Tunable Aggregation and Gelation of Thermoresponsive Suspensions

Apr 26, 2016 - Héloïse Thérien-Aubin , Yihe Wang , Katja Nothdurft , Elisabeth Prince , Sangho Cho , and Eugenia Kumacheva. Biomacromolecules 2016 ...
19 downloads 0 Views 6MB Size
Article pubs.acs.org/Biomac

Tunable Aggregation and Gelation of Thermoresponsive Suspensions of Polymer-Grafted Cellulose Nanocrystals Firas Azzam,†,‡ Eder Siqueira,†,‡ Sébastien Fort,†,‡ Roumaïssa Hassaini,†,‡,§,∥ Frédéric Pignon,§,∥ Christophe Travelet,†,‡ Jean-Luc Putaux,†,‡ and Bruno Jean*,†,‡ †

Université Grenoble Alpes, Centre de Recherches sur les Macromolécules Végétales (CERMAV), F-38000 Grenoble, France CNRS, CERMAV, F-38000 Grenoble, France § Université Grenoble Alpes, Laboratoire Rhéologie et Procédés (LRP), F-38000 Grenoble, France ∥ CNRS, LRP, F-38000 Grenoble, France ‡

S Supporting Information *

ABSTRACT: The colloidal stability together with the tunable aggregation and viscoelastic properties of thermoresponsive polymergrafted cellulose nanocrystals (CNCs) were investigated. TEMPO oxidation of CNCs followed by peptidic coupling in water were used to covalently graft thermosensitive Jeffamine polyetheramine M2005 chains onto the surface of CNCs. The resulting polymer-decorated particles (M2005-g-CNCs) exhibited new colloidal properties, by their ability to perfectly redisperse in water and organic solvents such as toluene, dichloromethane or DMF after freeze-drying. In addition, they presented an enhanced thermal stability when compared to that of sulfated or TEMPO-oxidized CNCs. Dynamic light scattering experiments were used to demonstrate that the thermally induced aggregation of M2005-g-CNCs was fully reversible and reproducible over many temperature cycles and that, most interestingly, the aggregation number could be tuned by varying the ionic strength and/or the pH of the medium, making the suspension multiresponsive. This property arises from the variations of the sign (attractive or repulsive) and the range of the different types (entropic, electrostatic, hydrophobic) of interaction forces between the thermosensitive polymer-decorated nanoparticles. The variation of the viscoelastic properties of M2005-g-CNCs suspensions as a function of temperature, probed by oscillatory rheology measurements of more concentrated suspensions, revealed a reversible temperature-triggered liquid-to-gel transition. Such enhanced functionalities pave the way to the design of advanced CNC-based materials benefiting both from the intrinsic characteristics of these biosourced particles and the new properties imparted by the stimuli-sensitive grafted chains.



thermoplastic nanocomposites.13 Moreover, while sulfate esters enable electrostatic interactions that are beneficial to the dispersion in aqueous media, they make colloidal stability highly sensitive to ionic strength, prevent dispersions in nonpolar media, and lead to disordered gel-like structures at high concentration. A key challenge to circumvent these drawbacks is a precise tuning of the surface properties of CNCs. Such a control can be achieved by grafting stimuli-sensitive polymers, which will impart both steric stabilization to the derivatized nanoparticles in various media and responsive properties triggered by external stimuli such as temperature, pH or light. As reported in the literature, the preparation of stimulisensitive CNCs actually represents a promising way to design smart biosourced materials, in particular for biomedical applications.14 Different polymer grafting strategies have been undertaken.15−18 The “grafting onto” method has been tested

INTRODUCTION The use of cellulose nanocrystals (CNCs) as renewable building blocks for the rational design of functional materials is now commonly recognized.1−4 This potential arises from the unique properties of these nanoparticles. CNCs are indeed crystalline nanorods resulting from the acid hydrolysis of abundant cellulose microfibrils. CNCs display a high and tunable aspect ratio (depending on the cellulose source), a low density (1600 kg m−3), a high specific surface area ranging from 150 to 300 m2 g−1,5,6 excellent mechanical properties,7−9 and the ability to self-organize into liquid crystalline phases.10 In addition, CNCs are no longer only produced at laboratory scale, as pilot-plant quantities are now commercially available in Canada and in the USA.11 However, the development of highperformance CNC-based materials is restricted by some limitations. CNCs are usually prepared by sulfuric acid hydrolysis of cellulose from various sources (wood pulp, cotton linters, algae, tunicates, etc.).12 This treatment introduces sulfate esters on the surface of CNCs, which significantly lowers their degradation temperature, thus restraining their use in © XXXX American Chemical Society

Received: March 7, 2016 Revised: April 24, 2016

A

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

2000 g mol−1, and its LCST ∼ 16 °C. Other chemicals were purchased from Sigma-Aldrich. Deionized water was used for all experiments. Preparation of Cellulose Nanocrystals by Acid Hydrolysis. To prepare CNC suspensions, cotton linters were hydrolyzed according to the method described by Revol et al. by treating the almost pure cellulosic substrate with 65 wt % sulfuric acid during 30 min at 63 °C.33 The suspensions were washed by repeated centrifugations, dialyzed against distilled water until constant conductivity of the dialysis bath and ultrasonicated for 4 min with a Branson Digital sonifier. After these treatments, the suspensions were filtered through 8 μm and then 1 μm cellulose nitrate membranes (Sartorius). At the end of the process, ∼3 wt % stock suspensions were obtained. Carboxylation of Cellulose Nanocrystals by TEMPO Oxidation. Cellulose nanocrystals resulting from sulfuric acid hydrolysis of cotton linters were subjected to TEMPO-mediated oxidation as previously reported.34−36 NaBr (1.588 g, 15.4 mmol) and TEMPO (135 mg, 0.86 mmol) were added to 500 mL of a 1 wt % cellulose nanocrystals suspension (30.8 mmol anhydroglucose units) and magnetic stirring was applied for 1 h. 21.1 mL of a 1.46 M NaOCl solution (1 equiv per anhydroglucose unit) was then added dropwise under stirring to the cellulose suspension. During NaOCl addition, the pH was maintained at 10 by using a 0.5 M NaOH solution. When the pH became constant, the TEMPO reaction was stopped by adding methanol (20 mL), and the pH was adjusted to 7 with 0.5 M HCl. The resulting suspensions were washed three times with distilled water by successive centrifugation at 20 000g and redispersion. NaCl (0.5 M) was added to facilitate the nanoparticle separation, and the washing procedure was repeated three times using a 0.1 N HCl solution. After dialysis against distilled water, a colloidal suspension stable over years was obtained. Polymer Grafting by Peptidic Coupling in Water. The grafting of amine-terminated Jeffamine polyetheramine M2005 was achieved through peptidic coupling according to Bulpitt and Aeschlimann.37 The pure polymer was added (Np moles per carboxyl unit measured by conductometry) to a 1 wt % carboxylated CNC suspension and stirred until dissolution. The reaction was performed at 4 °C to ensure that the macromolecules were under good solvent conditions. The pH was adjusted to 7.5−8.0 before the addition of 2 mL of an aqueous solution containing N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide hydrochloride (EDAC) and N-hydroxysuccimide (NHS) (NEDAC and NNHS mol per carboxyl group, respectively). The reaction was carried out with NP = NEDAC = NNHS = 4. The reaction lasted 24 h at room temperature under stirring while maintaining the pH of the mixture at 7.5−8.0 using 0.5 M NaOH or 0.5 M HCl. The pH was finally lowered to 1−2 by addition of 0.5 M HCl, and the resulting suspension was dialyzed against distilled water to remove excess reagents including nongrafted Jeffamine. Unless otherwise stated, the pH of the suspension was about 7.5. The degree of substitution was measured by conductometric titration to 0.05 mol polymer per mol anhydroglucose unit, and this value was confirmed by solid-state NMR, as described in our previous work.32 Assuming homogeneous grafting on the cellulose surface, this moderate grafting density makes lateral interaction between neighboring polymer chains on the surface of the CNCs impossible. The grafted chains are therefore in the socalled mushroom regime where their dimension, in good solvent conditions, is close the radius of giration of the chains (Rg ∼ 1.2 nm for Jeffamine polyetheramine M2005). It follows that the thickness of the polymer shell around the nanocrystals is about 1.5 nm. The estersulfate and carboxylate contents of the native, TEMPO-oxidized and polymer-grafted CNCs measured by elemental analysis and conductometric titration, respectively, are given in Table S1 in Supporting Information. A description of the reaction schemes is given in Supporting Information Figure S1. Jeffamine polyetheramine M2005grafted CNCs will be referred to as M2005-g-CNCs. Redispersion Test. Forty milligrams of freeze-dried M2005grafted CNCs was mixed with 4 mL of solvent to achieve a CNC concentration of 10 g L−1 and sonicated five times for 2 min with a 3 mm microtip probe. Vials containing the suspensions were kept in an

by Araki et al. and Kloser et al. with epoxy-terminated polyethylene glycol (PEG) chains19,20 and by Mangalam et al. with DNA oligonucleotides using peptidic coupling on carboxylated CNCs.21 In the former case, sterically stabilized CNCs were produced, and in the latter case, complementary strands of DNA grafted to separate populations of CNCs could hybridize to form duplex structures. “Grafting from” reactions were alternatively used. For example, thermoresponsive poly(N-isopropylacrylamide) chains were grafted onto CNCs via surface-initiated living radical polymerization.22−25 Zeinali and co-workers grafted dual-temperature and pH-sensitive random copolymers of N-isopropylacrylamide and acrylic acid (AA) onto CNCs via RAFT polymerization.26 pH-Sensitive poly(4vinylpyridine) chains or PAA-polyelectrolyte brushes-grafted CNCs were synthesized by Kan et al. and Majoinen et al. by surface-initiated graft polymerization with the initiator ceric(IV) ammonium nitrate and surface-initiated controlled radical polymerization, respectively.27,28 Poly(dimethylaminoethyl methacrylate)-grafted CNCs were prepared by Yi et al. and Tang et al. using free radical polymerization, and dualresponsive Pickering emulsion stabilization was achieved in the latter work.29,30 Random copolymer grafts consisting of dimethylaminoethyl methacrylate and napthyl-functionalized methacrylate were attached onto the surface of CNCs via surface-initiated ATRP and combined with PVA-derived chains by host−guest chemistry to yield stiff and self-healing hydrogels.31 We contributed to the field by grafting thermosensitive Jeffamine polyetheramines using the grafting-onto strategy via peptidic coupling after TEMPO oxidation.32 In this study, we unambiguously demonstrated the formation of stable covalent amide bonds between the polymer and the CNCs. The grafting density, quantitatively measured using conductometry and solid-state NMR, was sufficiently high to induce a steric stabilization of the CNCs in water that prevented flocculation at high ionic strength and made the CNCs surface-active. Additionally, a very interesting feature displayed by the surfacemodified CNCs was their fully reversible thermal aggregation behavior when the samples were heated above the lower critical solution temperature (LCST) of the polymer. In the present article, we show additional properties of Jeffamine polyetheramine-decorated CNCs related to their colloidal, aggregation, and rheological behaviors. To this end, characterization techniques such as dynamic light scattering (DLS), transmission electron microscopy (TEM), and rheology measurements were used. The results show that such a surface grafting of polymers broadens the properties of CNCs, making them responsive not only to temperature but also to ionic strength and pH, while increasing their thermal stability and redispersibility in various types of solvents. Such enhanced functionalities pave the way to the design of advanced CNCbased materials benefiting both from the intrinsic characteristics of the CNCs such as light weight and the new properties imparted by the stimuli-sensitive grafted chains.



EXPERIMENTAL SECTION

Materials. Cotton linters were provided by Buckeye Cellulose Corporation and used as the cellulose source without any further purification. The thermosensitive polymer used in this study was the Jeffamine polyetheramine M2005, which was donated by the Hunstman Corporation. It is a statistical amine-terminated copolymer of ethylene oxide (EO) and propylene oxide (PO) with an EO/PO monomer composition equal to 6/29. Its molecular weight is about B

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules



ice bath during sonication, and the samples were allowed to cool for 4 min between two sonication treatments. Thermogravimetric Analysis. The thermogravimetric analysis of the freeze-dried samples was performed with a Setaram 92−12 apparatus at a heating rate of 2 °C/min under dry nitrogen. Transmission Electron Microscopy (TEM). The general method to prepare dry specimens for TEM was the following: drops of ca. 0.001 wt % CNC suspensions were deposited onto carbon-coated TEM grids freshly glow-discharged in an EasiGlow plasma cleaning system (Pelco). After 2 min, the liquid in excess was blotted away with filter paper and, prior to drying, a drop of a 2% uranyl acetate solution was deposited on the specimen. After 2 min, the stain in excess was wicked off, and the remaining thin liquid film was allowed to dry. This protocol was adapted to account for the LCST of the grafted thermosensitive polymer (∼16 °C). One group of TEM specimens were prepared in a cold room from a CNC suspension and uranyl acetate solution both maintained at 4 °C. Another series of specimens were prepared from a CNC suspension and negative stain solution both preheated at 40 °C. Cryo-TEM specimens were prepared as well by quench-freezing thin films of CNC suspensions formed on NetMesh lacy carbon films (Pelco) into liquid ethane, using a Leica EM-GP fast-freezing workstation equipped with a temperature- and humidity-controlled chamber. The humidity was set to 85% r.h. For the experiment below the polymer LCST, it was not possible to cool down the chamber below room temperature, and thus the suspension (∼0.1 wt %) was kept at 4 °C prior to deposition on a precooled lacy carbon film. The thin liquid film was rapidly quench-frozen to prevent any significant warming up of the suspension. For the experiment above the polymer LCST, a drop of the suspension prewarmed at 40 °C was deposited on the grid, which was placed inside the EM-GP chamber previously equilibrated at 40 °C. After blotting, the liquid film was quench-frozen. The specimens were mounted in a Gatan 626 specimen holder cooled with liquid nitrogen, transferred into the microscope, and observed at low temperature (−176 °C), under low illumination. Negatively stained and quench-frozen specimens were observed using a Philips CM200 “Cryo” microscope operating at 80 kV. The images were recorded on Kodak SO163 films. Dynamic Light Scattering (DLS). DLS experiments were carried out with a Malvern NanoZS instrument. Unless otherwise stated, all measurements were made at a well-controlled (±0.05 °C) temperature at a backscattering detection angle of 173°. The intensity size distribution was obtained from the analysis of the correlation function using the multiple narrow mode algorithm of the Malvern DTS software. Viscoelasticity Measurements. Viscoelastic properties of polymer-grafted CNCs suspensions were carried out in the oscillatory mode with a TA Instruments Rheometer (ARG2) with a cone and plate geometry mode (angle 4°, diameter 20 mm, gap 113 μm). A Peltier plate allowed varying the temperature (temperature ramp 1 °C min−1) to determine the evolutions of the storage and loss moduli G′ and G′′ from 8 to 40 °C. The atmosphere around the sample was saturated with water to avoid evaporation during the temperature ramp. To define the linear domain, strain sweep procedures were conducted in the range of 0.006 to 0.1 at a frequency of 1 Hz to define the strain corresponding to the linear domains of the deformation. Then, the frequency sweeps in the range of 0.1−100 rad s−1 at strain of 0.06, defined from the linear domains, were performed. During oscillatory tests, the dynamic shear moduli (G′, the storage modulus and G″, the loss modulus) were measured as a function of temperature at a fixed frequency of 0.5 or 1 Hz and a strain amplitude of 0.06. Several measurements were done successively to verify that, after a temperature ramp test, the moduli values were at the same level, confirming the nonevaporation of the sample. Several strain and frequency sweeps were performed at different specific temperatures during temperature ramp tests, in order to assess that the measurements were well in the linear domain all along the temperature range from 8 to 40 °C in increasing and decreasing temperature ramps.

Article

RESULTS AND DISCUSSION

Redispersion in Water and Organic Solvents. As shown in our previous study, thermosensitive Jeffamine polyetheramines can be successfully covalently grafted onto CNCs using the “grafting onto” strategy through peptidic coupling after TEMPO oxidation.32 When the reaction is performed in water, a degree of substitution (DS) of about 0.05 is obtained, which corresponds to a moderate grafting density and suggests that grafted polymer chains are in a mushroom regime without polymer−polymer interactions between vicinal chains, rather than in a brush-like conformation. Even if the DS value is rather low, the surface attachment of polymer chains onto CNCs drastically improves the colloidal stability of the nanoparticles in water since it was shown that polyetheramine-grafted nanocrystals remained stable for months upon addition of 1 M NaCl, whereas nongrafted CNC dispersions immediately became turbid.32 Here, we have tested the redispersion of freeze-dried M2005g-CNCs in various organic solvents following the protocol described in the Experimental Section. Photographs of the samples 2 h after redispersion clearly showed that a very good dispersion was achieved in the following solvents: water, DMF, ethanol, dichloromethane, and toluene (Figure 1).

Figure 1. Dispersibility of M2005-g-CNCs in various media: water (a), DMF (b), ethanol (c), dichloromethane (d), and toluene (e).

Turbid samples were observed in the case of THF and 1butanol, showing incomplete dispersion in these solvents (Supporting Information Figure S2). No dispersion was obtained for DMSO and cyclohexane as evidenced by the phase-separated samples (Supporting Information Figure S2). Further investigations were undertaken in the case of toluene. As a blank experiment, TEMPO-oxidized CNCs were redispersed in toluene but even after extended sonication, no dispersion could be achieved leading CNC aggregates to quickly sediment (Supporting Information Figure S3). On the contrary and in addition to the naked-eye stability displayed in Figure 1, M2005-g-CNCs redispersed in toluene also exhibited birefringence under shaking (visual observation when the sample was placed between crossed polars), which is a good criterion for the absence of large aggregates. Furthermore, as shown in Figure 2a, the hydrodynamic diameter of M2005-gCNCs in toluene is very close to that of the same particles initially dispersed in water, showing that the dispersion quality in toluene is as good as in water. It has to be noted that the suspensions were stable in toluene for several months. Finally, the TEM micrograph of M2005-g-CNCs initially dispersed in toluene clearly showed well-dispersed rodlike nanoparticles with a similar appearance and dimensions as those of the initial CNCs in water (150 × 22 × 6 nm3 on average),32,38 proving again the absence of aggregates and the effective redispersion of C

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

sulfuric acid hydrolysis is detrimental to the thermal stability of the particles that decreases with increasing sulfate content. This behavior was ascribed to a desulfatation phenomenon that generates sulfuric acid, thus accelerating the thermal degradation of cellulose. In our case, TEMPO oxidation performed in alkaline conditions resulted in a significant removal by a factor 2 of sulfate esters that led to a strong increase in thermal stability (Supporting Information Table S1). Polymer grafting carried out at pH 7.5−8 does not affect the remaining sulfate esters to a great extent, but polymer chains might hinder thermally induced desulfatation, therefore improving further the thermal stability of polymer-grafted CNCs compared with that of oxidized ones. Similar enhanced thermal stability of stimuli-sensitive polymer-grafted CNCs were also reported by Yi et al. and Zeinali et al.26,29 These results reveal another beneficial effect of the polyetheramine grafting that makes it possible to preserve colloidal stability while strongly improving the thermal stability of the nanoparticles. Repeatability and Imaging of the Temperatureinduced Aggregation. As we previously reported from DLS and optical density data,32 M2005-g-CNC suspensions exhibited a clear reversible thermal aggregation behavior when the samples were heated above the polymer LCST (∼16 °C). As shown in Figure 4, a remarkable reversibility and

Figure 2. (a) Size distribution of M2005-g-CNCs initially dispersed in water and redispersed in toluene after freeze-drying. The concentration of both suspensions was 0.1 wt %. (b) TEM micrograph of unstained M2005-g-CNCs from a dispersion in toluene.

freeze-dried M2005-g-CNCs in toluene (Figure 2b). All these results prove that polyetheramine grafting induces steric stabilization that considerably widens the redispersion capability of CNCs into solvents of various polarities. This ability of surface-modified CNCs to redisperse in apolar solvents was previously reported after grafting of PEG19 or polystyrene,39 adsorption of surfactants40 and silylation.41 However, as developed below, the grafting of thermosensitive polyetheramines provides additional interesting properties. Thermal Stability. Sulfated CNCs, TEMPO-oxidized CNCs, M2005-g-CNCs, and M2005 polymer were investigated by thermogravimetric analysis to probe the effect of the successive chemical reactions on the thermal stability of the nanoparticles. The results after normalization are compared in Figure 3.

Figure 4. Hydrodynamic diameter variation exhibited by a 0.5 wt % M2005-g-CNC aqueous suspension submitted to 4 °C−40 °C−4 °C temperature cycles. Dotted lines are only guides for the eyes.

reproducibility of the phenomenon was evidenced when the 4 °C−40 °C−4 °C temperature cycles were repeated. This result shows that no degradation or loss of the thermally induced aggregation ability occurs when multiple temperature variations are applied, showing the robustness of the system. The temperature-induced aggregation was characterized by TEM and cryo-TEM. Samples were first prepared from suspensions kept at a temperature of 4 °C (about 12 °C lower than the LCST of the thermosensitive polyetheramine). The corresponding TEM (Figure 5a) and cryo-TEM (Figure 5b) micrographs show fairly well dispersed rodlike nanoparticles, which are similar to the initial bare CNCs, as previoulsy reported, which consist of a few laterally associated elementary crystallites (never disaggregated ribbon-like clusters).32,38 Both observations show that polymer grafting did not alter the morphology of the initial CNCs and that, at a temperature below the LCST of the polyetheramine, the suspension contained individual particles, in agreement with DLS data. The grafted polymer chains were not visualized since

Figure 3. Thermogravimograms of sulfated CNCs (S-CNC), TEMPO-oxidized CNCs (TO−CNC), Jeffamine polyetheramine M2005 (M2005) and M2005-g-CNCs.

The onset of mass loss occurs at about 180 °C for sulfated CNCs, 250 °C for oxidized CNCs and 280 °C for polymergrafted CNCs, showing that thermal stability of the nanoparticles was shifted to much higher temperatures after the oxidation reaction and to even higher temperatures after grafting. For M2005-g-CNCs, a two-step decomposition was observed that evidenced the mixed character of such hairy colloids. The TGA curve for this sample could tentatively be ascribed to a superposition of the features of the M2005 polyetheramine and unmodified cellulose, which usually proceeds in the 320−360 °C range. This very clear improvement of the thermal stability might be related to the sulfate content of the samples. As shown by Roman and Winter,13 the presence of sulfate esters on the surface of CNCs prepared by D

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Under such conditions, the interactions between polymerdecorated CNCs become attractive and aggregation takes place. Lateral aggregation is logically favored since the polymer shell probably mostly covers the sides of the CNCs and not their ends due to a higher number of reactive groups. For a low particle concentration such as the one used during the temperature-cycle experiment in Figure 4, hydrophobic attractive interactions that develop at 40 °C result in finitesize aggregates with a hydrodynamic diameter between 700 and 800 nm. TEM observations, carried out under lower concentration conditions also display finite size aggregates consisting of 10 to 40 M2005-g-CNCs. Most probably, these aggregates are electrostatically stabilized by remaining charges on the outer surface of the aggregates. DLS data show that a complete disaggregation was achieved each time the sample was cooled, even though some hysteresis had previously been observed.32 Tuning of the Temperature-Induced Aggregation. To further probe the precise origin of the thermally induced aggregation, the effect of salt addition was investigated. To this end, the hydrodynamic diameter during a 4 °C−40 °C−4 °C temperature cycle was measured for a 0.5 wt % M2005-g-CNC suspension where NaCl had been added to reach a concentration of 50 mM, and the results were compared with those obtained without added salt. The results reported in Figure 6 show that the addition of 50 mM NaCl had no effect

Figure 5. TEM micrographs of M2005-g-CNCs prepared from aqueous suspensions at 4 °C (a,b) and 40 °C (c,d), respectively: (a,c) negatively stained preparations; (b,d) cryo-TEM observations.

they are too small and their surface density is too low to generate any detectable contrast in the TEM images. The TEM image of a negatively stained preparation from a suspension kept at 40 °C (Figure 5c) shows aggregates with a length ranging from 300 to 700 nm, and a width from 75 to 200 nm. Moreover, the significant accumulation of stain around the bulkier aggregates (dark outline) suggests that they may have a somewhat cylindrical shape. Given the average dimensions of the initial CNCs (150 × 22 × 6 nm3), as previously reported,38 these aggregates likely result from the lateral assembling of M2005-g-CNCs; the width of these aggregates approximately corresponds to that of 4 to 10 individual particles, while their length is 2 to 4 times that of initial CNCs. The cryo-TEM image of the suspension quench-frozen at 40 °C (Figure 5d) also reveals laterally aggregated CNCs, in good agreement with the image of the negatively stained specimen, thus allowing the discard of possible drying artifacts. The size of the individual aggregates is more difficult to evaluate as they often overlap in the embedding film of amorphous ice. Interestingly, in the cryo-TEM image (Figure 5d), the CNCs constituting the aggregates, and exhibiting a dark contrast, do not seem to be contiguous. They are separated by a thin region of lighter contrast, which may correspond to the thin and less dense corona of polymer surrounding the particles. However, Fresnel effects prevent a reliable estimation of dimensions. This observation cannot clearly be seen in the image of negatively stained aggregates as the capillary forces exerted during drying combined to polymer chain collapse due to dehydration may promote a stronger compaction of the CNCs. DLS and TEM results can be explained by temperaturetriggered changes in the interaction forces between the nanoparticles. For temperatures below the LCST of Jeffamine polyetheramine M2005 (e.g. 4 °C in Figure 4), the grafted polymer chains are in good solvent conditions and generate entropic repulsion forces between the nanoparticles. The measured hydrodynamic diameter therefore has a low value about 160 nm that corresponds to the hydrodynamic size of individual polymer-decorated nanocrystals. Accordingly, images of both dry and quench-frozen specimens show individualized objects. When the temperature is increased above the LCST (e.g., 40 °C in Figure 4), polymer chains collapse due to poor solvent conditions and turn from hydrophilic to hydrophobic.

Figure 6. Hydrodynamic diameter variation exhibited by a 0.5 wt % aqueous M2005-g-CNC suspension submitted to a single 4 °C−40 °C−4 °C temperature cycle with no added salt (●) and with addition of 50 mM (NaCl) (○). *This data point is a rough estimate since such a high hydrodynamic diameter is beyond the range accessible by DLS with full confidence (see text for more details).

on the hydrodynamic diameter when the temperature was below the LCST (4 °C), but a dramatic increase in size is observed when the temperature was increased above the LCST (40 °C). It must be noted that the data point marked with an asterisk, which corresponds to a suspension with added salt at 40 °C, is only a rough estimation and should be taken as such. It was visually observed that under such conditions the sample was turbid and contained particles in the order of tens of microns. Such large sizes are beyond the range accessible by DLS with full confidence. This precaution being taken, the conclusion that the addition of 50 mM NaCl greatly increases the number of particles per aggregate (aggregation number, Nagg) can nevertheless be unambiguously drawn. E

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules To explain the variation in size with temperature in the presence of 50 mM NaCl, electrostatic interactions have to be considered in addition to the polymer−polymer interactions. Actually, as previously reported,32 only a fraction (about 30%) of the carboxylic acid groups generated on the surface of the CNCs after TEMPO oxidation reacts during the peptidic coupling reaction with amine-terminated polymer chains. Consequently, about 70% of the carboxylic acid groups are still present after polymer grafting. At neutral pH, these groups are in their carboxylate form (pKa ∼ 4.5) and therefore need to be considered as negative surface charges inducing electrostatic interactions. For temperatures below the LCST, such electrostatic repulsions add to the entropic repulsion forces between polymer chains. The increase in ionic strength upon salt addition will screen the charges and decrease the Debye length characterizing the range of the repulsion. As shown in our previous study32 and in Figure 6, this charge screening and consequent decrease in the range of electrostatic repulsion forces is not detrimental to colloidal stability since the entropic repulsion forces between swollen polymer chains still ensure strong repulsions between the nanoparticles. The stability was indeed even maintained when monovalent salt concentrations as high as 1 M were added.32 Above the LCST, the situation is different since there is now a competition between opposing forces: electrostatic repulsive interactions resulting from the carboxylate groups and hydrophobic attractive interactions between the collapsed polymer chains. At low ionic strength (no added salt), the increase of the temperature above the LCST induces a moderate aggregation since long-range electrostatic repulsions tend to oppose the hydrophobic interactions. As a consequence, in the case of a 0.5 wt % M2005-g-CNC suspension, a 5 to 6-fold increase in the hydrodynamic diameter is observed. However, in the presence of 50 mM NaCl, the weakening of the electrostatic interactions favors the aggregation, resulting in large bundles 10 to 30 times larger than those observed in the absence of added salt. A variation of the ionic strength is therefore a way to control the extent of the thermally induced aggregation. A low or high ionic strength results in a small or large Nagg value, respectively. Interestingly, a full reversibility to the initial individual particles size was obtained after cooling the sample even when large aggregates were formed. Since the thermally induced aggregation of M2005-g-CNCs is ionic strength-sensitive, it is expected that it is a fortiori pHsensitive. Carboxylic acids are indeed weak acids with a pKa around 4.5. At low pH values, e.g., pH 3, protonation occurs and neutral COOH groups are quantitatively obtained. At higher pH values, e.g., pH 7.5, the negatively charged COO− form has to be considered. It is thus possible not only to screen the carboxylate ions by adding salt but also to suppress them and make the nanoparticles neutral by decreasing the pH. To test the effect of low pH on the temperature-induced aggregation of Jeffamine polyetheramine-decorated CNCs without increasing the ionic strength, ion-exchange resins were used to lower the pH of the suspension to 3. The sample was then subjected to a similar DLS experiment during 4 °C− 40 °C−4 °C temperature cycles. Results are shown in Figure 7. At pH 3 and when the temperature was kept below the LCST (4 °C in this example), no effect of the suppression of charges was observed since the measured hydrodynamic diameter of about 150 nm was that of the individual polymer decorated nanoparticles. This result shows that in the absence of

Figure 7. Hydrodynamic diameter variation exhibited by a 0.1 wt % aqueous M2005-g-CNC suspension at pH 3 submitted to 4 °C−40 °C−4 °C temperature cycles. Straight lines are only guides for the eyes.

electrostatic interactions, entropic repulsion forces between polymer chains in good solvent are sufficient to ensure colloidal stability of individual nanoparticles. However, when the temperature is increased above the LCST, a very strong aggregation of M2005-g-CNCs, much more pronounced than at pH 7.5 (Figure 4) was observed in Figure 7. At pH 3 and T = 40 °C, electrostatic interactions are minimized and hardly counterbalance the attractive forces between collapsed polymer chains, leading to the formation of very large aggregates. These results show that the thermally induced aggregation of M2005g-CNCs can also be controlled by the pH of the suspension. As shown in Supporting Information Figure S4, when the particle concentration was as low as 0.02 wt %, high pH conditions prevented temperature-induced aggregation from taking place. In this case, electrostatic repulsions are strong enough to overcome the attractive interactions between the polymer chains. However, when the charges are suppressed at low pH, aggregation at 40 °C can occur. Thermoreversible Gelation. As shown in Figure 8, an increase of the temperature from 4 to 40 °C of a 1.6 wt %

Figure 8. Photographs of a 1.6 wt % M2005-g-CNC suspension at 4 °C (a) and 40 °C (b).

M2005-g-CNC suspension turns the sample from a transparent bluish liquid at low temperature to a turbid white gel at high temperature. This result, at a macroscopic level, indicates that at higher concentrations, not only the particle aggregation takes place, but also that a connectivity between aggregates can occur, resulting in a change of the rheological properties of the sample. To gain more insight into this temperature-induced gelation, the variation of the viscoelastic properties of M2005-g-CNC F

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

down from 34 °C toward the LCST, the disruption of the formed network is more and more pronounced, leading to a sharp decrease of G′ and a moderate decrease of G′′. An interesting phenomenon, which is emphasized during the last part of the cooling step from 16 to 8 °C, is a sudden increase of G′ and G′′ followed by a decrease until the viscoelastic values reach the same values as the initial conditions before warming and cooling. One explanation of this sudden increase in G′ and G′′ could be attributed to local reorganizations of the particle due to the balance between electrostatic interactions between CNC surfaces and polymer−polymer interactions until an equilibrium state is reached. This bump in the viscoelastic curves has been measured repeatedly on different samples at different concentrations during these kinds of warming and cooling procedures, as shown in Supporting Information Figure S5 for a higher CNC concentration. When the temperature reaches the starting temperature of 8 °C, initial values of both G′ and G′′ are recovered. This result demonstrates a full reversibility of the viscoelastic behavior. Overall, the viscoelastic behavior of the sample can be controlled by the temperature, and reversible transitions from liquid to gel state can be induced. In other words, the grafting of Jeffamine polyetheramine M2005 onto CNCs leads to thermoresponsive gelation of the suspension, providing that the sample concentration is higher than 1 wt %.

suspensions was determined as a function of temperature. The results are shown in Figure 9.

Figure 9. Storage modulus G′ and loss modulus G″ as a function of temperature of a 5.8 wt % M2005-g-CNC. Experiments were performed at 0.06 strain and at an angular frequency of 1 Hz. The temperature was increased from 8 to 40 °C in panel a and decreased back from 40 to 8 °C in panel b.

At 8 °C, the storage modulus G′ is lower than the loss modulus G″, showing that under these conditions the sample behaves as a viscous fluid. The same behavior was observed by Ureña-Benavides and co-workers for suspensions of sulfated bare nanocrystals of concentration lower than 10.4 vol %.42 Below the LCST of the polymer, entropic repulsive interactions between particles make the aggregation or connection between particles unfavorable, resulting in liquid-like features. When the temperature is increased toward the LCST of the polymer, G′ increases strongly, while G″ first keeps about the same value and then increases slowly. When T > 18 °C, G′ still increases but tends to plateau when T > 30 °C, while G″ is more or less constant. Consequently, G′ becomes higher than G″ when T > 18 °C, and the difference between both quantities increases but tends to stabilize for a G″/G′ ratio equal to ∼3. This observation shows that the sample turns from a viscous fluid to an elastic gel when the temperature reaches 18 °C. This temperature value is very close to the LCST of Jeffamine polyetheramine M2005 and the onset of aggregation of M2005g-CNCs through hydrophobic attractive interactions between collapsed polymer chains on the surface of the CNCs.32 In addition to the aggregation described above, interaggregate connection is also possible at this higher particle concentration. The sample thus undergoes a temperature-induced viscous liquid to gel transition when the temperature is increased above the LCST of the polymer. As G′ and G′′ increases regularly from 18 to 40 °C, one possible explanation is that the CNC assembled together in aggregated bundles of limited size connect with one another through interactions between their outer polymer corona to form a structural network that strengthens with temperature increase. This hypothesis of aggregates connection is in accordance with the fact that when the sample is cooled down from 40 °C, G′ decreases strongly at a temperature higher than the LCST, while a moderate decrease of G″ is observed. This abrupt decrease of G′ could be attributed to the disconnection of the aggregated subdomains that were held together by a limited number of connection points, while the internal aggregate structure is maintained by the cohesion between collapsed polymer chains as the temperature is higher than the LCST. This is a possible explanation why during the cooling down G′ becomes lower than G″ at T = 34 °C, which is 18 °C above the LCST of Jeffamine polyetheramine M2005. When the sample is cooled



CONCLUSION The presence of covalently grafted thermosensitive polymer chains onto the surface of CNCs drastically changes the colloidal, thermal, aggregation, and viscoelastic properties of these biosourced nanoparticles. Polymer chains first act as a stabilizing agent that enables the redispersion of freeze-dried samples in water and various organic solvents like toluene, dichloromethane, DMF, and ethanol. Second, the polymer corona surrounding the particles plays the role of a protective layer that enhances the thermal stability of the nanoparticles. Beyond the reproducible and reversible character of the temperature induced aggregation of M2005-g-CNCs that was shown, results prove that the aggregation number can be finely tuned by changing the pH or the ionic strength of the medium. Namely, Nagg will be maximum at low pH (whatever the ionic strength), minimum at high pH and low ionic strength, and intermediate at high pH and high ionic strength. Such an accurate control of the aggregation arises from the evolution of the sign and range of the different types of interaction forces between the particles. For temperatures below the LCST, entropic repulsions between the swollen polymer chains at the CNC surface ensure colloidal stability in the form of individualized particles, even when electrical charges are screened or minimized by increasing the ionic strength or decreasing the pH. Above the LCST, hydrophobic attractive interactions between the collapsed polymer chains oppose electrostatic repulsions between remaining carboxylate groups that can be screened by adding salt or suppressed by decreasing the pH, leading in the last case to enhanced aggregation. At higher concentration, a thermoreversible gelation of the polymer-decorated CNC suspension was evidenced by viscoelastic measurements and attributed to the formation and disruption of connections between aggregates that form above the polymer LCST. The system investigated here therefore displays a high degree of versatility and provides a wide variety of new functionalities to the CNCs. The sensitivity of the aggregation properties to temperature, ionic strength, G

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

(18) Larsson, E.; Sanchez, C. C.; Porsch, C.; Karabulut, E.; Wågberg, L.; Carlmark, A. Eur. Polym. J. 2013, 49, 2689. (19) Araki, J.; Wada, M.; Kuga, S. Langmuir 2001, 17, 21. (20) Kloser, E.; Gray, D. G. Langmuir 2010, 26, 13450. (21) Mangalam, A. P.; Simonsen, J.; Benight, A. S. Biomacromolecules 2009, 10, 497. (22) Zoppe, J. O.; Habibi, Y.; Rojas, O. J.; Venditti, R. A.; Johansson, L. S.; Efimenko, K.; Osterberg, M.; Laine, J. Biomacromolecules 2010, 11, 2683. (23) Zoppe, J. O.; Osterberg, M.; Venditti, R. A.; Laine, J.; Rojas, O. J. Biomacromolecules 2011, 12, 2788. (24) Zoppe, J. O.; Venditti, R. A.; Rojas, O. J. J. Colloid Interface Sci. 2012, 369, 202. (25) Hemraz, U. D.; Lu, A.; Sunasee, R.; Boluk, Y. J. Colloid Interface Sci. 2014, 430, 157. (26) Zeinali, E.; Haddadi-Asl, V.; Roghani-Mamaqani, H. RSC Adv. 2014, 4, 31428. (27) Kan, K. H. M.; Li, J.; Wijesekera, K.; Cranston, E. D. Biomacromolecules 2013, 14, 3130. (28) Majoinen, J.; Walther, A.; McKee, J. R.; Kontturi, E.; Aseyev, V.; Malho, J. M.; Ruokolainen, J.; Ikkala, O. Biomacromolecules 2011, 12, 2997. (29) Yi, J.; Xu, Q.; Zhang, X.; Zhang, H. Cellulose (Dordrecht, Neth.) 2009, 16, 989. (30) Tang, J.; Lee, M. F. X.; Zhang, W.; Zhao, B.; Berry, R. M.; Tam, K. C. Biomacromolecules 2014, 15, 3052. (31) McKee, J. R.; Appel, E. A.; Seitsonen, J.; Kontturi, E.; Scherman, O. A.; Ikkala, O. Adv. Funct. Mater. 2014, 24, 2706. (32) Azzam, F.; Heux, L.; Putaux, J.-L.; Jean, B. Biomacromolecules 2010, 11, 3652. (33) Revol, J.-F.; Godbout, L.; Dong, X. M.; Gray, D. G.; Chanzy, H.; Maret, G. Liq. Cryst. 1994, 16, 127. (34) Denooy, A. E. J.; Besemer, A. C.; Vanbekkum, H. Recueil Des Travaux Chimiques Des Pays-Bas-Journal of the Royal Netherlands Chemical Society 1994, 113, 165. (35) Saito, T.; Isogai, A. Biomacromolecules 2004, 5, 1983. (36) Habibi, Y.; Chanzy, H.; Vignon, M. R. Cellulose (Dordrecht, Neth.) 2006, 13, 679. (37) Bulpitt, P.; Aeschlimann, D. J. Biomed. Mater. Res. 1999, 47, 152. (38) Elazzouzi-Hafraoui, S.; Nishiyama, Y.; Putaux, J.-L.; Heux, L.; Dubreuil, F.; Rochas, C. Biomacromolecules 2008, 9, 57. (39) Morandi, G.; Heath, L.; Thielemans, W. Langmuir 2009, 25, 8280. (40) Bonini, C.; Heux, L.; Cavaillé, J.-Y.; Lindner, P.; Dewhurst, C.; Terech, P. Langmuir 2002, 18, 3311. (41) Goussé, C.; Chanzy, H.; Excoffier, G.; Soubeyrand, L.; Fleury, E. Polymer 2002, 43, 2645. (42) Ureña-Benavides, E. E.; Ao, G. Y.; Davis, V. A.; Kitchens, C. L. Macromolecules 2011, 44, 8990.

and pH make the polymer-decorated particles multiresponsive, which paves the way to their use as smart building blocks for the preparation of advanced CNC-based materials.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications Web site and contains a schematic description of the chemical modification of CNCs, data about sulfate and carboxylate contents, images of redispersion tests in various solvents and intensity size distribution measured by DLS of a 0.02 wt % M2005-g-CNC suspension in water at pH 3.5 and pH 7.6 measured at 8, 20, and 40 °C. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biomac.6b00344. (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Funding

The authors are grateful to the French Ministry of Higher Education and Research for the Ph.D. grant of F.A. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Hunstman Corporation is acknowledged for the generous gift of Jeffamine polyetheramine samples. REFERENCES

(1) Eichhorn, S. J. Soft Matter 2011, 7, 303. (2) Habibi, Y.; Lucia, L. A.; Rojas, O. J. Chem. Rev. 2010, 110, 3479. (3) Peng, B. L.; Dhar, N.; Liu, H. L.; Tam, K. C. Can. J. Chem. Eng. 2011, 89, 1191. (4) Tingaut, P.; Zimmermann, T.; Sèbe, G. J. Mater. Chem. 2012, 22, 20105. (5) MatosRuiz, M.; Cavaille, J.-Y.; Dufresne, A.; Gerard, J.-F.; Graillat, C. Compos. Interfaces 2000, 7, 117. (6) Angles, M. N.; Dufresne, A. Macromolecules 2001, 34, 2921. (7) Eichhorn, S. J.; Young, R. J.; Davies, G. R. Biomacromolecules 2005, 6, 507. (8) Sturcova, A.; Davies, G. R.; Eichhorn, S. J. Biomacromolecules 2005, 6, 1055. (9) Eichhorn, S. J.; Dufresne, A.; Aranguren, M.; Marcovich, N. E.; Capadona, J. R.; Rowan, S. J.; Weder, C.; Thielemans, W.; Roman, M.; Renneckar, S.; Gindl, W.; Veigel, S.; Keckes, J.; Yano, H.; Abe, K.; Nogi, M.; Nakagaito, A. N.; Mangalam, A.; Simonsen, J.; Benight, A. S.; Bismarck, A.; Berglund, L. A.; Peijs, T. J. Mater. Sci. 2010, 45, 1. (10) Revol, J.-F.; Bradford, H.; Giasson, J.; Marchessault, R. H.; Gray, D. G. Int. J. Biol. Macromol. 1992, 14, 170. (11) Bras, J.; Chauve, G. In Handbook of Green Materials: Processing Technologies, Properties and Applications; Oksman, K., Mathew, A. P., Bismark, A., Rojas, O., Sain, M., Ed.; World Scientific Publishing Co.: Singapore, 2014. (12) Chauve, G.; Fraschini, C.; Jean, B. In Handbook of Green Materials: Processing Technologies, Properties and Applications; Oksman, K.; Mathew, A. P.; Rojas, O.; Sain, M. Eds.; World Scientific Publishing Co.: Singapore, 2014; Vol. 1, p 73. (13) Roman, M.; Winter, W. T. Biomacromolecules 2004, 5, 1671. (14) Jorfi, M.; Foster, E. J. J. Appl. Polym. Sci. 2015, 132, 41719. (15) Eyley, S.; Thielemans, W. Nanoscale 2014, 6, 7764. (16) Habibi, Y. Chem. Soc. Rev. 2014, 43, 1519. (17) Carlmark, A.; Larsson, E.; Malmström, E. Eur. Polym. J. 2012, 48, 1646. H

DOI: 10.1021/acs.biomac.6b00344 Biomacromolecules XXXX, XXX, XXX−XXX