J. Phys. Chem. B 2008, 112, 2809-2817
2809
Porosity Inherent to Chemically Crosslinked Polymers. Poly(N-vinylimidazole) Hydrogels V. Calvino-Casilda,† A. J. Lo´ pez-Peinado,† E. Vaganova,‡ S. Yitzchaik,‡ I. E. Pacios,§ and I. F. Pie´ rola*,§ Department Quı´mica Inorga´ nica y Quı´mica Te´ cnica, Facultad de Ciencias, UniVersidad a Distancia (UNED), 28040 Madrid, Spain, Department of Inorganic and Analytical Chemistry, Hebrew UniVersity of Jerusalem, IL-91904 Jerusalem, Israel, and Department de Ciencias y Te´ cnicas Fisicoquı´micas, Facultad de Ciencias, UniVersidad a Distancia (UNED), 28040 Madrid, Spain ReceiVed: NoVember 6, 2007; In Final Form: December 17, 2007
Swollen polymer networks exhibit multiscale pores filled with solvent. Such porosity, inherent to crosslinked polymers, determines some of their most relevant physical properties and applications. In this research, several samples of chemically crosslinked poly(N-vinylimidazole) were synthesized with the same permanent crosslinking density at two different conversions, and their inherent porosity was characterized on freezedried specimens by SEM, TEM and nitrogen physisorption. It was thus found that all of the samples showed pores, both on the nanometer and the micrometer scales, whose dimensions were mostly equal to or larger than the mesh size of the primary polymer network (22 nm) and whose volume and specific surface decreased with increasing conversion. Micropores have, in all cases, a very minor contribution. Samples synthesized with the largest comonomer concentrations show quasi-spherical mesopores (90 nm average diameter at any conversion) and macropores (from 5 to 10 µm with increasing conversion), whereas the mesopores of samples synthesized with the largest crosslinker ratios were channel-like (150 nm) and the macropores were interconnected contiguous voids (3 µm). Samples with intermediate compositions exhibit the lowest porosity due, mostly, to interconnected mesopores. The differences in shape were ascribed to the mechanism of phase separation, taking place during polymerization, even for samples that are transparent following polymerization. The inherent porosity is a significant source of spatial inhomogeneity, which contributes to the increase in turbidity. Light scattering decreases with increasing ionization when the degree of protonation is greater than 10%. An important consequence of the inherent porosity is that the degrees of swelling determined either gravimetrically or through size measurements are not equivalent.
Introduction Crosslinked polymers may contain pores, the diameters of which vary from nanometer to millimeter in length scale. These pores have different structures (closed or interconnected) and different shapes (fibrillar networks, honeycomb-like, spherical, cylindrical, channel-like, etc.), depending on several variables, such as the porogen and crosslinker agents used in feed or the previous thermal history of the sample.1-10 Polymer networks, even if they are synthesized without using porogen agents, may also exhibit pores (inherent porosity) that are filled with solvent in the swollen state. Upon drying, these pores collapse,2,10 unless they are processed using special methods.5,10 Porosity determines numerous applications of polymers that are used as membranes, in chromatography, catalysis, cell culturing or other biomedical devices. In order to control the porous structure of crosslinked polymers, it is necessary to know first the characteristics of their inherent porosity. Porosity is the volume of the voids or the interstices of a material per unit mass of dry material. Only voids must be * To whom correspondence should be addressed. E-mail: ipierola@ ccia.uned.es. † Department Quı´mica Inorga ´ nica y Quı´mica Te´cnica, Facultad de Ciencias, Universidad a Distancia. ‡ Department of Inorganic and Analytical Chemistry, Hebrew University of Jerusalem. § Department de Ciencias y Te ´ cnicas Fisicoquı´micas, Facultad de Ciencias, Universidad a Distancia.
considered for bulk samples, such as those studied here. When classifying pores according to size in the field of porous materials, several general categories, with not well-defined crossover, are used. Thus, pores on the nanometer scale are called micropores (below 2 nm in diameter) or mesopores (2 to 50 nm), whereas pores with larger size are called macropores (0.05 to 10 µm) or superpores (10 to 1000 µm). Micropores and mesopores (also called nanopores in the context of polymer materials11) are visualized by HR-TEM (high-resolution transmission electron microscopy), whereas macropores and superpores can be observed by means of SEM (scanning electron microscopy). The size and shape of pores have important implications for the properties of crosslinked polymers, e.g., mesopores, macropores, and superpores are effective reservoirs for solvent uptake.12 The elastic modulus of some hydrogels is not affected by the increased porosity until the pores become interconnected.13 The diffusion rate and permeability of small compounds through hydrogels with channel-like pores oriented parallel to the flow are significantly higher.4 Furthermore, macroporous and superporous hydrogels exhibit rapid swelling or shrinkage in response to external stimulus and enhanced release of drugs.2,6,7 Optical properties are also affected by the size and number of pores. Materials that are homogeneous on a length scale greater than 1/20th of the wavelength of visible light (above 20 nm) are transparent. However, samples showing voids having widths larger than 20 nm, surrounded by a polymer
10.1021/jp7106473 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/19/2008
2810 J. Phys. Chem. B, Vol. 112, No. 10, 2008
Calvino-Casilda et al.
TABLE 1: Concentrations of Crosslinker and Bifunctional Comonomer Employed in the Synthesis of Crosslinked PVI Samplesa C
sample code
[VI] (mol/L)
[BA] (mol/ L)
[BA] (mol/L)2
Y-2h (%)
Y-44h (%)
PVI-6.1 PVI-3.9 PVI-2.8 PVI-1.8
6.07 3.86 2.76 1.66
0.039 0.063 0.086 0.137
0.238 0.247 0.245 0.246
23.6 34.8 43.3 54.4
80.4 79.3 86.8 77.9
T
a The yield of the polymerization of samples synthesized during 2 h (Y-2h) or during 44 h (Y-44h) is also shown.
matrix with different refractive index, scatter light and eventually the material becomes translucent or opaque.14 Spatial inhomogeneity is an intrinsic characteristic of polymer networks synthesized by conventional radical crosslinking copolymerization. To a large extent, it determines the thermodynamic, mechanical, and transport properties of the networks.15-17 Heterogeneous gels display high turbidity (intense scattering of light), and opacity is frequently considered to be the clearest symptom of inhomogeneity. In other words, a transparent sample is usually considered to be homogeneous, without voids or other types of heterogeneous density distribution. The spatial inhomogeneities are interpreted as being due to the heterogeneous distribution of crosslinker18,19 or as the result of phase separation during the formation of the polymer gels.20 It is also remarkable that the ionization of polymer networks, which is a reversible process, is one of the methods used to suppress inhomogeneties19 and that upon drying, opaque hydrogels become, in some cases, transparent. All of these facts will be helpful in discussing the relationship between porosity and heterogeneity in PVI hydrogels. The current work focuses on the inherent porosity of pH sensitive poly(N-vinylimidazole) hydrogels21-24 (PVI) synthesized by radical crosslinking copolymerization in aqueous solution, with different comonomer concentrations. Two variables are particularly important in determining the porosity of the polymer networks, the length of post-gel reactions8 and the degree of cross-linking.4 In order to reduce the number of variables, several actions were undertaken in the synthesis strategy: no porogen was used,2 the degree of conversion was controlled,8 and the total comonomer concentration (CT) and crosslinker concentration were varied while keeping their product constant.23 In this way, the permanent crosslinking density of samples synthesized at total conversion was the same but their morphology was totally different, depending on CT. The characterization of the porosity was carried out using several techniques (namely, TEM, SEM, and nitrogen adsorption) and the consequences of the inherent porosity for the swelling behavior and optical properties were analyzed. Experimental Section Poly(N-vinylimidazole) hydrogel samples (PVI) were synthesized by radical crosslinking copolymerization of N-vinylimidazole (VI) and N,N′-methylene-bis-acrylamide (BA) in aqueous solution, with 2,2′-azobis(isobutyronitrile) (AIBN) (9.8 × 10-3 M) as initiator. N-vinylimidazole (Aldrich) was distilled under reduced pressure, just prior to use. Water was distilled and deionized by a Milli-Q system from Millipore and AIBN (Fluka) was recrystallized from methanol. BA (Aldrich) was used as received. The concentrations of VI and BA (Table 1) in the feed varied for each sample, with total comonomer concentrations, CT, ranging from 1.8 to 6.1 M and BA concentration ([BA]) from 0.14 to 0.04 M in such a way that
the product CT ×[BA] is approximately constant at 0.244 ( 0.004. Henceforth, samples will be denoted as PVI-CT and the corresponding [BA] can be determined as 0.244/CT. The feed mixtures were poured into glass tubes, bubbled with argon for 10 min, and introduced into a thermostatic bath at 70 °C. Polymerization took place for 2 h (2h-PVI-CT) or 44 h (44h-PVI-CT). Ten minutes after being introduced into the bath, the tubes were inverted to check the movement of their contents, and in all cases, it was found that samples did not flow, indicating that gelation had already taken place. After synthesis, the hydrogel cylinders were taken out of the molds; one 1-cm thick sample from the top and another from the bottom were ruled out and the central part was cut into disk-like pieces of 12.8 mm diameter (the internal diameter of the tubes employed in the synthesis) and less than 1 mm thick. In addition to the samples described in Table 1, other samples synthesized with the same protocol for previous works,8 were also employed. For swelling measurements in methanol, ethanol, and deionized water, samples were immersed in the swelling solvent immediately following polymerization (without previous drying). During the next 10 days, the swelling solvent was frequently changed. Once the samples had reached a constant weight (mh), we measured their diameter (Dh). Samples were then dried slowly in air during 40 days to prevent deformation. During 10 additional days, they were dried under vacuum at room temperature. The diameter and mass (Do, mo) of the dry slabs were then measured. Prior to swelling measurements in HCl aqueous solutions, samples were washed exhaustively in deionized water and dried. Specimens of about 10 mg of dried gel with known mass (mo) and size (Do), were immersed in a volume (V) of HCl solutions of known initial pHi (around 10 mL), required to have in each case the same effective polymer concentration, Cef ) mo/MoV, with Mo being the molecular weight of the monomeric unit. The effective polymer concentration is not a real concentration, since the polymer is not dissolved but simply immersed. In this work, Cef was within the range 10.5 ( 1 mM. One month later, samples were weighed (mh) and their sizes (Dh) and light scattering intensities (vide infra) as well as the equilibrium pH of the baths (pHeq) were measured. The swelling solution was not replaced in order to avoid artifacts due to changes of pH. HCl solutions without polymer were kept under the same conditions as those containing the polymer samples in order to control possible variations of pHi depending on time. pHi values are employed in plots and text. The degree of protonation of the polymer was determined as24
R)
10-pHi - 10-pHeq Cef
(1)
The equilibrium degree of swelling, S, was determined as the ratio of the weight of the swelling solvent (mh - mo), to the weight of the dry gel (mo) (S ) (mh - mo)/mo). The polymer volume fraction in the equilibrium swollen state, V2, taking into account that swelling was isotropic and therefore any dimension changes in the same proportion, was determined as
V2 ) (Do/Dh)3
(2)
The swelling degree, S, and polymer volume fraction, V2, are related to the densities of the xerogel (F2) and the swelling solvent (F1):
V2 ) (1 + F2S/F1)-1
(3)
Porosity of Crosslinked Polymers Measurements of pore diameters performed with a caliper are less precise than gravimetric measurements, and therefore, the experimental uncertainty of V2 calculated with eq 2 was around 15% for samples swollen in HCl aqueous solutions or 5% for samples swollen in single solvents (in this case, the dry pellets were larger). The uncertainty was below 2% when V2 was calculated with eq 3. SEM measurements were carried out in a JEOL JSM 6400 or in a Hitachi S-3000N electron microscope and HR TEM measurements in a Tecnai-F20 G2. Solid specimens were prepared by freeze-drying (Heto CT/DW60E) the gel samples swollen at equilibrium in deionized water. For SEM measurements, they were coated with gold by means of a Balzers SCD 004 sputter-coater. Surface and bulk structures were observed. Inner parts of the specimens were revealed by simply cutting them with a bistoury or by cryogenic fracture. Several specimens of each sample were studied and a number of micrographs of any relevant portion were taken for each specimen. The textural characterization of all samples was accomplished by gas adsorption.25-27 The adsorption isotherms for N2 at -196 °C were measured with a Micromeritics ASAP 2010 equipment. From the N2 adsorption isotherms, the specific surface area (SBET) was calculated by applying the BrunauerEmmett-Teller (BET) method,26 and the external surface area (Sext) and micropore volume (V1) were calculated by the t-plot method.27 The pore size distributions and the total pore volumes (diameter less than 400 nm) (Vp) were calculated by means of the Density Functional Theory (DFT) method.25 In all of this calculus, the nitrogen molar volume was 34.67 cm3‚mol-1 and the nitrogen molecular area was 0.162 nm2‚molecule-1. Light scattering was measured with an SLM-Aminco Bowman AB2 spectrofluorometer configured for front face excitation. The preparation of samples synthesized at high conversion and swollen in HCl aqueous solutions was the same as that for protonation and swelling measurements. For light scattering measurements of PVI samples synthesized with different times of polymerization, the dry specimens were swollen by immersion in deionized water for one week. Swollen samples were placed in the fluorometer holder for solid samples between glass plates, which form a 30° angle with the incident beam. In this way, the overlap of light scattering (observed at 120°) with reflection (150°) was avoided. The excitation wavelength was set at 550 nm, a nonabsorbing wavelength, and the emission wavelengths were run between 540 and 560 nm. The scattering band thus observed was integrated. This method yields results with relatively large uncertainty due to the roughness of the polymer surface. However, it has several advantages for the soft materials employed here: (i) since there is no light transmission, it is available both for transparent and opaque samples; (ii) the observed light scattering does not depend on the sample thickness; and moreover, (iii) the samples are prepared by exactly the same protocol as that employed for the other measurements, which ensures that their characteristics are always the same. Results and Discussion PVI hydrogel samples synthesized (i) with different feed concentrations such that the product CT×[BA] is the same among them and (ii) under reaction conditions (e.g., 44 h at 70 °C) which result in almost total conversion, have the same permanent crosslinking density,8,23 in accordance with the polymer network model with pendant vinyl groups proposed by Tanaka.28 Following DSC measurements of the glass transition temperature, the permanent crosslinking density νe(Tg) of PVI samples
J. Phys. Chem. B, Vol. 112, No. 10, 2008 2811 synthesized at total conversion with the protocol described in the Experimental Section, was found to be 0.046 ( 0.006 mol/ L.23 The permanent crosslinking density of the same samples, synthesized with lower conversions, is larger.8 Polymer chains joined in cross-linking points form the primary or molecular polymer network. The mesh size of the primary polymer network, i.e., the average distance between adjacent knots, is one of the controlling parameters affecting the transport of drugs and proteins through crosslinked polymers.6 It can be calculated in terms of the permanent crosslinking density, νe, through the expression29
[
]
2F2 ξ ) V2-1/3 CN νe M o
1/2
l
(4)
where CN is the characteristic ratio and l is the bond length of the polymer skeleton. The characteristic ratio depends on Kθ, a parameter proportional to the unperturbed dimensions, as30
CN )
( ) Kθ φ
2/3
Mo 2l2
(5)
where φ represents the universal constant of the Flory formulation. As far as we are aware, Kθ has not been reported for PVI but it may be determined by making use of some published data31 on the intrinsic viscosity ([η]) in ethanol at 20 °C of 10 PVI monodisperse fractions of known molecular weight, M, by means of the Stockmayer-Fixman equation32
[η]/M 1/2 ) Kθ + 0.51φBM1/2
(6)
where B represents the second virial coefficient. Thus, we find Kθ ) 0.092 ( 0.050 cm3g-3/2mol1/2 and CN ) 9.9. For dry compact samples V2 ) 1 and therefore, the mesh size of polymer networks at total conversion is ξ ) 11.5 nm whereas for total conversion PVI samples swollen in deionized water, V2 ) 0.14 (on average) and ξ ) 22 nm. Smaller ξ values are expected for samples at low conversion. Freeze-dried samples reproduce the spatial distribution of the swollen polymer network with only a very slight contraction. Therefore, in a homogeneous sample, randomly distributed pores with 10 to 20 nm diameter, should be expected for SEM and TEM images.33 Pore diameters that are very similar to the mesh size and which decrease when the degree of crosslinking increases were previously observed for poly(vinyl alcohol) derivative hydrogels.34 Nevertheless, that is not the morphology observed for PVI. High-resolution TEM of freeze-dried samples provides imaging of the mesopores of the polymer networks in the swollen state. Figure 1, parts A and B, shows ovoid or quasi-spherical pores with diameters in the range of 10 to 100 nm. This type of pores was observed for samples synthesized with large CT whereas samples synthesized with small CT and large crosslinker concentration show elongated channel-like pores (Figure 1, parts C and D) in the scale of TEM measurements. Since pores with diameters close to 22 nm (the mesh size of the primary polymer network in the swollen state) were scarcely observed (Figure 1, parts A and B), it must be concluded that either such small nanopores collapse upon freeze-drying or disappear throughout postgel reactions when they are filled by polymer material, thereby forming compact nonporous regions. SEM micrographs of the same samples (Figure 2) show that the shape of pores is the same as observed by TEM (Figure 1) in spite of the quite different scales: 6 to 60 µm for SEM and 50 nm for TEM. As the total comonomer concentration increases (right to left in Figure 2) or the conversion increases (upper to
2812 J. Phys. Chem. B, Vol. 112, No. 10, 2008
Calvino-Casilda et al.
Figure 1. TEM micrographs of freeze-dried PVI samples: (A) 2h-PVI-6.1, (B) 44h-PVI-6.1, (C) 2h-PVI-1.8, and (D) 44h-PVI-1.8. The scale bar represents 50 nm.
lower row), a larger proportion of closed quasi-spherical voids appears. Samples with limiting compositions (PVI-6.1, PVI1.8) give quite similar micrographs in different regions while samples PVI-3.9 and PVI-2.8 show diverse morphologies depending on the observed region, as represented by Figure 2, parts B, C, and F, or G. In all cases, the regular spatial distribution of the polymer material may be described as a network formed by filaments or films surrounding voids. Such networks are secondary polymer networks because their filaments or walls are not single polymer chains and their knots are not crosslinker units, as in the primary polymer network; secondary networks are formed by entities larger than single chains and crosslinker units and give place to voids larger than the mesh size of primary networks. Secondary networks with mesh size around 0.5 µm (mesopores) were frequently observed in low or high conversion samples synthesized with intermediate crosslinker ratios (Figure 2, parts B, C, and F). Other isolated voids with larger size (about 3 µm diameter) interrupt the secondary network. The honeycomb-like morphology, formed by closed voids with 5 to 10 µm diameter and films with 0.3 µm thickness as walls, appear at intermediate compositions (Figure 2G) as well as in low conversion samples with high CT
(Figure 2A). The honeycomb transforms to closed vesicular voids (10 µm) separated by thick walls (5 µm) in samples obtained with the largest conversion and high CT (Figure 2E). If the average size of pores increases, as for sample PVI-6.1 with increasing conversion, then the number of pores per unit area decreases; in this way, the total volume of macropores (PVI-6.1) and mesopores (PVI-3.9 and PVI-2.8) decreases. At the other feed composition limit, the micrographs of sample PVI-1.8 show the secondary network surrounding voids of about 3 µm diameter. In that case, the number of voids per unit volume is so large that they are contiguous and therefore they are interconnected. The only remarkable change with increasing conversion is that walls are more completed and holes connecting next voids are less frequent. Previously reported changes of morphology of dry hydrogels (from interconnected spherical pores to tubular or channel-like pores) were ascribed to differences in the crosslinking density.4 However, in the case considered here, the observed morphology is slightly different, and the permanent crosslinking density of all the samples is the same. The differences in the morphology of PVI may be ascribed to differences in the inhomogeneous spatial distribution of the polymer formed throughout polym-
Porosity of Crosslinked Polymers
J. Phys. Chem. B, Vol. 112, No. 10, 2008 2813
Figure 2. SEM micrographs of freeze-dried PVI samples: (A) 2h-PVI-6.1, (B) 2h-PVI-3.9, (C) 2h-PVI-2.8, (D) 2h-PVI-1.8, (E) 44h-PVI-6.1, (F) 44h-PVI-3.9, (G) 44h-PVI-2.8, and (H) 44h-PVI-1.8.
Figure 3. SEM micrographs representative of the morphology of PVI samples on the submicrometric scale. The sample is PVI-3.9 synthesized during (A) 2 h and (B) 44 h.
TABLE 2: Textural Characteristics of PVI-CT Samples Synthesized for 2 h (low conversion) or for 44 h (high conversion), Determined through Nitrogen Adsorption Isotherms Y-2h S
PVI-6.1 PVI-2.8 PVI-1.8
Y-44h V
V
(m2/g)
(m2/ g)
(cm3/ g)
(cm3/g)
(m2/ g)
(m2/ g)
(cm3/ g)
(cm3/ g)
45.7 14.9 20.2
44.1 14.8 13.5
0.318 0.043 0.116
0.0001 0.0002 0.006
6.3 5.3 13.4
1.0 4.4 4.5
0.066 0.012 0.052
0.002 0.0003 0.004
BET
S
ext
V
V
p
1
erization, i.e., to differences in the phase separation mechanism, which depends on the feed composition irrespective of the crosslinking density. The cross-linking polymerization is always, by itself, concomitant with a phase separation process on the nanometer scale and for some systems, also on larger scales. It was suggested that quasi-spherical holes were developed by encapsulation of water droplets during polymerization.35,36 If such water droplets were in contact, holes would be intercon-
S
BET
S
ext
p
1
nected36 and otherwise they would be closed as in Figure 2E. Throughout post-gel reactions, the polymer grows into droplets filling the holes and thus, the total volume of the holes decreases8 and the wall thickness increases (Figure 2, parts A and E). In both TEM and SEM micrographs, closed quasispherical holes are more frequently observed for samples synthesized with the largest comonomer concentrations. Small water droplets whose size is close to the mesh size of the primary
2814 J. Phys. Chem. B, Vol. 112, No. 10, 2008 polymer network, may be responsible for the smallest holes observed by TEM, while the formation of large droplets or holes suggests phase separation by microsyneresis.20 In the other limit of feed composition, for low CT and high [BA], high-resolution SEM micrographs (Figure 3) show that crosslinked polymers are formed by polymer spheres,9,10 joined into filaments, which fuse to form a film. The thickness of polymer spheres, filaments and films in Figure 3 appears to be about 100 nm, i.e., 10× the average mesh size of the dry polymer network and 5× that of the network swollen in water. Therefore, each sphere contains several elastic chains and knots. The primary polymer network forms the elements of the secondary network, which has larger dimensions. For these samples, TEM shows the channels connecting voids of the secondary network and instead, SEM shows a clearer view of the filaments of the secondary network that, with increasing CT and conversion, become the closed walls. Nitrogen adsorption was used to evaluate the textural characteristics of freeze-dried samples. The obtained isotherms were of type I, except for samples 2h-PVI-6.1 and 44h-PVI2.8 which gave type II isotherms, according to the classification of Brunauer-Deming-Deming-Teller (BDDT).37 The type I isotherms correspond to systems with high affinity nitrogenadsorbent while isotherms of type II are characteristic of nonporous solids. Table 2 summarizes the textural parameters of PVI samples studied here. Because of the upper limit of pore size accessibility for N2 adsorption, macropores larger than 400 nm and superpores are not observed with this technique. The specific surface and total volume of pores are low but still on the same scale as those of typically porous materials such as polymer networks based on high internal phase emulsions,36 silica based xerogels38 or macroporous beads,39 and 1 order of magnitude below those of cross-linked polymers with high intrinsic porosity (PIM).5 The volume of PVI xerogels as compact materials (air-dried) is 1/F2 ) 0.82 cm3/g, and therefore, the total volume of pores represents from 28% (2h-PVI-6.1) to 1.4% (44h-PVI.2.8) of the total volume of the freeze-dried samples. It is remarkable that SBET and Sext have almost the same values (Table 2), and V1 is negligible with regard to Vp (except perhaps for sample PVI-1.8), which indicates that freeze-dried PVI samples exhibit quite small amounts of micropores (diameter less than 2 nm) or ultramicropores (diameter less than 0.7 nm), unlike PIM.5 Another conclusion drawn from Table 2 is that both the porosity (Vp) and the specific surface area (SBET) undergo a marked decrease throughout the postgel reactions. Such a decrease is particularly large for sample PVI-6.1, likely because that is the sample with the largest difference in yield between polymerization times of 2 or 44 h (Table 1). Swelling in deionized water is consistent with this result, as shown in Figure 4: S is much larger for the low-conversion samples, which show larger porosity (Table 2), and the largest difference between low and high conversion samples corresponds to PVI-6.1 (Figure 4). The dependence of SBET on feed composition does not parallel that of Vp. Although for the high conversion samples, the largest volume of pores is found for PVI-6.1, the specific surface is maximum for PVI-1.8. This is in accordance with the different pore shape that was observed in TEM and SEM micrographs (Figures 1 and 2). The surface to volume ratio (SBET/Vp) is minimum for PVI-6.1, both for low and high conversion samples, because their pores are closed and quasi-spherical, while mesopores of PVI-1.8 and PVI-2.8 are open and more similar in shape to cylindrical channels.
Calvino-Casilda et al.
Figure 4. Degree of swelling at equilibrium of PVI in deionized water for samples synthesized with different feed compositions during 1.5, 2, or 44 h.
The pore size distributions, obtained by applying the DFT method, are depicted in Figures 5 and 6. Samples PVI-6.1 and PVI-1.8 (Figure 5, parts A and C) are rather homogeneous, showing a broad unimodal pore size distribution centered at 86 and 148 nm, respectively. In Figure 5, parts A and C, the maximum remains almost invariant throughout the polymerization, i.e., the size of mesopores does not change. However, comparing the absolute values of the incremental volume in the case of samples with low (Figure 6A) and high conversion (Figure 6B), it is obvious that the volume of mesopores decreases significantly throughout polymerization, which must necessarily be due to the descent in the number of mesopores per unit mass. Only pore widths larger than the mesh size of the primary polymer network, 10 to 22 nm, are relevant to the distributions of Figure 5, parts A and C. Alternatively, the sample PVI-2.8, synthesized with intermediate feed composition, has a more heterogeneous pore size distribution, showing many different modes or secondary maxima (Figure 5B). While its profile at low conversion resembles that of PVI-6.1, at high conversion, the maximum shifts to larger pore widths and the distribution resembles more closely that of PVI-1.8 (Figure 5B). In view of the normalized distribution (Figure 5B), it might be concluded that, in this case, the relative contribution of micropores is important but the absolute values of the incremental volume shown in Figure 6, reveal that the volume of micropores is, as for the other samples, extremely low. In view of the fact that the optical appearance of PVI hydrogels varies from transparency to opacity, light scattering measurements were performed with respect to several variables. They afforded the following results: (a) light scattering decreases slightly with increasing conversion, the same as previously observed for polyacrylamide hydrogels;40 (b) immediately following polymerization, PVI samples synthesized at total conversion with the largest crosslinker concentrations are opaque, whereas those obtained with the largest total comonomer concentrations are transparent; (c) once they have reached the swelling equilibrium, all the samples are opaque or at least highly translucent; and (d) upon drying, all of them become transparent. (e) Samples synthesized with larger crosslinker concentration in the feed are more turbid (Figure 7A) and (f) the dependence of light scattering intensity on the polymer degree of protonation shows a maximum for low R values (Figure 7B).
Porosity of Crosslinked Polymers
J. Phys. Chem. B, Vol. 112, No. 10, 2008 2815
Figure 5. Pore size distributions normalized to the maximum value, determined from nitrogen adsorption measurements on freeze-dried PVI samples: (A) PVI-6.1, (B) PVI-3.9, and (C) PVI-1.8.
Figure 6. Pore size distributions determined from nitrogen adsorption measurement on freeze-dried PVI samples synthesized with different feed compositions during: (A) 2 h or (B) 44 h.
Figure 7. Light scattering intensity (LS) at 550 nm, of 44h-PVI-CT samples swollen at equilibrium in aqueous HCl solutions with different pHi, plotted as a function of (A) the cross-linker concentration in the feed, [BA], and of (B) the equilibrium degree of protonation, R.
These optical properties arise from the different refractive indices of the solvent and polymer network and may be understood by taking into account (i) the existence of scattering domains (pores filled with solvent) having dimensions in the range of one tenth of the wavelength of visible light (above 20 nm) and (ii) the direct dependence of the light scattering intensity on the size and number of the scattering domains.41,42 This suggests that with increasing pore size, while keeping the number of pores per unit mass constant (as occurs from immediately following polymerization toward swelling equilibrium and from dry toward swollen samples), the turbidity increases. Increasing conversion, the total pore-volume and specific surface decrease (Table 2) and as a consequence, turbidity decreases too.
It is clear from Figure 7 that the crosslinker concentration in the feed determines light scattering more strongly than the degree of protonation. Figure 7A shows that, for high conversion samples swollen at equilibrium in HCl aqueous solution with different pHi, the light scattering intensity increases monotonously with crosslinker concentration in the feed. For any sample, i.e., for any given crosslinker ratio, the light scattering intensity reaches a maximum for low degrees of ionization, around 10% (Figure 7B). Such trends may not be explained by changes of the degree of swelling or, equivalently, by changes of polymer concentration in the swollen state. For total conversion samples, increasing the crosslinker concentration, the area of the external surface increases (Table 2) and therefore the polymer-solvent interface increases as well. That could
2816 J. Phys. Chem. B, Vol. 112, No. 10, 2008
Calvino-Casilda et al.
Figure 8. Polymer volume fraction of PVI samples synthesized with CT ×[BA] = 0.24 and swollen in methanol (MeOH), ethanol (EtOH), deionized water and HCl aqueous solutions with different initial pH. (A) Samples were synthesized with two different feed compositions (CT ) 6.2 and 3.7 M) at different conversions ranging from 10 to 75%. (B) and (C) samples were synthesized with nine different feed compositions at total conversion. Error bars correspond to 5% in plots A and B and to 15% in plot C (see Experimental Section). The straight lines represent equality of V2 determined through gravimetric or dimensional measurements.
explain the dependence of light scattering intensity on the feed composition (Figure 7A). More difficult to understand is the decrease in light scattering with increasing degree of ionization above the maximum, an effect that was observed previously for similar systems.19 We can only speculate that counterion condensation taking place at degrees of protonation21,24 above 30%, modifies the characteristics of the polymer external surface, e.g., the refractive index. Swelling increases both the diameter and the mass of polymer slabs. The polymer volume fraction (V2) in the equilibrium swollen state of hydrogel slabs is frequently determined through gravimetric measurements1 with eq 3, while for microgels or nanogels, size measurements, often performed with optical microscopy or dynamic light scattering, are usually applied.43 Figure 8 compares those two methods of determining V2 for PVI slabs. Swelling degrees in methanol, ethanol, deionized water, HCl aqueous solutions of different initial pH (always with 0.01 M polymer effective concentration) of PVI samples synthesized with CT ×[BA] = 0.24, used here to calculate V2 with eq 3, were previously reported.8 Values of V2 determined through size measurements (reported here for the first time) are significantly larger than those determined gravimetrically with F2 ) 1.225 g/cm3 (picnometrically determined with acetone as the non-solvent).8 Xerogel densities ranging from 0.4 to 1.2 g/cm3 would be necessary in order to match both sets of results through eq 3. Such low values of the density, typical of foams, are meaningful because the density of the linear polymer in the bulk, picnometrically determined44 is 1.156 g/cm3 and PVI xerogels are more dense than the corresponding linear polymer.45,46 Very few previously reported results on hydrogels synthesized without porogen, account for both dimensional and gravimetric swelling measurements and the discrepancy of results for may also be observed for them.39,46-48 Hence, it must be concluded that to some extent, the solvent uptake increases the mass without increasing correspondingly the volume of the swollen hydrogel. This effect suggests the existence of pores filled with solvent that behaves as an external solvent, that is to say, that being outside the primary polymer network, it does not modify the mixing or the elastic osmotic swelling pressure. Such an effect, previously suggested for macroporous hydrogels in macroscopic slabs,2 may be more clearly understood for macroporous beads synthesized with a diluent as porogen agent: the solvent fills both the pores and the interstitial voids between beads (external solvent) and as a consequence, the swelling ratio determined gravimetrically is
even 6-fold that determined through dimensional measurements.39 This argument implies that only dimensional measurements of the polymer volume fraction are strictly valid for inclusion in the balance of the osmotic swelling pressure components, i.e., in the swelling equation. It would be interesting to establish the critical size of pores delimiting the crossover between behaviors of internal or external solvent. This will require further research. Conclusions Two sets of chemically crosslinked PVI hydrogels were synthesized during 2 or 44 h of polymerization. The protocol employed guarantees that all of the high conversion samples have the same permanent crosslinking density and different morphology. The research on these samples reveals that crosslinked polymers synthesized without pore forming agents exhibit pores in the swollen state whose size extends through several scales and that they can be observed in freeze-dried samples by means of several techniques. Nitrogen adsorption, SEM, and TEM afford coherent results, which support the following conclusions. The contribution of micropores is negligible or a minor contribution. For intermediate feed compositions, the porosity is minimal and corresponds mostly to mesopores, whereas for the two limit feed compositions, mesopores and macropores were observed. Porosity suffers a drastic descent throughout postgel reactions because (i) for intermediate feed compositions, the smallest mesopores disappear filled by new polymer material, and as a consequence, the number of mesopores per unit mass decreases and the average pore width increases; (ii) for the two limit feed compositions, the number of mesopores of any size decreases without changing the average mesopore width; and (iii) for the samples with the largest comonomer concentration, the number of macropores per unit mass decreases while increasing the average macropore width. The width of pores is mostly equal or larger than the mesh size of the primary or molecular polymer network. The variable determining the number, size, and shape of the pores is the feed composition employed in synthesis, better than the cross-linking density. Samples synthesized with large CT show closed quasi-spherical pores in all the scales, from the nm to several µm and mesopores form part of the thick and apparently compact walls of macropores. Samples synthesized with large crosslinker concentration show channel-like mesopores, which connect neighbor macropores formed by contiguous quasi-
Porosity of Crosslinked Polymers spherical voids with incomplete walls. Samples with intermediate compositions are more heterogeneous in the size of mesopores, which are interconnected, and show a very small proportion of macropores as closed quasi spherical voids. The differences in shape were ascribed to the mechanism of phase separation, taking place throughout polymerization. In view of these results, only pores larger in size or in number per unit mass, than those corresponding to the inherent porosity, may be ascribed to porogenic agents or methods. Everything points to the identification of scattering domains and pores filled with solvent: (i) throughout postgel reactions, the porosity diminishes and LS diminish too; (ii) following polymerization, transparent samples have swelling degrees below the equilibrium and exhibit pores small with regard to the wavelength of the incident light, while opaque samples are closer to equilibrium and have larger pores; (iii) increasing swelling by approaching equilibrium, the size of the pores filled with solvent increases and transparent samples become translucent; and (iv) upon drying, the pores collapse and samples homogenize thus becoming all them transparent. An important consequence of the existence of inherent porosity on swelling is that gravimetric measurements are not equivalent to dimensional measurements. It was suggested that only polymer volume fractions determined through size measurements are strictly valid to be considered in the balance of the osmotic swelling pressure components, i.e., in the swelling equation. Acknowledgment. This work was supported by DGI (Spain) under Grant Nos. CTQ2004-05706/BQU and CTQ2007-61007/ BQU. SEM measurements were carried out in the Centro de Microscopı´a Electro´nica “Luis Bru” of the Universidad Complutense and in Facultad de Ciencias, UNED. References and Notes (1) Geissler E., Ed. Functional Networks and Gels; Wiley-VCH: Weinheim, Germany, 2003; Macromolecular Symposia, Vol. 200. (2) Pastoriza, A.; Pacios, I. E.; Pierola, I. F. Polym. Int. 2005, 54, 1205. (3) Aroca, A. S.; Fernandez, A. J. C.; Ribelles, J. L. G.; Pradas, M. M.; Ferrer, G.; Pisis, P. Polymer 2004, 45, 8949. (4) Silva, R. D.; Ganzarolli, de Oliveira, M. Polymer 2007, 48, 4114. (5) Maffei, A. V.; Budd, P. M.; McKeown, N. B. Langmuir 2006, 22, 4225. (6) Ende, M. T. A.; Peppas, N. A. J. Controlled Release 1997, 48, 47. (7) Caykara, T.; Kucuktepe, S.; Turan, E. Polym. Int. 2007, 56, 532. (8) Pacios, I. E.; Pierola, I. F. Macromolecules 2006, 39, 4120. (9) Pacios, I. E.; Horta, A.; Renamayor, C. S. Macromolecules 2004, 37, 4643. (10) Pacios, I. E.; Pastoriza, A.; Pierola, I. F. Colloid Polym. Sci. 2006, 285, 263. (11) Wu, J. J.; Gross, A. F.; Tolbert, S. H. J. Phys. Chem. B 1999, 103, 2374. (12) Landry, M. R. Thermochim. Acta 2005, 433, 27.
J. Phys. Chem. B, Vol. 112, No. 10, 2008 2817 (13) Liu, Q.; Hedberg, E. L.; Liu, Z.; Bahulekar, R.; Meszlenyi, R. K.; Mikos, A. G. Biomaterials 2000, 21, 2163. (14) Ugur, S.; Elaissari, A.; Yargi, O.; Peckcan, O. Colloid Polym. Sci. 2007, 285, 423. (15) Weiss, M.; Van, Vliet, T.; Silberberg, A. J. Polym. Sci., Polym. Phys. Ed. 1979, 17, 2229. (16) Hsu, T. P.; Ma, D. S.; Cohen, C. Polymer 1983, 24, 1273. (17) Baselga, J.; Herna´ndez-Fuentes, I.; Pierola, I. F.; Llorente, M. A. Macromolecules 1987, 20, 3060. (18) Yamane, Y.; Ando, I.; Buchholz, F. L.; Reinhardt, A. R.; Schlick, S. Macromolecules 2004, 37, 9841. (19) Ikkai, F.; Shibayama, M. J. Polym. Sci. Part B: Polym. Phys. 2005, 43, 617. (20) Dusek, K.; Prins, W. AdV. Polym. Sci. 1969, 6, 1. (21) Molina, M. J.; Gomez-Anton, M. R.; Pierola, I. F. Macromol. Chem. Phys. 2002, 203, 2075. (22) Molina, M. J.; Gomez-Anton, M. R.; Pierola, I. F. J. Polym. Sci., Part B: Polym. Phys. 2004, 42, 2294. (23) Pacios, I. E.; Molina, M. J.; Gomez-Anton, M. R.; Pierola, I. F. J. Appl. Polym. Sci. 2007, 103, 263. (24) Molina, M. J.; Gomez-Anton, M. R.; Pierola, I. F. J. Phys. Chem. B 2007, 111, 12066. (25) Oliver, J. P. J. Porous Mater. 1995, 2, 9. (26) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309. (27) Horvath, G.; Kawazoe, K. J. Chem. Eng. Jpn. 1983, 16, 470. (28) Bromberg, L.; Grosberg, A. Y.; Matsuo, E. S.; Suzuki, Y.; Tanaka, T. J. Chem. Phys. 1997, 106, 2906. (29) Peppas, N. A. Hydrogels in Medicine and Pharmacy; CRC Press: Boca Raton FL, 1986. (30) Flory, P. J. Statistical Mechanics of Chain Molecules; Interscience: New York, 1969. (31) Savin, G.; Burchard, W.; Luca, C.; Beldie, C. Macromolecules 2004, 37, 6565. (32) Zimm, B. H.; Stockmayer, W. H.; Fixman, M. J. Chem. Phys. 1953, 21, 1716. (33) Chiessi, E.; Cavallieri, F.; Paradossi, G. J. Phys. Chem. B 2007, 111, 2820. (34) Ruiz, J.; Mantecon, A.; Cadiz, V. J. Polym. Sci.: Part B: Polym. Phys. 2003, 41, 1462. (35) Wong, K. H.; Davis, T. P.; Barner-Kowollik, Ch.; Stenzel, M. H. Polymer 2007, 48, 4950. (36) Livshin, S.; Silverstein, M. S. Macromolecules 2007, 40, 6349. (37) Brunauer, S.; Deming, L. S.; Deming, W. E.; Teller, E. J. Am. Chem. Soc. 1940, 62, 1723. (38) Aguado-Serrano, J.; Rojas-Cervantes, M. L.; Lopez-Peinado, A. J.; Gomez-Serrano, V. Micropor. Mesopor. Mater. 2004, 74, 111. (39) Gomez, C. G.; Alvarez-Igarzabal, C. I.; Strumia, M. C. Polymer 2004, 45, 6189. (40) Kuru, E. A.; Orakdogen, N.; Okay, O. Eur. Polym. J. 2007, 43, 2913. (41) Perez-Dorado, A.; Pierola, I. F.; Baselga, J.; Gargallo, L.; Radic, D. Makromol. Chem.-Macromol. Chem. Phys. 1989, 190, 2975. (42) Kerker, M. The Scattering of Light and Other Electromagnetic Radiations; Academic Press: New York, 1969. (43) Kazakov, S.; Kaholek, M.; Gazaryan, I.; Krasnikov, B.; Miller, K.; Levon, K. J. Phys. Chem. B 2006, 110, 15107. (44) Morales, E.; Gomez-Anton, M. R.; Pierola, I. F. J. Appl. Polym. Sci. 1995, 58, 1945. (45) Valencia, J.; Pierola, I. F. Eur. Polym. J. 2001, 37, 2345. (46) Xue, W.; Huglin, M. B.; Liao, B. Eur. Polym. J. 2006, 42, 3015. (47) Xue, W.; Champ, S.; Huglin, M. B. Polymer 2001, 42, 3665. (48) Durmaz, S.; Okay, O. Polymer 2000, 41, 3693.
