2328
Biomacromolecules 2009, 10, 2328–2337
Porous Alginate Hydrogels: Synthetic Methods for Tailoring the Porous Texture Andrea Barbetta,* Elena Barigelli, and Mariella Dentini* Department of Chemistry, University of Rome “La Sapienza”, P.le A. Moro 5, 00185 Rome, Italy Received May 6, 2009; Revised Manuscript Received June 18, 2009
Alginate is a versatile, renewable biopolymer that has found numerous applications in diverse areas such as adsorbent materials of water pollutants and scaffolds for tissue engineering. In such kinds of applications the most convenient physical form of alginate-based materials is as porous matrices. The pore scale dimension has to be carefully engineered to meet the requirements posed by the specific application. The aim of this paper is to describe two synthetic methodologies that allow the preparation of alginate porous materials characterized by pores lying in well separated dimension ranges. One process is based on emulsion templating, which consists of dispersing an organic phase into an aqueous solution of alginate in the presence of a suitable emulsion stabilizer and locking in the structure of the continuous phase by chemical cross-linking. This approach required the preliminary degradation of alginate to reduce its molecular weight and, hence, the viscosity of the external phase of the concentrated emulsion. Porous matrices were characterized by pores and interconnects of about 10-20 and 2-5 µm, respectively, and a surface area of 230 m2/g. The second process consisted of replacing the organic, internal phase with a gas, namely, CO2, generated in situ the aqueous solution of alginate. The chemical reaction for CO2 generation, nature of the surfactant, and cross-linking method were carefully selected to give highly porous, stable matrices with pores and interconnects of the order of 300 and 80 µm, respectively.
Introduction Among polysaccharides, alginate, a block copolymer of 1,4β-D-mannuronate (M) and R-L-glucuronate (G) residues, is one of the most versatile from an application point of view. Two major fields of applications of alginate-based materials are as adsorbents for the elimination of heavy metals and organic pollutants from contaminated environments1-8 and as devices for tissue engineering and drug delivery.9-20 In the former type of application, alginate is one of the most extensively investigated biopolymers as it is inexpensive, nontoxic, and efficient. Insolubility in the majority of the organic solvents and the presence of numerous and different surface functionalities such as hydroxyl and carboxyl groups and excellent selectively for certain metals are the key features that make of alginate as a valid alternative to conventional methods of metal recovery.20,21 As a biomaterial, alginate has a number of advantageous features including biocompatibility and nonimmunogenicity, which are likely related to its hydrophilicity.22 Due to its nontoxicity, unique tissue compatibility, and biodegradability, alginate has been studied extensively in tissue engineering, including the regeneration of skin,23 cartilage,24,25 bone,26 liver,27 and cardiac tissue.28 A critical advantage is its gentle gelling behavior, which allows encapsulation of various substances with minimal trauma.29 A chemically modified alginate has been used clinically as a drug delivery vehicle for proteins that promote regeneration of mineralized tissue30 and as a carrier for transplanted cells.31 Furthermore, in recent years genetic engineering allowed to produce tailor-made alginates with definite relative amount of glucuronic and mannuronic residues arranged in the desired type of sequence.32-36 This opened up * To whom correspondence should be addressed. Tel.: +39-06-49913630 (A.B.); +39-06-49913630 (M.D.). Fax: +39-06-4457112 (A.D.); +39-064457112 (M.D.). E-mail:
[email protected] (A.D.); mariella.
[email protected] (M.D.).
the possibility to induce specific biological responses once the correlation between structure and biological activity is disclosed. From a practical point of view, in both kind of applications the more convenient physical form of the devices is as a porous support with the respective pore sizes lying in a different scale range. A suitable adsorbent for polar molecules or metal ions should present a high concentration of adsorption sites, high surface area, and an interconnected pore morphology for good circulation of the solution. To have a high surface area material, a significant fraction of mesopores (2-50 nm) should be present. In the case of scaffolds for tissue engineering applications it is generally accepted that pore and interconnects sizes should be at least 100 and 40 µm, respectively, to favor cell colonization of the entire scaffold volume, metabolic waste removal, etc. Two synthetic methodologies that allow satisfying the morphological requirements posed by the aforementioned type of applications is high internal phase emulsion (HIPE) and foam templating. A considerable number of publications dealt with the preparation and characterization of HIPE templated porous materials exhibiting high surface areas and with applications ranging from scavengers, catalysis support and photonic to quote a few.37-43 Unfortunately, so far, HIPE templating did not allow accessing to scaffolds characterized by pores and interconnects with average diameters >100 and 40 µm, respectively.44-50 As a consequence, there is the need of a complementary fabrication technique which permits the synthesis of materials characterized by larger pores and interconnects. It is well-known that gas blown plastics (foams) are often characterized by pores (either closed or open) of the order of 100s of µm. For instance, in polyurethane synthesis, the blowing gas (CO2) is developed in situ as a byproduct during the condensation reaction between an di-isocyanide and a dicarboxylic reagent. We envisaged that a similar approach could be adapted to our purposes. In other words, the liquid phase that represents the discontinuous phase
10.1021/bm900517q CCC: $40.75 2009 American Chemical Society Published on Web 07/10/2009
Porous Alginate Hydrogels
in HIPEs could be replaced by an inert gas generated within the aqueous solution of alginate. Such a gas development should be gradual to mimic the dropwise addition of the internal phase during HIPE making. The present article aims at showing how alginate porous materials characterized by very different scale dimension of pores and interconnects can be obtained by employing either a classical HIPE approach and a new, innovative high internal phase foam (HIPF) approach.
Experimental Section Materials. Two types of alginates were used in this work: Laminaria hyperborea (Stipe LF 10/60 S12727, with a fraction of guluronic acid, FG ) 0.65), which was kindly provided by Prof. Skjak-Bræk of the University of Trondheim, Norway, and a commercial alginate provided by Fluka, FG ) 0.40. Triton X-405 (solution 70% w/v, Aldrich), ethylendiaminetetraacetic acid disodium salt dihydrate (ETDA; Fluka), calcium chloride dihydrate (Fluka), calcium carbonate (Carlo Erba Analitycals), D-glucone-δ-lactone (DGL) (Fluka), 2-morpholinoethanesulfonic acid monohydrate (MES; Fluka), N-hydroxysuccinimide (NHS; Fluka), 1-ethyl-3,3-[3-(dimethylamino)propyl]carbodiimide hydrochloride (EDC; Fluka), toluene (Carlo Erba), dimethylsuphoxide (Carlo Erba), n-hexane (Carlo Erba), sodium chloride (Fluka), citric acid, tartaric acid, pluronic F-108, tyloxapol (all Adrich), and polyquaternium 10JR400 (Amerchol) were used as received. Mechanical Degradation of Alginate. Alginate degradation was carried out with high energy ball milling. About 80 g of alginate (Fluka) was placed inside a jar that was allocated in a ball mill ZOZ GmbHD5910 Kreuztal (a picture of the machine is reported in the Supporting Information, Figure S1). The apparatus was provided with a cooling system to keep the jar at room temperature and was connected to a vacuum pump to exclude oxygen during the degradation process. Iron balls (diameter 5 mm, ZOZ GmbH-D57482 Wenden Grinding Media) were used in a ratio of 10/1 g/g with respect to the weight of alginate. The angular velocity was set at 1000 rpm. At predetermined times, grinding was stopped and aliquots of alginate were taken out of the jar and subjected to molecular weight determination. Prior to use, the degraded alginate samples were dissolved in water and filtered with a series of descending pore size filters (from 3 to 0.45 µm) and successively dialyzed against water. Finally, water was removed by freeze-drying. PolyHIPE Synthesis. A total of 0.5 g of degraded alginate was dissolved in 2.2 mL of MES buffer (pH ) 4.5) in the presence of Triton X-405 (6.0% w/v). The solution was poured into a thermostated (18 °C), three necked, round-bottom flask provided with a dropping funnel. Toluene (the dispersed phase, 6.6, 12.5, or 19.8 mL) was added dropwise under stirring (300 rpm). At the end of the addition, stirring was prolonged for another 15 min to allow better homogenization of the HIPE. Afterward, 0.3 mL of a solution of EDC (EDC/alginateCOONa ) 2.5 mol/mol) and NHS (NHS/EDC ) 0.2 mol/mol) were added to the HIPE and stirring was maintained for 2 min. The HIPE was transferred into a polyethylene bottle that was placed inside an oven set at 40 °C for 24 h. After this period, the polyHIPE was soaked in DMSO, which was changed regularly (typically three times a day) for 1 week. This procedure was aimed at displacing toluene thoroughly. The polyHIPE was then Soxhlet extracted with water for 1 day and finally freeze-dried. Solid Foam Synthesis and Foam Stability Evaluation. Alginate from Laminaria hyperborea was used. A solution of 5% w/v (0.125 g in 2.5 mL) of alginate, pluronic F-108 (4.0 or 6.0% w/v), and sodium bicarbonate (2% w/v) was prepared and poured into the reactor (a scheme of the experimental apparatus is reported in Figure S2 of the Supporting Information). An equivalent amount, with respect to the moles of NaHCO3, of acid (tartaric, citric, and gluconic) was added while keeping the alginate solution under stirring (500 rpm). Stirring was continued for 15 min to allow CO2 to fully develop. Afterward,
Biomacromolecules, Vol. 10, No. 8, 2009
2329
the foam was frozen in liquid nitrogen and freeze-dried. The resulting solid foam was soaked in a solution 2.0 M of CaCl2 to induce the formation of the calcium-based physical gel for 24 h and then dialyzed against a solution of CaCl2 0.1 M. Finally, the solid foam was freezedried. Chemical cross-linking of the Ca2+-based solid foam was carried out with EDC/NHS (EDC/alginate-COONa ) 2.5 mol/mol, HNS/ EDC ) 0.2 mol/mol) in MES buffer (pH ) 4.5) under mild stirring for 24 h. The solid foams were dialyzed against water and then freeze-dried. The selection of the surfactant more effective for the stabilization of the CO2-in-water foam was carried out by measuring the decline of the foam volume inside a graduated cylinder as a function of time. Foams were prepared by placing 5 mL of solution (15% w/v of degraded alginate, NaHCO3, organic acid, and surfactant) in a glassstoppered 25 mL graduated cylinder. The cylinder was shaken 10 times by hand through an arc of 180°. The initial volume of foam produced was recorded and additional measurements were made every 2 min. The inspection of the size distribution of the foam bubbles was monitored by light microscopy observations (Leica Z16 APO). Characterization. Elemental Analysis. Elemental analysis was carried out by emission optical spectroscopy inductively coupled plasma (ICP-OES) using a VARIAN VISTA MPX CCD Simultaneous ICPOES equipped with an ultrasound nebulizer. Sample solutions with a concentration of alginate of 2.8 × 10-7 M in G and M residues were obtained by dilution of a stock solution with a concentration of 2.8 × 10-2 M. Solution pH was adjusted to 3.5 with HCl. Particle Size Determination. For the determination of particles size, a laser diffraction particle analyzer Beckam Coulter LS 13 320 was used. Alginate powders from the original sample and from the degraded product (40 h of milling) were suspended in n-hexane and analyzed. Average particle size and particle size distributions provided are the mean values from three independent measurements. Viscosity and Gel Permeation (GPC) Measurements. The intrinsic viscosity [η] was determined using the viscosity measuring unit AVS379 (Schott-Gerate, Hofheim, Germany), connected to a Viscodoser AVS20 piston buret (for automatic dilutions), to make automated measurements of the flow-through times in a capillary viscometer (Ubbelholde viscometer, Φ ) 0.53 mm, for dilution sequences). The viscometer was immersed in a precision water bath (transparent thermostat CT 1150, Schott Gerate, Hofheim, Germany) to maintain the temperature at 25 ( 0.1 °C. The solutions were prepared by dissolving the lyophilized Alg and Algdeg (starting alginate and the various alginate samples at different degrading times, respectively) with magnetic stirring for at least 24 h at room temperature, followed by filtration through a Millipore filter of 0.45 µm. Solutions had relative viscosities from about 1.2 to 2.0 to ensure good accuracy and linearity of extrapolation to zero concentration. The intrinsic viscosity, [η], was obtained by double extrapolation to zero concentration of Huggins’ and Kraemer’s equations, respectively
ηsp ) [η] + k[η]2C C (ln ηrel) ) [η] + k[η]2C C where ηrel and ηsp are the (dimensionless) relative and specific viscosities, k′ and k′′ are the Huggins’ and Kraemer’s coefficients, respectively, and C is the solution concentration. Alg and Algdeg were also subjected to GPC analysis. The GPC used is a modular system (Lab flow 4000) with a refractometer (Shimadzu RID-10A). The samples were run using two TSK-GEL GM-PW (30 × 7.5, 17 µm) columns and an aqueous solution of NaCl 0.01 N as the mobile phase. The flow was set at 0.8 mL/min and the temperature at 25 °C. Calibration of the GPC was achieved using pullulan standards. NMR Experiments. The samples (Alg and Algdeg ∼ 2 mg) were dissolved in 700 µL of phosphate buffered (pD ) 7) D2O solution, 0.01 M NaCl. 1H NMR experiments were performed on a Bruker AVANCE AQS 600 spectrometer operating at 600.13 and 150.92 MHz,
2330
Biomacromolecules, Vol. 10, No. 8, 2009
respectively, with a Bruker inverse multinuclear z-gradient probehead. Because at room temperature the 1H spectra appear broad and poorly resolved, spectra were recorded at 343 K; at this temperature, in fact, all resonances appear sufficiently resolved to be assigned. When necessary, to avoid some signal overlapping, the spectra were also run at 333 K. In all 1H spectra, a soft presaturation of the HOD residual signal was performed. Chemical shifts of 1H spectra are reported in ppm with respect to 2,2-dimethyl-2 silapentane-5-sulfonate sodium salt (DSS) used as an internal standard. The deconvolution of the 1H NMR spectra was performed using the SHAPE2000 (version 2.1) software package. Applying the deconvolution program, the areas Ii of peaks resonating in the 4.4-5.1 ppm spectral region are obtained along with errors ∆Ii. It is worth noting that the obtained errors are always well within 4-5% of the nominal Ii values. Therefore, the error on the area of each resonance is small, which is satisfactory. Because it is not possible to obtain errors on the calculated molar fractions, on the doublet and triplet frequencies, and on the average block lengths, straight from the deconvolution program, the propagation error theory has been applied. However, as is wellknown, this theory may overestimate the errors. As a consequence, errors reported in Table 3 are overestimated. In our notation, M ) mannuronic acid unit and G ) guluronic acid unit. Electron Microscopy. Scanning electron microscopy (SEM) images of both polyHIPEs and solid foams were obtained by using a LEO 1450 VP electron microscope. Samples were mounted on aluminum stubs using a carbon tape and sputter coated with a gold layer. Surface Area/Pore Size Distribution. After the samples were vacuum-dried at 473 K overnight, nitrogen adsorption/desorption isotherms were determined at 77.3 K using a Micromeritics ASAP 2010 apparatus. The specific surface areas were calculated by the BrunauerEmmet-Teller (BET) equation utilizing eight points from the adsorption isotherm collected over the relative pressure range 0.05-0.20 (P/P0). Pore size distributions were obtained from the analysis of the adsorption branch of the isotherm using the Barrett-Joyner-Halenda (BJH) method.51 N2 sorption measurements were carried out on both the freeze-dried samples or previously treated with a solution of CaCl2 (1.0 M). The latter type of sample was dehydrated with supercritical CO2 (scCO2). The Ca2+-polyHIPE was first dehydrated by immersion in a series of successive ethanol-water baths of increasing alcohol concentration (10, 30, 50, 70, 90, and 100%) for 15 min each. Finally, it was dried under supercritical CO2 conditions (74 bar, 31.5 °C) in a Polaron 3100 apparatus.
Results and Discussion Emulsion Templated Porous Materials. A major problem in the preparation of concentrated emulsions from polymeric aqueous solutions is their viscosity. Experience teaches47,48 that too high a viscosity of the continuous phase can hinder the dispersion of the internal phase, which precludes the possibility of incorporating it completely within the emulsion and obtaining a homogeneous polyHIPE. Commercial alginate is usually high molecular weight. This joined to its intrinsic stiffness and polyelectrolytic nature makes it impossible to employ it directly in the production of a HIPE (which by definition is characterized by a water to oil ratio of at least 0.75). Therefore, a necessary step preceding its employment is represented by its degradation to reduce substantially its molecular weight. To this end, we have employed mechanical degradation by high energy ball milling. To our knowledge this is the first time that this technique has been used in the degradation of a complex biopolymer such as alginate; therefore, it is of interest to give a full description of its effectiveness outlining advantages and disadvantages.
Barbetta et al.
Figure 1. Behavior of the viscosity (Mη), number (Mn), and weight (Mw) average molecular weights (A) and polydipsersity index (B) of alginate sample as a function of the degradation time.
Degradation of Alginate by Ball Milling. In ball milling, hard balls are placed inside a cylindrical container (jar) along with the bulk polymer. The cylinder is turned horizontally along its long axis to cause the balls to repeatedly tumble over one another, thereby fragmenting in the first instance the sample particles into smaller ones and then causing the breakage of the polymer chemical bonds. One of the main advantages of this technique is that it operates in the solid phase, thus allowing large quantities of polymer to be processed. Furthermore, it guarantees the complete recovery of the employed material. In comparison, hydrolytic degradation would necessitate dissolving the same amount of alginate in a very large amount of solvent (water, in our case), making the degradation process less practical. A problem associated with the ball milling technology is the erosion from the milling material and container walls during the milling process that inevitably contaminate the product with a very fine powder. For this reason, the degradation product needs purification by filtration and extensive dialysis. During milling, relevant energies are involved and internal frictions cause heating of the sample, therefore, radical species may be generated, which may induce side reactions with O2 or moisture or induce the formation of branched species through the combination of macroradicals. To avoid such side reactions, degradation was carried out under vacuum and the jar was refrigerated. The first parameter to be monitored during degradation was the molecular weight. At prefixed times, the milling process was stopped and a sample of alginate was taken out of the jar. The molecular weight was determined by viscometry and GPC j w, (Figure S2 of the Supporting Information). In Figure 1a, M j η are plotted as a function of the degradation time. j n, and M M Numerical values are reported in Table SI of the Supporting Information. It can be noted that the degradation is more efficient within the first 5 h. For more extended periods of time, molecular weights tend toward asymptotic values. In Figure 1b and in Table SI it can be seen that the polydipsersity index, I, is approximately constant during the first 4 h of milling. This may reflect the random nature of chain scission, which in the
Porous Alginate Hydrogels
Biomacromolecules, Vol. 10, No. 8, 2009
2331
Table 2. NMR Peak Assignements (ppm), Intensities, and Weight Percentages of Diads and Triads in Alg0h and Alg40h Alg0h ppm
intensity
weight %
intensity
weight %
MG5G GG5G M1M M1G MG5M GG5M G1G-MG1M
4.42 4.44 4.66 4.69 4.70 4.72 5.05
3256 6282 21365 13430 2580 2342 18007
4.8 9.3 31.8 20.0 3.8 3.8 26.8
3060 12624 33491 14942 8422 2447 29981
2.9 12.0 31.9 14.2 8.0 2.3 28.6
a
Figure 2. Particle size volume distributions of starting alginate (Alg0h) and degraded alginate after 40 h of milling (Alg40h). Table 1. Average Particle Dimensions and Standard Deviations (SD) and Normalized SD (CV) of Alg0h and Alg40h average SDa CVb d10c d50c d90c sample dimensions (µm) (µm) (%) (%; µm) (%; µm) (%; µm) Alg0h Alg40h
229 43
60.9 34
27 78
155 93
223 33
312 94
a Standard deviation. b CV ) (SD × 100)/average dimensions. c % of particles within prefixed dimension ranges (dx).
solid phase may involve different chains to a different extent. As the degradation proceeds, all chains undergo to a similar number of chain cuts, with the consequence that I at longer degradation times decreases. It may be wondered the reason the molecular weight tends for long degradation times to an asymptotic value. The efficiency of degradation depends on the rheology of the sample powder, which in turn is dependent on particles size. Therefore, the particle size distributions of starting alginate (Alg0h) and degraded alginate after 40 h of milling (Alg40h) were determined by laser diffraction. Such distributions are displayed in Figure 2. Both distributions are single peaked and from their comparison it is evident that that relative to Alg40h is shifted remarkably toward the low diameter side and is narrower with respect to Alg0h one. In Table 1, the corresponding average particle diameters and standard deviations (SD) are reported. The data of Table 1 support in more quantitative terms the qualitative analysis of Figure 2. The standard deviations of Alg40h and Alg0h are 34 and 61 µm, respectively. Furthermore 75% of Alg0h is constituted by particles of dimension in the range between 160-310 µm, while 54% of Alg40h contains particle of dimension in the range between 14-48 µm. It has been recognized52 that as the average particle dimension decreases the powder acquires a more pseudoplastic behavior with the consequence that the efficiency of energy transfer from the milling apparatus to particles fragmentation and chain degradation decreases progressively. Another aspect of concern in the mechanical degradation of complex biomacromolecules such as alginate is whether degraded alginate retains its starting primary structure. It is therefore useful to make a quantitative comparison between the sequences of G, M, and MG blocks in both Alg0h and Alg40h. To this end, 1H NMR spectra have been recorded for both Alg0h and Alg40h and analyzed (Figure S3a,b of the Supporting Information). The deconvolution of the anomeric region (4-5 ppm) allowed the estimation of the relative percentages of the different kind of diads and triads.53 From Table 2 it can be seen that no major differences characterize the primary structures of Alg0h and Alg40h. The most notable differences regard the relative amount of M1G diads and MG5M triads. The total
Alg40h
diads or triadsa
M: mannuronis residues; G: glucuronic residues.
Table 3. Results of ICP-OES on Alginate Alg0h and its Degradation Products Alg20 and Alg40h sample
Nia
Cra
Fea
Alg0h Alg20h Alg40h
0.2 8.6 364.7
ndb 2.8 116.3
ndb 23.9 1010.5
a Element content is expressed as ppm in 1.0 g of sample. detectable limit.
b
Below
content of M and G residues are ∼62 and 38% for Alg0h and 66 and 34% for Alg40h, These results indicate that the degradation takes place predominantly at the glycosidic bond (with some preference for the glycosidic bond between a mannuronic and glucuronic residues) and does not involve the glucopyranose ring. To evaluate the magnitude of contamination caused by the milling process on the sample of alginate, the quantitative determination of the metallic elements present in both the jar and milling balls (Fe, Cr, Ni) was carried out on Alg0h, Alg20h, and Alg40h by ICP-OES. Results are displayed in Table 3. It is evident that the content of metallic elements increased exponentially with the time of degradation. In fact, most of the contamination occurs in the last 20 h of milling when the degradation process becomes progressively less efficient. As a consequence, before using degraded alginate, it was necessary to purify it extensively. Filtration and dialysis (see Experimental Section) proved successful. Preparation of Emulsion Templated Porous Materials. The methodology for the preparation of HIPE templated porous materials is well established. Biopolymers of a different nature, that is, polysaccharides and proteins,44-48 have been successfully included in polyHIPE formulation. A general requirement underlying the employment of biopolymers as the main components of the external phase of a HIPE is the control over the continuous phase viscosity. The molecular weight of the biopolymer must be controlled to keep the viscosity of the continuous phase within values that allow the incorporation of large volume fractions of the internal phase (φ g 0.75). In this respect, the intrinsic stiffness of the biopolymer plays an important role. A flexible polymer such as dextran even of high molecular weight is immediately employable. On the contrary, a stiff polymer such as alginate must be subjected to a considerable reduction of its molecular weight to reduce the viscosity of its concentrated solutions to processable values. The various degraded alginate samples (Table SI) have been subjected to emulsification tests. The concentration of alginate in the continuous phase chosen was 15 or 20% w/v and φ (defined as the ratio of the organic phase volume to total emulsion volume) was set at 0.75, 0.85, or 0.90. Only Alg40h gave rise to macroscopically homogeneous HIPEs, the less degraded alginate samples (Algxh with x < 40 h) did not allow the full incorporation of the entire volume of the dispersed
2332
Biomacromolecules, Vol. 10, No. 8, 2009
Barbetta et al.
Figure 3. Scanning electron micrographs of Alg40h-based polyHIPEs obtained by varying the volume fraction of the dispersed phase (φ) and the concentration of polymer in the aqueous phase (Cp): (a) φ ) 0.75, Cp ) 15% w/v; (b) φ ) 0.90, Cp ) 15% w/v; (c) φ ) 0.85, Cp ) 20% w/v; (d) φ ) 0.90, Cp ) 20% w/v. Surfactant concentration (Triton X-405): 6% w/v.
phase. Thus, in the following of this section it is implicitly assumed that the alginate material used is Alg40h. The simplest way to modulate the morphology of polyHIPEs is by varying φ. If the volume of the internal phase is kept constant, the increase of φ will cause the thinning of the film of aqueous phase surrounding the droplets of the dispersed phase. On polymerization,54 this will bring about materials characterized, on the average, by larger interconnecting holes. In flow through applications this translates into a higher permeability. In Figure 3a-d, the SEM micrographs illustrate this conceptual approach to morphology tailoring. For instance, φ ) 0.75 and a polymer concentration, Cp, of 15% w/v (Figure 3a) voids and interconnecting holes are in the range of 3-15 and 1-8 µm. Increasing φ to 0.90 (Cp ) 15% w/v; Figure 4b) causes a considerable increase in both void and interconnect dimensions, 10-40 and 2-20 µm, respectively. In general, the viscosity of HIPEs (ηHIPE) is directly proportional to the viscosity of the external phase (ηe) and on φ. When the external phase contains a polymer, the relationship between ηe and polymer concentration (Cp) is obvious. Also, the effect of φ on ηHIPE is intuitive: when the droplets are concentrated above the random close-packing limit (φ g 0.74), they cannot move freely and are trapped by their neighbors and take the shape of a polyhedron. As a consequence, ηHIPE increases with φ, because part of the stress imposed will be dissipated as the work of
droplets deformation. The combined effect of φ and Cp is well illustrated by the morphology of the polyHIPEs shown in Figure 3c,d. When φ ) 0.85 and Cp ) 20% w/v (Figure 3c), large voids of the order of 20-70 µm are interdispersed into a more homogeneous porous structure consisting of much smaller voids. This effect is even more evident in the polyHIPE characterized by φ ) 0.90 and Cp ) 20% w/v, where voids as large as 50-100 µm appear. This is clearly the outcome of the combined effects of the increase of φ and Cp on the HIPE viscosity. During HIPE formation, as the amount of the added dispersed phase increases, the emulsion becomes more and more viscous and the shear stress provided is not effective enough to break the droplets of the final portion of the added internal phase into smaller ones. Another aspect of the matrices presented that needs to be evaluated is the textural morphology of the polyHIPE walls. In applications such as removal of pollutants and as supports in biocatalysis it would be desirable to have high surface materials. Valentin et al.55,56 have shown that the drying procedure of ionically cross-linked alginate gels is critical to preserve the porous texture of the gel. In this respect, extraction of solvent with scCO2 proved to be the most effective procedure. This technique is commonly processed with inorganic solids to achieve very high specific surface areas.57 On the contrary, freeze-drying caused the alginate chains to collapse over one another causing the disappearance of macro (>50 nm) and mesopores (5-50 nm). Both approaches were implemented with
Porous Alginate Hydrogels
Biomacromolecules, Vol. 10, No. 8, 2009
2333
Figure 4. SEM micrographs of EDC cross-linked aerogel (A) and of EDC and calcium cross-linked xerogel alginate polyHIPE (B). Nitrogen adsorption-desorption isotherms (C) and BHJ pore size distributions of alginate aerogel (black line) and xerogel polyHIPEs (red line).
the alginate-based polyHIPEs. Treatment with scCO2 must be preceded by exchange of water with a series of graded ethanol solutions. During such a process, the polyHIPE underwent to a remarkable shrinking. We hypothesized that this phenomena might be due to a level of chemical cross-linking not high enough to avoid the pore collapse phenomenon consequent to the exposure of highly hydrophilic gels to a nonsolvent. To confer a higher rigidity to the polyHIPEs, the ability of alginate to undergo physical cross-linking in the presence of divalent ions (Ca2+, Ba2+, Cd2+, Sr2+, etc.) was exploited. A polyHIPE (φ ) 0.75, Cp ) 15% w/v) was treated before subjecting them to exchange with ethanol with a concentrated solution of CaCl2 (1.0 M). We hoped in this way to increase the cross-linking density (chemical and physical) and obtain a more rigid material. It was indeed observed that when the Ca2+-polyHIPE were exposed to ethanol the degree of shrinkage was much less. The influence of the drying procedure (freeze-drying vs treatment with scCO2) on the morphology of the polyHIPEs is illustrated in Figure 4a,b. The difference between the two procedures of drying is clear when we observe the details of the polyHIPE walls. In the case of freeze-dried sample the walls appear completely smooth (Figure 4a). On the contrary, in the case of scCO2 dried sample (termed aerogel-polyHIPE), walls exhibit a porous texture. Further information about the textural properties of the polyHIPE-based alginate was obtained by nitrogen adsorptiondesorption hysotherms. The nitrogen adsorption-desorption hysotherms of N2 at 77 K of the freeze-dried and aerogelpolyHIPEs are presented in Figure 4c. All isotherms of Figure 4c are of type IV at the borderline with type II in the IUPAC classification. From their qualitative comparison it evident that N2 adsorption from the aerogel-polyHIPE is much higher than the freeze-dried polyHIPE and the hysteresis loop much broader. Specific surface areas were 237 and 10 m2/g, respectively. The pore size distributions show (Figure 4d) that in the case of freeze-dried polyHIPE no mesopores are present at all as it is
clearly observed in the relative SEM micrograph (Figure 4a). Evidently, the surface area measured comes mainly from the macrovoids. The aerogel-polyHIPE exhibited a pore size distribution characterized by a significant presence of mesopores. Nevertheless, the porosity is mainly in the macropore region. Gas Templated Solid Foams. While the emulsion templated alginate porous materials presented in the previous section may be useful in areas such as purification of wastewater and enzyme immobilization due to their excellent permeability and good mechanical stability, they are inadequate for applications such scaffolds for tissue engineering. In this area of application, a fundamental prerequisite a scaffold must possess is a porous texture composed of voids and interconnects of at least 100 and 40 µm, respectively. It has already been established that in the case of HIPE templated porous materials the upper limits of void and interconnect sizes do not overcome the values reported above.45,47 For this reason we searched for a related technique of porous materials production that would bring in this respect significant improvements. We wondered if the replacement of the internal liquid phase of the HIPE with a gas would represent the key factor to achieve a substantial increase of voids and interconnects sizes. Such an approach would be beneficial also under other aspects: the synthetic procedure and purification would be simplified because it does not involve extraction of the organic internal phase with another organic solvent followed with extensive washing with water; the avoidance of any organic solvents as the internal phase represents an advantage as far biocompatibility of the ensuing biomaterials is concerned. On the other end, a drawback of foam templating is represented by the inferior kinetic stability of foams with respect to HIPEs. For this reason, the choice of surfactant and the gas generating system is critical and must therefore be carefully selected.
2334
Biomacromolecules, Vol. 10, No. 8, 2009
Barbetta et al.
Table 4. pH of the Aqueous Solutions Containing the Indicated Reactants for the In Situ Development of Gas and Related Observations solutiona
pH
SA 2% SA + SN BC 2% BC + CA BC + TA BC + DGL
0.6 1.1 8.7 4.6 3.4 7.2
evaluation is given. As it can be seen, the reactions involving TA and, to a lesser extent, CA are the most promising. Another component of paramount importance is the surfactant whose function is to stabilize the foam for a time long enough to allow cross-linking to lock-in the structure of the external phase before destabilization process take place to any significant extent. Foams are systems kinetically less stable than emulsions. Due to the relatively large density difference between the gas phase and the liquid, the bubbles rise in the gravity field to the top of the liquid. There they form a more or less close-packed, honeycomb-like, bubble structure from which the liquid drains, forming foams that become dryer and dryer. The thin films formed can collapse for some reason, causing the coalescence of bubbles. The lifetime of a foam can be extended significantly through one of the following: (1) a high viscosity of the liquid phase, which retards drainage of the liquid from between the bubble interface, as well as providing a cushion effect to adsorb shocks resulting from random or induced motion; (2) a high surface viscosity, which also retards liquid loss from between interfaces and dampens film deformation prior to bubble collapse; (3) surface effects such as the Gibbs and Marangoni effects,60 which act to “heal” areas of film thinning due lo liquid loss; and (4) electrostatic and steric repulsion between adjacent interfaces due to the adsorption of ionic and nonionic surfactants, polymers, and so on. Most of the guidelines 1-4 can be fulfilled by using a combination of a high molecular weight alginate jη ) and a surfactant. Alginate from Laminaria hyperborea (M 2.3 × 105) was used. Foaming properties of a given surfactant are necessarily characterized by at least two parameters: one specifying the ease of foam production under specified conditions and one specifying the persistence of this foam under quiescent conditions. Shaking the surfactant solution in a cylinder and observing for 20 min the decline in foam volume is a simple procedure satisfying these conditions. A certain number of surfactants, either low and macromolecular weight (Figure S5 of the Supporting Information), were tested and their performances were evaluated. Results are summarized in Table 5. As it can
experimental observations intense development of N2 fair development of CO2 intense development of CO2 modest development of CO2
a SA, sulfamic acid; SN, sodium nitrite; BC, sodium bicarbonate; CA, citric acid; TA, tartaric acid; DGL, D-glucono-δ-lactone.
Choice of the Gas Generating System. There are essentially two inert gas that can be exploited for foam formation: N2 and CO2. N2 can be generated in situ the alginate solution by mean of the oxidative/reductive reaction:
H2NSO3H + NaNO2 f N2 + NaHSO3 + H2O
(a)
CO2 can be generated by virtue of the acid/base reaction:
RCOOH + NaHCO3 f RCOONa + H2O + CO2 (b) where R can be D-glucono-δ-lactone, citric acid, and tartaric acid (see Supporting Information, Figure S4, for chemical formulas). It is well-known58,59 that alginates undergo gelation in acidic environment (pH < 3); as a consequence, it was necessary to make a screening of the final equilibrium pH for each reaction ((a) and (b)). Results are summarized in Table 4. As expected, the equilibrium pH of reaction (a) is rather low and is thus inapplicable for our purpose. On the contrary, all reactions based on the development of CO2 (b) give rise to a pH > 3 and are thus exploitable. Another aspect to be evaluated regards the intensity of CO2 development. In Table 4, such a qualitative
Table 5. Remarks on the Kinetics of Foams Decay as a Function of Surfactant Type and its Concentrationa surfactant PF
PQ-SDS
SDS
TY
TR
a
% w/w
foaming
2 min
10 min
4
intense
no variation
no variation
6
intense
no variation
no variation
8
intense
no variation
no variation
0.1-0.01
scarce
0.4-0.04
scarce
0.8-0.08 4 6
modest intense intense