Report

Hierarchical Porous Polymer Scaffolds from Block Copolymers

+ See all authors and affiliations

Science  02 Aug 2013:
Vol. 341, Issue 6145, pp. 530-534
DOI: 10.1126/science.1238159

A Complicated Scaffold, Simply

Materials with tailored pore structures can be useful as catalysis supports and for lightweight materials. When preparing medical scaffolds, restrictive preparation conditions have to be met, which can prohibit multistep preparation procedures. Sai et al. (p. 530) describe a method for making porous polymers containing both relatively large (several microns) interconnecting pores and a second population of ∼ tens of nanometer pores. The process exploits spinodal decomposition of a block copolymer blended with small-molecule additives and requires a simple washing step with water, methanol, or ethanol.

Abstract

Hierarchical porous polymer materials are of increasing importance because of their potential application in catalysis, separation technology, or bioengineering. Examples for their synthesis exist, but there is a need for a facile yet versatile conceptual approach to such hierarchical scaffolds and quantitative characterization of their nonperiodic pore systems. Here, we introduce a synthesis method combining well-established concepts of macroscale spinodal decomposition and nanoscale block copolymer self-assembly with porosity formation on both length scales via rinsing with protic solvents. We used scanning electron microscopy, small-angle x-ray scattering, transmission electron tomography, and nanoscale x-ray computed tomography for quantitative pore-structure characterization. The method was demonstrated for AB- and ABC-type block copolymers, and resulting materials were used as scaffolds for calcite crystal growth.

Hierarchically porous scaffolds provide synergies between mechanical properties, transport properties, and enhanced surface area (1). Integrating mesoscale (2 to 50 nm) porosity with three-dimensional (3D) continuous macropores (>50 nm) is of particular importance because it combines high specific surface area with high flux and pore accessibility desired, for example, in catalytic conversions. Potential applications range from catalysis to separation technology to bioengineering. Among polymeric materials, block copolymer (BCP) self-assembly is known to offer access to mesoscale-ordered structures with tunable size and morphology through control over molecular parameters such as block chemistry, sequence, and molar mass (2). Specific methods have been developed to form mesopores, including chemical block removal (35) and swelling with sacrificial components (69). The strong interest in hierarchical polymer scaffolds has resulted in specific strategies for structure generation at multiple length scales using BCPs, such as confined self-assembly in preformed macroscale templates (1012) and nonsolvent- or polymerization-induced phase separation (1316). However, when combined together these approaches often require specific chemistries, only work in narrow synthesis parameter windows, or rely on multiple tedious steps that limit their general use (8). Moreover, quantitative structural assessments of nonperiodic porosity remains challenging.

A well-studied physical phenomenon in polymer science is the spinodal decomposition of polymer blends (17, 18). By driving a multicomponent polymeric mixture to a supersaturated state through control of temperature or through quick solvent evaporation, a continuous interface at the micrometer scale emerges upon phase segregation. A facile and versatile, yet unexplored approach for generating hierarchical porosity would be to induce spinodal decomposition in a BCP-additive blend that would separate into an additive-rich phase and a BCP-rich phase, where one block gets selectively swollen by the additive (Fig. 1). Rinsing out both the additive-rich phase and the additive swelling of the BCP block with the same selective solvent enables hierarchical pore formation. To render the process more relevant for applications, it is highly desirable for this solvent to be water or other protic solvents and for the swollen block to be polyethylene oxide (PEO), endowing the final material with antifouling properties well established for PEO (19, 20). Well-defined structure formation would benefit from (i) the BCP to be strongly segregating to ensure structural integrity of the BCP phase during additive removal; (ii) one block having a high glass transition temperature, Tg, to ensure mechanical stability; and (iii) a relatively small additive to maximize BCP swelling and, in particular, its removal by rinsing.

Fig. 1 Schematic for the synthesis method and ternary phase diagram.

Synthesis of hierarchically porous polymer scaffolds with ordered mesostructure using the SIM2PLE method. Red color on the surface of the pores suggests PEO lining. Schematic ternary phase diagram shows paths to hexagonal and network mesostructures via solvent evaporation.

As a first example of spinodal-decomposition–induced macro- and mesophase separation plus extraction by rinsing (SIM2PLE), we chose a widely used strongly segregating amphiphilic BCP, polystyrene-block-polyethylene oxide (PS-b-PEO), and a PEO oligomer (o-PEO), as a water- or alcohol-soluble small additive, to form a mechanically stable film through solvent evaporation-induced phase separation. A 36.6-kg/mol PS-b-PEO containing 13.8 weight percent (wt %) PEO was synthesized via sequential anionic polymerization according to previously reported procedures (21). The BCP was then mixed at a ratio of roughly 1:1 with the o-PEO additive with molar mass of 400 g/mol, and the mixture was dissolved in xylene at 10 total wt %, followed by solvent evaporation at 130°C on a hot plate covered with a hemispherical dome. During the evaporation period, the clear solution turned cloudy, indicating macrophase separation–induced scattering of visible light. After xylene evaporation was complete, as indicated by mass loss, the resulting white film was immersed in the protic solvents water, methanol, or ethanol to remove the o-PEO. Drying the film yielded a lightweight material with a highly opaque appearance (22).

Figure 2, A to D, shows scanning electron microscopy (SEM) images of the film cross section after removal of o-PEO via rinsing in methanol. Randomly distributed porosity is observed on the micrometer length scale throughout the film (Fig. 2, A and B). These macropores, albeit broadly distributed in size, form an interconnected network characteristic of co-continuous structures obtained via spinodal decomposition (14). Within the polymer struts, hexagonally arranged cylindrical mesopores are observed that have a radius of 10 ± 2 nm (N = 112), as estimated from an analysis of the SEM images (Fig. 2, C and D, and fig. S1). These mesopores are preferentially aligned parallel to the macropore walls, and a fraction of pores are observed to be accessible from the macropores (Fig. 2C). Rinsing with other protic solvents (water and ethanol) resulted in the same structures (fig. S1).

Fig. 2 Hexagonal hierarchical material characterization.

(A to D) SEM images at different length scales of a fractured cross section of a bulk hierarchically porous BCP film after removal of o-PEO in methanol. (E) GPC traces of as-made and rinsed samples. Each curve is normalized in refractive index (RI) detector response at the peak height of the PS-b-PEO peak and calibrated for elution volume at the PS-b-PEO peak. a.u., arbitrary units. (F) SAXS patterns of the as-cast film and films after o-PEO removal through rinsing in protic solvents. Curves for methanol-rinsed and ethanol-rinsed samples are shifted vertically by 102 and 104 upward, respectively. Vertical lines correspond to expected peak positions for a lattice with p6mm symmetry with primary peak position of q* = 0.154 nm−1.

Removal of o-PEO from the bulk film is confirmed by comparing gel-permeation chromatography (GPC) results with N,N-dimethylformamide (DMF) as an eluent for the as-cast film and films rinsed with water, methanol, or ethanol (Fig. 2E). From the refractive index detector response, 90 to 95% of the o-PEO is removed by rinsing the as-cast film with these protic solvents for 2 hours at room temperature (compare peaks on the right at ~35 ml). Successful removal of the oligomeric additive corroborates the high degree of interconnected macro- and mesoporosity throughout the structure as observed in SEM.

Further evidence for easy accessibility of, and removal of short-chain o-PEO from, mesopores via rinsing was established via small angle x-ray scattering (SAXS). Removal of o-PEO should lead to higher electron density contrast and thus appearance of higher-order reflections. Figure 2F shows the SAXS patterns of an as-made film as well as films after rinsing with water, methanol, or ethanol. The pattern for the as-cast film shows a weak primary peak at q* = 0.154 nm−1; q denotes the scattering vector magnitude and is defined as q = 4πsinθ/λ, where θ is half of the scattering angle, and λ is the x-ray wavelength. After 2 hours of soaking in protic solvents at room temperature, a set of reflections consistent with a two-dimensional hexagonal lattice (p6mm symmetry) appears, with the identical primary peak position, q*, to the as-cast film. From the primary peak positions using p6mm symmetry, a channel-to-channel distance of 47.1 nm can be calculated, suggesting substantial swelling of the PEO block by o-PEO when compared with results on the parent BCP film exhibiting a diffuse scattering peak at q* = 0.25 nm−1 (fig. S2). Both GPC and SAXS results suggest that water and methanol are slightly more effective in o-PEO additive removal than ethanol, which may be due to their smaller size.

One of the advantages of working with BCPs is the versatility in precisely controlling nanostructures. We found that mesopore morphologies can be tuned by controlling the casting temperature. Figure 3, A to D, shows the macro- and mesostructural characteristics as this temperature is reduced from 130° to 100°C while using otherwise identical conditions. The low-magnification SEM image in Fig. 3A reveals the pore structure differences on the μm length scale when compared with the image in Fig. 2A. Macropores with sizes as large as 5 to 10 μm are observed for films cast at 100°C. On the mesoscale, fourfold symmetry projections for the pore arrangement in cross sections of the polymer scaffold are observed (Fig. 3, B and C), suggesting a cubic symmetry for the mesostructure formed. In contrast to the hexagonal mesopores, the cubic mesopore structures are isotropic and thus allow for even easier access from the macropores.

Fig. 3 Cubic hierarchical material characterization.

(A to C) SEM images of film cross sections of PS-b-PEO/o-PEO blends cast from xylene at 100°C at increasing magnifications. Images show macroporosity (A), interconnected mesopores accessible from the macropores (B), and fourfold symmetry in the mesoscale porosity (C). (D) SAXS patterns of as-made and rinsed film cast at 100°C. Spectrum for the rinsed film is shifted in intensity compared with results from as-made films. Tick marks correspond to expected peak positions for a lattice with Embedded Image symmetry with the first peak position corresponding to (110) reflection at 0.183 nm−1. (E) TEM tomographic reconstruction of the polymer scaffold. The material is shown in bright colors. (F) Representative projection images generated from proposed DD and I-WP structure models and tomographic reconstruction. Scale bars indicate 20 nm.

In order to corroborate the mesoscale structural difference observed in SEM, SAXS patterns were obtained for the films cast at 100°C (Fig. 3D). Similar to the films prepared at 130°C, SAXS patterns of as-cast samples showed weak peaks, whereas after rinsing with protic solvents patterns exhibited strong higher-order scattering peaks while retaining the primary peak position of the as-cast film. For example, after rinsing with methanol, films show a set of reflections with the ratios of (q/q*)2 = 1, 3/2, 4/2, 6/2, 9/2, 10/2, 12/2, 14/2, and 17/2 and the first-order peak location at q* = 0.183 nm−1, which is consistent with a cubic symmetry of aspect 4 with lattice spacing of 48.7 nm (23), typically associated with a double diamond (DD) network morphology. We used transmission electron tomography to generate a 3D reconstruction of the mesostructure (Fig. 3, E and F, and fig. S3). The reconstruction shows a cubic periodic network structure most consistent with either DD or Schoen’s I-WP morphology (Fig. 1) but did not enable a definite assignment (22). Although the packing frustration in fourfold (DD) or eightfold (I-WP) nodes generally precludes such highly crowded network structure formation in neat BCP systems, BCP/homopolymer blends are predicted by self-consistent field theory to form DD structures through homopolymer segregation in the nodes (2426).

We have applied the SIM2PLE method to a variety of systems. We first varied the solvent from xylene to anisole, a more-polar and hydrogen-bonding solvent that dissolves PEO better and has a higher boiling point. The resulting mesostructure (fig. S4) is similar to that obtained from xylene, indicating that the mesoporous structures for this pair of solvents are not as sensitive to the solvent choice as they are to the change in casting temperature. We have further varied the BCP system from PS-b-PEO to poly(4-tert-butyl)styrene-block-polyethylene oxide (PtBS-b-PEO) and to the triblock terpolymer polyisoprene-block-polystyrene-block-polyethylene oxide (PI-b-PS-b-PEO). For PtBS-b-PEO, preliminary results of the SIM2PLE method with oligomeric poly(acrylic acid) as the additive and tetrahydrofuran as solvent gave similar hierarchical structures and accessible pores after rinsing with ammonium hydroxide-containing aqueous solution (fig. S5). The higher glass-transition temperature of PtBS (Tg = 138°C) relative to PS (Tg = 100°C) allows for standard sterilization procedures, for example, autoclaving, to be used on these materials for potential biological applications. Successful translation to PI-b-PS-b-PEO (fig. S6) suggests a path to membranes with improved toughness (15). Last, preliminary experiments with a 226-kg/mol molar mass PS-b-PEO-L yielded highly pliable films with multiscale porosity, suggesting that this method can also be translated to highly entangled polymers (fig. S7).

Nanoscale x-ray computed tomography, nanoCT, was used to image the nonperiodic 3D macroporous structure on micrometer length scales. This technique requires no alteration to the samples before imaging and can reconstruct a relatively large volume of ~1 mm3. Figure 4A shows a 3D rendering of x-ray absorption contrast in a PS-b-PEO/o-PEO–based sample cast from xylene at 130°C with hexagonal mesostructure, revealing the co-continuous nature of the macrostructure. Strut-thinning processes (27) on the interface toward either the polymer region or toward the pore region yielded fully connected skeletal networks throughout the film thickness, confirming a micrometer-scale bicontinuous network consistent with the suggested spinodal decomposition mechanism (Fig. 4B). A characteristic feature size of 5 to 10 μm is detected from the node-to-node distance distribution of the skeletal network as well as from a weak correlation peak at 7.6 μm in the radial distribution plot of the 3D fast Fourier transform (FFT) (Fig. 4, C and D). Furthermore, the skeletal network analysis provides information on the degree of complexity in the population distribution of struts per node shown in Fig. 4E: Although a large fraction of nodes are found to be trivalent, we also observed nodes that are more crowded. Similar analysis was performed on a sample cast at 100°C with cubic mesostructure (fig. S8). The distribution of struts per node shifts to a larger number for the 100°C sample compared with the 130°C sample. Accessibility of, and transport through, network pores is of critical importance to structure replication processes (28). Figure 4, F and G, shows SEM images of calcium carbonate (calcite) crystallized within the hierarchical scaffold after template removal (22). Whereas the image in Fig. 4F depicts typical facets of calcite, the Fig. 4G image shows mineral replica features consistent with growth in macro- and hexagonally arranged cylindrical mesopores. Growth into the template on both length scales has occurred over hundreds of nanometers, implying transport of crystal precursors through the film.

Fig. 4 Macrostructure characterization and materials use as scaffold.

(A to E) A 3D tomographic reconstruction of the macrostructure of a film of system PS-b-PEO/o-PEO/xylene/130°C with hexagonal mesostructure, using nanoCT. (A) Isosurface visualization. (B) Skeletal networks for (A) of the polymeric (blue) and the porous (red) regions. (C) Node-to-node distance distribution of the porous network from a 136-μm cubic region. (D) Radial distribution function of the 3D FFT volumetric data from (C). (E) Population distribution of struts per node for (C) (red columns), as well as an identically generated network of a sample cast at 100°C with cubic mesostructure (black columns). Mono- and divalent nodes arise from analysis artifacts on the edges of the volume. (F) Low-magnification SEM image showing the rhombohedral facets consistent with the crystal habit of the calcite polymorph. Right-hand portion of the crystal with the highly textured surface was grown in contact with the polymer scaffold. (G) High-magnification image of the mineral replica of the polymer. The larger features, associated with growth into macropores, show {104} facets, which are all uniformly aligned, indicating an epitaxial relationship to the parent crystals.

The presence of o-PEO played a dual role in the formation of film porosity. We speculate that the o-PEO, a precipitating solvent for the majority PS block of the BCP, induces spinodal decomposition of the initially single-phase solution as the solvent evaporates, consistent with the observed opacity in the films. Preliminary mixing experiments of PS-b-PEO and o-PEO in xylene suggest that the miscibility gap extends far out on the side of the additive and goes up to about 40 wt % solvent (Fig. 1) (22). The macrophase separated o-PEO provides continuous macroporous domains. Evaporation at different temperatures leads to a different quench depth into the spinodal decomposition region, resulting in different amounts of residual o-PEO in the PEO block. Deeper quenches (higher T, faster evaporation) lead to less additive in the BCP-rich domains (Fig. 1), resulting in a change in ordered mesostructure, for example, from network to hexagonal, consistent with our experimental observations (29, 30). Because o-PEO is not a strongly associating swelling agent, rinsing at room temperature for a short period of time is sufficient to form the final hierarchically porous structure. This is in contrast to the often harsh bond-cleaving conditions required for etching block copolymer domains (35). Use of highly amphiphilic BCPs such as PS-b-PEO, PtBS-b-PEO, or PI-b-PS-b-PEO prevents substantial intrusion of rinsing solvents into the hydrophobic part of the scaffold, contributing to mesostructural integrity. We also note that, because the PEO block in the BCP is not decomposed, the pores are lined with PEO chains, providing wettability and possibly antifouling properties of the walls.

We have developed a facile and versatile one-pot approach for the preparation of hierarchical macro- and mesoporous polymer scaffolds via spinodal decomposition of BCP/small additive mixtures from solution and have introduced techniques for their quantitative structural characterization. The SIM2PLE method combines ease of preparation with high degree and choice of ordering within the macroporous structure. It further replaces more demanding decomposition or chemical transformation steps to induce macro- and mesoporosity by a simple rinsing step with protic solvents like water or alcohols. It thus provides advantages over multiple-step fabrication methods currently used for integrating nanoscale porosity into macroscopic scaffolds. We have already shown it to work for different BCPs, small molar mass additives, solvents, and protic rinsing agents. Because the method is based on general thermodynamic principles, it may provide a powerful conceptual approach to generate hierarchical materials.

Supplementary Materials

www.sciencemag.org/cgi/content/full/341/6145/530/DC1

Materials and Methods

Figs. S1 to S9

Table S1

References (31, 32)

Movies S1 and S2

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: This work was supported by the NSF Single Investigator Award (DMR-1104773). K.W.T. gratefully acknowledges the Singapore Energy Innovation Programme Office for a National Research Foundation graduate fellowship. This work made use of the SEM, transmission EM, and polymer characterization facility of the Cornell Center for Materials Research (CCMR) with support from the NSF Materials Research Science and Engineering Centers (MRSEC) program (DMR-1120296) and CHESS, which is supported by the NSF and the NIH/National Institute of General Medical Sciences under NSF award DMR-0936384. This work also used the Xradia VERSA XRM-500 instrument in the Cornell University Biotechnology Resource Center (BRC) Multiscale CT Imaging Facility. The authors gratefully acknowledge S. Strogatz (Cornell University) for fruitful discussions; J. Song (Cornell University) for providing the triblock terpolymer; and R. Dorin, R. Li, and J. Kim (Cornell University) for experimental assistance and discussion.
View Abstract

Related Content

Subjects

Navigate This Article