Mesoporous Hydrogels: Revealing Reversible Porosity by

Mar 22, 2010 - The observed critical cross-linker ratio for pore stability compared favorably with a simple estimate of the critical cross-linker dens...
1 downloads 8 Views 2MB Size
pubs.acs.org/Langmuir © 2010 American Chemical Society

Mesoporous Hydrogels: Revealing Reversible Porosity by Cryoporometry, X-ray Scattering, and Gas Adsorption Jens Weber*,†,‡ and Lennart Bergstr€om† †

Department of Materials and Environmental Chemistry, Arrhenius Laboratory, Stockholm University, S-106 91 Stockholm, Sweden, and ‡Max Planck Institute of Colloids and Interfaces, Department of Colloid Chemistry, Research Campus Golm, Potsdam, D-14424 Germany Received January 20, 2010. Revised Manuscript Received March 10, 2010 Mesoporous poly(2-hydroxyethyl methacrylate-co-ethylene glycol dimethacrylate) networks, with cross-linker contents ranging from 100 to 5 mol %, were prepared using a hard-templating approach. The imbibition of the silica pellets with a monomer/cross-linker mixture resulted in mesoporous gels with a pore size of ∼10 nm, which corresponds well with the average size of the fumed silica particles (10-11 nm). The highly cross-linked materials showed permanent surface areas of up to 230 m2 g-1 and porosities up to ∼33 vol %. The porosity of the hydrogels was investigated in both dry and water-saturated state by nitrogen sorption, cryoporometry, and small-angle X-ray scattering (SAXS). It is only polymeric materials that contain 50 mol % or more of the cross-linker that showed a significant porosity after evaporative drying. Freeze-drying is able to preserve the porosity also for hydrogels of intermediate cross-linker content, but the pores of the materials of low cross-linker content collapses completely upon solvent removal. The observed critical cross-linker ratio for pore stability compared favorably with a simple estimate of the critical cross-linker density needed to make the material sufficiently stiff to withstand the Laplace pressure during solvent removal. Analysis of the hydrogels in the water swollen state revealed that gels having cross-linker contents down to 5 mol % still possessed mesoporosity. The pores got less defined at very low cross-linker contents, while their size was rather constant at intermediate to high cross-linking densities. Closed pores could be reopened upon swelling, which suggests that the observed pore collapse upon drying may be at least partly reversible.

Introduction Polymer (hydro)gels are of interest in a number of applications, for example, biomolecule separation,1,2 microfluidic devices, sensors, and indicators,3-5 and in controlled release of drugs and other active molecules.2,6,7 The nanoscale architecture, macroscopic morphology, and porosity are parameters that play a pivotal role in determining the mechanical and optical properties and in the control of the diffusion kinetics of different kinds of solutes. Polymeric hydrogels with nanoscale architecture can be produced in different forms, for example, as thin films and as discrete micro- and nanogels,8,9 where the diffusion path is relatively small. The particle size has also a strong effect on the response time for stimulus-sensitive gels which can undergo volume-phase transitions.10,3,11 While macroporous polymer gels, that is, gels *To whom correspondence should be addressed. Telephone: þ49-3315679569. Fax: þ49-331-5679502. E-mail: [email protected]. (1) Kumar, A.; Srivastava, A.; Galaev, I. Y.; Mattiasson, B. Prog. Polym. Sci. 2007, 32, 1205–1237. (2) Roy, I.; Gupta, M. N. Chem. Biol. 2003, 10, 1161–1171. (3) Richter, A.; Paschew, G.; Klatt, S.; Lienig, J.; Arndt, K.; Adler, H. P. Sensors 2008, 8, 561–581. (4) Eddington, D. T.; Beebe, D. J. Adv. Drug Delivery Rev. 2004, 56, 199–210. (5) Zhang, C.; Xing, D.; Li, Y. Biotechnol. Adv. 2007, 25, 483–514. (6) Oh, J. K.; Drumright, R.; Siegwart, D. J.; Matyjaszewski, K. Prog. Polym. Sci. 2008, 33, 448–477. (7) Hoare, T. R.; Kohane, D. S. Polymer 2008, 49, 1993–2007. (8) Hegewald, J.; Schmidt, T.; Eichhorn, K.; Kretschmer, K.; Kuckling, D.; Arndt, K. Langmuir 2006, 22, 5152–5159. (9) Mendrek, S.; Mendrek, A.; Adler, H.; Dworak, A.; Kuckling, D. Macromolecules 2009, 42, 9161–9169. (10) Shibayama, M.; Tanaka, T. Adv. Polym. Sci. 1993, 109, 1–62. (11) Li, Y.; Tanaka, T. Annu. Rev. Mater. Sci. 1992, 22, 243–277. (12) Buchmeiser, M. R. Polymer 2007, 48, 2187–2198. (13) Okay, O. Prog. Polym. Sci. 2000, 25, 711–779. (14) Izaak, T. I.; Vodyankina, O. V. Russ. Chem. Rev. 2009, 78, 77–88.

10158 DOI: 10.1021/la100290j

with pore sizes larger than 50 nm, are well-known,12-16 the preparation of mesoporous polymers that contain stable pores of sizes between 2 and 50 nm remains a challenge. While porosity in the zeolite and material science community is generally defined as the permanent porosity of the dried materials, the porosity of polymer gels is often evaluated in the solvent swollen state. It is a general observation that the mesopores in polymeric materials frequently collapse when the gel is dried or subjected to certain solvents.17-22 Pore collapse is related to the stiffness and flexibility of the polymeric material; it is frequently observed that the pores only are stable above a critical degree of cross-linking.17-19,21 Furthermore, it has been shown that the drying protocol has a major influence on the preservation of the porosity.23 However, a detailed understanding of the stability of the pores in polymeric materials is lacking, and the number of systematic studies is sparse. Mesoporous polymers can be synthesized via templating approaches and selective removal of certain components or segments.24 Another possible soft-templating pathway is the use (15) Svec, F. J. Sep. Sci. 2004, 27, 747–766. (16) Hentze, H. P.; Antonietti, M. Curr. Opin. Solid State Mater. Sci. 2001, 5, 343–353. (17) Weber, J.; Bergstrom, L. Macromolecules 2009, 42, 8234–8240. (18) Guo, F.; Andreasen, J. W.; Vigild, M. E.; Ndoni, S. Macromolecules 2007, 40, 3669–3675. (19) Cavicchi, K. A.; Zalusky, A. S.; Hillmyer, M. A.; Lodge, T. P. Macromol. Rapid Commun. 2004, 25, 704–709. (20) Muralidharan, V.; Hui, C. Macromol. Rapid Commun. 2004, 25, 1487–1490. (21) Weber, J.; Antonietti, M.; Thomas, A. Macromolecules 2007, 40, 1299– 1304. (22) Szewczykowski, P. P.; Andersen, K.; Schulte, L.; Mortensen, K.; Vigild, M. E.; Ndoni, S. Macromolecules 2009, 42, 5636–5641. (23) Hasegawa, J.; Kanamori, K.; Nakanishi, K.; Hanada, T.; Yamago, S. Macromol. Rapid Commun. 2009, 30, 986–990. (24) Olson, D. A.; Chen, L.; Hillmyer, M. A. Chem. Mater. 2008, 20, 869–890.

Published on Web 03/22/2010

Langmuir 2010, 26(12), 10158–10164

Weber and Bergstr€ om

of well-defined phase separation processes in controlled radical polymerizations to obtain mesoporous polymers.25,23 The use of hard-templates, for example, mesostructured silica, for the synthesis of mesoporous polymers17,21,26-30 has several advantages, as it does not require the use of (sometimes) expensive block copolymers or the limitations of narrowly defined conditions for phase separation. Recently, we used mesoporous pellets, derived from fumed silica, as a hard-template for the synthesis of mesoporous poly(styrene).17 In this study, we show how the hard templating route using mesoporous silica pellets can be extended toward mesoporous hydrogels, namely, poly(2-hydroxyethyl methacrylate-co-ethylene glycol dimethacrylate) networks possessing pore sizes of ∼10 nm. We characterized the polymer hydrogels in both the dry and wet (swollen) state by gas sorption, small-angle X-ray scattering (SAXS), and cryoporometry and use the results to discuss the effect of cross-linker concentration on the pore stability. While nitrogen sorption can only probe the porosity in the dry state, cryoporometry and SAXS also allow an investigation of the porosity in the swollen state. Cryoporometry, i.e. the analysis of the freezing point depression of solvents confined in mesopores, will be used in combination with the N2 adsorption and SAXS results to discuss the effect of cross-linker concentration on the pore stability.

Materials and Methods Materials. Fumed silica (primary particle size = 14 nm, Aldrich S5505), ethylene glycol dimethacrylate (EGDMA, 98%), 2-hydroxyethyl methacrylate (HEMA, 97%), and 2,20 azobis(2-methylpropionitrile) (AIBN, 98%) were purchased from Aldrich. EGDMA and HEMA were purified by passing through a column of basic activated alumina prior to use. Synthesis of Mesoporous Hydrogels. Approximately 150 mg of fumed silica were pressed into a pellet using a hydraulic pellet press at an average uniaxial pressure of 9 MPa. An amount of 1.5 g of the monomer mixture containing 2 wt % AIBN was deaerated by bubbling with nitrogen for a few minutes. The silica pellets were infiltrated with this mixture, and excess monomer was removed. The saturated pellets were transferred to 10 mL flasks, which were purged with argon. After sealing the flask with a septum, polymerization was initiated by heating to 65 C for 18-20 h. After cooling to room temperature, the inorganic/ organic hybrid material was broken into smaller pieces and the silica was removed by treatment with a 1 M sodium hydroxide solution for 4 days.31 Afterward the sodium hydroxide treatment, the mesoporous hydrogels were purified by extensive washing with deionized water and finally dried under vacuum at room temperature or by freeze-drying. Methods. Nitrogen sorption isotherms were measured at 77K using a Micromeritics ASAP 2020 machine after degassing the samples under vacuum for at least 20 h. Fourier Transform infrared spectra were acquired using a Varian 670-IR spectrometer equipped with a GoldenGate attenuated total reflectance (ATR) device. Thermogravimetric analysis (TGA) was performed under air using a Perkin-Elmer TGA7 instrument. (25) Hasegawa, J.; Kanamori, K.; Nakanishi, K.; Hanada, T.; Yamago, S. Macromolecules 2009, 42, 1270–1277. (26) Deryzo-Marczewska, A.; Goworek, J.; Zgrajka, W. Langmuir 2001, 17, 6518–6523. (27) Goltner, C. G.; Weissenberger, M. C. Acta Polym. 1998, 49, 704–709. (28) Johnson, S. A.; Ollivier, P. J.; Mallouk, T. E. Science 1999, 283, 963–965. (29) Kim, J. Y.; Yoon, S. B.; Kooli, F.; Yu, J. S. J. Mater. Chem. 2001, 11, 2912– 2914. (30) Thomas, A.; Goettmann, F.; Antonietti, M. Chem. Mater. 2008, 20, 738– 755. (31) In principle, there is no need to break the pellet into pieces for removal of the silica. This was done in order to isolate parts of the hybrid material for analytical purposes.

Langmuir 2010, 26(12), 10158–10164

Article Differential scanning calorimetry (DSC) analysis was performed using a Mettler-Toledo DSC1 instrument. For a typical cryoporometry experiment, ∼5 mg of water swollen polymer and 2.5 μL of water were sealed in an aluminum pan. DSC experiments were conducted at varying heating rates (0.5-2 K/min). Typically, the samples were first frozen to -50 C, then heated to -0.25 C, cooled to -50 C again, and heated to 25 C. After the DSC measurements, the pans were weighed. The water was evaporated under vacuum until constant weight, and the weight of the dry sample was determined. Densities of the gels were measured using a pycnometer (volume = 5 cm3) at 20 C. Cyclohexane was used as liquid, as it is a nonsolvent for poly(HEMA). SAXS curves were recorded at room temperature with a NONIUS rotating anode instrument (4 kW, Cu KR) with pinhole collimation and a MARCCD detector (pixel size = 79) which was calibrated using silver behenate. The distance between the sample and detector was 74 cm, covering a range of the scattering vector s = 2/λ sin(θ) = 0.04-0.7 nm-1 (2θ = scattering angle, λ = 0.154 nm). The observed patterns were corrected for empty-beam scattering. The 2D diffraction patterns were transformed into a 1D radial average of the scattering intensity using the Fit2D software. The SAXS patterns, which were used for quantitative analysis, were corrected for scattering contributions arising from 3D electron density fluctuations as suggested by Perret and Ruland before being further analyzed.32 Measured intensities are not absolute but intercomparable, as comparable sample amounts and measurement times were used. The Porod length was determined numerically from the corrected scattering patterns using following relation:32 lp ¼

R¥ 2π 0 s2 IðsÞ ds π3 lim s4 IðsÞ

ð1Þ

s f¥

The Porod length is related to the pore size Ælporeæ and to the wall size Ælwallæ through the following relation, where φ is the porosity. 1 1 1 ¼ ¼ lp φÆlwall æ ð1 - φÞÆlpore æ

ð2Þ

Ælporeæ is the number-averaged pore size, that is, an average over all possible distances between the pore walls. Ælwallæ is derived similarly and describes an average over all the possible lines, which could be laid through a pore wall, connecting different pores.

Results and Discussion Mesoporous hydrogels have been synthesized by adapting a hard-templating routine previously developed for the synthesis of mesoporous polystyrene.17 Commercial, nanoparticulate fumed silica powder was pressed into pellets. The pellets contain a large amount of interstitial voids, and previous studies17 using nitrogen sorption and SAXS showed that the pellets possess a well-defined mesoporosity with an average porosity φ of 0.69 and a pore size Ælporeæ of about 24 nm. Silica/polymer hybrid materials were obtained by thermal polymerization of the monomer/initiator mixture within the mesopores of the silica pellet. The hybrid materials contained between 52 and 56 wt % polymer as determined by TGA. The measured polymer content in the as-polymerized inorganic/organic hybrid materials corresponds well with the estimated polymer content that would result from complete filling of the pores of the hard template (∼54 wt %). This suggests that the monomer/initiator mixture is able to imbibe the entire volume of the porous silica pellet. (32) Perret, R.; Ruland, W. Kolloid Z. Z. Polym. 1971, 247, 835–843.

DOI: 10.1021/la100290j

10159

Article

Weber and Bergstr€ om

Figure 2. SAXS patterns of freeze-dried mesoporous poly(2hydroxyethyl methacrylate-co-ethylene glycol dimethacrylate) networks. The shown scattering patterns were corrected for the empty beam but not for 3D electron density fluctuations (see Supporting Information Figure S2 for those). The patterns were plotted without vertical offset; that is, the intensity difference between individual patterns is a qualitative one.

Figure 1. Nitrogen sorption isotherms (a) and pore size distribution (BJH analysis, adsorption branch) (b) of freeze-dried and evaporative dried mesoporous poly(2-hydroxyethyl methacrylateco-ethylene glycol dimethacrylate) networks.

The silica was removed by etching with aqueous 1 M sodium hydroxide solution at room temperature. IR spectroscopy studies of the remaining polymer suggest that no significant hydrolysis of the acrylates is induced by this procedure (Supporting Information Figure S1). Hydrogels with varying cross-linker (EGDMA) content (100, 65, 50, 37, 25, 17.5, 10, and 5 mol % of the total monomer content) were prepared and named according to the cross-linker content; that is, the material named HG100 contains 100 mol % EGDMA, the material HG50 contains 50 mol % EGDMA, and so forth. Analysis of the Porosity in the Dry State. Nitrogen sorption isotherms have been determined for the synthesized polymeric materials after the water was removed by conventional evaporative drying and freeze-drying, respectively. Figure 1 shows the measured nitrogen sorption isotherms as well as the derived pore size distributions (PSDs). The nitrogen sorption results clearly show that it is only polymeric materials that contain 50 mol % or more of EGDMA that showed a significant porosity. The pores collapse or become completely inaccessible upon solvent removal for all the materials with lower cross-linking ratios, irrespective of the applied drying method. The porosity of HG100 (SBET ∼ 230 m2g -1, Vpore ∼ 0.48 cm3 -1 g ) and HG65 (SBET ∼149 m2 g-1, Vpore ∼0.33 cm3 g-1) showed only little sensitivity to the applied drying method. This is somewhat intuitive, as the fully cross-linked material has the highest stiffness, and consequently, it should provide the highest resistance against pore collapse or closure. The impact of the applied 10160 DOI: 10.1021/la100290j

drying method is much higher in the case of lower cross-linking densities, as those materials possess a higher flexibility. Freezedrying can help to preserve the loose porous structure in such cases. Indeed, it was found that freeze-drying of the material HG50 yielded a sample with a specific surface area of 112 m2 g-1, in contrast to only 62 m2 g-1 when the water has been removed by evaporative drying. The insensitivity of the fully cross-linked gel to the applied drying method is, however, in partial disagreement with results of Hasegawa et al., who showed that even for fully cross-linked poly(acrylamide) gels there is an influence of the drying protocol.23 We speculate that the size of the polymeric pore walls might play a major role in the preservation of the permanent porosity, but further studies are necessary to clarify this issue. Interestingly, Figure 1 shows that the pore size distribution was centered around 9-10 nm for all the materials, irrespective of the cross-linker concentration and the employed solvent removal technique. This pore size corresponds well with the average size of the fumed silica particles (10-11 nm by SAXS), which proves that the replication process was successful. In order to elucidate the phenomenon of pore collapse in more detail, we performed SAXS measurements of freeze-dried and evaporative dried materials. SAXS is a versatile method for the analysis of mesoporous materials.17,21,33 Figure 2 shows that the porous polymers that had been freeze-dried display a strong scattering signal with a pronounced Porod behavior, that is, a decay of the scattering intensity with a power of -4, for materials with cross-linker contents down to 17.5 mol %. The observation of a Porod behavior is typical for the presence of a two-phase system (such as a porous material), irrespective of the presence of any well-defined pore geometry. For HG10, only a weak scattering signal was observed and it was not possible to evaluate any information from the observed pattern. This was also the case for the material HG5 (data not shown). The evaporative dried samples also displayed a Porod behavior for materials with a cross-linker content above 17.5 mol % (see Supporting Information, Figure S2). (33) Smarsly, B.; Groenewolt, M.; Antonietti, M. Prog. Colloid Polym. Sci. 2005, 130, 105–113.

Langmuir 2010, 26(12), 10158–10164

Weber and Bergstr€ om

Article

Table 1. Permanent Porosity Characteristics of Materials HG100-HG25a entry

SBET (FD) [m2 g-1]b

SBET (ED) [m2 g-1]c

lp (FD) Vpore (FD) Dpore (FD) φ [nm] [cm3 g-1] [nm]d

HG100 230 224 0.48 9.19 0.33 6.56 HG65 149 140 0.33 8.95 0.26 6.87 HG50 112 63 0.23 8.25 0.19 5.95 HG37 6.6 n.m. 0.01 8.16 0.01 6.3 HG25 4.3 3.2 0.01 0.01 6.84 a Materials HG17.5-HG5 did not feature any permanent porosity that could be probed by gas sorption. b FD: freeze-dried. c ED: evaporative dried. d Determined using the BJH model (adsorption branch).

From the SAXS data of the HG25-HG100 materials, it was possible to determine the Porod length lp, which is a characteristic length of the two-phase system under investigation. The Porod length is connected to Ælporeæ and Ælwallæ through the porosity φ.17,33 However, the porosity is difficult to determine independently on materials with partially collapsed or nonaccessible pores; hence, we have only calculated the pore and wall sizes for the fully crosslinked material, HG100. The calculated values of Ælporeæ = 9.3 nm, Ælwallæ = 19.1 nm, and SSAXS = 190.2 m2 g-1 indicate that the templating procedure indeed yielded an inverse replica of the pressed silica pellet, as the average wall size of the silica pellet Ælwallæ = 10.7 nm corresponds well with the pore size of the porous polymer. Table 1 shows that, for materials having cross-linker contents between 25 and 100 mol %, the Porod length was rather constant (lp ∼ 6-7 nm). This suggests that the pores, which do not collapse, keep their size. In other words, the pore collapse or closure can be regarded as a critical phenomena; that is, a pore is either collapsing completely or remains unaffected upon solvent removal. This type of critical mesopore collapse behavior has been observed previously,17-20 and is possibly related to an inhomogeneous cross-linking density which is a common feature of polymer networks synthesized by free radical polymerization.34,35 The combination of the SAXS and nitrogen adsorption data also suggests that porous polymers of intermediate cross-linking density possess closed porosity in the dry state. Closed pores cannot be probed by nitrogen adsorption, but they still give rise to a strong signal in the SAXS measurements as they provide scattering contrast. In order to find further evidence for the hypothesis of closed porosity, we measured the densities of selected samples using a pycnometer and cyclohexane as a nonsolvent. The densities of the porous materials were compared with the measured densities of the corresponding bulk materials of the same composition. For the material HG65, the density of the mesoporous sample was slightly lower than the bulk density (Fporous = 0.96 g cm-3, Fbulk = 1.1 g cm-3), which indicates already some amount of closed porosity. In the case of intermediate cross-linking densities, the difference between the densities was even larger. For HG25, we found Fporous = 1.1 g cm-3 and Fbulk = 1.34 g cm-3. If the cross-linker density was further lowered, the difference between the porosities got lower (HG17.5: Fporous =1.26 g cm-3, Fbulk = 1.4 g cm-3), and in the case of the lowest cross-linking density, that is, in the case of complete pore collapse, both samples had the same density (HG5: Fporous =Fbulk =1.38 g cm-3). It should be noted that the densities of the weakly cross-linked materials are in good agreement with reported densities of poly(HEMA).36 (34) Bastide, J.; Leibler, L. Macromolecules 1988, 21, 2647–2649. (35) Shibayama, M. Bull. Chem. Soc. Jpn. 2006, 79, 1799–1819. (36) Young, C.; Wu, J.; Tsou, T. J. Membr. Sci. 1998, 146, 83–93.

Langmuir 2010, 26(12), 10158–10164

Further evidence for the presence of closed porosity comes from solvent penetration experiments (Supporting Information Figures S5 and S6). With the exception of HG5 and HG10, which are mainly transparent, all of the dried gels are opaque. If subjected to cyclohexane, which cannot swell the gels but can enter open pores, materials with open porosity (e.g., HG65) became transparent due to the qualitatively matching refractive indices of the polymer gel and the penetrant. On the contrary, materials with closed porosity (e.g., HG25 and HG17.5) remained opaque when subjected to cyclohexane. This can be interpreted as indirect proof of a nonaccessible microstructure. If the porosity was fully collapsed (e.g., material HG5), the material was transparent in the dry state as well as if dispersed in cyclohexane. Unfortunately, we were not able to quantify the amount of closed porosity yet, but future studies will focus on this task. It should be noted that this closed porosity might be related to the so-called “traced” porosity, which was reported for other examples of mesoporous polymers.17,18 Theoretical Background of Pore Collapse. Muralidharan and Hui analyzed the stability of nanoporous elastic materials and derived on the basis of Lame’s solution for stresses in cylinders with an infinite shell and the well-known Laplace pressures associated with nanoporous materials a stability criterion.20 For the case of incompressible materials, they proposed a pore stability criterion: 2Erp >1 3γ

ð3Þ

where E is the modulus of the material, rp is the pore radius, and γ is the surface free energy of the material. Assuming the surface free energy γ of the hydrogels is ∼70 mN m-1,37 and the pore radius is ∼5 nm, this indicates that the modulus should be higher than ∼21 MPa for the pores to be stable in the mesoporous poly(2-hydroxyethyl methacrylate-co-ethylene glycol dimethacrylate) networks. The above estimate is based on the assumption that the drying process is a two-step process, where first the mesopores are emptied before the water is removed from the pore walls.38 We have made an attempt to estimate the critical cross-linker density needed to yield the necessary modulus from a simple relation between the modules, E, and the corresponding average molecular weight of polymer chains between the network junctions, Mc. Mc 

3FRT E

ð4Þ

where R is the gas constant, T is the absolute temperature, F is the polymer density, and E is the modulus. This relation is valid for elastomers. Based on the postulated two-step drying process, we assume the water saturated hydrogel pore walls to be in a rubberlike state, except they are fully crosslinked. Assuming the density of the swollen networks to be fixed at 1.1 g cm-3 and the temperature to be 298 K, one arrives at a critical Mc value of ∼400 g mol-1. As the molecular weight of one HEMA molecule is 130 g mol-1, this suggests that approximately every fourth molecule should be a cross-linker. This result (37) Andrade, J.; Ma, S.; King, R.; Gregonis, D. J. Colloid Interface Sci. 1979, 72, 488–494. (38) This assumption is based on analogy to the desorption of adsorbates (e.g., nitrogen) from bimodal porous systems. Generally, the adsorbate is first removed from bigger pores as the impact of curvature and wall-adsorbate interactions is lower for those.

DOI: 10.1021/la100290j

10161

Article

Weber and Bergstr€ om

Figure 3. Cryoporometry data on mesoporous poly(2-hydroxyethyl methacrylate-co-ethylene glycol dimethacrylate) networks containing different contents of cross-linker. Typical heating (a) and cooling traces (b) of water swollen materials as observed by DSC (heating rate 2 K min-1; sealed pans); (c) pore size distributions (PSDs) of the hydrogels, determined from the melting curves using relation 7; (d) average pore radius of mesoporous hydrogels in dependence on the cross-linker content (determined using eq 6 with δm = 1.1 nm).

corresponds reasonably well with the experimental observation of no detectable permanent porosity already for materials of intermediate cross-linking density. The Porosity in the Solvent-Saturated State. It is important to also determine the porosity of the materials in the wet, solvent swollen state. This can be done by scattering methods as well as by cryoporometry, which relies on the analysis of the freezing or melting point depression of a solvent confined within the pores. The shift of the phase transition temperature in mesoporous materials depends on the pore curvature and is expressed in the Gibbs-Thomson equation (eq 5): ΔTm ðrp Þ ¼

1 2T0 γsl νl rp Δhb

ð5Þ

where γsl is the interfacial tension of the solid/liquid interface, νl is the molar free volume of the liquid, and Δhb the melting enthalpy in the unconfined state at the bulk coexistence temperature T0. There are two major techniques for cryoporometry experiments: differential scanning calorimetry (DSC)39 and nuclear magnetic resonance (NMR).40 While the DSC directly probes the thermal properties of the solvent, the NMR method makes use (39) Landry, M. Thermochim. Acta 2005, 433, 27–50. (40) Petrov, O. V.; Furo, I. Prog. Nucl. Magn. Reson. Spectrosc. 2009, 54, 97– 122.

10162 DOI: 10.1021/la100290j

of the different mobilities of molecules in the liquid or solid state and the respective correlation of the signal intensity to determine the phase transition. Although significant progress has been made in the area of cryoporometry during the past decade, studies on the use of cryoporometry as a versatile technique for the analysis of mesoporous polymers are sparse.39-43 In this study, we have used DSC to evaluate the melting and freezing of water inside the mesopores while in contact with bulk ice.39 Additionally, we performed SAXS measurements of water swollen gels in order to crosscheck and verify the cryoporometry results. Figure 3a and b shows that it was possible to identify welldefined peaks due to the freezing or melting of water confined in nanopores for all materials except for HG5. The melting/freezing temperature only decreases slightly with decreasing cross-linking density except for the HG10 polymer that shows a significant decrease in the freezing and melting temperature compared to the porous materials prepared at higher cross-linker concentrations. While the melting point depression is slightly higher than that for samples of higher cross-linking density, the freezing of confined (41) Wang, J.; Gonzalez, A. D.; Ugaz, V. M. Adv. Mater. 2008, 20, 4482–4489. (42) Ishikiriyama, K.; Todoki, M.; Kobayashi, T.; Tanzawa, H. J. Colloid Interface Sci. 1995, 173, 419–428. (43) Iza, M.; Woerly, S.; Danumah, C.; Kaliaguine, S.; Bousmina, M. Polymer 2000, 41, 5885–5893.

Langmuir 2010, 26(12), 10158–10164

Weber and Bergstr€ om

Article

water took place just at very low temperatures, that is, lower than -40 C. This points to the presence of very small domains of pure water, eventually coupled with some kind of antifreezing activity of dangling polymer chains. The freezing traces show a non-Gaussian peak shape with a steep increase at the onset of the freezing process. This suggests that these measurements were affected by a supercooling effect.44,45 The freezing temperatures were slightly increased by a reduction of the cooling rate from 2 to 1 K min-1, but the peak shape was still non-Gaussian. As a consequence, we did not perform any detailed evaluation of the freezing experiments but concentrated on the melting traces. Those were not sensitive to a change of the heating rate from 2 to 1 K min-1 (see Supporting Information Figure S4), which suggests that the observed melting process was near equilibrium. It is possible to determine the pore radius as well as a pore size distribution from the DSC traces.39 The pore radius is given in eq 6, which was empirically developed and optimized using combined thermoporometry and nitrogen sorption data on mesoporous silica gels:39,46,47 rp ¼ -

33:30 þ 0:32 þ δm ΔT

ð6Þ

ΔT denotes the melting point depression in Kelvin, taken as the difference between the peak melting temperature of confined water and the onset temperature of the melting peak of bulk water.39,44 δm denotes the thickness of the nonfreezing interfacial layer and can take values between 0.5 and 2.2 nm. If not stated otherwise, we assume the layer thickness to be 1.1 nm, which corresponds to ∼3 layers of water.42 The pore size distribution dVp/drp can be determined by the following relation 7:39 dVp dQ dt dðΔTÞ 1 ¼ dt dðΔTÞ drp mΔHf ðTÞFðTÞ drp

Scheme 1. Summary of the Porosity Analysis in the Dry and Swollen Statea

ð7Þ

where dQ/dt is the heat flow curve, dt/d(ΔT) is the scanning rate of the experiment, m is the mass of the dry sample, ΔHf(T) is the temperature dependent heat of fusion, and F(T) is the temperature dependent density of the probe fluid. The quantity d(ΔT)/drp is determined from eq 6. It should be noted that the integration of the obtained PSD could yield the pore volume of the material. However, contrary to the reliability of cryoporometry in the case of pore size and PSD, a complete agreement of pore volumes determined by cryoporometry and nitrogen sorption is not expected. Cryoporometry is known to underestimate pore volumes, especially in the case of small pores.39 Exemplarily, the pore volume obtained from integration of the PSDs is lower for the cryoporometry technique compared to nitrogen sorption (Vp(DSC) = 0.33 cm3 g-1 versus Vp(BJH) = 0.45 cm3 g-1 for material HG100). Consequently, the calculation of pore volumes was set aside. On the other hand, the qualitative discussion of determined pore sizes and PSDs is not affected by those findings. Figure 3d shows that the pore radius is around 5-6 nm for porous polymers with cross-linking densities between 17.5 and (44) Jahnert, S.; Chavez, F. V.; Schaumann, G. E.; Schreiber, A.; Schonhoff, M.; Findenegg, G. H. Phys. Chem. Chem. Phys. 2008, 10, 6039–6051. (45) Schreiber, A.; Ketelsen, I.; Findenegg, G. H. Phys. Chem. Chem. Phys. 2001, 3, 1185–1195. (46) Ishikiriyama, K.; Todoki, M.; Motomura, K. J. Colloid Interface Sci. 1995, 171, 92–102. (47) Ishikiriyama, K.; Todoki, M. J. Colloid Interface Sci. 1995, 171, 103–111.

Langmuir 2010, 26(12), 10158–10164

Figure 4. SAXS patterns of selected water swollen mesoporous hydrogels. The patterns were corrected for the empty beam scattering. Please note that the SAXS patterns are vertically offset, as they all possessed a similar qualitative scattering intensity.

a

(a) In the case of high to intermediate cross-linking density, permanent porosity is detectable with the pore radius rpore dictated by the used template; (b) in the case of intermediate to low cross-linking density, no permanent porosity can be detected due to pore closure or collapse; (c) in the case of high cross-linking density, the pores are filled with water, but no significant swelling of the pore wall takes place. δ denotes the nonfreezing interfacial layer. (d) Materials of intermediate cross-linking density show well-defined water domains; additionally, the pore walls are also swollen to some extent; (e) materials of low cross-linking density do feature a biphasic structure; however, they possess an ill-defined interface between the swollen polymeric walls and the pure water domains.

100%. Generally, a slight decrease in the pore size can be observed for the materials with low cross-linking density, most notable for the material HG10. The pore sizes and PSDs (see Figure 3c and d) of highly crosslinked materials (HG100-HG50) determined by cryoporometry are in very good agreement with the pore sizes determined by nitrogen sorption and SAXS in the dry state. As such highly DOI: 10.1021/la100290j

10163

Article

cross-linked materials do not show a significant swelling (bulk swelling degrees Qm ∼ 1.05-1.15), it can be expected that the pores in the swollen/solvent-saturated state and the dry state for the porous polymers above a critical cross-linking are relatively well-defined and not much affected by swelling or solvent removal. In the case of HG10, there is a huge hysteresis between the melting and freezing peaks, and in the case of HG5 no freezing of water confined in mesopores can be observed within the chosen temperature range. Before discussing these findings, it is necessary to introduce the results of a SAXS analysis of selected swollen mesoporous hydrogels (see Figure 4). While the water-saturated mesoporous hydrogels HG100HG17.5 show comparable scattering patterns with a pronounced Porod decay, which are reminiscent of the profiles observed in the dry state, the scattering profiles of HG10 and HG5 do not feature such behavior. Although possessing qualitatively a similar scattering intensity as the higher cross-linked materials (Supporting Information Figure S3), the pattern deviates from the Porod law at high s-values. This might be due to the stronger swelling, which most probably results in higher electron density fluctuations as well as to a less defined interface. Nevertheless, the presence of a high scattering intensity can be assigned to the presence of a two-phase system. Hence, we postulate that even for those mesoporous hydrogels with a low cross-linker content there are water-rich domains and polymer-rich domains on the mesoscale. However, due to the stronger swelling degree, the pure-water cavities might be too small to allow a well-defined freezing, comparable to the finding that water does not freeze in silica pores having diameters smaller than ∼2.8-2.9 nm.44 Second, it is possible that dangling or loose chains reach into the voids and prevent nucleation.

Conclusions We have demonstrated an easy and versatile pathway for the synthesis of mesoporous hydrogels. The size of the mesopores could be controlled by the particle size of the silica template, and the replication yielded the inverse of the pressed pellet. The materials of varying cross-linking density were characterized with regard to their permanent and reversible porosity using multiple methods. It could be shown that a minimum cross-linker content of 50 mol % was needed to preserve a significant fraction of the open porosity of the hydrogels. Freeze-drying leads to a higher porosity for materials of 50 mol % EGDMA content compared to evaporative drying, while no influence of the drying protocol could be observed in the case of fully cross-linked materials. Hydrogel materials of intermediate cross-linking densities (37-17.5 mol %), which did not possess any permanent porosity that could be probed by nitrogen sorption, showed nevertheless prominent signals when analyzed by SAXS. This suggests that the materials possess closed porosity, probably caused by elastic

10164 DOI: 10.1021/la100290j

Weber and Bergstr€ om

rearrangements of the water-swollen gels upon drying. Further evidence for the presence of closed porosity was obtained from density measurements and nonsolvent penetration experiments. Materials with low cross-linking densities (5-10 mol %) did finally not feature any porosity, neither by nitrogen sorption nor by SAXS. Analysis of the water-swollen gels by cryoporometry and SAXS revealed that all hydrogels were two-phase systems. Cryoporometry (using DSC) resulted in pore sizes of about 10-12 nm, with only little dependency on the cross-linking density. These values correspond well with the pore sizes determined by nitrogen sorption for the highly cross-linked materials, which corroborates that the cryoporometry by DSC is sound. This makes cryoporometry a powerful technique for the analysis of biphasic hydrogels, as was also shown by Gonzalez et al. recently.41 Most interestingly, also hydrogel materials with low crosslinking densities (5-10 mol %) for which there was no indication of the preservation of a mesophase upon drying did feature a biphasic structure upon reswelling in water as shown by DSC (HG10) and SAXS (HG5 and HG10). This is in accordance with a recent report of polyisoprene elastomers with reversible nanoporosity,22 where a similar effect was demonstrated. Scheme 1 summarizes the results. In summary, the here presented approach offers access to nanosized hydrogels. As the synthetic approach is expected to be tolerant to a large variety of monomers, a large number of functional materials (stimulus-sensitive, sensing functionalities, controlled release) can be expected to benefit from the nanostructure. Furthermore, the reversible porosity and strong influence of the solvent means that the use of nitrogen sorption alone is often not sufficient to analyze and characterize porous polymers that often find their use in liquid-based applications. Future studies will target on an analysis of the diffusion of guest molecules through the mesoporous hydrogels, which might help to get a better understanding of the materials. In addition, SANS studies are planned in order to investigate the swelling-deswelling process and the associated pore closure or collapse. Acknowledgment. We thank Anwar Ahniyaz (YKI Stockholm, Sweden) for providing access to DSC. The Stockholm University research center on porous materials, EXSELENT, is acknowledged for partial financial support. J.W. acknowledges financial support from the German Research Foundation (DFG, Grant Number: WE-4504/1-1) Supporting Information Available: Further analytical data (FTIR, DSC, and SAXS data; photographs of cyclohexane penetration experiments). This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2010, 26(12), 10158–10164