3 times with ethanol to remove ammonium hydroxide. The remained MCNCs were centrifuged at 4500 rpm for 20 minutes, and dried in an oven overnight. The dried samples were stored for characterization. The resulting 5 MCNC samples were denoted as MCNC0.01, MCNC0.05, MCNC0.1, MCNC0.5, and MCNC1, which stand for MCNC prepared with 0.01, 0.05, 0.10, 0.50, and 1.00 wt% CNC respectively.

2.3. Stabilization of Pickering emulsion

The as-prepared MCNCs were used to prepare oil-in-water (O/W) emulsion using palm oil. Firstly, MCNC sample (0.05 wt%) were re-dispersed in water. Oil phase (red palm superolein) of a fixed volume fraction (φ_oil = 0.3) were then added to the MCNC dispersion, and emulsified using a computer-aided ultrasonic horn at room temperature for 3 minutes.

2.4. Characterization

Particle size and surface morphology of the sonochemically prepared MCNC composites were analyzed using Hitachi SU2010 field emission scanning electron microscope (FE-SEM) (Hitachi, Japan) at 15 kV. ImageJ were also used to examine the particle size of MNPs on CNC. Functional groups available were examined through Fourier transform infrared spectroscopy (FTIR) over a range of 550–4000 cm⁻¹ on a FTIR spectrophotometer (Nicolet iS10, Thermo Scientific, USA) equipped with diamond probe. Magnetivity of MCNCs was measured via vibrating sample magnetometry (VSM) (Lakeshore 7400 Series). Zeta potential of MCNC samples was measured using Zetasizer Nano ZS 90 (Malvern instruments, UK), at 25 °C. Surface wettability of MCNC composites was analyzed using a contact-angle goniometer (Ramé-hart, USA). An Optical microscope (Nikon Eclipse TS100, Nikon Instruments Inc., USA) was employed to visualize the MCNC stabilized Pickering emulsion droplets.

2.5. Statistical analysis

Analysis of variance (ANOVA) were conducted using Prism software and p < 0.05 was considered as statistically significant. Data from histogram generated by ImageJ were also analysed via ANOVA using Prism software.

3. Results and discussion

In the present study, we prepared the MCNC composites by employing one-step ultrasound-aided in situ co-precipitation method. FTIR was carried out to confirm the interaction between the iron oxide and CNCs. As shown in Fig. 1, the spectra pattern of MCNC was similar to the combined spectra of both pure CNC and MNPs, with slight changes on certain wavelengths. Some important characteristics functional groups of sulfuric acid hydrolyzed CNC were observed at 3334 cm⁻¹ (–OH hydrogen bonding), 2896 cm⁻¹ (CH₂ bonding), 1651 & 1430 cm⁻¹ (symmetric and asymmetric COO– stretching), 1011 cm⁻¹ (C–O bonding), 664 cm⁻¹ (C–OH bonding), 610 cm⁻¹ (C–C–O bonding).²⁸ Peak at 561 cm⁻¹ in MCNC corresponded to the Fe–O bond stretching of Fe₃O₄.^28–30 As can be seen from Fig. 1, the signal at 664 cm⁻¹ assigned to the C–OH of CNC was absent in FTIR spectra of the MCNC, indicating the reaction between the Fe ions and the hydroxyl groups of CNCs. This FTIR studies confirms the suggested mechanism on the formation of MCNC illustrated in Scheme 1.

Fig. 1 FTIR spectra of (a) MNPs, (b) CNCs, and (c), MCNC composite.

Scheme 1 Schematic illustration of the formation of MCNC.

STEM analysis was performed to evaluate the MNP coverage on CNC template with varying CNC concentration and the results were presented in Fig. 2. In this study, MNPs and CNCs were used as control (Fig. 2a and g). The microscopic analysis illustrated that the MNPs prepared without the presence of CNC as template appears to be agglomerates of very irregular shape (Fig. 2a) due to their high surface energy and magnetic dipole interaction. Fig. 2g revealed the interlinked network of CNC formed from its original rod-like structure. In contrast, the presence of CNC led to much better dispersion of MNPs, as shown in Fig. 2b–f. It was found that highest MNPs coverage was achieved with MCNC0.01 sample (Fig. 2b). However, MCNC0.01 suffered from severe MNPs aggregation (Fig. 2b inset). Despite of slightly lowered MNPs coverage, a much improved dispersion of MNPs was observed at MCNC0.05 (Fig. 2c inset). Besides that, Fig. 2c–f showed that the dispersity of MNPs gradually improved when CNC concentration increased from 0.05 to 1.00 wt%. This STEM analysis further confirmed the role of CNC as an effective dispersant for MNPs.

Fig. 2 STEM images of (a) pure MNPs, (b) MCNC0.01, (c) MCNC0.05, (d) MCNC0.1, (e) MCNC0.5, (f) MCNC1, and (g) pure CNCs. The inset shows the images at higher magnification. All scale bars are representing 100 nm.

The microscopic images (Fig. 2) illustrated that MNPs diameter decreases with increasing CNC concentration. Based on ImageJ analysis, the average particle size of MNPs reduced significantly (P < 0.05) from 16 nm to 12 nm when CNC concentration increased from 0.01 to 0.05 wt% (Fig. 3a). Interestingly, negligible changes was observed in MNPs diameter when more than 0.05 wt% of CNC was used as template (Fig. 3a). The main reason for this size reduction was due to the insufficient amount of CNC for stabilizing all MNPs at concentration of 0.01 wt%. In this study, MCNC0.05 led to a good dispersion for MNPs with average particle diameter around 12 nm. In addition, using the XRD data (Fig. S1†) the mean particles size for MCNC0.05 sample was also estimated to be around 9 nm utilizing Scherrer equation. This is similar to the diameter obtained from the ImageJ analysis (Fig. 3a and S2†). Besides that, it was noticeable that the counts of MNPs on MCNC samples reduces when CNC concentration increases. Thus the sample MCNC0.01 results in slight aggregation of MNPs, while MCNC1 shows most scattered MNPs loading. The microscopic images implies that lower CNC concentration will resulted in higher MNPs loading and vice versa.

Fig. 3 (a) MNPs Feret diameter in different MCNC samples, standard error of mean of triplicate readings were represented by the error bars in each graph, different alphabetic letters was significantly different at P ≤ 0.05 by Bonferroni's multiple comparison test. (b) MNPs loading numbers count on CNC in different MCNC samples.

The sign and magnitude of the surface charge is an important parameter in determining the stability of a dispersion and zeta potential is commonly used to predict and control dispersion stability. Fig. 4 shows the zeta potential values of MCNC samples. The result shows that CNC possess very high surface charge (−70.3 ± 2.0 mV) as a result of ester sulfate groups introduced during acid-hydrolysis process in native CNC production. MNPs, on the other hand, have much lower surface charge (−14.6 ± 0.7 mV) and thus it has great tendency to particle agglomeration. At fixed concentration of MNP, one can observe that the surface charges of MCNC reduce significantly from −22.2 ± 0.2 to −59.7 ± 2.5 mV with increasing CNC concentration. It was found that the MCNC samples exhibited stable surface charged starting from CNC content of 0.05 wt% (see Fig. 4).³¹ This trend is mainly due to the presence of more and more anionic surface functional groups when more and more CNC is used. These results were in good agreement with our STEM results. At increasing CNC concentration, the ratio of Fe ions to CNC functional groups reduced greatly, rendering more and more un-reacted surface functional groups remained after the synthesis. This study implies that higher concentration of CNC enhances the electrostatic stability of MCNC, thereby improves its dispersion stability. To visualize the data, photograph of suspension of pure MNPs, pure CNCs, and all MCNC samples were presented in Fig. 5. It was obvious that the pure MNPs settled down quickly (Fig. 5a), while the pure CNC remain dispersed in water medium (Fig. 5g). For the MCNC samples, the nanocomposites settled down when the CNC concentration was at 0.01 wt% (Fig. 5b). The suspensions with CNC concentration of 0.05 wt% or above, on the other hand, were stable towards sedimentation (Fig. 5c–f). This correlate well with our zeta potential results, where stable zeta potential were observed starting from CNC content of 0.05 wt%.³¹

Fig. 4 Surface charge of CNC, MNP, and all MCNCs of different CNC concentration.

Fig. 5 Photograph of water dispersion of (a) pure MNPs, (b) MCNC0.01, (c) MCNC0.05, (d) MCNC0.1, (e) MCNC0.5, (f) MCNC1, and (g) pure CNC. Images captured 10 min after preparation.

The surface wettability of MCNC nanocomposite was presented in Table 1. It is generally reported that pristine CNCs are amphiphilic cellulose fibers that easily disperse in water medium,³² while pure MNPs are typically hydrophilic particles.³³ Based on Table 1, it has been observed that all MCNC composites appear to be more hydrophobic as compared to the two control samples, namely pristine CNC and MNPs. The results showed that the contact angle of the composites increased with increasing CNC concentration from 0.01 wt% to 0.05 wt%. Intriguingly, the contact angle decreases considerably from 51.87 ± 0.17 to 43.42 ± 0.36° as CNC concentration were further increased from 0.05 wt% to 1.00 wt%. This phenomenon was mainly because at fixed concentration of MNPs, the surface wettability of composite reduces with increasing amount of CNC. When increasing the CNC concentration progressively from 0.05 to 1.00 wt%, the wetting ability of MCNC is directed by the excess amount of CNC. Further increase in CNC concentration to 1.00 wt% resulted in substantial increase in the amount of hydrophilic driving hydroxyl groups on the CNC surface. Thus, this led to a more hydrophilic behavior of MCNC composite, as witnessed by reduced contact angle at MCNC1. To support this, our STEM analysis showed a drastic fall in MNP content from samples MCNC0.05 to MCNC1, with the lowest amount of MNPs observed in MCNC1 (Fig. 3b). The noticeable trend of composite surface wettability was primarily attributed to the changes in the amount of hydrophilic groups introduced as a result from different CNC concentration. The hydroxyl groups (–OH⁻) play an important role in the formation of MCNC nanocomposites. As illustrated in Scheme 1, it is proposed that three Fe ion would require four oxygen ions in order to form a single MNP coated on CNCs, one –OH⁻ group of CNC possesses, however, only one oxygen ions. Thus, four –OH⁻ groups are required to initiate the deposition of a single MNP on the CNC surface. One could envision that embedment of MNPs onto the cellulose matrix led to a great reduction in total hydrophilic groups available on CNC surface. However, the overall hydrophilicity will become higher with excess MNPs deposition on the CNC surfaces. In literature, a similar stepwise formation of MNP had been reported by Awwad and Salem³⁴ on the preparation of MNPs stabilized by carob leaf extract.

Table 1 Water–air contact angle of CNC, MNP, and MCNCs samples (see Fig. S3 for contact angles images) different alphabetic letters was significantly different at P ≤ 0.05 by Bonferroni's multiple comparison test

Sample	Water–air contact angles (°)
MNP	21.17 ± 0.35^a
MCNC0.01	44.98 ± 0.23^b
MCNC0.05	51.87 ± 0.17^c
MCNC0.1	49.10 ± 0.15^d
MCNC0.5	44.84 ± 0.17^b
MCNC1	43.42 ± 0.36^e
CNC	36.72 ± 0.25^f

---

Fig. 6 presented the magnetization curve of MCNC with varying concentration of CNC. It was apparent that all MCNC samples exhibited superparamagnetic properties, with negligible coercivity, as manifested by the inset. An evident decrease in the saturation magnetization (M_s) of the MCNC composite was observed as concentration of CNC increases. This might due to the fact that at fixed concentration of MNPs, the elevated amount of CNC in the dispersion gradually reduces the magnitude of M_s of the MCNC. In literature, studies showed that the M_s of MNPs at single domain (<30 nm) reduced gradually as its particles size decreased due to spin canting effect at the surface.^35,36 In the present study, one can, however, notice that the particles size remained unchanged despite a marked reduction in M_s from CNC 0.10 wt% to 1.00 wt%. One possible explanation for this observation was because MNPs were immobilized firmly in the CNC network, preventing them from particle aggregation.

Fig. 6 Magnetization plot of (a) (black) MNPs, (b) (yellow) MCNC0.01, (c) (purple) MCNC0.05, (d) (cyan blue) MCNC0.1, (e) (green) MCNC0.5, and (f) (red) MCNC1. The inset shows the negligible coercivity in MCNC.

As part of the current study, formation of Pickering emulsions using developed MCNC0.05 sample has been attempted. As shown in Fig. 7a, the use of pristine CNC did not lead to the formation of stable emulsions. The CNC-stabilized emulsions suffered from severe creaming, with a distinct oil layer forming at the top layer. This could be explained by the fact that the presence of sulfate half-ester groups promotes the electrostatic repulsion between the CNC particles at the interface, thus affects their ordering at the oil–water interfaces, leading to unstable emulsion.³⁷ Kalashnikova et al.³² observed that emulsion exhibited poor colloidal stability when CNC with greater surface charge densities was used as stabilizer. In this study, MCNCs were found to favor the formation of stable oil-in-water (O/W) emulsions, as evidenced in Fig. 7b. The embedment of hydrophilic MNPs onto the CNC surface affects the contact angle and thus the wettability of MCNC composites and therefore its emulsifying performance.

Fig. 7 Pickering emulsion stabilized with (a) CNC (b) MCNC0.05.

4. Conclusions

In the present study, superparamagnetic MCNC nanocomposites were successfully synthesized in situ via ultrasonication. The MNPs deposited on CNCs were found to be uniformly dispersed and have particle diameter below 20 nm. The study revealed that the CNC concentration greatly affects the dispersion stability, magnetic strength and wetting behaviour of MCNC nanocomposites. The STEM images showed that high degree of MNP coverage was achieved with slight particle agglomeration at a low CNC concentration of 0.01 wt%. The particle size of MNPs reduced when CNC concentration was increased to 0.05 wt%. Further increment of CNC concentration above 0.05 wt% did not promote the size changes of MNPs but lead to significant reduction of M_s from 30.798 to 1.625 emu g⁻¹. The dispersion stability of MCNC improved with increasing CNC content. Contact angle measurement indicated that MCNC was more hydrophobic than pristine CNC upon the deposition of MNPs on CNC surface. MCNCs prepared with lower CNC concentration of 0.05 wt% yield a highest water–air contact angle with its recorded value of 51.87 ± 0.17°, showing that the surface wettability of magnetic cellulosic bionanocomposites can be modulated by manipulating the CNC concentration. The developed MCNC has been successfully exploited for stabilization of palm-oil based Pickering emulsion and it represents a significant progress towards developing related MCNC-based controlled drug delivery systems suitable for nutraceutical and pharmaceutical applications.

Conflict of interest

The authors declare no competing financial interest.

Acknowledgements

Author is grateful to the support from the HDR Scholarship from the School of Engineering, Monash University Malaysia. The authors acknowledge Advanced Engineering Platform (AEP), Monash University for the financial support for sample analysis. B. H. Ong acknowledges the Ministry of Higher Education (FP064-2014A) and SATU Joint Research Scheme (RU022F-2014).

Notes and references

1. D. Kim, Y. Zhang, J. Kehr, T. Klason, B. Bjelke and M. Muhammed, J. Magn. Magn. Mater., 2001, 225, 256–261.
2. J. Zhang and R. Misra, Acta Biomater., 2007, 3, 838–850.
3. A. Silva, T. Oliveira, J. B. Mamani, S. Malheiros, L. Malavolta, L. Pavon, T. Sibov, E. Amaro Jr and L. Gamarra, Int. J. Nanomed., 2011, 591–603.
4. A. Zubarev and L. Iskakova, Phys. A, 2004, 343, 65–80.
5. A. Szpak, G. Kania, T. Skórka, W. Tokarz, S. Zapotoczny and M. Nowakowska, J. Nanopart. Res., 2012, 15, 1372–1379.
6. M. Lattuada and T. Hatton, Langmuir, 2007, 23, 2158–2168.
7. B. Kharisov, H. Dias, O. Kharissova, A. Vázquez, Y. Peña and I. Gómez, RSC Adv., 2014, 4, 45354–45381.
8. Y. Ye, T. Kong, X. Yu, Y. Wu, K. Zhang and X. Wang, Talanta, 2012, 89, 417–421.
9. K. Liu, J. Nasrallah, L. Chen, L. Huang and Y. Ni, Carbohydr. Polym., 2015, 126, 175–178.
10. W. Wu, Q. He and C. Jiang, Nanoscale Res. Lett., 2008, 3, 397–415.
11. M. Carmen Bautista, O. Bomati-Miguel, M. del Puerto Morales, C. Serna and S. Veintemillas-Verdaguer, J. Magn. Magn. Mater., 2005, 293, 20–27.
12. B. Gaihre, S. Aryal, M. Khil and H. Kim, J. Microencapsulation, 2008, 25, 21–30.
13. D. Kim, M. Mikhaylova, F. Wang, J. Kehr, B. Bjelke, Y. Zhang, T. Tsakalakos and M. Muhammed, Chem. Mater., 2003, 15, 4343–4351.
14. A. Zhu, L. Yuan and T. Liao, Int. J. Pharm., 2008, 350, 361–368.
15. J. Peng, Q. Liu, Z. Xu and J. Masliyah, Adv. Funct. Mater., 2012, 22, 1732–1740.
16. R. Batmaz, N. Mohammed, M. Zaman, G. Minhas, R. Berry and K. Tam, Cellulose, 2014, 21, 1655–1665.
17. T. Huq, S. Salmieri, A. Khan, R. Khan, C. Le Tien, B. Riedl, C. Fraschini, J. Bouchard, J. Uribe-Calderon and M. Kamal, et al., Carbohydr. Polym., 2012, 90, 1757–1763.
18. F. Hu, S. Fu, J. Huang, D. P. Anderson and P. R. Chang, in Polysaccharide-Based Nanocrystals: Chemistry and Applications, ed. J. Huang, P. R. Chang, N. Lin and A. Dufresne, Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim, Germany, 2014, pp. 15–62.
19. R. Moon, A. Martini, J. Nairn, J. Simonsen and J. Youngblood, Chem. Soc. Rev., 2011, 40, 3941–3994.
20. Y. Habibi, L. Lucia and O. Rojas, Chem. Rev., 2010, 110, 3479–3500.
21. R. Xiong, C. Lu, Y. Wang, Z. Zhou and X. Zhang, J. Mater. Chem. A, 2013, 1, 14910–14918.
22. S. Chin, A. Binti Romainor and S. Pang, Mater. Lett., 2014, 115, 241–243.
23. T. Anirudhan and S. Rejeena, J. Appl. Chem., 2014, 2014, 1–10.
24. S. Cao, X. Li, W. Lou and M. Zong, J. Mater. Chem. B, 2014, 2, 5522–5530.
25. Y. Li, H. Zhu, H. Gu, H. Dai, Z. Fang, N. Weadock, Z. Guo and L. Hu, J. Mater. Chem. A, 2013, 1, 15278–15283.
26. C. Olivier, C. Moreau, P. Bertoncini, H. Bizot, O. Chauvet and B. Cathala, Langmuir, 2012, 28, 12463–12471.
27. N. Ljungberg, C. Bonini, F. Bortolussi, C. Boisson, L. Heux and J. Y. Cavaillé, Biomacromolecules, 2005, 6, 2732–2739.
28. K. Mahmoud, E. Lam, S. Hrapovic and J. Luong, ACS Appl. Mater. Interfaces, 2013, 5, 4978–4985.
29. N. Lin and A. Dufresne, Nanoscale, 2014, 6, 5384–5393.
30. S. Suganuma, K. Nakajima, M. Kitano, D. Yamaguchi, H. Kato and S. Hayashi, J. Am. Chem. Soc., 2008, 130, 12787–12793.
31. Nanocomposix Online, http://50.87.149.212/sites/default/files/nanoComposix%20Guidelines%20for%20Zeta%20Potential%20Analysis%20of%20Nanoparticles.pdf (accessed May 3, 2016).
32. I. Kalashnikova, H. Bizot, B. Cathala and I. Capron, Biomacromolecules, 2012, 13, 267–275.
33. J. Zhou, X. Qiao, B. Binks, K. Sun, M. Bai, Y. Li and Y. Liu, Langmuir, 2011, 27, 3308–3316.
34. M. A. Awwad and M. N. Salem, Nanosci. Nanotechnol., 2013, 2, 208–213.
35. A. Roca, M. Morales, K. O'Grady and C. Serna, Nanotechnology, 2006, 17, 2783–2788.
36. S. Linderoth, P. Hendriksen, F. Bødker, S. Wells, K. Davies, S. Charles and S. Mørup, J. Appl. Phys., 1994, 75, 6583–6585.
37. K. Fleming, D. Gray and S. Matthews, Chem.–Eur. J., 2001, 7, 1831–1836.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16109j

---