Introduction

Materials comprising two or more phases assembled in periodic, nanoscale structures represent a substantial area of current research. In particular, bicontinuous structures are distinctive arrangements where each phase is independently continuous in three dimensions. If one of the phases is removed, the resultant material will contain a three-dimensionally interconnected network of pores with inherently high surface area. Such materials are key to many applications, including catalysis,1 separations and drug delivery,2 electronic devices3 and gas storage.4 Therefore, there is a strong motivation to develop new or improved strategies for the synthesis of bicontinuous, nanoporous materials with well-defined pore shapes and sizes.

Nanocasting, a nanoscale extension of traditional casting or molding techniques, has emerged as a powerful tool for precisely manipulating nanostructure.5 A solid, nanoporous material or a nanostructured array of objects containing a prescribed void space is used as a template. The void space is infiltrated with a liquid precursor to the desired product. Through an external stimulus such as temperature or pH the precursor is converted to the desired product. Finally, the template is selectively removed by chemical or thermal means, and the infiltrated product is generated, possessing the inverse structure of the template. These steps are shown schematically in Figure 1.

Figure 1
figure 1

Schematic representation of the general nanocasting process. A nanoporous template (green) is infiltrated with a liquid precursor (red). After infiltration, the precursor is converted to a product (gray) through an external stimulus. Pores are generated in the product by thermal degradation of the template or exposure to a selective chemical agent (turquoise). Adapted with permission from Jones and Lodge.6 Copyright 2010 American Chemical Society.

Several factors dictate whether a particular combination of product and template can be successfully nanocasted. For a liquid precursor to spontaneously infiltrate a nanoporous template, a positive capillary pressure must exist; this implies that the contact angle of the precursor with the pore walls must be less than 90°.7 Thus, the surface chemistry of the template is a critical consideration in nanocasting, as templates with specific pore wall functionality may resist infiltration by a substantially incompatible liquid. Another important concern is the relative stability of the template and product materials. The template must be insoluble in, or otherwise unaffected by, the infiltrated precursor, and must survive the conditions necessary for conversion to product. Furthermore, a route must exist by which the template can be selectively removed or degraded, without disturbing the structural integrity of the product. Then, the product must have sufficient connectivity and mechanical strength to support the framework defined by the inverse of the template. These aspects are controlled to a large degree by intrinsic properties, for example, modulus, and the means by which the precursor is transformed to the product. Maximum connectivity is obtained when there are no byproducts from the latter; syntheses conducted in dilute solution are typically challenging to apply. In situations where the entire infiltrated volume of precursor is converted to product, that is, there are no solvents or byproducts, density changes can, nevertheless, still hinder perfect structural replication of the template.

Nanocasting is a complementary synthesis technique—it is dependent on the availability of materials with well-defined nanostructures, produced by alternative techniques. A variety of chemistries have been successfully employed as templates, including surfactant-structured sol–gel materials,8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56 block polymer-derived nanoporous materials57, 58, 59, 60, 61, 62, 63 and colloidal materials.64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84 The nanocasting approach summarized here is based on polymeric bicontinuous microemulsions (BμEs). This choice provides a pathway complementary to the various synthetic strategies noted above, and enables the production of bicontinuous materials with highly tailorable chemistry and uniformly shaped, ca. 100nm pores. In this review, we will summarize our recent application of the BμE template to the production of nanoporous polymers, ceramics and hierarchically structured materials. A BμE is found under very specific conditions in fluid mixtures containing immiscible substances and a compatibilizing agent.85, 86 The morphology consists of independent networks of the immiscible substances, which are both three-dimensionally continuous and contain extensive interfacial area and near-zero mean interfacial curvature. These interfaces are not correlated over long length scales, that is, the structure is disordered; however, they exhibit local periodicity on the order of tens of nm to hundreds of nm. The compatibilizing agent resides primarily at the interface.

BμEs are not unique to polymeric systems; they were first discovered in mixtures of oil and water stabilized by surfactants.87 Bodet et al.,88 as well as Jahn and Strey,89 provided direct proof of the existence of BμEs in oil–water–surfactant systems using freeze fracture electron microscopy. Teubner and Strey90 provided a robust means to quantitatively characterize the inherent periodicity of a microemulsion by using scattering experiments. The intensity, I, of electromagnetic radiation scattered by a microemulsion exhibits a broad maximum as a function of scattering angle, or wave vector q, following the form:

where a2, c1 and c2 are coefficients of a Landau expansion of the free energy density of the system. In particular, c1<0 for microemulsions, reflecting the negative interfacial tension characteristic of such fluids. From this description of the scattering, two characteristic length scales arise—the domain spacing d and the correlation length ξ:

The value of d corresponds to the average separation between adjacent identical domains; in other words, d is equal to the sum of the average widths of the water and oil domains. ξ corresponds to the length over which interfaces are correlated; larger values of ξ imply a fluid with a longer range of ordering.

Ternary polymer blends of two immiscible homopolymers A and B and the corresponding A–B diblock copolymer constitute high-molecular-weight analogs of oil–water–surfactant mixtures. In such blends, the diblock copolymer prefers to segregate at the interface between A and B, thereby lowering the interfacial tension.91 Since the BμE morphology occurs in oil–water–surfactant mixtures possessing comparable amounts of oil and water and a small-to-moderate fraction of surfactant, an equivalent morphology could be reasonably expected in polymeric mixtures with comparable amounts of A and B and a small-to-moderate fraction of A–B. Moreover, the chemical equivalence of the homopolymers and the corresponding blocks of the copolymer produces a temperature-independent copolymer solubility, which tends to widen the temperature range over which a BμE is stable.

Initial studies regarding ternary blends of this nature were primarily theoretical.92, 93, 94, 95 The anticipated isopleth (φA=φB, fA≈0.5 and NA=NB, where φ, f and N denote volume fraction, volumetric copolymer composition and volumetric degree of polymerization, respectively) phase diagram is reproduced schematically in Figure 2.93 At high temperature, A and B are miscible and the predicted equilibrium state is a single-phase, homogeneous melt over the entire composition range. At low temperature and high copolymer content, the blend is dominated by the micro-phase separation of the copolymer into alternating lamellae of A and B. As equal amounts of A and B homopolymer are progressively added, the lamellae swell until their spacing diverges. Further increases in homopolymer content result in coexisting A-rich and B-rich phases, as there is insufficient copolymer to suppress macroscopic phase separation. The microemulsion regime intervenes between the lamellar and multi-phase states in oil–water–surfactant mixtures. Interestingly, mean-field theory does not predict the existence of a BμE in polymeric blends; rather, the transition between lamellae and macro-phase separation occurs directly at an unbinding transition.96 This unbinding transition intersects the line defining the phase transitions between the high-temperature, homogeneous and low-temperature, micro- or macro-phase-separated states at an isotropic Lifshitz point.97 Broseta and Fredrickson calculated that the Lifshitz point occurs when:

Figure 2
figure 2

Isopleth phase diagram for symmetric A/B/A–B blends, in which NA=NB≈0.2NA−B, φA=φB and fA=0.5. Mean-field theory anticipates a Lifshitz point where the line of order–disorder transitions meets the line of critical points, and an unbinding transition separating lamellae from regions of phase separation. Experimentally, both of these features are masked by a channel of bicontinuous microemulsion, shown in red.

where φH is the volume fraction of homopolymers (φH=φA+φB=1−φA−B), α is the ratio of homopolymer size to copolymer size (α=NH/NA−B <1) and χ is the segment–segment interaction parameter.98 In the context of free energy density, a Lifshitz point corresponds to a2=c1=0, which, according to equation (2), implies that Iq4. For a nearly symmetric blend of polyethylene (PE), poly(ethylene-alt-propylene) (PEP) and poly(ethylene-b-ethylene-alt-propylene) (PE–PEP), this scattering behavior was experimentally confirmed at the composition predicted by equation (3).93

Bates et al.86 first demonstrated the existence of BμEs in PE/PEP/PE–PEP blends using a combination of rheology, scattering and microscopy. Their experimentally determined isopleth phase diagram for this system is reproduced in Figure 3a. The disordered, lamellar and two-phase regions predicted by mean-field theory are reproduced; however, the unbinding transition is preempted by a narrow channel of BμE. A theoretical study of A/B/A–B blends using fluctuation theory, that was published in the same year as this seminal work, indicated that the Lifshitz point is destroyed.99 Nevertheless, the composition and temperature of the Lifshitz point predicted by mean-field theory can be used as a guide to locate the phase space over which a BμE exists.100 Before the unbinding transition is reached, that is, before the lamellar spacing diverges, composition fluctuations destroy the swollen, destabilized lamellae, creating a thermodynamically stable, bicontinuous structure.

Figure 3
figure 3

Isopleth phase diagrams for (a) polyethylene/poly(ethylene-alt-propylene)/poly(ethylene-b-ethylene-alt-propylene), (b) poly(ethylethylene)/poly(dimethylsiloxane)/poly(ethylethylene-b-dimethylsiloxane) and (c) polyethylene/poly(ethylene oxide)/poly(ethylene-b-ethylene oxide). L, H and PS indicate lamellae, a hexagonally symmetric single phase and macroscopic phase separation, respectively. The phase transitions were determined by rheology ( × ), small-angle neutron scattering (SANS) (•, ) and cloud point measurements (□). The dashed lines demarcate the channel over which a bicontinuous microemulsion is stable. Reprinted with permission from Hillmyer et al.101 Copyright 1999 American Chemical Society.

Later experimental work demonstrated that the general phase behavior depicted in Figure 3a is universal for symmetric, A/B/A–B, polymeric systems.101 Symmetric blends of poly(ethylethylene) (PEE), poly(dimethylsiloxane) (PDMS) and poly(ethylethylene-b-dimethylsiloxane) (PEE–PDMS), and PE, poly(ethylene oxide) (PEO) and poly(ethylene-b-ethylene oxide) (PE–PEO) were investigated, as well as the PE/PEP/PE–PEP system. In addition, a value of α≈0.2 was maintained, in an attempt to match the phase-transition temperatures at the composition extremes of the isopleth phase diagram, for experimental convenience. Mean-field theory predicts the order–disorder transition of a symmetric diblock copolymer (φH=0) and the critical point of a symmetric, binary homopolymer blend (φH=1) to occur at χNA−B=10.5 and χNH=2, respectively, resulting in χNH/χNA−B=α≈0.2 (if one assumes a single χ applies to both copolymers and homopolymers).102 Across these systems, the copolymer molecular weights ranged from 108.3 kg mol−1 for PE–PEP to 12.7 kg mol−1 for PEE–PDMS to 2.13 kg mol−1 for PE–PEO. Figure 3, which depicts the phase diagrams from this work, clearly shows the consistency in phase behavior across these systems. A channel of BμE was observed in all three cases. Also evident from these phase diagrams is the fact that the BμE channel can exist despite a substantial mismatch102 between the phase transition temperatures at the composition extremes of the diagram. Collectively, these results establish that BμEs can be found in A/B/A–B polymeric systems of arbitrary chemistry, provided the aforementioned symmetry constraints are appropriately satisfied.

The significant volume of published research conducted toward understanding ternary polymer blends and polymeric BμEs has laid the foundation for their use as efficient precursors to nanoporous materials. The capture of a polymeric BμE in a solid form is straightforward in principle, provided the BμE channel intersects a solidification transition of one of the blend components, such as a glass transition or crystallization point. Through a judicious selection of blend components, one of the three-dimensionally continuous networks comprising a polymeric BμE can then be voided by selective chemical or thermal means. Zhou et al.103, 104 reported the only previous example of a porous material derived from a polymeric BμE. The BμE in a polystyrene (PS)/polyisoprene (PI)/poly(styrene-b-isoprene) (PS–PI) blend was first trapped by vitrification of the PS domains. The resultant material was exposed to the vapor of S2Cl2 to crosslink the PI and, finally, the PS homopolymer was removed by extraction with hexane. Scanning electron microscopy (SEM) images of a fracture surface of the final, monolithic product are shown in Figure 4. The distinctive BμE structure is retained and imprinted into the pore structure that results from removal of the PS homopolymer. From Figure 4, the pores appear to be uniform in size and interconnected in three dimensions. Indeed, the pores were filled with an ionic liquid and the high ionic conductivity measured of the resulting composite proved the continuity of the pore system across the monolith. In addition, the pore-size distribution calculated from nitrogen sorption measurements exhibited a single peak centered at a pore diameter of 43 nm. The value of d calculated for the starting BμE from small-angle X-ray scattering data was 80 nm, slightly less than twice the average pore diameter. These results were consistent with the expected structure of a porous material generated by solvation of homopolymer; the copolymer is insoluble in the extracting solvent and, therefore, the PS block of the copolymer remained as a coating along the pore walls.

Figure 4
figure 4

Scanning electron microscopy images of the fracture surface of nanoporous, crosslinked PI monolith derived from a polystyrene/polyisoprene/poly(styrene-b-isoprene) bicontinuous microemulsion. The left and right images are low- and high-magnification views, respectively. Reprinted with permission from Zhou et al.103 Copyright 2006 American Chemical Society.

In the remainder of this review, the BμE-forming polymer system that has been used as a nanocasting template comprises the original PE/PEP/PE–PEP system. The choice of this system was motivated by several factors. First, the polymers can be synthesized by a straightforward anionic polymerization/hydrogenation procedure105, 106 and are stable indefinitely under ambient conditions. In addition, the semi-crystalline nature of PE allows the BμE morphology to be captured in a solid material.107 Meanwhile, to convert a solidified BμE into a nanoporous material, the disparity in solubility between PE and PEP can be exploited to remove the latter from the structure.108 The remaining PE comprising the nanoporous material melts above 100 °C and is insoluble in any liquid below 70 °C, thus imparting sufficient robustness to the material for infiltration and conversion of nanocasting precursors in the pores. However, the PE can be dissolved in a range of solvents at moderate temperature and can be thermally degraded above 500 °C and, hence, can be readily removed as a nanocasting template.

Experimental procedure

Polymer synthesis and characterization

The synthetic route used to prepare the PE–PEP diblock copolymer is shown in Scheme 1; analogous routes were used to prepare the PE and PEP homopolymers. As a precursor to PE, PEP and PE–PEP, anionic polymerization was used to prepare 1,4-polybutadiene (PB), 1,4-PI and poly(1,4-butadiene-b-1,4-isoprene) (PB–PI), respectively. 1,3-Butadiene (Aldrich, St Louis, MO, USA, 99+%) was purified by two successive vacuum distillations into evacuated flasks containing dried n-butyl lithium (Aldrich, supplied as a 2.5-M solution in hexanes), with stirring for at least 30 min at 0 °C. 1,3-Isoprene (Aldrich, 99%) was deoxygenated by three successive freeze–pump–thaw cycles and then further purified by a procedure identical to that used to purify 1,3-butadiene. After purification, both monomers were vacuum-distilled into evacuated, flame-dried burettes and stored at −78 °C. Sec-butyl lithium (Aldrich, supplied as a 1.4-M solution in cyclohexane) was used as received. Methanol (Mallinckrodt, St Louis, MO, USA, 99.8%) was deoxygenated by at least three successive freeze–pump–thaw cycles. Typically, between 10 and 50 g of monomer and 0.5 and 1 l of solvent were used. The reaction was terminated by the addition of excess methanol via an argon-purged, gas-tight syringe. The resulting solution was precipitated in a 3:1 methanol:isopropanol mixture and allowed to settle. The supernatant was decanted and the polymer was dried successively under a nitrogen stream and vacuum. The polydienes were stored in air at −23 °C.

Catalytic hydrogenation/deuteration

To convert the polydiene precursors PB, PI and PB–PI to the corresponding polyolefins—PE, PEP and PE–PEP, respectively—unsaturation was removed by a catalyzed reaction with H2 or D2. The polydiene was dissolved in 500 ml of cyclohexane (VWR, 99%). The polymer solution was added to a thick-walled, 1l stainless-steel reactor, along with the catalyst and a stir bar. The heterogeneous catalyst consisted of platinum and rhenium supported on silica (Dow Chemical, Midland, MI, USA) and was added at 20% of the polymer weight. The reactor was purged with argon for 20 min under stirring and then sealed. After sealing, the reactor was pressurized with approximately 500 psi of H2 (Airgas, Radnor, PA, USA, ultra-high purity grade) or D2 (Cambridge Isotope Laboratories, Andover, MA, USA, 99.8%) and the temperature was raised to 170 °C and held for 12 h. The PI homopolymer and PB–PI copolymer were reacted with H2 to give fully hydrogenous PEP and PE–PEP, respectively, whereas different samples of the PB homopolymer were reacted with H2 or D2 to give fully hydrogenous PE or partially deuterated PE, respectively. Partially deuterated PE was necessary to provide neutron scattering contrast in PE/PEP/PE–PEP blends for the determination of phase behavior by small-angle neutron scattering. After the 12-h reaction time, the reactor was cooled to room temperature for PEP and to 70 °C for PE and PE–PEP. The solution was then filtered at that temperature to remove the catalyst.

Size exclusion chromatography

Size exclusion chromatography measurements were performed in tetrahydrofuran (THF) with Phenomenex (Torrance, CA, USA) Phenogel columns packed with porous beads of 5 × 103, 5 × 104 and 5 × 105 Å pore sizes. Solutions were prepared at concentrations ranging from 1 to 7 mg ml−1. The flow rate was controlled at 1 ml min−1 via an Alltech (Deerfield, IL, USA) 426 HPLC positive displacement pump. The eluate was analyzed successively by a Wyatt Technology (Santa Barbara, CA, USA) Dawn DSP multi-angle laser photometer with a 633-nm helium–neon laser and a Wyatt Technology Optilab DSP interferometric refractometer. The molecular weights of PE, PEP and PE–PEP were calculated from the molecular weights of PB, PI and PB–PI, respectively, by accounting for the known quantity of additional hydrogen or deuterium. The PE, PEP and PE–PEP were determined to have number-average molecular weights (Mn) of 23.0 (based on a fully hydrogenated repeat unit), 22.5 and 101 kg mol−1 and polydispersity indices of 1.05, 1.02 and 1.07, respectively.

Nuclear magnetic resonance spectroscopy

Proton nuclear magnetic resonance spectroscopy was used to determine stereochemistry, copolymer composition and hydrogenation efficiency. The diene precursors and PEP were dissolved in deuterated chloroform at concentrations ranging from 1% to 5% by weight, and spectra were acquired using a Varian (Palo Alto, CA, USA) UNITY 300 MHz spectrometer at room temperature. For PE and PE–PEP, spectra were acquired using a Varian INOVA 300 MHz spectrometer at 80 °C with deuterated toluene as a solvent. Typically, 128 scans were performed and delay and acquisition times of 20 and 3 s were employed, respectively. The PE homopolymer and PE block of PE–PEP contain 85% and 87% linear ethylene units, respectively, while the PEP homopolymer and PEP block of PE–PEP contain 96% and 95% ethylene-alt-propylene units, respectively. The PE–PEP contains 50% PE by volume. Finally, the PE, PEP and PE–PEP were determined to be 100% saturated, within experimental resolution.

Preparation of nanoporous PE

Blends of PE, PEP and PE–PEP were prepared by dissolving specific volumes of the three polymers in benzene at 70 °C and then freeze–drying to remove the benzene. Through a combination of rheological measurements, small-angle neutron scattering and transmission electron microscopy (TEM), the isopleth for this system was mapped and the phase space over which a BμE is stabilized was determined.109 Nanoporous PE monoliths were derived from BμE-forming blends by first heating in ampules, typically to 130 °C, and holding for 2 h. Next, they were immersed in liquid nitrogen to rapidly quench the sample below the crystallization temperature, Tc, of PE. A rapid quench is necessary to restrict the crystallization of PE within the domain structure of the BμE.107, 110 Slower rates of cooling lead to large-scale crystallization, which significantly disrupts or destroys the melt structure. The solidified monoliths were recovered from the ampules and pores were created by soaking in THF at room temperature for at least 3 h. The THF was always replenished once during the soaking process. Since PEP is soluble in THF, whereas PE and PE–PEP are not, the PEP homopolymer was removed from the monoliths. After soaking, the nanoporous PE monoliths were dried under vacuum.

Scanning electron microscopy

A Hitachi (Tokyo, Japan) S-900 microscope was used at operating voltages ranging from 2 to 5 kV. For nanoporous PE, very small rectangular bars (1 mm × 1 mm × several mm) were cut from the monoliths with a razor blade. These bars were then submersed in liquid nitrogen and fractured. The fracture surfaces were mounted on metal stubs via a conductive carbon tape and coated with 2 nm of Pt using a VCR indirect ion-beam sputtering system. A typical fracture surface of the nanoporous PE used throughout this work as a nanocasting template is shown in Figure 5. The three-dimensionally continuous pore structure provided by the BμE from which the material is derived is plainly evident. Similar procedures were used in the SEM analysis of nanoporous monoliths generated by nanocasting with nanoporous PE as template.

Figure 5
figure 5

Scanning electron microscopy images of a typical bicontinuous microemulsion-derived, nanoporous polyethylene (PE) monolith used as a nanocasting template throughout this work. This particular monolith was created by annealing a PE/poly(ethylene-alt-propylene) (PEP)/poly(ethylene-b-ethylene-alt-propylene) blend with φH=0.85 at 127 °C for 2 h, quenching in liquid nitrogen and removing PEP homopolymer by solvation in tetrahydrofuran. The images were taken after fracturing the monolith while submerged in liquid nitrogen. The right two images are a higher-magnification view of the central portion of the region shown in the corresponding image to the left. The scale bars, from left to right, correspond to 3 μm, 750 nm and 300 nm, respectively.

Transmission electron microscopy

A FEI (Hillsboro, OR, USA) Tecnai G2 F30 field emission gun instrument was used, operated at 300 kV accelerating voltage. Images were obtained with a Gatan (Pleasanton, CA, USA) Ultrascan 4000 × 4000 pixel charge-coupled device camera. Samples were cut into 85–200-nm sections at −150 to −80 °C using a Leica (Wetzlar, Germany) UC6 cryogenic ultramicrotome. These sections were then dispersed onto an uncoated copper mesh grid.

Porosity measurements

The porosities of monoliths were probed by two different methods—nitrogen sorption and differential scanning calorimetry (DSC) thermoporometry. Nitrogen sorption is the most commonly employed means to characterize the porosity of nanoporous solids. The technique exploits the fact that a gas confined within a pore condenses at a pressure less than its bulk saturation pressure.111 The measurements were conducted at 77 K using a Quantachrome Instruments (Boynton Beach, FL, USA) Autosorb-1 or Autosorb iQ. The measured samples comprised monoliths that had been broken into several pieces. Pore-size distributions were calculated from the desorption branch of the measured isotherms using the method of Barrett, Joyner and Halenda.112 DSC thermoporometry is a much less frequently used technique that, similar to nitrogen sorption, exploits the fact that a liquid confined within a pore freezes—or, equivalently, a solid confined within a pore melts—at a lower temperature than the bulk freezing point of the substance. Brun et al.113 were the first to use calorimetry to monitor this depression in freezing point of a confined liquid and its correlation with pore size. As detailed in an extensive review by Landry,114 a typical experiment involves the hermetic encapsulation of a small amount of the porous solid of interest combined with a volume of wetting liquid greater than the pore volume of the solid. At present, cyclohexane was used as the probe liquid. Measurements were conducted using a TA Instruments (New Castle, DA, USA) Q1000 DSC. The sample was first rapidly cooled well below the freezing temperature of cyclohexane (6 °C) and then warmed at 0.1 °C min−1 from 2 to 7 °C. The obtained DSC traces were converted to pore-size distributions using the procedure of Landry114 with appropriate modifications.6

Results

Nanoporous thermosetting polymers

Thermosets are rigid, highly crosslinked, polymeric materials, synthesized by the condensation polymerization of multi-functional monomers or the chain polymerization of monomers containing multiple vinyl groups. Epoxy resins are typically formed by the reaction of molecules with epoxy and amine functionalities; polyurethanes are formed by the reaction of molecules with isocyanate and alcohol functionalities. Owing to the highly crosslinked structure, thermosets have among the highest moduli, strengths, and thermal and chemical stabilities of any polymeric materials. For several advanced applications, there is the need for easily processable, porous solids possessing such adventitious properties. Porous epoxies, polyurethanes and similar materials have been proposed for use in separations,115 encapsulation and electrical insulation of components in microelectronics,116, 117, 118 battery separators119 and stimuli-responsive textiles.120

As a first example of the utility of BμE-derived, nanoporous PE as a nanocasting template, we have generated nanoporous epoxy and polyurethane monoliths.6 In the former case, glycidyl end-capped poly(bisphenol A-co-epichlorohydrin) (Aldrich, Mn348 g mol−1) and 4,4′-methylenedianiline (Aldrich) were used as monomers, mixed in a stoichiometric molar ratio of 2:1 poly(bisphenol A-co-epichlorohydrin):4,4′-methylenedianiline (see Scheme 2). To facilitate uptake by the PE template, the monomers were first dissolved in chloroform or THF at 50% by weight. Then, PE monoliths were added to the solution and allowed to rest for at least 1 h. The solvent was removed successively under N2 and vacuum over the course of 2 days. After removing the monoliths and cleaning excess monomer from the surface, the weight increase was consistent with filled pores. The filled monoliths were heated to 90 °C in air overnight to complete the reaction of the monomers. Subsequent heating to 150 °C for several hours caused no visual or dimensional change in the monolith, despite the fact that this temperature is well above the melting temperature of PE. This indicates the successful crosslinking of the epoxy within the PE pores. The final nanoporous epoxy was generated by soaking the monoliths in toluene at 70 °C overnight to dissolve PE, followed by removal of the toluene under vacuum.

For the nanoporous polyurethane, toluene diisocyanate (Aldrich) and glycerol propoxylate (Aldrich, Mn266 g mol−1) were used as monomers, mixed in the stoichiometric molar ratio of 3:2 toluene diisocyanate:glycerol propoxylate (Scheme 2). The monomers were directly mixed and stirred until homogeneous. Nanoporous PE monoliths were then added and allowed to rest for 2 h. This time roughly coincides with the gel time of the mixture; the PE monoliths were simply removed from the mixture just before the gel point was reached. After cleaning the excess monomer, the weights of the monoliths were consistent with the filled pores. Unlike the epoxy case, the PE pores are readily filled without adding a co-solvent. The reaction of the monomers was again completed by heating to 90 °C in air overnight. The final nanoporous polyurethane was generated by soaking the monoliths in toluene at 70 °C overnight to dissolve PE, followed by removal of the toluene under vacuum.

Both systems provided excellent nanoporous monoliths, as documented by the SEM images of fracture surfaces of epoxy and polyurethane in Figures 6 and 7, respectively. The images clearly show that the BμE structure is templated into the pore structure of these materials, resulting in disordered, three-dimensionally continuous pore networks. The epoxy and polyurethane monomers were chosen to yield very tightly crosslinked networks, that is, a low average molecular weight between crosslinks. This choice was intended to minimize the swelling of the networks by toluene during dissolution of the PE template and, thus, avoid a decrease in the modulus of the networks such that porosity could no longer be supported. Presumably, if the molecular weight of the poly(bisphenol A-co-epichlorohydrin) or glycerol propoxylate monomers was progressively increased, at some critical value the nanocasting procedure would cease to reproduce the BμE structure in these porous materials. This underscores a limitation of the nanocasting procedure used, namely that the synthesized product must be sufficiently intractable or have otherwise limited interaction with a good solvent for PE. DSC thermoporometry measurements indicated that both materials possess uniform pore-size distributions centered at 100 nm, consistent with the domain spacing of the original BμE from which the monoliths are derived,6 as described in more detail later. Furthermore, internal surface areas of 30–35 m2 g−1 were measured.

Figure 6
figure 6

Scanning electron microscopy images of a fracture surface of nanoporous epoxy monolith at different magnifications. The scale bars, from left to right, correspond to 2 μm, 500 nm and 300 nm, respectively.

Figure 7
figure 7

Scanning electron microscopy images of a fracture surface of nanoporous polyurethane monolith at different magnifications. The scale bars, from left to right, correspond to 2 μm, 500 nm and 300 nm, respectively.

Nanoporous poly(3,4-ethylenedioxythiophene) (PEDOT)

PEDOT is an important, commercially available conducting polymer and its synthesis, properties and applications have been studied extensively.121, 122 PEDOT possesses excellent chemical and thermal stability, unique electrochemical and spectroscopic properties, and inherently high conductivity. Consequently, it has been suggested for use in a wide range of applications, including antistatic coatings, capacitors, thin-film transistors, and electroluminescent and photovoltaic devices.123, 124, 125, 126, 127, 128, 129, 130 Owing to the demand for porous materials with good stability, high conductivity and large internal surface area in sensor, battery, capacitor and storage technologies,131 it is desirable to develop PEDOT materials possessing controlled porosity. To this end, a high-surface-area PEDOT electrode has been demonstrated as an effective O2 reduction catalyst in fuel cells and batteries,132 and macroporous PEDOT films have been used as counter-electrodes in dye-sensitized solar cells.133

We have demonstrated nanoporous PE as a template by which nanoporous PEDOT monoliths are conveniently nanocast.134 EDOT monomer can be polymerized electrochemically or by initiation with a suitable oxidant,135 such as the salts iron(III) chloride135 and iron(III) tris-p-toluenesulfonate136 (Fe(OTs)3). Oxidation of PEDOT chains leaves the final product in a doped, conducting state. PEDOT synthesized by this method is insoluble, but can be processed into transparent films with conductivities approaching 1000 S cm−1.137 The synthetic route to PEDOT employed in our work was the chemical oxidative polymerization of EDOT (Aldrich) initiated by Fe(OTs)3 (Aldrich). The polymerization is extremely rapid and is typically conducted in solution.136 To adapt this polymerization to the present nanocasting method, several experimental challenges had to be solved. The solution-based synthesis of PEDOT within the pores of a nanoporous PE monolith is not a viable means to template the structure of the latter, because the majority component—the solvent—is removed after polymerization. Ideally, Fe(OTs)3 would be directly dissolved in the liquid EDOT monomer and the mixture then infiltrated into nanoporous PE and allowed to react. This is impossible, however, due to the extremely rapid kinetics of polymerization and the high loadings of Fe(OTs)3 required for sufficient doping of the final PEDOT.137 Therefore, nanoporous PE monoliths were first impregnated with pure EDOT monomer. A solution of 72 wt% Fe(OTs)3 in isopropanol, with an absolute amount of Fe(OTs)3 corresponding to a Fe(OTs)3: EDOT molar ratio of 2:1, was prepared at 80 °C. While at 80 °C, the EDOT-filled monoliths were exposed to this solution. After a few hours, the resulting black monoliths were removed, washed with excess isopropanol and dried under vacuum. The final nanoporous PEDOT was generated by soaking the monoliths in toluene at 70 °C overnight to dissolve PE, followed by vacuum removal of the toluene.

In Figure 8, SEM images of a fracture surface of a typical nanoporous PEDOT monolith are presented. The PEDOT is quite sensitive to beam damage in the microscope, of which the thin fiber-like objects seen at higher magnification are a direct result. Nevertheless, the images clearly show that the BμE structure is templated into the pore structure of the PEDOT, resulting in a disordered, three-dimensionally continuous pore network. This result further demonstrates that the nanocasting procedure can be successfully applied when the desired product is not a crosslinked network, but rather is simply insoluble in a good solvent for PE. Nitrogen sorption measurements indicated a relatively uniform pore-size distribution centered roughly at 100–200 nm.134 The internal surface area of the material calculated by the method of Brunauer, Emmett and Teller138 was 26 m2/g. Both of these values are comparable to those of the nanoporous thermosets, highlighting the consistency with which the nanocasting technique replicates the initial BμE structure.

Figure 8
figure 8

Scanning electron microscopy images of a fracture surface of nanoporous PEDOT monolith at different magnifications. The scale bars, from left to right, correspond to 2 μm, 500 nm and 300 nm, respectively.

Amorphous ceramics

As a third example of the utility of BμE-derived, nanoporous PE as a nanocasting template, we have generated nanoporous, high-temperature, ceramic monoliths.6, 139 In the synthesis of nanoporous silicon carbonitride, a commercially available polysilazane, Ceraset 20 (KiON Specialty Polymers, Huntingdon Valley, PA, USA), was used as a precursor. Polysilazanes possess low glass-transition temperatures,140 and thus Ceraset 20 is a liquid at room temperature. Its general chemical structure141 is shown in Scheme 3. Blends of Ceraset 20 and 1 wt% dicumyl peroxide (Acros Organics, Geel, Belgium) were prepared by direct dissolution. The nanoporous PE monoliths were added to the Ceraset/peroxide blends so as to infiltrate and fill the PE pores. After allowing the nanoporous PE monoliths to rest in the Ceraset/peroxide blends overnight, the monoliths were removed and excess blend was cleaned from the surface. Gravimetry confirmed that the pores had spontaneously filled with precursor. The monoliths were heated to 90 °C for 48 h to crosslink the Ceraset via reaction of peroxide fragments with the vinyl substituents of the polysilazane. Finally, the composite monoliths of PE and crosslinked Ceraset were converted to the final nanoporous ceramic product by heating under flowing N2 to 1000 °C at a rate of 2 °C min−1. The monoliths were held at 1000 °C for 2 h and then cooled to room temperature at the rate of 2 °C min−1. It is important to note that all steps of this synthesis procedure were conducted in air, with the exception of the aforementioned pyrolysis step.

SEM images of a fracture surface of a typical nanoporous silicon carbonitride monolith are shown in Figure 9. Once again, the BμE structure has been templated into the pore structure of the nanocast material. To confirm that the pores in this material result from complete degradation of the PE template during the pyrolysis and conversion of the Ceraset precursor to silicon carbonitride, a sample of PE was subjected to thermogravimetric analysis. This analysis indicated that PE is completely degraded above 500 °C when heated in N2. The Ceraset–ceramic conversion is accompanied by an over-twofold isotropic increase in the density of the material. A comparison of the SEM images in Figure 9 indicates a decrease in both the pore size and pore wall size of the ceramic, relative to the PE template (Figure 5), which is undoubtedly a consequence of this shrinkage. It is remarkable that the BμE structure and monolith integrity are so well-preserved after synthesis, considering the drastic physical changes that are incurred. Presumably, the slow temperature ramp rate employed during pyrolysis is crucial in this regard, so as to avoid prohibitive accumulation of thermal stress. The present nanoporous ceramic exhibited excellent thermal stability when heated in air to 1000 °C, as indicated by zero weight loss during thermogravimetric analysis, as well as by subsequent SEM analysis, whereby an identical pore structure was observed.

Figure 9
figure 9

Scanning electron microscopy images of a fracture surface of nanoporous silicon carbonitride monolith at different magnifications. The scale bars, from left to right, correspond to 2 μm, 500 nm and 300 nm, respectively.

To characterize the pore-size distribution of the nanoporous ceramic quantitatively, both nitrogen sorption and DSC thermoporometry were used. The pore-size distributions obtained from both techniques are simultaneously displayed in Figure 10. The data in both cases show a relatively narrow distribution in pore sizes centered at 80 nm. The domain spacing of the starting BμE is ca. 160 nm. To a first approximation, the pore size of the ceramic should correspond to the sum of the size of the PE domains and the size of the PEP block of the PE–PEP block copolymer, as these are the components of the starting ternary polymer blend that comprise the nanoporous PE template. The ternary blend consists of 57.5 vol.% PE homopolymer and PE–PEP block copolymer, so the expected pore size by this approximation is 160 nm × 0.575, or 92 nm. Of course, it is reasonable to expect the actual pore size to be less than this estimate, given the previously discussed density change of the ceramic during synthesis. Therefore, this estimated pore size is in good agreement with the experimental value. In Figure 10, the upper and lower accessible pore sizes by nitrogen sorption and DSC thermoporometry are also apparent. Nitrogen sorption indicates negligible pore volume in the ceramic from pores with sizes covering the range of 1–50 nm, but is insensitive to pores with sizes >100 nm. Conversely, DSC thermoporometry indicates negligible pore volume from pores with sizes covering the range of approximately 200 nm up to 10 μm, but provides essentially no information on pores with sizes <40 nm.

Figure 10
figure 10

Pore-size distribution of nanoporous ceramic determined by nitrogen sorption (open symbols) and differential scanning calorimetry (DSC) thermoporometry (solid line). The latter data are plotted as a solid line owing to the high density of data points and to highlight the dearth of nitrogen sorption data at pore sizes >100 nm. The inset shows the DSC thermoporometry data plotted as individual data points.

Hierarchically structured materials

There is widespread interest in preparing self-assembled materials with two or more characteristic length scales.142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203 As described herein, nanocasting provides simple pathways to such hierarchically structured materials by the self-assembly of multi-phase precursors within rigid templates. In 2004, Zhou et al.204 described a new synthetic route to mesostructured SiO2 monoliths through the condensation of tetramethylorthosilicate with the ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate—[BMIM][BF4]—as a solvent. After gelation, the ionic liquid was selectively extracted. The resulting SiO2 material contained a bicontinuous, disordered network of wormlike mesopores with uniform pore sizes of 2.5 nm and a periodicity of 5 nm, that is, pore wall thicknesses of 2.5 nm. We have adapted this unique route to mesostructured SiO2 to our nanocasting strategy, thereby enabling the production of hierarchically structured SiO2.205 The choice of ionic liquid as a solvent is critical, as its non-volatility and active role in the mesoscale templating of the condensing SiO2 network enable successful nanocasting. Comparable volumes of [BMIM][BF4] and tetramethylorthosilicate were mixed and infiltrated into nanoporous PE monoliths, after which the mixture was exposed to an acid catalyst to initiate the sol–gel reaction of tetramethylorthosilicate. TEM images of the hierarchically structured material obtained after the completion of gelation are shown in Figure 11. From these images, two distinct, periodic arrangements of matter are apparent. First, a comparatively large-scale, disordered, bicontinuous structure is observed, consisting of interpenetrating networks of untextured and textured phases. This arrangement corresponds to the BμE structure provided by the nanoporous template, with the untextured and textured phases demarcating the PE and the infiltrated gel, respectively. Indeed, the periodicity of this arrangement is on the order of 100–200 nm, consistent with the domain spacing of the starting BμE. Second, a comparatively small-scale, disordered, bicontinuous structure is observed, confined within the textured phase and consisting of interpenetrating networks of bright and dark phases. This arrangement corresponds to the expected mesoscale structure of the infiltrated gel, with the bright and dark phases demarcating the [BMIM][BF4] and SiO2, respectively. Overall, these TEM images show that the nanocasting strategy enables the straightforward production of hierarchical, tricontinuous, polymer–ceramic–ionic liquid composite monoliths having two characteristic length scales.

Figure 11
figure 11

Transmission electron microscopy images of hierarchically structured polyethylene (PE)–SiO2–1-butyl-3-methylimidazolium tetrafluoroborate ([BMIM][BF4]) monolith obtained by synthesis of mesostructured SiO2 within nanoporous PE. The PE appears as the untextured phase, while the SiO2 and [BMIM][BF4] appear as the dark and bright components of the textured phase, respectively. Scale bars indicate 100 nm (left) and 50 nm (right). Reprinted with permission from Jones and Lodge.205 Copyright 2011 American Chemical Society.

Two independent pore networks can be rendered in the material shown in Figure 11 through sequential voiding of the PE and [BMIM][BF4] domains with selective solvents. Subsequent SEM and TEM analysis (not shown) revealed a hierarchically porous material with a structure identical to that shown in Figure 11, except that the PE and [BMIM][BF4] domains were, as expected, converted to three-dimensionally continuous 100 and 6nm pore networks, respectively. A more quantitative description of the pore characteristics of this hierarchically porous SiO2 was provided by nitrogen sorption measurements, as shown in Figure 12. The isotherm exhibits two distinct steps along both the adsorption and desorption branches, with the typical hysteresis that accompanies capillary condensation and evaporation. This is a direct reflection of the hierarchical arrangement of pores, as the lower and higher relative pressures of the steps can be attributed to the small and large pores generated by the removal of [BMIM][BF4] and PE, respectively. Indeed, the pore-size distribution shown in the inset of Figure 12 possesses two distinct maxima at pore diameters slightly less than 10 nm and slightly greater than 100 nm. By comparison with, for example, Figure 10, the larger pores present in the SiO2 material are consistent with the expected size generated by the nanocasting technique. In contrast to the previously mentioned examples, the simultaneous presence of smaller mesopores within the walls of the large pore framework confers exceptionally high internal surface area to the material (570 m2 g−1), as assessed by the method of Brunauer, Emmett and Teller. We anticipate that this unique hierarchically porous material could be an excellent candidate for applications requiring both high surface area and low mass transport barriers, owing to the highly interconnected nature of the pores.

Figure 12
figure 12

Nitrogen adsorption/desorption isotherm of hierarchically porous SiO2 obtained by nanocasting of mesostructured SiO2 with nanoporous PE as a template. The inset shows the pore-size distribution of the material calculated from the desorption branch of the isotherm using the method of Barrett, Joyner and Halenda.112 Reprinted with permission from Jones and Lodge.205 Copyright 2011 American Chemical Society.

We have also generated novel hierarchically structured materials that are entirely polymeric through the infiltration and self-assembly of poly(isoprene-b-2-vinylpyridine) (PI–P2VP) diblock copolymers within nanoporous PE.206 As a model example, a PI–P2VP copolymer having Mn=41 kg mol−1, a narrow molecular-weight distribution and 18% PI by vol. was used. The equilibrium bulk morphology of the copolymer consists of hexagonally packed cylinders of PI of 13 nm diameter and 31 nm spacing, as assessed by small-angle X-ray scattering and TEM. PE monoliths were first allowed to imbibe a concentrated solution of copolymer in THF. With the monoliths submerged, the THF was slowly evaporated over a week or more. This procedure not only induces self-assembly of the PI–P2VP confined by the PE pore network, but also maximizes the copolymer loading within the pores. The resultant composite monoliths of PE and PI–P2VP were then exposed to the vapor of diiodobutane to effect crosslinking of the P2VP component. Separate experiments performed on bulk PI–P2VP confirmed that this procedure does not significantly alter its self-assembled structure.

The PI–P2VP confined and self-assembled within nanoporous PE exhibits a morphology that is markedly different from its bulk counterpart. Representative TEM images of the hierarchically structured PE/PI–P2VP composite are shown in Figure 13. As in the previous case (Figure 11), two distinct, periodic arrangements of matter are observed. The comparatively large-scale, disordered, bicontinuous structure provided by the BμE template is apparent at low magnification (Figures 13a and b). The bright, untextured and dark, textured networks correspond to the PE and PI–P2VP, respectively. Within the latter, micro-phase separation between PI and P2VP is readily observed at higher magnification (Figures 13c and d). The particular sample shown was stained with OsO4 before imaging; hence, the PI domains appear darker than the diiodobutane-crosslinked P2VP domains. The PI–P2VP exhibits three distinct regions of micro-phase separation as a function of local proximity to the PE/copolymer interface. First, this interface is wetted exclusively by PI, as indicated by the thin dark layer immediately adjacent to the bright PE domains. This is a consequence of the preference of PI over P2VP to form an interface with polyolefins, which is easily rationalized by considering the relative values of χ for this system.207, 208 Second, a much thicker layer, rich in P2VP, occurs adjacent to the PI wetting layer. This P2VP-rich layer is likely enforced by the depletion of PI to the PE/copolymer interface, coupled with the relatively low volume fraction of PI constituting the copolymer. This phenomenon is similar to reported observations in diblock copolymer thin films that exhibit preferential wetting of one block to a substrate.209 Third, along the core of the channels occupied by PI–P2VP, cylinders, discs and helices comprising PI are dispersed in a continuous P2VP matrix. The PI objects are clearly aligned parallel to the PE/copolymer interface, that is, the confining pore walls of the PE template. However, it is difficult to resolve the particular connectivity of many of the PI objects, primarily due to the overlapping effect created by the projection of the three-dimensional sample volume into a two-dimensional image.

Figure 13
figure 13

Transmission electron microscopy images of hierarchically structured polyethylene (PE)/poly(isoprene-b-2-vinylpyridine) (PI–P2VP) monolith obtained by confined self-assembly of PI–P2VP within nanoporous PE. The P2VP is crosslinked by diiodobutane and the PI is stained with OsO4. The relative brightness of the domains follows the order PE>P2VP>PI. Scale bars indicate (a) 400 nm, (b) 200 nm, (c) 100 nm and (d) 50 nm. Reprinted with permission from Jones and Lodge.206 Copyright 2011 American Chemical Society.

The self-assembly of diblock copolymers confined within nanoporous materials has, in fact, been recently studied in detail.210 The perturbation of morphologies relative to the bulk is indeed a universal feature under strong confinement. A significant portion of the published work has focused on the behavior of bulk-cylinder-forming copolymers within the cylindrical pores of anodic aluminum oxide.62, 63, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222 A critical parameter that can be correlated to the observed morphology is the ratio of the confining pore diameter (D) to the bulk copolymer period (L0). For example, Dobriyal et al.219 examined the morphology of PS–PB in anodic aluminum oxide over a D/L0 range from 0.92 to 2.22. With increasing D/L0, stacked discs, stacked torus-like structures, single helices, double helices, combinations of rings, cylinders and helices, and triple helices of the minority component were identified. Furthermore, hexagonally packed cylinders of the minority component have been identified at even larger D/L0 values.212 In all cases, the minority domains were aligned parallel to the long axis of the pores. The morphology of PI–P2VP confined within nanoporous PE described above is consistent with the previous work. Based on the bulk, equilibrium period of PI–P2VP and the pore-size distribution of PE, determined by small-angle X-ray scattering and nitrogen sorption measurements, respectively, the estimated range of D/L0 for this system is 1.9–3.2. Therefore, a variety of cylindrical, disc-like and helical PI domains are expected. Moreover, the confining pores, despite being highly tortuous and interconnected, are locally cylindrical in shape. Thus, the preference for the PI domains to align parallel to the pore walls is further expected, and is indeed evident from Figure 13. Through the bicontinuous structure of the confining template, self-assembled block copolymer morphologies representative of two-dimensional confinement are obtained in a three-dimensionally continuous manner. More importantly, these results demonstrate that our nanocasting strategy provides a simple pathway to unique, hierarchically structured polymeric materials.

Discussion/future prospects

The recent work summarized in this review establishes the following important conclusions. A wide variety of solid materials, including polymeric thermosets, conducting polymers, block copolymers, amorphous ceramics and mesoporous silica, can be readily prepared as nanoporous monoliths, based on a single polymeric BμE template. The resulting porosities, average pore sizes and pore-size distributions replicate the corresponding features of the template with good fidelity. The initial template is derived from an equilibrium-structured fluid based on a ternary polymer bend, and is thus highly reproducible. The existence of this phase has previously been demonstrated to be universal in appropriately designed systems, thereby indicating that templates could be prepared from a rich variety of starting polymer materials. Given the flexibility in choice of template and in choice of target solid material, there is no obvious limit to the kinds of materials that could be employed. Furthermore, the desirability of nanoporous materials with three-dimensionally connected, ca. 100nm pores for multiple applications, ranging from separations and catalysis to batteries and fuel cells, suggests that much further work can be done. Possible future directions to consider include the following:

  1. i)

    Surface functionalization of the pores: This could be achieved either by installing appropriately reactive groups in the initial copolymer surfactant or by direct modification of the product surface.

  2. ii)

    Tuning of the average pore size: In principle, the domain size of the initial structured fluid is sensitive to blend composition and temperature, and can be tuned over the approximate range 50–250 nm. Whether this tunability can be transferred to the resulting nanoporous product remains to be established.

  3. iii)

    Production of new bicontinuous C/D materials: In the work summarized here, an A/B template led to a nanoporous C material. These pores could themselves be filled with the precursor to a new solid, D. A candidate system could include a hole-conducting C polymer and an electron-conducting D phase, with photovoltaic materials in mind.

scheme 1

Synthetic procedure for preparation of poly(ethylene-b-ethylene-alt-propylene) diblock copolymer by anionic polymerization of poly(1,4-butadiene-b-1,4-isoprene), followed by catalytic hydrogenation.

scheme 2

Chemical structures of monomers glycidyl end-capped poly(bisphenol A-co-epichlorohydrin) (top left), 4,4′-methylenedianiline (bottom left), toluene-2,4-diisocyanate (center) and glycerol propoxylate (right). The particular grade of toluene diisocyanate used is an 80:20 mixture of 2,4:2,6 isomers.

scheme 3

Chemical structure of Ceraset 20 polysilazane used as a precursor in the synthesis of SiCN ceramic.