Cold Gelation of Alginates Induced by Monovalent Cations

2966

Biomacromolecules 2010, 11, 2966–2975

Cold Gelation of Alginates Induced by Monovalent Cations C. Karakasyan,† M. Legros,† S. Lack,‡ F. Brunel,‡ P. Maingault,‡ G. Ducouret,† and D. Hourdet*,† Physico-Chimie des Polyme`res et des Milieux Disperse´s (UMR 7615, UPMC-CNRS-ESPCI), 10 rue Vauquelin, 75005 Paris, France, and Laboratoires Brothier, Z.A. B.P. 26, 49590 Fontevraud L’Abbaye, France Received July 11, 2010; Revised Manuscript Received September 13, 2010

A new reversible gelation pathway is described for alginates in aqueous media. From various samples differing by their mannuronic/guluronic content (M/G), both enthalpic and viscoelastic experiments demonstrate that alginates having a high M content are able to form thermoreversible assemblies in the presence of potassium salts. The aggregation behavior is driven by the low solubility of M-blocks at low temperature and high ionic strength. In semidilute solutions, responsive assemblies induce a strong increase of the viscosity below a critical temperature. A true physical gel is obtained in the entangled regime, although the length scale of specific interactions between M-blocks decreases with increasing density of entanglements. Cold setting takes place at low temperatures, below 0 °C for potassium concentrations lower than 0.2 mol/kg, but the aggregation process can be easily shifted to higher temperatures by increasing the salt concentration. The self-assembling process of alginates in solution of potassium salts is characterized by a sharp gelation exotherm and a broad melting endotherm with a large hysteresis of 20-30 °C between the transition temperatures. The viscoelastic properties of alginate gels in potassium salts closely depend on thermal treatment (rate of cooling, time, and temperature of storage), polymer and salt concentrations, and monomer composition as well. In the case of alginates with a high G content, a similar aggregation behavior is also evidenced at higher salt concentrations, but the extent of the self-assembling process remains too weak to develop a true gelation behavior in solution.

Introduction Alginate is a water-soluble linear polysaccharide extracted from brown seaweed and consisting of 1-4 linked R-L-guluronic (G) and β-D-mannuronic (M) acid residues. It may be regarded as a block copolymer with M and G units arranged in a nonregular blockwise pattern of varying proportions of homopolymeric GG, MM, and heteropolymeric alternating MG sequences.1-3 Alginates are one of the main structural components in marine brown algae and, like pectins in plants, they play an important physiological role in the intercellular matrix, forming hydrogels in the presence of divalent cations and bringing mechanical strength and flexibility.4 For alginates and pectins, the gelation properties are induced by G-blocks (guluronic in alginates and galacturonic in pectins) that are able to bind cooperatively divalent cations forming the so-called eggbox model described earlier by Grant et al.5 This intrinsic ability to form ionotropic gels in the presence of multivalent cations, combined with their specific biological properties, biocompatibility, and low toxicity, have led to important developments in food, medicine, and pharmacy. For instance, their gelling, viscosifying, and stabilizing properties have been used for decades to improve and modify the texture of foods or to prepare wound dressing or dental impression materials.6-9 There is actually growing amount of literature dealing with applications of alginate in drug release or tissue engineering where they form very useful membranes and matrices.10-14 As commercial alginates are produced from different brown seaweed sources, their molecular weight and composition, that * To whom correspondence should be addressed. Tel.: +33 (0)1 40 79 46 43. Fax: +33 (0)1 40 79 46 86. E-mail: [email protected]. † Physico-Chimie des Polyme`res et des Milieux Disperse´s. ‡ Laboratoires Brothier.

is, the proportion of polymeric sequences (GG, MM, and GM), vary widely according to geographical, seasonal, and growth conditions and strongly impact the solution properties and gelation ability.3 Contrary to other polysaccharide forming gels in an aqueous environment, like agarose, carrageenan, gellan, or pectins, gels of calcium alginate are almost temperature independent,15 and this makes them very suitable in the design of stable soft materials as an immobilization matrix for living cells. Nevertheless, as the ionotropic gelation is very fast, it could be necessary to introduce Ca2+ under inactivated form (CaCO3 or complexed Ca2+) followed by addition of the slowly hydrolyzing D-gluconoδ-lactone (GDL) to get homogeneous samples of a well-defined shape.3,16 Alginates may also form acid gels at pH values below the pKa of their uronic residues (pKa ) 3.4 and 3.6 for M and G units, respectively), but again, the pH must be lowered in a controlled fashion with GDL, as direct addition of acid to sodium alginate leads to an instantaneous precipitation rather than a gel.1 Apart from the specific binding of multivalent cations, which has attracted most of the research activity on alginates, the literature remains very scarce with other ionotropic gelation, and in particular, there are very few works reporting the ability of alginate to form gels in the presence of monovalent cations. Of course, it is well-known that any change of ionic strength in an alginate solution has a profound effect on polymer behavior, especially on the polymer chain conformation, its solubility, and its solution properties. For instance, Haug has reported earlier that alginate may be precipitated and fractionated to give a precipitate enriched in mannuronate residues using high concentrations of inorganic salts like potassium chloride.17,18 In that case, we can imagine that the border between precipitation and gelation is not very sharp as phase separation generally

10.1021/bm100776b  2010 American Chemical Society Published on Web 10/08/2010

Cold Gelation of Alginates

Biomacromolecules, Vol. 11, No. 11, 2010

Table 1. Characteristics of Alginates sample

Mn (kg/mol)

Mw (kg/mol)

Ð

[η] (mL/g)

M/G

ALG-1 ALG-2 ALG-3 ALG-4

40 175 160 85

200 415 385 225

5 2.4 2.4 2.6

350 400 470 340

1.6 0.6 0.5 0.5

gives rise to a rich polymer phase of high viscosity. We must also mention the reference work of Seale et al.19 which have reported that sodium ions may be capable of stabilizing intermolecular “egg-box” junctions analogous to those formed with divalent cations but of considerably shorter time-scale and lower binding energy. Under forcing conditions (high concentrations of Na+; high content of polyguluronate, freezing and thawing), a gel-like structure has been evidenced by the authors at room temperature but not really investigated from a viscoelastic point of view. They conclude that alginate can interact in solution with univalent cations according to three different modes: (a) ion-pair formation with carboxyl groups of mannuronate and isolated guluronate residues; (b) specific site-binding to contiguous guluronate residues; and (c) co-operative “eggbox” binding, particularly of Na+, between poly-L-guluronate chain sequences. Recently, we found that alginates with a high mannuronic content were able to form thermotropic gels under ionized form in the presence of monovalent cations. To the best of our knowledge, this is the first example of cold gelation in alginate solutions, and the aim of the present paper is to describe this new and original behavior, taking into account the influence of salt and polymer concentrations and the impact of the thermal treatment on the reversible sol/gel transition and related viscoelastic properties.

2967

experiments were carried out in the linear viscoelastic regime, which was determined for each sample by a preliminary stress sweep at 1 Hz. The temperature was controlled by a Peltier Plate temperature system that provided fast and precise adjustment of the temperature during heating and cooling stages with a temperature accuracy of (0.1 °C. Two oscillation temperature sweeps have been performed, either at constant frequency equal to 1 Hz or at multiple frequencies (1, 2.2, 4.6, and 10 Hz). We have checked that the shear stress value equal to 2 Pa applied during the tests is in the linear viscoelastic plateau in the temperature range 2-60 °C. Particular care was taken to avoid the drying of the sample by using a homemade cover that prevents water evaporation during the experiment. In these conditions, storage and elastic moduli, as well as complex viscosity, were determined between 2 and 60 °C. The following standard procedure was applied for all the samples after a waiting time of 30 min at 2 °C: step 1, heating from 2 to 60 °C; step 2, cooling from 60 to 2 °C, with a waiting time of 10 min at 2 °C; step 3, heating from 2 to 60 °C. Heating and cooling scans were performed at 2 °C · min-1 for all the samples, but slower rates (0.5 and 1 °C · min-1) were also used to investigate the impact of heating and cooling procedures on the sol/gel transition. The reproducibility was checked for each experiment. DSC. Phase transition of alginates in aqueous solutions was studied by differential scanning calorimetry (DSC) with a microDSCIII from Setaram. Solutions of approximately 0.6-0.8 g, equilibrated with a reference filled with the same weight of solvent, were submitted to temperature cycles between -5 and 60 °C, with heating and cooling rates of 0.1 to 0.5 °C · min-1. For each sample, the standard procedure was the following: (i) cooling from 20 to -5 °C (0.5 °C · min-1); (ii) isotherm at -5 °C (30 min); (iii) step 1, heating from -5 to 60 °C (0.5 °C · min-1); (iv) isotherm at 60 °C (10 min); (v) step 2, cooling from 60 to -5 °C (0.5 °C · min-1); (vi) isotherm at -5 °C (30 min); (vii) step 3, heating from -5 to 60 °C (0.5 °C · min-1); and repetition of cycles (stages iv-vii), applying different heating and cooling rates.

Results Experimental Section Chemicals. All inorganic salts, potassium carbonate (K2CO3 from Aldrich), potassium chloride (KCl from Fluka), and calcium chloride (CaCl2 from Prolabo) were used as received without further purification. The same is true for ammonium citrate (Citrate) purchased from Aldrich. Water was purified with a Millipore system combining inverse osmosis membrane (Milli-RO) and ion exchange resins (Milli-Q). Sodium alginates (ALG) were kindly given by Brothier Laboratories (Fontevraud, France). The average molar masses, molar mass dipersity (Ð), and intrinsic viscosity ([η]) were characterized by SEC using a Viscotek system equipped with three Shodex OH-pak columns equilibrated at T ) 35 °C in aqueous solution (NaNO3 0.5 mol/L). The three detectors in line (refractometry, viscometry, and light scattering) allow an absolute characterization of the molar masses after an initial calibration of the columns with PEO standards. The macromolecular characteristics are reported in Table 1, as well as the average composition of mannuronic/guluronic (M/G) obtained by 1H NMR in D2O at 80 °C. Preparation of Alginate Solutions. All the studies were performed with semidilute solutions, either in pure water or in salted media: K2CO3 or KCl. For each solution, the polymer was allowed to dissolve in the salted medium on a shaking table (250 rpm) at room temperature for at least 24 h, until homogeneous solutions were obtained. If necessary, solutions were heated at 50 °C for a short time to complete the dissolution process. The polymer concentrations are given in weight percentage (wt%) and the salt concentration is expressed in moles of salt per kg of solution (m). Rheology. The viscoelastic properties of alginates were studied in semidilute solutions using a stress-controlled rheometer (Haake RS600) equipped with a rough stainless steel cone (diameter 35 mm, angle 2°, truncature 103 µm) and smooth stainless steel Peltier plate. The

Sol/Gel Transition of Alginates in the Presence of Potassium Carbonate. As previously discussed, gelation of alginates can be induced either at low pH, playing with hydrogen bonds between carboxylic acids and hydroxyl groups, or above the pKa (under ionized form) by complexation of carboxylate groups using multivalent cations. In the present study we started our investigations on alginate aqueous solutions containing a relatively high concentration of potassium carbonate: 0.3 m. The choice of this salt originates from our initial work on thermoresponsive alginates, which describes the associating properties of these polysaccharides grafted with poly(ethylene oxide-copropylene oxide) side chains [PEPO]. As reported in a preceding paper,20 the associating properties of these graft copolymers in semidilute solution were observed at high temperature in pure water, and the self-assembling process of PEPO side chains in aqueous solution was shifted at lower temperatures by adding salts that strongly impact the solubility of PEPO. This was originally done with potassium sulfate in the former study (mainly below 0.3 m), but we also reproduced the same experiments with potassium carbonate and we found that high concentrations of salt also greatly impact the solution behavior of the backbone itself. When placed in conditions of high ionic strength, alginates can be reasonably dissolved with time (24 h) at room temperature and form macroscopically homogeneous solutions. Nevertheless, depending on alginate and salt concentration, all the polymer chains are not molecularly dispersed, even in dilute conditions, and a short heating stage of a few minutes above 50 °C is often necessary to complete the dissolution process. For that reason we have applied a specific procedure to

2968


Figure 1. Temperature dependence of the complex viscosity (f ) 1 Hz) for alginate solutions (C ) 5 wt %) in [K2CO3] ) 0.3 m: ALG-1 (b), ALG-2 (2), ALG-3 (O), and ALG-4 (4). For ALG-1, the viscosity behavior is also shown during pretreatment (step 1: from 2 to 60 °C; see insert), cooling (step 2: from 60 to 2 °C), and heating (step 3: from 2 to 60 °C; scanning rates ) 2 °C/min).

investigate the viscoelastic properties of alginate solutions and their temperature dependence. As already described in the Experimental Section, the solutions undergo four different scanning steps: isotherm at 2 °C during 30 min, step 1, from 2 to 60 °C, step 2, from 60 to 2 °C (waiting time of 10 min at 2 °C), and step 3, from 2 to 60 °C (scanning rates ) 2 °C/min). An example is given in Figure 1 for different alginates in [K2CO3] ) 0.3 m. For most alginate samples, characterized by a high guluronic content (M/G < 1), the viscosity simply follows an Arrhenian behavior with an activation energy for a viscous flow of about 25-27 kJ/mol, as expected for aqueous polymer solutions at this concentration. In these cases, the three scanning steps give exactly the same results. By comparison, the behavior of ALG-1 is much more complex in [K2CO3] ) 0.3 m. During the first heating step (step 1 in the inset of Figure 1), the viscosity of ALG-1 decreases more rapidly than that of previous samples and more abruptly between 20-25 °C and then 35-40 °C. During the cooling, step 2, the viscosity slightly increases but remains very low between 60 and 10 °C and then rapidly increases between 10 and 2 °C (almost two decades). During the second heating, step 3, the solution remains highly viscous up to 30 °C and then the viscosity falls down and recovers its initial low level above 40 °C. This first set of data clearly points out that the alginate sample ALG-1, having a high mannuronic content, is able to form “thermoreversible gels” or at least thermoviscosifying solutions in the presence of potassium carbonate with an important hysteresis of more than 20 °C. It is noteworthy to mention that the same solution becomes slightly turbid after cooling in the “gel state”. If we exclude the first heating step (step 1), which takes into account the history of the sample (preparation and storage), the viscoelastic data are highly reproducible from one solution to another and fully superimposable when applying several cooling/heating cycles at the same scanning rate. As the cooling rate impacts the gelation kinetics, we will concentrate on the following steps 2 and 3, which are kinetically controlled at 2 °C/min. To investigate more accurately the transition behavior of ALG-1 in [K2CO3] ) 0.3 m, we have plotted in Figure 2 the temperature dependence of dynamic moduli determined at constant frequency (1 Hz). As we can see, this representation of G′ and G′′ clearly enlightens the sol/gel transition of the formulation with a liquid behavior at high temperature (60 °C), where the loss modulus is more than 20 times higher than the elastic one, and a gel-

Karakasyan et al.

Figure 2. Temperature dependence of elastic (G′: O) and loss (G′′: b) moduli (f ) 1 Hz) of ALG-1 solution: C ) 5 wt % and [K2CO3] ) 0.3 m. Heating and cooling rates ) 2 °C/min: 60-2 °C (steps 2) and 2-60 °C (steps 3).

Figure 3. (a) Temperature dependence of tan δ for ALG-1 solution (C ) 5 wt % and [K2CO3] ) 0.3 m). The symbols are related to experiments performed at different temperatures and different frequencies (f ) 1 Hz (b), 2.2 Hz (O), 4.6 Hz (2), and 10 Hz (4)) during cooling (step 2: 60-2 °C) and heating (step 3: 2-60 °C), while the gray thick line corresponds to scanning experiments performed at 1 Hz (cooling and heating rates ) 2 °C/min). (b) First derivative of the temperature dependence of log(tan δ) calculated from the scanning experiment.

like behavior at very low temperature (2 °C) where oppositely G′ is more than five times higher than G′′. In between these two temperatures, the solution exhibits a cold gelation with an important hysteresis. If we use G′ ) G′′ as a simple criterion to define the limits between elastic and liquid states, two transition temperatures can be determined: one for the formation of the gel at low temperature (Ts-g ) 6.5 °C; cold gelation) and the other one for the collapse of elastic properties upon heating (Tg-s ) 32 °C). These transitions are emphasized in Figure 3 where the temperature dependence of the loss tangent (tan δ ) G′′/G′) has been plotted using two different set of experiments. The first set is the scanning procedure already applied to characterize the dynamic moduli and complex viscosity in Figures 1 and 2. The scanning plot of tan δ versus temperature and its first derivative (gray thick lines in Figure 3a,b) allow to define the borderlines of the transitions: Tass ) 8.5 °C, which corresponds to the beginning of the association process observed during cooling, Tdis-1 ) 28 °C and Tdis-2 ) 38.5 °C that are, respectively, the beginning and the end of the dissociation process observed upon heating. All these temperatures, which are related to the experimental procedure used in dynamic



2969

Table 2. Transition Temperatures of ALG-1 in Potassium Carbonate Aqueous Solutions from dynamic experiments C (wt%)

[K2CO3] (m)

1 2 3 5 7 9 5 5 5 5

0.3 0.3 0.3 0.3 0.3 0.3 0.2 0.3 0.4 0.5

Ts-g (°C) 4.5 6.5 6.5 4 3 6.5 10.5 17

Tass (°C)

Tdis-1 (°C)

9 9.5 9.5 8.5 6.5 6.5 17 28 33 42.5

from DSC

Tg-s (°C)

Tdis-2 (°C)

Tg-s - Ts-g (°C)

30.5 34 32 29 27.5

41 41 41 38.5 34.5 35

26 27.5 25.5 25 24.5

32 38.5 44.5

38.5 45 54

25.5 28 27.5

measurements (heating and cooling rates ) 2 °C/min) are reported in Table 2. Concerning the exact value of the sol/gel transition itself, we can use, as previously discussed, the criterion tan δ ) 1, although this definition is rather arbitrary, to separate the liquid state from the gel state, as the experiments were carried out at a fixed frequency (1 Hz). If we follow the arguments of Winter and Chambon,21 the critical point would theoretically correspond to the invariancy of tan δ with frequency. Such an experiment was tested by measuring the frequency dependence of dynamic moduli at different temperatures. Despite the short frequency range investigated (1-10 Hz) and the large gap between the temperatures under analysis (∆T = 5 °C), we can see in Figure 3a that this condition is fulfilled at about Ts-g ) 8 °C upon cooling and Tg-s ) 35 °C upon heating. These critical temperatures are very close to the previous ones obtained at tan δ ) 1 or with those that we can get from the first derivative of the semilogarithmic plot of tan δ versus temperature (T1 ) 6.5 °C and T2 ) 34 °C in Figure 3b). In the following we will keep the simple criterion tan δ ) 1 to approximate the sol/gel transition of alginate solutions. As it will be discussed later, we can notice that in the gel state the dynamic modulus (G′) is almost constant in the whole frequency range investigated (10-3 to 100 rad/s). That means that the relaxation time of alginate chains becomes very high after cold setting and that the gel keeps its macroscopic integrity and does not flow for at least several hours. DSC. Using the same solution of ALG-1 (C ) 5 wt % in [K2CO3] ) 0.3 m), we have reproduced the previous scanning experiments by DSC but at different rates (between 0.1 and 0.5 °C/min), as the scanning procedure is limited by the thermal inertia of inox cells filled with a large amount of solution (0.6-0.8 g). From typical thermograms given in Figure 4, we can see that cold gelation is an exothermic process that slightly depends on the rate of cooling. The beginning of the transition varies from 11 to 9 °C when the cooling rate increases from 0.1 to 0.5 °C/min, respectively, while the transition enthalpy remains independent of the scanning rate and close to 5 J · g-1. On the other hand, the “melting” process gives a broad endotherm that does not depend on the heating rate (between 0.1 and 0.5 °C/min), with a maximum located at 34 °C. Nevertheless, the endotherm is strongly correlated to the history of the sample and more specifically to dissolution and cold gelation processes. For gelation experiments performed at a fixed temperature after cooling down the solution from 60 °C, we can see in Figure 4 that the endotherm is exactly the same if the solution is maintained for half an hour or more at -5 or 2 °C. In these conditions, we can reasonably assume that the thermodynamic transition is fully achieved. On the other hand, the association process is far from complete when gelation

TMcool (°C)

∆Hgel (J/g)

TMheat (°C)

6.3

-5

37.1

6.7

-5.1

34.2

4.9 -2 6.7 13.5 21.5

-4.5

30.2 21.7 34.2 43.1 50.7

-5.1 -5 -4

is carried out at higher temperatures, 8 or 10 °C, even for longer times. The impact of the scanning rate on cold gelation is also pointed out in Figure 5a with the temperature dependence of the loss tangent. In that case, an increase in the scanning rate slightly shifts the gelation threshold to lower temperatures with a small increase of the elastic modulus or relative decrease of the loss tangent. From a general point of view, the comparison in Figure 5b between calorimetric and viscoelasticity experiments carried out at the same scanning rate (0.5 °C/min) clearly points out the tight correlation between thermodynamic and macroscopic events. Influence of Polymer Concentration. As ALG-1 is able to form gels in [K2CO3] ) 0.3 m, we investigate the impact of

Figure 4. Thermograms of ALG-1 solution (C ) 5 wt % and [K2CO3] ) 0.3 m) obtained at different scanning rates during cooling (step 2): 0.5 °C/min (solid line), 0.3 °C/min (long dash), 0.2 °C/min (short dash), 0.1 °C/min (dotted) and at the same scanning rate (0.5 °C/min) during heating (step 3) after 30 min at -5 °C (thick line), 30 min at 2 °C (4), 120 min at 8 °C (b), or 120 min at 10 °C (O).

Figure 5. Inset a: Influence of scanning rate on viscoelastic properties of ALG-1 (C ) 5 wt %, [K2CO3] ) 0.3 m, and f ) 1 Hz). (b) Comparison between enthalpic and viscoelastic transitions investigated at the same scanning rate (0.5 °C/min during cooling (step 2: 60-2 °C) and heating (step 3: 2-60 °C).

2970


Karakasyan et al.

Figure 6. Temperature dependence of dynamic moduli (G′ (O) and G′′ (b)) of ALG-1 solutions ([K2CO3] ) 0.3 m) at different polymer concentrations (f ) 1 Hz).

Figure 7. Temperature dependence of the loss tangent (tan δ; f ) 1 Hz) of ALG-1 solutions ([K2CO3] ) 0.3 m) at different polymer concentrations during cooling (step 2: 60-2 °C) and heating (step 3: 2-60 °C): C ) 3 (0); 5 (9); 7 (4); and 9 wt % (2).

polymer concentration on viscoelastic properties (Figure 6). If we exclude the solution at 1 wt %, which shows an abrupt transition of viscosity upon cooling, but no gelation (G′ < G′′), all the other solutions prepared at higher concentrations clearly exhibit a sol/gel transition upon cooling. They also show a true gel behavior at low temperature with G′ independent of the frequency and dominating G′′. A rapid comparison of viscoelastic data also displays that the increase in the elastic modulus during gelation is very high for C ) 2 wt % (more than three decades) and then progressively decreases with increasing concentration. At 9 wt %, the increase in G′ between 10 and 2 °C is less than two decades. This characteristic feature is emphasized in Figure 7, where we can see that the thermal loop of the loss tangent is progressively reduced by increasing the polymer concentration. To point out how kinetics can influence the formation of physical networks, we have followed with time the evolution of dynamic moduli at 2 °C after a scanning experiment performed between 60 and 2 °C at 2 °C/min. As shown in Figure

Figure 8. Time dependence of dynamic moduli (G′: hollow symbols) and G′′ (filled symbols) after cold gelation of ALG-1 solutions in [K2CO3] ) 0.3 m from 60 to 2 °C at 2 °C/min: C ) 3 (3,1), 5 (0,9), 7 (O,b), and 9 wt % (4,2). The frequency dependence of moduli, obtained after 1 h setting at 2 °C, is also shown for C ) 5 wt %.

8, the elastic modulus clearly displays a time dependence that increases with increasing concentration, while the loss modulus remains almost constant with time. Kinetic effects on gelation are not negligible, especially at 9 wt %, where G′ increases from 1500 to 3400 Pa after 1 h, but they are much weaker than in the case of gelatin22 and they cannot explain on their own the lower enhancement of viscoelastic properties already described from Figures 6 and 7. Here the steric hindrance coming from chain entanglements is very likely responsible for the slowing down of the gelation process with increasing concentration. We have also plotted in Figure 8 the frequency dependence of dynamic moduli determined for C ) 5 wt % at 2 °C after 1 h setting at this temperature. As we can see, the elastic


Figure 9. Concentration dependence of the plateau modulus (G0) at 2 °C and the Newtonian viscosity (η0) at 60 °C for ALG-1 solutions prepared in [K2CO3] ) 0.3 m. G0 values were determined just after reaching 2 °C (3) or after a 1 h rest at 2 °C (4).

modulus G′ is almost constant in the whole frequency range investigated, and the value of the plateau modulus (G0 ) G′ in the gel state) can be used to describe the intrinsic properties of the physical network. In this association process, we can conclude that the lifetime of local interactions taking place between alginates dramatically increases the terminal relaxation time of individual chains. From the data given in Figure 8, we can argue that, once the gel is formed by cooling, it is able to retain its macroscopic integrity at least for hours or even more. While G0 is the right parameter to describe the polymer properties in the gel sate, it becomes meaningless in the liquid state, and only the zero shear viscosity (η0) has to be considered in that case. Conversely, η0 diverges and becomes meaningless in the gel state. At high temperature, typically above 40 °C, the alginate solutions are in the “sol” state and exhibit a Newtonian behavior at least for shear rates lower than 100 s-1. That means that, in these conditions (T > 40 °C), the complex viscosity calculated at f ) 1 Hz (η*) can be fully identified with the zero shear viscosity (η* ) η0) and can be used to describe the solution properties in the liquid state. Using the data given in Figure 6, we have plotted in Figure 9 the concentration dependence of the polymer solution in the liquid state (η0 at 60 °C) and in the gel state (G0 at 2 °C). In the latter case, we have also included the data obtained after 1 h setting at 2 °C, which are assumed to be closer to the equilibrium (see Figure 8). Two scaling relations are obtained for concentrations above 1 wt %. At high temperature, the Newtonian viscosity of ALG-1 solutions follows a concentration dependence with a scaling exponent of 3.2. In the scaling theory of polyelectrolyte solutions developed by Dobrynin et al.,23 the prediction for scaling exponents are 1/2 (in pure water) and 5/4 (at high ionic strength) in the unentangled regime and 3/2 (in pure water) and 15/4 (at high ionic strength) in the entangled regime. By comparison with our experimental data we can obviously conclude that ALG-1 solutions become entangled above 1 wt % and that alginates behave as neutral chains due to the high screening of electrostatic interactions in [K2CO3] ) 0.3 m. Similarly, we obtain a scaling exponent of 2.6 for the elastic modulus in the gel state (2 °C), which is quite close to the theoretical value of 9/4 predicted for entangled polyelectrolyte solutions at high ionic strength (3/2 in pure water). Such agreement obtained in the gel state means that the new physical cross-links that have been created during cooling do not significantly modify the elasticity of the network, which continues to be dominated by entanglements (G0 = CRT/Ne, with Ne equal to the number of monomers in an entanglement strand). As theoretically described in the sticky reptation


2971

Figure 10. Thermograms of ALG-1 solutions in [K2CO3] ) 0.3 m obtained at different concentrations using the same scanning rates (0.5 °C/min) during cooling (step 2) and heating (step 3) after a 30 min setting at -5 °C: C ) 2 (medium dash), 5 (solid line), and 9 wt % (dotted).

Figure 11. Temperature dependence of the loss tangent (tan δ at f ) 1 Hz) of ALG-1 solutions (C ) 5 wt %) at different salt concentration: [K2CO3] ) 0.2 (O), 0.3 (b), 0.4 (0) and 0.5 m (9).

model,24,25 the main change of viscoelastic properties originates from a large increase in the terminal relaxation time of chains that scale in the associating regime with the number of physical interactions per chain (number of stickers) and their lifetime. It is worth mentioning that a lot of gelling systems, either synthetic or natural,26-28 exhibit scaling relations in the gel state, with a scaling exponent close to 2 (G ∼ C2). While similar criterions based on screening length or entanglements have been used to account for this behavior, other theories developed for rigid or cellular networks also support this concentration dependence.29,30 The thermograms of aqueous solutions of ALG-1 given in Figure 10 and their characteristics reported in Table 2 confirm the concentration dependence of the sol/gel transition. This is particularly obvious during melting, where the maximum of the endotherm is progressively shifted to higher temperature with decreasing concentrations: TM,heat ) 30, 34, and 37 °C for C ) 9, 5, and 2 wt %, respectively. These calorimetric data are in line with viscoelastic measurements where the end of the melting process (Tdis-2 in Table 2) also decreases from 41 to 35 °C between 1 and 9 wt %. This feature can be attributed to the reduced mobility of the chains in the entangled regime, due to physical constraints, and to the formation of smaller aggregates (smaller junction zones) with increasing alginate concentration and decreasing correlation length (ξ ∼ C-3/4 in conditions of high ionic strength). Influence of Salt Concentration. As shown in Figure 11, the concentration of potassium carbonate mainly impacts the transition temperature and much less the viscoelastic properties. At low salt concentration, typically for [K2CO3] e 0.2 m, the sol/gel transition is not really observed in the experimental

2972


Figure 12. Thermograms of ALG-1 solutions (C ) 5 wt %) obtained at different salt concentrations using the same scanning rates (0.5 °C/min) during cooling (step 2) and heating (step 3): [K2CO3] ) 0.2 m (solid thin line), 0.3 m (medium dash), 0.4 m (dotted), and 0.5 m (thick line).

conditions set up for rheology. Nevertheless cold gelation readily takes place, particularly for ALG-1 in [K2CO3] ) 0.2 m when the solution is stored for several hours in the fridge at T ) 1-2 °C. This can be observed visually by the formation of a turbid gel in these conditions and experimentally by DSC, with the exotherm starting at about 0 °C when the solution is cooled down at 0.5 °C/min (Figure 12). As reported in Table 2, the transition temperatures increase with increasing salt concentration. This very general behavior, already reported for other charged gelling biopolymers like carageenans or gellan,31-33 is attributed to the screening of electrostatic repulsions between polyelectrolyte backbones. In the case of κ-carageenan, studied in potassium chloride solutions,31 it was shown that double helices, formed upon cooling, only start to form larger aggregates responsible for the gelation process above a critical salt concentration (Cs ) 7 × 10-3 eq/ L). Similarly, it was observed that the transition temperatures and the hysteresis were also increasing with salt concentration. In the case of alginates, where we have mainly worked in the high salt concentration regime ([K2CO3] ) 0.1-0.5 m), we do not really observe a strong dependence of the hysteresis with salt concentration. The width of the hysteresis obtained from viscoelastic data remains almost constant (Tg-s - Ts-g = 25-28 °C) between [K2CO3] ) 0.2 and 0.5 m, while calorimetry gives a similar weak dependence in the same interval (TMheat - TMcool = 24 to 29 °C). The salt dependence of associating properties of ALG-1 solutions can be summarized by comparing Figures 11 and 13, where the loss tangent determined at different temperatures during the cooling process has been plotted versus the salt concentration. For the polymer solution at 5 wt %, one can find either a critical temperature for a given salt concentration or conversely a critical salt concentration for a given temperature in the range 0-100 °C. The effects of salt and temperature are nevertheless not really comparable in terms of structure and dynamics of the physical network. For instance, the thermograms given in Figure 12 show that, with an increasing amount of added salt, the sol/gel transition takes place on a larger range of temperatures and the transition becomes less exothermic with increasing amounts of added salt. In other words, the cooperativity of the associating process decreases with ionic strength. Moreover, while the value of the plateau modulus remains almost constant for alginate gels prepared at T ) 2 °C in [K2CO3] ) 0.3 and 0.4 m (G0 ) 700-800 Pa), it drops to about 200 Pa when the same solution is prepared in [K2CO3] ) 0.5 m (see Figure 14). Obviously, there is a critical balance for controlling the structure/

Karakasyan et al.

Figure 13. Salt dependence (K2CO3) of the loss tangent (tan δ; f ) 1 Hz) of ALG-1 solutions (C ) 5 wt %; step 3) at different temperatures (°C): T ) 2 (O), 10 (b), 20 (4), 30 (2), 32 (3), 35 (1), 38 (0), 40 (9), 42 (]), 45 ([), 50 (f), and 60 (`).

properties relationship of alginate gels. Indeed, while a critical potassium salt concentration is needed to obtain the associating behavior above 0 °C, too much salt will extend the aggregation process from a microscopic scale to a macroscopic one, with a catastrophic phase separation expected at high salt concentration. Influence of the M/G Composition of Alginates. In Figure 1 we have initially shown that most alginate samples, except ALG-1, were not forming gels by cold setting in [K2CO3] ) 0.3 m. To explore the limits of these systems, we tried to boost the cold gelation process by increasing the salt concentration as much as possible. The results of this study performed in [K2CO3] ) 0.5 m are given in Figure 14. If we exclude ALG-1, which clearly shows a cold gelation process around 20 °C, the other samples, which are predominantly guluronic, display a very weak thermothickening effect at lower temperatures: 15 °C for ALG-2 and 10 °C for ALG-3 and ALG-4. These weak macroscopic transitions are also correlated with low energies, as determined by DSC: ∆H ) 4, 0.9, 0.35, and 0.3 J · g-1 for ALG-1, ALG-2, ALG-3, and ALG4, respectively. Obviously there is a close correlation between the gelation ability of alginates, the heat of the association process, and the amount of mannuronic units in the sample. Influence of the Nature of Salt. While potassium carbonate is able to drive specific interactions between alginates upon cooling, the nature of salt remains an important issue in this association process. As the cation itself is known to strongly impact the gelation behavior in polysaccharide solutions,34 we keep the potassium and simply investigate the effect of the anion. As shown in Figure 15, the viscoelastic properties of ALG-1 solutions at 5 wt % are not strongly modified as soon as we work at constant potassium concentration ([K+] ) 0.6 m) using either potassium carbonate or potassium chloride. In these conditions, where all carboxylic units can be considered as fully ionized, the pH of the solutions, about 7 for [KCl] ) 0.6 m and 11.7 for [K2CO3] ) 0.3 m, does not really influence the association process. The other point that has been considered in this section was the possibility that some residual calcium ions could influence the cold gelation mechanism. To investigate this hypothesis, ammonium citrate was added to the original solution of ALG-1 in [K2CO3] ) 0.3 m. Indeed, citrate is well-known to compete with alginate in the calcium complexation and it is often used to delay, hinder, or ruin the gel formation.3 In the present experiment, citrate was added in a stoichiometric amount with respect to uronic units. As observed in Figure 15, the viscoelastic properties of the solution of ALG-1, and typically



2973

Figure 14. Temperature dependence of dynamic moduli (G′ (O) and G′′ (b) at 1 Hz) of ALG solutions (C ) 5 wt %) in [K2CO3] ) 0.5 m (cooling and heating rates (60-2-60 °C) ) 2 °C/min) and thermograms obtained during cooling (0.5 °C/min).

Figure 15. Temperature dependence of the loss tangent (f ) 1 Hz) of ALG solutions (C ) 5 wt %) in different salt conditions: [K2CO3] ) 0.3 m (b); [K2CO3] ) 0.3 m + [citrate] ) 0.25 m (O); and [KCl] ) 0.6 m (gray triangle). Heating and cooling rates ) 2 °C/min: 60-2 °C (step 2) and 2-60 °C (step 3).

the gelation loop, are simply shifted toward higher temperature (∆T ) 4-6 °C) as a result of an increasing ionic strength. Similarly, the addition of a small amount of calcium chloride into a 5 wt % solution of ALG-4 (Ca2+/COO- ) 1/8), initially dissolved in [K2CO3] ) 0.5 m, does not change at all the viscoelastic behavior previously reported in Figure 14.

Discussion Rheology and calorimetry are complementary techniques that have been widely used to investigate biopolymers like gelatin, carrageenans, agarose, galactomannans, and other gelling systems.22,35-37 The main purpose is generally to correlate the macroscopic modifications induced by changing the temperature to enthalpic events as phase separation, crystallization, or conformational transition. In the framework of this new mech-

anism of cold gelation in alginates, we could draw some legitimate comparisons with a lot of systems, like pectins, for example, which have been shown to form gels in the presence of monovalent sodium or potassium salts.38,39 Nevertheless, we will pay more attention in the following to galactomannans, which afford some interesting similarities with mannuronic-rich alginates in terms of covalent architecture based on mannan blocks. For instance, the general structure of galactomannans, polysaccharides issued from plant seeds like locust bean, guar, or carob gums, is based on a 1,4-linked β-D-mannan backbone with 1,6-linked R-D-galactose side groups. Depending on the concentration, galactomannans having a high mannose content are known to form isolated aggregates or reversible gels at low temperature or through freeze/thaw cycling. For galactomannans, there is evidence for a correlation between the structure (i.e., degree of galactose substitution and the amount and length of unsubstituted mannose regions) and the ability to form gels at low water activity.37,40-44 The general model for the gel structure generally takes into account dense junction zones, issued from the aggregation of unsubstituted regions of the mannan backbone, alternating with solvated domains comprising more heavily substituted regions of the macromolecule. If we take into account the regular M-blocks in both galactomannans and alginates as the key feature for their solution similarities, the important difference between these two copolymers is the presence of ionizable carboxylic groups on mannuronic residues. At the solid state and under acid conditions, polymannuronic block adopts a flat ribbon-like two-fold conformation that is similar to that of other β-1,4 diequatorially linked hexosans, such as cellulose, chitin, and mannan, of course. On the other hand, under salt form, polymannuronic adopts a more extended and ribbon-like three-fold structure in the solid state, while polyguluronic chains maintain a buckled two-fold conformation in both solid and gel states under acid and salt forms.45-47 These

2974


Figure 16. Proposed model for cold gelation of high M-alginates in salt conditions.

differences are attributed to the high calculated energy difference between the two ordered forms (21 and 32) in the case of G-blocks (7.1 kJ · mol-1) compared to M-blocks (3.3 kJ · mol-1).48 Moreover, in pure water and under ionized form, the carboxylate groups and their mobile counterions provide a higher solubility to alginates compared to galactomannans, but this ionic contribution progressively vanishes when working at high ionic strength, as it is the case in the present study. Taking into account that at high salt concentration G-blocks keep a higher water solubility than M-blocks,2,17,18 we can draw some parallels between the general model used to describe the gel structure of galactomannans and the cold gelation behavior of ALG-1 (see Figure 16). From the hysteresis observed between cold setting and melting, we can reasonably assume that the gelation proceeds through a nucleation and growth process that is typical from first order transitions. As proposed in Figure 16, it would correspond to the self-association of M-blocks into critical clusters. Here, the trans-conformation of M-blocks from a random coil into a more rigid ribbon-like structure having a higher propensity to self-aggregate could be an interesting nucleation mechanism, but we clearly lack information to fully support this hypothesis. Whatever is the initiation process, the dense junction zones formed upon cooling will behave as physical knots, while the other G and MG blocks will maintain the solubility of the overall structure, preventing the polysaccharide network from a macroscopic phase separation at intermediate salt concentrations. Due to the polyelectrolyte behavior of alginates, the polymer chains are originally repulsive in pure water, but the attractive potential between M-blocks progressively increases with increasing ionic strength, leading to physical gels of enhanced thermal stability (increase of TMcool and TMheat). The association process taking place in salty conditions is observed for a large range of polymer concentrations, but it turns out that a true gel state can be obtained only in the entangled regime. This presumes that intrachain associations are likely to occur and to dominate the viscoelastic properties at low concentration, even in the semidilute unentangled regime. Nevertheless, it appears that kinetics and thermodynamic processes become less favorable at high concentration due to steric constraints inside the entangled polymer matrix. The decreasing enthalpy observed with increasing concentration in Figure 10 points out the formation of associating clusters of smaller size. From setting experiments carried out at different temperatures (Figure 4), it was shown that there was a broad range of melting

Karakasyan et al.

points involving aggregates of different sizes and M-blocks of different lengths. While the rate of cooling does not really impact the melting process and the morphology of clusters when the solution is cooled down at sufficiently low temperatures, the situation becomes different when the association takes place at “higher” temperatures. For instance, when an isothermal gelation is carried out at 10 °C for several hours, the melting endotherm becomes smaller, as it is for the fraction of aggregated chains, but it mainly corresponds to the superstructures of the highest melting points. That means that the small clusters cannot be formed in these conditions and probably that the length of M-blocks is the critical parameter in the nucleation and growth process. Finally, in the case of low M/G alginates, one can assume that not only the fraction of M-blocks but also their average length are smaller, providing a weaker responsivity to ionic strength and temperature. In that case, the association process can be driven only at very high ionic strength, but the density of physical cross-links remains too small to induce gelation, even in the entangled regime (Figure 14).

Conclusion Alginates with a high mannuronic content are able to form thermoreversible assemblies in aqueous solutions in the presence of monovalent cations. The aggregation behavior is driven by the low solubility of M-blocks at low temperature and high ionic strength. In semidilute solutions, the self-assembling process of alginate chains triggered by cooling gives rise to a strong increase in the viscosity below a critical temperature. A true gel behavior is obtained in the entangled regime, although the length scale of specific interactions between M-blocks decreases with the increasing density of entanglements. Cold setting takes place at low temperature, below 0 °C for potassium concentrations lower than 0.2 molar, but gelation can be easily shifted at higher temperatures by increasing the salt concentration. The self-assembling setup observed during cooling is assumed to follow a nucleation and growth mechanism that is responsible for the large hysteresis observed between gelation and melting temperatures. The viscoelastic properties of potassium alginate gels closely depend on thermal treatment (rate of cooling, time, and temperature of storage) and polymer and salt concentrations. While most of calorimetric and viscoelastic experiments have been performed in the presence of potassium carbonate to draw the outlines of this new gelation mechanism, a lot of questions remain open concerning the existence of a conformational transition of M-blocks in the vicinity of the gel point as well as the real impact of salts (nature of cations for instance) or other cosolutes on thermodynamic, self-assembling, and viscoelastic properties. Such investigations are currently in progress and will be reported in forthcoming papers.

References and Notes (1) Haug, A.; Larsen, B.; Smidsrød, O. Acta Chem. Scand. 1966, 20, 183– 190. (2) Haug, A.; Larsen, B.; Smidsrød, O. Acta Chem. Scand. 1967, 21, 691– 704. (3) Draget, K. I. In Handbook of Hydrocolloids; Phillips, G. O., Williams, P. A., Eds.; CRC Press LLC: Boca Raton, FL, 2000; Chapter 22, pp 379-395. (4) Andresen, I.-L.; Skipnes, O.; Smidsrød, O.; Østgaard, K.; Hemmer, P. C. Some biological functions of matrix components in benthic algae in relation to their chemistry and the composition of seawater; ACS Symposium Series 48; American Chemical Society: Washington, DC; 1977; pp 361-381. (5) Grant, G. T.; Morris, E. R.; Rees, D. A.; Smirk, P. J. C.; Thom, D. FEBS Lett. 1973, 32, 195–198.

Cold Gelation of Alginates (6) Brownlee, I. A.; Seal, C. J.; Wilcox, M.; Dettmar, P. W.; Pearson, J. P. In Alginates: Biology and Applications; Rehm, B. H. A., Ed.; Microbiology Monographs Series 13; Springer Verlag: New York, 2009; pp 211-228. (7) Paul, W.; Sharma, C. P. Trends Biomater. Artif. Organs 2004, 18, 18–23. (8) Qin, Y. Polym. Int. 2008, 57, 171–180. (9) Cook, W. J. Biomed. Mater. Res. 1986, 20, 1–24. (10) Wang, L.; Shelton, R. M.; Cooper, P. R.; Lawson, M.; Triffitt, J. T.; Barralet, J. E. Biomaterials 2003, 24, 3475–3481. (11) Wong, M. In Biopolymer Methods in Tissue Engineering; Hollander, A. P., Hatton, P. V., Eds.; Methods in Molecular Biology Series 238; Humana Press Inc.: Totowa, NJ, 2004; pp 77-86. (12) Ma, P. X. In Scaffolding in Tissue Engineering; Ma, P. X., Elisseeff, J. , Eds.; CRC Press: Boca Raton, FL, 2006; Chapter 2, pp 13-26. (13) Augst, A. D.; Kong, H. J.; Mooney, D. J. Macromol. Biosci. 2006, 6, 623–633. (14) Sachan, N. K.; Pushkar, S.; Jha, A.; Bhattcharya, A. J. Pharm. Res. 2009, 2, 1191–1199. (15) Williams, P. A.; Phillips, G. O. In Handbook of Hydrocolloids; Phillips, G. O., Williams, P. A. , Eds.; CRC Press LLC: Boca Raton, FL, 2000; Chapter 1, pp 1-20. (16) Stokke, B. T.; Draget, K. I.; Smidsrød, O.; Yuguchi, Y.; Urakawa, H.; Kajiwara, K. Macromolecules 2000, 33, 1853–1863. (17) Haug, A. Acta Chem. Scand. 1959, 13, 601–603. (18) Haug, A. Acta Chem. Scand. 1959, 13, 1250–1251. (19) Seale, R.; Morris, E. R.; Rees, D. A. Carb. Res. 1982, 110, 101–112. (20) Karakasyan, C.; Lack, S.; Brunel, F.; Maingault, P.; Hourdet, D. Biomacromolecules 2008, 9, 2419–2429. (21) Winter, H. H.; Chambon, F. J. Rheol. 1986, 30, 367–382. (22) Djabourov, M.; Leblond, J.; Papon, P. J. Phys. (Paris) 1988, 49, 319– 332. (23) Dobrynin, A. V.; Colby, R. H.; Rubinstein, M. Macromolecules 1995, 28, 1859–1871. (24) Leibler, L.; Rubinstein, M.; Colby, R. H. Macromolecules 1991, 24, 4701–4707. (25) Rubinstein, M.; Semenov, A. N. Macromolecules 2001, 34, 1058– 1068. (26) Ferry, J. D. AdV. Protein Chem. 1948, 4, 1–78.


2975

(27) Fazel, Z.; Fazel, N.; Guenet, J.-M. J. Phys. (France) 1992, 2, 1745– 1754. (28) Rochas, C.; Landry, S. In Gums and Stabilizers for the Food Industry, version 4; Phillips, G. O., Wedlock, D. J., Williams, P. A., Eds.; IRL Press: Oxford, U.K., 1988; p 445. (29) Jones, J. L.; Marques, C. M. J. Phys. (Paris) 1990, 51, 1113–1127. (30) Gibson, L. J.; Ashby, M. F. In Cellular Solids: Structure and Properties; Pergamon Press: Elmsford, NY, 1988. (31) Rochas, C.; Rinaudo, M.; Landry, S. Carbohydr. Polym. 1990, 12, 255–266. (32) Nickerson, M. T.; Paulson, A. T. Carbohydr. Polym. 2004, 58, 15– 24. (33) Williams, P. A. Struct. Chem. 2009, 20, 299–308. (34) Rochas, C.; Rinaudo, M. Biopolymers 1980, 19, 1675–1687. (35) Rochas, C.; Rinaudo, M. Carbohydr. Res. 1982, 105, 227–236. (36) Guenet, J.-M. In ThermoreVersible Gelation of Polymers and Biopolymers; Guenet, J.-M., Ed.; Academic Press: New York, 1992. (37) Richardson, P. H.; Clark, A. H.; Russell, A. L.; Aymard, P.; Norton, I. T. Macromolecules 1999, 32, 1519–1527. (38) Yoo, S.-H.; Fishman, M. L.; Savary, B. J.; Hotchkiss, A. T. J. Agric. Food Chem. 2003, 51, 7410–7417. (39) Wehr, J. B.; Menzies, N. W.; Blamey, F. P. C. Food Hydrocolloid. 2004, 18, 375–378. (40) Dea, I. C. M.; Morris, E. R.; Rees, D. A.; Welsh, E. J. Carbohydr. Res. 1977, 57, 249–272. (41) Dea, I. C. M.; Clark, A. H.; McCleary, B. V. Carbohydr. Res. 1986, 147, 275–294. (42) Dea, I. C. M. Pure Appl. Chem. 1989, 61, 1315–1322. (43) Rinaudo, M. Food Hydrocolloid. 2001, 15, 433–440. (44) Patmore, J. V.; Goff, H. D.; Fernandes, S. Food Hydrocolloid. 2003, 17, 161–169. (45) Mackie, W. Biochem. J. 1971, 125, 89, part 4. (46) Kohn, R. Symp. Carbohydr. Chem. 1974, 371–397. (47) Rees, D. A.; Welsh, E. J. Angew. Chem., Int. Ed. Engl. 1977, 16, 214–224. (48) Braccini, I.; Grasso, R. P.; Pe´rez, S. Carbohydr. Res. 1999, 317, 119–130.

BM100776B

Cold Gelation of Alginates Induced by Monovalent Cations

Recommend Documents