Report

Reticulated Nanoporous Polymers by Controlled Polymerization-Induced Microphase Separation

See allHide authors and affiliations

Science  15 Jun 2012:
Vol. 336, Issue 6087, pp. 1422-1425
DOI: 10.1126/science.1221383

Abstract

Materials with percolating mesopores are attractive for applications such as catalysis, nanotemplating, and separations. Polymeric frameworks are particularly appealing because the chemical composition and the surface chemistry are readily tunable. We report on the preparation of robust nanoporous polymers with percolating pores in the 4- to 8-nanometer range from a microphase-separated bicontinuous precursor. We combined polymerization-induced phase separation with in situ block polymer formation from a mixture of multifunctional monomers and a chemically etchable polymer containing a terminal chain transfer agent. This marriage results in microphase separation of the mixture into continuous domains of the etchable polymer and the emergent cross-linked polymer. Precise control over pore size distribution and mechanical integrity renders these materials particularly suited for various advanced applications.

Mesoporous metals, ceramics, and polymers are tremendously useful in the fields of catalysis, templating, and separations (1). Those containing both a percolating pore structure and a continuous matrix structure (i.e., bicontinuous structures) are highly desirable from the standpoint of pore accessibility and structural isotropy. Although metal and ceramic nanoporous materials of this type have dominated the landscape (2, 3), nanoporous polymers (4, 5) have received attention due to the exquisite synthetic control of their functional attributes using contemporary polymerization mechanisms and functionalization protocols (6).

Block polymers are an intriguing class of self-assembling hybrid macromolecules that are versatile precursors to nanoporous polymers with well-defined pore structures, sizes, and surface functionality (7). However, there are three key issues that currently limit the utility of nanoporous materials from block polymers in separation applications (8). The straightforward preparation of large-area samples with a bicontinuous pore structure has only been achieved in a few cases (912). The drive to smaller and smaller pore sizes (e.g., below ~10 nm) has been challenging due to fundamental block polymer thermodynamic issues and high Laplace pressures that can lead to pore collapse (13). In addition, the typically poor mechanical properties of nanoporous polymers generated from block polymers limit how they can be implemented in practice (10). Here, we present a new block polymer-based strategy for the preparation of nanoporous polymers that simultaneously address all three of these issues.

We combined polymerization-induced phase separation with in situ block polymer formation (14, 15) by polymerizing a mixture of multifunctional monomers in the presence of a macromolecular chain transfer agent (macro-CTA) that is initially soluble in the monomer mixture but incompatible with the emergent cross-linked polymer. Covalent connection of the macro-CTA to the polymerizing mixture of monomers favors the formation of structured materials with length scales that are restricted by the macromolecular dimensions of the block polymers formed. Importantly, the reversible addition-fragmentation chain transfer (RAFT) polymerization process (16) allows for linear increases in molar mass. Thus, the majority of the growing block polymers reach the point of becoming incompatible in the polymerizing monomer medium and microphase-separate over a small time interval, giving a homogeneous cross-linked, mechanically robust, and nanostructured composite. The resultant bicontinuous structure is prevented from coarsening by virtue of the cross-linking of the matrix. By employing a chemically etchable block as the macro-CTA, nanoporous monoliths with a percolating pore structure can be generated. Relying on microphase separation of a block polymer obtained by controlled polymerizations, this strategy is distinguished from other polymerization-induced phase separation methods that result in the formation of bicontinuous but macrophase-separated monoliths (5, 17) or that generate microphase-separated conetworks by uncontrolled polymerization mechanisms (15, 18). A schematic of the overall process is given in Fig. 1.

Fig. 1

Schematic depiction of nanoporous monolith generation by a controlled polymerization-induced microphase separation process. (A) Initially, a macro-CTA containing the etchable polymer (yellow) is dissolved in the mixture (light blue) of a monomer and a cross-linker. RAFT copolymerization allows controlled growth of chains (dark blue) generating a block polymer structure; the growing chains are also being cross-linked in situ (B). At a critical conversion, microphase separation occurs over a small time interval, and cross-linking arrests the emergent to generate and arrest a bicontinuous structure with a nanoscopic length scale (C). Subsequent removal of the etchable polymer produces percolating nanopores in a cross-linked and mechanically robust matrix (D).

Polylactide (PLA) with a trithiocarbonate chain transfer agent at one terminus (PLA-CTA) was chosen as the chemically etchable macro-CTA, and styrene (S)/divinylbenzene (DVB) was selected as the cross-linkable monomer mixture. Importantly, PLA and poly(styrene) (PS) are characterized by a large Flory-Huggins interaction parameter and thus their block polymers should form well-segregated structures with sharp interfaces in the ordered state (19). The number-average molar masses of the PLA-CTA samples were 11, 22, and 41 kg mol–1 and are labeled PLA-CTA-11, PLA-CTA-22, and PLA-CTA-41, respectively (fig. S1) (20). Similar PLA-CTA agents are effective for controlled RAFT polymerization of S, yielding well-defined PLA-PS block polymers (21, 22). Initiation of the S/DVB mixture leads to rapid incorporation of the PLA-CTA due to the high chain transfer constant for the trithiocarbonate group. Continued growth of the S/DVB copolymer results in the formation of a PLA-P(S/DVB) block polymer in the presence of the remaining unreacted S/DVB. Because of pendant double bonds stemming from the DVB incorporation, this growing P(S/DVB) block is ultimately incorporated into a cross-linked network, and microphase separation of the PLA block from this emergent cross-linked matrix occurs. We posit that as the S/DVB mixture continues to polymerize, the S and DVB in the swollen PLA-rich regions of the mixture migrates to the growing matrix, ultimately leading to relatively pure PLA domains.

In an exemplary case, 30 weight percent (wt %) PLA-CTA-22 was dissolved in a 4/1 molar mixture of S/DVB to give a homogeneous solution that was optically transparent. Heating to 120°C (with or without added radical initiator) for 5 hours results in a transparent, insoluble monolithic solid, as shown in Fig. 2. When hydroxyl-terminated PLA (PLA-OH) or a mixture of PLA-OH with CTA was used instead of PLA-CTA, macrophase separation was evident (fig. S2). Consistent with related cross-linked PS/DVB resins, the monolith exhibited an ultimate tensile strength of 38 MPa, an ultimate elongation of ~14%, and a Young’s modulus of 560 MPa (fig. S3). Small-angle x-ray scattering (SAXS) analysis of the sample gave a single broad reflection centered at a scattering vector q = 0.3 nm–1, consistent with a microphase-separated, but disorganized, structure with compositional heterogeneities on a ~21-nm-length scale (fig. S4). Observation of the fractured surface after coating with Pt (~1 nm) by scanning electron microscopy (SEM) did not reveal obvious nanostructuring. However, selective staining of the PS domain with RuO4 gave images that support the presence of a nanostructure consistent with the SAXS data (fig. S5). Differential scanning calorimetry (DSC) analysis of this and all other samples indicated a single thermal transition attributed to the glass transition of PLA plasticized with residual unreacted S/DVB (fig. S6). Similar samples prepared with PLA-CTA-11 and PLA-CTA-41 gave cross-linked monoliths with nearly identical physical characteristics. However, SAXS analysis revealed principal domain dimensions of 15 nm and 27 nm, respectively.

Fig. 2

Image of the cross-linked precursor (left) and the nanoporous monolith (right) generated from PLA-CTA-11 (32 wt %) and azobisisobutyronitrile (0.16 equivalents relative to PLA-CTA) in a 4/1 S/DVB mixture that was heated to 120°C for 5 hours.

To remove the PLA, the monolithic samples were immersed in 0.5 M methanol (40% by volume)/water solution of NaOH. Complete PLA removal was indicated by a mass loss corresponding to the initial weight fraction of PLA used in the synthesis, absence of the PLA C=O vibration in the infrared spectrum, and disappearance of glass transition corresponding to PLA in the DSC thermogram (figs. S6 and S7). An image of a monolith synthesized using PLA-CTA-11 after PLA etching is shown in Fig. 2. The change in sample dimension upon etching was minimal, and the transparency post-etching is consistent with retention of a nanostructured material. A related sample prepared from PLA-CTA-22 exhibited an ultimate tensile strength of 6.4 MPa, an ultimate elongation of ~9%, and a Young’s modulus of ~170 MPa, indicating robust mechanical properties after PLA removal (fig. S3).

Elimination of PLA in all samples resulted in dramatically increased SAXS intensity compared with the precursor material, but no change in the position of the principal peak was observed (fig. S4). These data strongly support generation of a porous structure similar to the nanostructured morphology present in the as-synthesized monoliths. An SEM image of a cryo-fractured surface of a cross-linked, nanoporous material prepared using PLA-CTA-22 revealed that interconnected nanopores were evenly distributed through the material (Fig. 3). The structure is similar to inorganic nanoporous structures such as Vycor glass, where bicontinuous structures are produced by a spinodal decomposition process that is kinetically restricted from coarsening due to the viscosity of the medium; subsequent selective leaching results in nanopore formation (3). Characteristics of the scattering data from the nanoporous materials described here were consistent with other materials formed by spinodal decomposition processes (fig. S8) (23, 24). Image analysis of Fig. 3 gave an estimate of average pore diameter ≈ 9 ± 2 nm (after accounting for a ~1-nm Pt coating). This is in contrast to conventional porous P(S/DVB) and its analogs that adopt macroporous structures consisting of a network of agglomerated microgel particles (17).

Fig. 3

A representative SEM image of the nanoporous monolith. The sample was coated with ~1 nm of Pt. The sample was derived from 32 wt % of PLA-CTA-22 in a 4/1 S/DVB mixture that was heated to 120°C for 5 hours and etched in a basic solution.

SEM images of the nanoporous materials indicated that the pore diameter could be controlled by the molar mass of PLA-CTA used (fig. S9); pores down to 6 nm were evident in voided monoliths derived from PLA-CTA-11. Nitrogen adsorption experiments were consistent with these observations and provided more information on the average pore size and distribution (Fig. 4, A and B, and fig. S10). All the nanoporous materials produced H2-type hysteresis curves very similar to the adsorption isotherms observed for Vycor glass, supporting the presence of a three-dimensionally connected porous network (fig. S11). At low relative pressures, the desorption branch was not coincident with the adsorption branch and thus resulted in an “open” isotherm. One possible explanation for this behavior is that nitrogen remains dissolved within the polymer matrix (25). Pore size distributions based on Barrett-Joyner-Halenda (BJH) (26) analysis of the desorption branch gave a mean pore diameter of 6 nm when 32 wt % of PLA-CTA-22 was used (Fig. 4B). A total pore volume of 0.37 mL g–1 was estimated from these data. Using densities of PLA and P(S/DVB) of 1.25 g cm–3 and 1.05 g cm–3, respectively, the initial volume of PLA in the cross-linked monolith was estimated to be 0.38 mL g–1, in a good agreement with the pore volume in the nanoporous version (fig. S12). The surface area for the sample generated using PLA-CTA-22 was determined to be 210 m2 g–1 by Brunauer-Emmett-Teller (BET) analysis (27). Nanoporous samples prepared using PLA-CTA-11 and PLA-CTA-41 gave pore diameters of 4 and 8 nm, and surface areas of 250 and 170 m2 g–1, respectively (table S1). Based on the Flory-Huggins interaction parameter for PLA-PS (19), we estimate that pore diameters as low as 2 nm will be accessible. The nanoporous monoliths were more thermally stable compared with non-cross-linked nanoporous polystyrene that collapsed above 94°C with a coincident exothermic transition in the DSC thermogram (19). No such exothermic transition was found for these cross-linked materials up to 150°C (fig. S6). The thermal stability was further supported by 83% retention of the pore volume for the sample derived from a PLA-CTA-41 sample after heating for 1 hour at 100°C (fig. S13).

Fig. 4

(A) Nitrogen adsorption isotherm of the nanoporous monolith at 77.3 K. The sample was derived from 32 wt % of PLA-CTA-22 in a 4/1 S/DVB mixture that was heated to 120°C for 5 hours and etched in a basic solution. Filled squares, adsorption branch; empty circles, desorption branch. (B) Pore size distribution as a function of pore diameter (d) based on BJH analysis of the desorption branch in Fig. 4A. (C) Evolution of SEC traces for the first 30 min of polymerization of S and DVB (4/1) in the presence of PLA-CTA-22 (32 wt %) at 120°C. Intensities were obtained by refractive index detector responses, and molar masses were relative to polystyrene standards. (D) Evolution of SAXS peak intensity (I) over polymerization time (t) during the polymerization of S and DVB (4/1) in the presence of PLA-CTA-22 (31 wt %) at 120°C. The peak intensity at q = 0.3 nm–1 was normalized by the intensity at t = 0. The dashed line indicates exponential increase of I at the onset of phase separation obtained by linear least-square fitting of data points from 35 to 70 min.

The microphase separation during the polymerization was investigated by combination of reaction kinetics and in situ SAXS studies (Fig. 4, C and D, and figs. S14 to S16). When a 4/1 S/DVB mixture in the presence of 32 wt % PLA-CTA-22 was heated to 120°C, soluble polymers were obtained for the first 30 min of the polymerization. The apparent molar masses of the polymers generated over the first 20 min, as characterized by size exclusion chromatography (SEC), gradually shifted to higher values, suggesting controlled growth of P(S/DVB) from PLA-CTA by the RAFT process (Fig. 4C). Shoulders at high molar mass were consistent with dimerized and trimerized PLA-P(S/DVB) molecules with low P(S/DVB) content. Over the next 5 min, coupling reactions produced very high molar mass species (~103 kg mol–1). 1H nuclear magnetic resonance spectra of the soluble polymers supported formation of PLA-P(S/DVB) with pendent vinyl groups (fig. S15 and tables S2 and S3) (28). The mixture gelled after 40 min (70% gel fraction) and solidified after 90 min (90% gel fraction).

In situ SAXS analysis of a similar polymerization mixture at 120°C (31 wt % of PLA-CTA-22 in a 4/1 S/DVB mixture) showed the appearance and subsequent growth of a scattering peak at q = 0.3 nm–1 after 35 min that saturated after 80 min, indicating microphase structuring of the emergent PLA-P(S/DVB) block polymer (fig. S16, A and B). The position of the scattering peak was essentially constant throughout the polymerization as cross-linking arrested further coarsening of the microphase-separated structure. The scattering intensity increased exponentially at early times, reminiscent of early-stage spinodal decomposition (Fig. 4D), and further analysis of this data supports microphase separation driven by in situ block polymer formation (fig. S16C) (29, 30).

Bicontinuity of the nanoporous monolith was also supported by gas and liquid transport measurements. Gas permeation profiles of methane, nitrogen, and carbon dioxide through a nanoporous monolith synthesized using PLA-CTA-41 were consistent with Knudsen diffusion, assuming 8-nm pores (from BJH analysis) and incorporating a tortuosity factor of 1.7 (fig. S17 and table S4). This tortuosity value was consistent with the literature value for other bicontinuous materials (31). In addition, water was able to pass through a similarly produced 310-μm-thick nanoporous sample with a permeability of 0.45 L h–1 m–2 bar–1. Using the data obtained from the nitrogen adsorption experiment, we deduced a similarly valued tortuosity of 1.4 (fig. S18).

Nanoporous polymers with sub-10-nm pores were prepared by a controlled polymerization-induced microphase separation method. Although the specific combination of PLA-CTA with S and DVB was presented, similar nanoscopic bicontinuous structures should be accessible if controlled cross-linking polymerization can induce microphase separation. Considering the utility of bicontinuous phases in various applications, the versatility of block polymer self-assembly, and the chemical tunability of block structures and interfacial functionality, this strategy can find even broader application because of the added advantages of simple synthesis, processability, and precise domain size control.

Supplementary Materials

www.sciencemag.org/cgi/content/full/336/6087/1422/DC1

Materials and Methods

Figs. S1 to S18

Tables S1 to S4

References (3237)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: The authors thank Dow Chemical Company for financial support. Partial support for this work was also provided by the National Science Foundation (DMR-1006370). We acknowledge E. L. Cussler, J. Rzayev, and M. Schulze for helpful M. Soma for tensile testing, input, and C. Frethem and F. Zhou for SEM imaging of the cross-linked precursor. Parts of this work were carried out in the College of Science and Engineering Characterization Facility, University of Minnesota, a member of the NSF-funded Materials Research Facilities Network. Synchrotron SAXS data were acquired at the DuPont-Northwestern-Dow Collaborative Access Team (DND-CAT) located at Sector 5 of the Advanced Photon Source (APS). DND-CAT is supported by E. I. DuPont de Nemours & Co., The Dow Chemical Company, and Northwestern University. Use of the APS, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under contract DE-AC02-06CH11357. The authors are declared to be inventors on a patent filed by the University of Minnesota and the Dow Chemical Company related to the work presented here.
View Abstract

Navigate This Article