Anisotropic and Interconnected Nanoporous Materials from Randomly

Jun 6, 2017 - Department of Polymer Science and Engineering, University of Massachusetts Amherst, 120 Governors Drive, Amherst, Massachusetts ...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/Macromolecules

Anisotropic and Interconnected Nanoporous Materials from Randomly End-Linked Copolymer Networks Di Zeng, Alexander Ribbe, and Ryan C. Hayward* Department of Polymer Science and Engineering, University of Massachusetts Amherst, 120 Governors Drive, Amherst, Massachusetts 01003-9263, United States S Supporting Information *

ABSTRACT: Microphase separation within randomly end-linked copolymer networks (RECNs) provides access to disordered bicontinuous morphologies over a wide composition range of the constituent network strands. Here, we rely on end-linking of telechelic hydroxyl-terminated polystyrene (PS) and poly(D,Llactide) (PLA) chains of equal molecular weight, with a tetrafunctional isocyanate cross-linker in a good solvent for both strands, followed by solvent removal to induce microphase separation, and finally etching of the PLA phase to yield nanoporous materials. Transmission electron microscopy (TEM) tomographic reconstructed 3D images along with gravimetric measurements and small-angle X-ray scattering (SAXS) indicate the formation of highly interconnected structures over a range of ∼40−70 vol % of PLA, while N2 adsorption measurements indicate narrowly distributed pore sizes that can be tuned by varying the strand molecular weights. Stretching of the PS/PLA copolymer networks above the glass transition temperatures of both components prior to etching the PLA phase provides a straightforward means to introduce controlled anisotropy into the 3D interconnected porous materials.



INTRODUCTION Materials with nanometer-scale pores of controlled size and shape are of interest for a wide variety of applications, including separations, catalysis, sensing, energy storage, and drug delivery.1,2 In many cases, interconnected porous structures where both the matrix and the pores form three-dimensionally (3D) percolating networks are beneficial because the matrix provides mechanical robustness, while the highly interconnected pores facilitate mass transport and provide large surface areas and porosities, without the need to obtain a high degree of global alignment as generally required for materials with cylindrical pores.3 Routes to generate polymeric membranes with continuous pore structures can broadly be divided into two approaches: equilibrium self-assembly and kinetically trapped phase separation. With regards to equilibrium routes, the self-assembly of diblock copolymers into double gyroid nanostructures, followed by selective etching of the minority domain, represents an attractive route to porous monoliths with percolating pores of precisely defined sizes.3−7 However, as this structure is formed only over a narrow range of polymer composition (∼5 vol %),8 precisely controlled synthesis and processing conditions are generally required. To kinetically trap bicontinuous structures in systems undergoing phase separation, a wide variety of methods have been employed, including phase inversion techniques widely used to make commercial filtration membranes, 9,10 solvent11−13 or crystallization14 induced phase separation, sol−gel processes,15,16 and crosslinking during melt mixing.17 Recently, polymerization induced phase separation (PIPS) of mixtures including a degradable © XXXX American Chemical Society

polymer has been established as a facile route to interconnected porous materials with well-controlled pore sizes.18,19 However, as in all examples of kinetically trapped structures, fine control over the relative rates of phase separation and structural arrest is generally required to obtain the desired structure. Moreover, interconnected nanoporous polymers reported to date generally possess isotropic 3D structures, whereas the ability to introduce controlled anisotropy within porous structures can provide interesting and useful anisotropic transport,20 optical,21 and mechanical22 properties. Cross-linked copolymer networks containing two or more immiscible network strands, including amphiphilic polymer conetworks (APCNs),23 interpenetrating networks (IPNs),24 and networks formed by PIPS,25 are well-known to microphase separate into bicontinuous structures under appropriate conditions.18,26−29 However, in many cases, the lack of wellcontrolled strand molecular weights and the sensitivity of network structure to the reaction conditions make it difficult to determine a priori where bicontinuous structures will be found. In this regard, randomly end-linked copolymer networks (RECNs) formed by reaction of telechelic linear polymer chains with multifunctional cross-linkers in a good solvent for all components present a particularly interesting class of materials, since they provide comparatively well-controlled network parameters such as strand stoichiometry and Received: January 2, 2017 Revised: May 18, 2017

A

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules length.30−34 For example, Walker et al.30 recently showed that following solvent removal, RECNs of polystyrene (PS) and poly(ethylene oxide) (PEO) undergo microphase separation into disordered bicontinuous morphologies over a wide range of network compositions (spanning >30 vol %), thus providing a simple and robust means to achieve this desirable morphology. Despite these advantages, however, this approach has not yet been applied to generate nanoporous polymeric materials. In the current report, we demonstrate that interconnected porous materials based on RECNs can be formed from PS and the degradable polymer poly(D,L-lactide) (PLA). We find that the bicontinuous window spans over 30 vol % for samples with strand number-average molecular weights Mn of 11 and 10 kg/ mol for PS and PLA, respectively, and that by selectively etching the PLA phase, free-standing materials with highly interconnected pores can be generated. Further experiments show that pore sizes can be tuned simply by adjusting the designed molecular weight of precursors, in reasonable agreement with the scaling relationship d ∼ Mn0.5 predicted by de Gennes for randomly cross-linked blends.35 Theoretical predictions by Panyukov and Rubinstein also suggests that anisotropic structures can be generated in such networks by stretching,36 and while this has been confirmed for cross-linked block copolymers,37 it has not been tested for randomly crosslinked blends, nor used to generate anisotropic porous structures. Here, we show that stretching the PS/PLA networks above the glass transition temperatures of both components gives rise to aligned structures, which after selective etching of PLA results in a stable interconnected porous structures with a high degree of anisotropy.

with respective Mn values for PS and PLA of 6 and 5 kg/mol, 11 and 10 kg/mol, or 33 and 34 kg/mol. Detailed sample preparation procedures are provided in the Experimental Methods section. Briefly, PS and PLA are dissolved in tetrahydrofuran (THF) in a nitrogen-filled glovebag to a total polymer concentration of 30 wt % (or 15 wt % in the case of 33 kg/mol PS and 34 kg/mol PLA, as 30 wt % solutions of these polymers undergo phase separation), followed by addition of tetraisocyanatosilane with NCO/OH = 1.4. This modest excess of isocyanate groups is selected as it gives networks with gel fractions typically >95 wt % (though only >90 wt % for the 33 kg/mol PS and 34 kg/mol PLA system), compared to values of less than 80 wt % for NCO/OH = 1.0 or 1.2. In situ IR spectroscopy measurements during end-linking (Figure S1) reveal that the conversion of NCO reaches nearly 100% with NCO/OH = 1.4; presumably a fraction of the isocyanate groups react with water, thereby lowering the effective functionality of the cross-linkers. Because of screening of unfavorable interactions between PS and PLA strands by the nonselective solvent THF, the effective interaction parameter χeff in solution is below that needed to drive micro- or macrophase separation. Thus, the network formed by end-linking is ideally globally homogeneous in composition, with nearly random attachment of PS and PLA strands at each junction point. In reality, composition fluctuations present in solution can initially evolve as higher molecular weight branched polymers are formed by endlinking, before ultimately becoming locked into the gelled network, therefore leading to some spatial heterogeneity in composition. However, the solutions remain completely homogeneous and transparent throughout end-linking and gel formation, indicating that these heterogeneities are limited to small sizes and/or differences in composition. Following gelation, the resulting monolithic samples are first swelled in clean THF to remove the sol fraction and then dried under N2 to evaporate solvent. The removal of solvent leads to a large increase in χeff, toward its bare value of χ ≈ 0.2 in the absence of solvent, thereby driving microphase separation within the network, as illustrated in Figure 1. In the dry, microphase-separated state, the materials are slightly cloudy, as shown in Figure S2a, probably due to the trapped composition fluctuations on length scales approaching the wavelength of light. The microphase-separated network is then immersed in a 2 M solution of NaOH in water/methanol (6/4, v/v) for a defined length of time at room temperature to etch the PLA domains. Samples are named following the convention SMPSLMPLA-X, where the subscripts refer to the molecular weights of the respective strands and X represents the percentage by weight of PLA initially present in solution prior to end-linking. We first characterize the formation of interconnected nanoporous materials from RECNs consisting of 11 kg/mol PS and 10 kg/mol PLA. The extent of removal of PLA domains is studied by gravimetry, Fourier transform infrared spectroscopy (FTIR), and differential scanning calorimetry (DSC), as shown in Figure 2a−c for sample S11KL10K-55. Gravimetry shows that the weight loss reaches 0.55, equal to the feeding ratio of PLA in the network, within ∼1 day of etching, and remains subsequently unchanged even after 16 days, suggesting that the PLA component is removed quantitatively. Compared with the PS/PLA networks (black lines), the complete disappearance of both the −CO stretching peak from the ester group in PLA (1751 cm−1) from the FTIR spectrum



RESULTS AND DISCUSSION Our approach to form RECNs relies on urethane end-linking chemistry using two different dihydroxyl-terminated polymers homogeneously mixed in solution along with a tetraisocyanate small molecule cross-linker (Figure 1). We choose PS and PLA

Figure 1. Schematic depiction of the preparation of interconnected porous samples from hydroxyl end-functionalized precursor polymers.

as a model system to study the formation of interconnected porous materials for a number of reasons. The driving force for segregation between PS and PLA in the absence of solvent is reasonably high, corresponding to a Flory−Huggins parameter (χ) of about 0.2 at 20 °C,38 thus enabling microphase separation for fairly low molecular weights and therefore access to small pore sizes. Furthermore, PLA is noncrystalline and can easily be selectively etched under basic conditions (pH > 7), while PS provides a high modulus structural phase. We focus on RECNs that are nearly symmetric with regards to strand molecular weight, in particular three different systems B

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 2. Characterization of sample S11KL10K-55. (a) Weight loss against time when immersed in a 2 M solution of NaOH in water/methanol (6/4, v/v). (b) Fourier transform infrared spectroscopy (FTIR) measurements before (black) and after (red) etching. (c) Differential scanning calorimetry (DSC) measurements of S11KL10K-55 before (black) and after (red) etching and pure PLA network (blue). (d) Small-angle X-ray scattering (SAXS) characterization before (black) and after (red) etching. (e) Scanning electron microscope (SEM) characterization; 1 nm of Au was coated before imaging. (f) Nitrogen adsorption isotherm at 77.3 K. (g) TEM tomographic reconstruction of the porous sample with the PS matrix shown in dark colors. (h) The same reconstruction following 3D flood filling of interconnected pores, with the unfilled (disconnected porous) portions shown in dark colors. (i) Size distributions of pore (left image and red bars) and matrix (right image and green bars) domains calculated by finding the diameter of the largest sphere that fits within the respective domain at each point in the tomogram.

(Figure 2b) and the PLA glass transition at 38 °C from the DSC curves (Figure 2c) for PS porous monoliths (red lines) further support the complete removal of PLA. In the FTIR spectrum of porous PS, as seen in Figure S3, the carbonyl stretch from the urethane linkage is observed at 1740 cm−1, rather than at 1710 cm−1, suggesting that there is limited hydrogen bonding between urethane groups.39 The structures of the network are investigated by small-angle X-ray scattering (SAXS). As seen in Figure 2d, S11KL10K-55 shows a clear and broad scattering peak at q* = 0.20 nm−1, corresponding to d-spacing = 2π/q* = 31 nm, consistent with a disordered microphase-separated structure. Notably, as shown in Figure S4, similar SAXS patterns are obtained when samples are prepared under different conditions (i.e., with the network formed at an elevated temperature to increase the end-linking kinetics or as a thin film to increase the rate of solvent evaporation), suggesting that this method is not highly sensitive to the kinetics of reaction or solvent evaporation. After extracting PLA phases, the corresponding porous monolith exhibits a scattering peak with the same shape and d-spacing,

suggesting that the characteristic structure has not changed; however, the scattering intensity is increased due to the greater contrast in electron density between PS and vacuum compared to that between PS and PLA. This increased contrast also makes it possible to detect a weak higher order peak centered at 0.60 nm−1, corresponding to 3q*. The presence of such weak secondary reflections has also been noted in many other disordered bicontinuous structures,18,40−44 including from similar end-linked networks.30 Although the origin of these higher-order peaks is still not fully understood, within late stage spinodal structures they have been interpreted in terms of the structure factor defined by locally ordered regions41,42 or as a signature of the negative Gaussian curvature interfaces.43,44 In the high q region, the intensity decrease corresponding to q‑4 for porous monoliths, indicating sharp interfaces between PS matrix and pores.45 As seen in Figure S2b, such nanoporous PS monoliths are opaque because of the large refractive index contrast between the PS walls and the pores. Interestingly, the porous material is also soluble in THF, indicating that degradation of PLA is sufficient in this case to reduce the C

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 3. (a) Gravimetric characterization of S11KL10K samples. Images are taken for sample S11KL10K-51 and S11KL10K-80 after PLA is etched. (b) Porosities calculated from N2 adsorption isotherm measurements for samples prepared from S11KL10K networks with different compositions. Increasing the PLA content wPLA = 0.46 to 0.70 yields a linear increase in porosity from 0.40 to 1.05 cm3/g as a function of wPLA/(1 − wPLA), indicating that the monoliths have similar interconnected pore structures, while the lower porosity at wPLA = 0.75 suggests partial pore collapse. (c) SAXS characterization of S11KL10K-46, S11KL10K-70, and S11KL10K-75.

cross-link density below the gel point. In situ SAXS measurements during heating (Figure S5) reveal that the porous structure is stable up to roughly 100 °C, corresponding to the glass transition temperature of PS, but then collapses, as expected for a nanoporous system below the gel point. Unfortunately, as the current materials are composed of modest molecular weight polystyrene, they are fairly brittle, making it difficult to determine their mechanical properties. The pore size distribution is next studied by a nitrogen adsorption isotherm measurement, as shown in Figure 2f. Based on Barrett−Joyner−Halenda (BJH)46 analysis, the average pore size is 20 nm. The pore volume is calculated to be 0.55 cm3/g by the nitrogen uptake at P/P0 = 0.95, which is somewhat smaller than the ideal value of 0.98 cm3/g based on respective densities for PS and PLA of 1.05 and 1.25 cm3/g. We suspect that this difference reflects the collapse of micropores within the PS walls generated by etching of PLA chains “trapped” within the PS phase or perhaps the influences of cracks in the monolith formed during solvent evaporation. The surface area is calculated by Brunauer−Emmett−Teller (BET) analysis47 to be 134 m2/g, comparable to other porous materials with similar pore size and porosity. 48 The morphology of the porous monolith is further characterized by scanning electron microscopy (SEM), as shown in Figure 2e. The PS phase is clearly well percolated but is penetrated by an interconnected and disordered network of pores, with a characteristic spacing of ∼30 nm, in agreement with the SAXS measurement. Transmission electron microscope (TEM) tomography is used to provide a more detailed characterization of the nanoporous structure, as shown in Figure 2g. In performing the reconstruction, the threshold between pores (bright) and PS walls (dark) is chosen to match the porosity measured by N2 adsorption. The tomogram reveals a disordered morphology with a well-defined length scale similar to 30 nm and a high degree of interconnectivity between both PS and porous phases. As seen in Figure 2i, we characterize the distribution of domain sizes from the tomogram by calculating the diameter of the largest sphere at each point that fits completely within the domain and contains the point.49 From this analysis, the average sizes of pore and matrix phases are determined to be 13.5 and 16.8 nm, respectively. Although this pore size is slightly smaller than the value of 20 nm determined from nitrogen adsorption, such a discrepancy is reasonable, since BJH analysis assumes cylindrical pores, while the real pore geometries are highly branched. By applying 3D flood filling

analysis, we identify that less than 0.5 vol % of the pores are not connected to the percolating porous region within the reconstruction volume, as shown in Figure 2h. Since these regions are located almost exclusively along the boundaries, it is likely that they are indeed connected to the percolating pore network through connections outside of the reconstruction volume. The complete removal of the PLA phase and the interconnecting pore structure, along with the PS phase maintaining its structural integrity, clearly indicates that this randomly end-linked network is bicontinuous with both phases percolating in 3D. To determine the composition window over which bicontinuous structures can be formed in PS−PLA RECNs, we consider S11KL10K networks prepared using a variety of composition, expressed in terms of the weight fraction of the constituent polymers. As seen in Figure 3a, wPLA is varied from 0.31 to 0.85, corresponding to volume fractions φPLA of 0.27 to 0.83. SAXS measurements reveal a similar scattering pattern in all cases, i.e., a broad scattering peak with a d-spacing of 30−31 nm (Figure S6), indicating disordered microphase-separated morphologies over the full range of composition. Unfortunately, these scattering patterns do not provide any clear evidence as to the bicontinuity of the samples. Thus, we instead turn to a combination of gravimetry, porosimetry, FTIR, and SAXS measurements to unambiguously determine percolation of both phases. The gravimetric data, plotted as the mass loss after etching Δm normalized by the initial sample mass m0, vs the initial loading of PLA wPLA, exhibit three distinct regions, as shown in Figure 3a. When wPLA ≤ 0.41, values of Δm/m0 points fall well below the line of y = x, indicating that PLA is only partially etched and revealing the presence of isolated PLA domains embedded within the PS matrix that are not part of a percolating network and therefore inaccessible to the basic solution. Indeed, FTIR measurements in Figure S7 indicate that PLA domains still exist even after one month of etching. Surprisingly, ∼10 wt % of the samples (wPLA = 0.30 and 0.35) can still be extracted in this nonpercolating regime, presumably due to cracking of the monolith that provides access to PLA domains neighboring these surfaces. Above wPLA = 0.41, however, the fraction of PLA extracted increases rapidly with increasing PLA content, corresponding to the percolation of PLA domains. When 0.46 ≤ wPLA ≤ 0.70 (0.42 ≤ φPLA ≤ 0.66), the values Δm/m0 fall upon the line y = x, representing essentially quantitative removal of PLA and indicating that both PS and PLA are fully percolating, allowing D

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 4. Dependence of RECN structure on strand molecular weight for nearly symmetric systems (φPLA = 0.51 and MPS ≈ MPLA). (a) SAXS data for porous monoliths with different values of network strand molecular weight (black: 11 kDa; red: 21 kDa; blue: 67 kDa). (b) The variation of dspacing with Mn is consistent with the prediction d ∼ Mn0.5 by de Gennes.35 (c−e) SEM images of symmetric S6KL5K (c), S11KL10K (d), and S33KL34K (e) after complete removal of PLA domains, showing an increase in pore size with increased Mn of the precursor polymers.

complete accessibility of base to the PLA phase, while all portions of the PS phase remain part of a single monolith following etching. For all samples in this region, the consistency of SAXS patterns before and after etching, as shown in Figure 3c for S11KL10K-46 and S11KL10K-70, also indicates that PS domains are highly percolated, providing mechanical integrity. In Figure S8, FTIR and SEM also support the complete removal of PLA and persistence of a fully interconnected structure. In addition, N2 adsorption measurements (Figure 3b) reveal that increasing the PLA content wPLA = 0.46 to 0.70 yields a linear increase in porosity from 0.39 to 1.03 cm3/g as a function of wPLA/(1 − wPLA), as expected for materials with similar interconnected pore structures, further supporting that they have bicontinuous morphologies with both percolating PS and PLA domains in this composition window. When the PLA content is further increased to wPLA > 0.75, the PS domains lose percolation, as evidenced by a second rapid increase in the value of Δm/m0 to 1. After etching, samples in this region break up into small granules or even powders, as seen for S11KL10K-80 in Figure 3a, which we report as a normalized mass loss of 1. Notably, when wPLA = 0.75 (sample S11KL10K-75), though gravimetry appears to indicate a bicontinuous morphology, the SAXS pattern (Figure 3c) shows a broadening of the scattering peak and shift to larger q* after extracting PLA domains, indicating that the nanostructure is not completely maintained upon etching. Also in Figure S9a, compared to S11KL10K-70, the calculated pore size distribution for this sample shows a broadening and an increase in the proportion of small pores, although almost half of the porosity is retained, as indicated in Figure 3b. Presumably, this reflects either a partial pore collapse during solvent evaporation after etching due to the low content

of PS and tenuous percolation of these domains or the physical attachment of isolated PS domains to the remaining percolating structure, due to the poor dispersibility of PS particles in the etching solution. We thus take this sample to be at the border between a bicontinuous structure and one with dispersed PS domains. Although assessing the 3D structure from 2D TEM images is difficult, samples prior to etching (Figure S10) show projected contrast consistent with an evolution from dispersed PLA domains, to a bicontinuous morphology, and finally to dispersed PS domains as the PLA content is increased. Taking all of these data together, we estimate the region for bicontinuous structures as wPLA ≈ 0.44−0.75 (φPLA ≈ 0.40− 0.72), as denoted by the shaded gray region in Figure 3a. Notably, the width of the bicontinuous region is similar to that of φPEO ≈ 0.30−0.65 reported previously for PS−PEO RECNs.30 Next, we study the ability to control the characteristic size of interconnected porous structures by adjusting the strand molecular weight for a series of nearly symmetric RECNs (φPLA = 0.51 and MPS ≈ MPLA). As seen in Figure 4a, porous monoliths for each sample show a clear peak by SAXS, with no significant change in q* or shape of the pattern upon removal of PLA domains (Figure S11), indicating a robustly percolating structure in each case. Meanwhile, S6KL5K and S11KL10K show nearly identical scattering patterns when normalized by q*, indicating that the structures are quite similar. The normalized pattern for S33KL34K shows a broader primary peak than the others. Although the origin of this difference requires further study, we speculate that it may reflect a more heterogeneous network structure due to the lower initial concentration of polymer in solution in this case (15 wt % for S33KL34K vs 30 wt E

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

S11KL10K-55 (φPLA = 0.51) shows a scattering ring of uniform intensity, indicating an isotropic structure. Subsequent deformation of the network above its Tg will stretch the chains and therefore favor the formation of an oriented morphology.36 Indeed, after stretching and etching, the 2D SAXS pattern reveals quite different scattering behaviors along the directions parallel and perpendicular to the stretching direction. Specifically, in the perpendicular direction, strong scattering peaks are observed, while in the parallel direction, the scattered intensity is greatly reduced in intensity and shifted to smaller q. This observation indicates that highly anisotropic pores are formed by uniaxial deformation of the RECNs above Tg of the component strands. Consistent with these scattering results, SEM imaging of the resulting anisotropic porous PS monolith in Figure 5b shows that the PS domains and pores are both aligned along the stretching direction, although still percolating in 3D through the monolith. This anisotropic behavior of RECNs is qualitatively in agreement with the predictions of Panyukov and Rubinstein36 and similar to previous findings on the alignment of cross-linked block copolymers subjected to uniaxial stretching (although in that case, an ordered lamellar structure was obtained and not converted to a porous material).37,51 In Figure 5c, to further analyze the 2D SAXS pattern of stretched sample, the scattering intensity integrated over the range q = 0.01−0.03 nm−1 (which covers the scattering peak) is plotted as a function of azimuthal angle θ. Two clear peaks are seen, centered at 90° and 270° (with full widths at half-maximum of 52°), corresponding to the strong scattering from domains aligned along the scattering direction. The degree of orientation can be quantified in terms of the order parameter P2 = (1/2)(3⟨cos2 θ⟩ − 1) with

% for S6KL5K and S11KL10K) or perhaps a larger magnitude of trapped concentration fluctuations due to the higher molecular weights of the strands. However, as shown in Figure 4b, the variation in domain spacing (d = 2π/q*) with Mn is consistent with the dependence d ∼ Mn0.5 predicted by de Gennes35 for randomly cross-linked copolymer networks approaching the microphase separation transition and also previously reported for PS/PEO RECNs.30 In all cases, the appearance of a higherorder peak (3q* for S6KL5K and S11KL10K, but 2q* and 4q* for S33KL34K) and the intensity decrease corresponding to q‑4 in the high-q region for porous monoliths are consistent with a bicontinuous morphology with sharp interfaces.45 Further, N2 adsorption isotherm measurements (Figure S12) reveal wellcontrolled pore sizes varying from 9 to 34 nm with increasing M, with respective decreases in BET surface area from 191 to 102 m2/g and nearly constant porosities of ∼0.55 cm3/g. In Figure 4c−e, SEM images of porous monoliths also clearly show uniform pores with sizes that increase with Mn. Given the high χ between PS and PLA, we expect that further reductions in pore size to ≈7 nm should be possible with a reduction in strand molecular weight to ≈3 kg/mol (based on the predicted onset of microphase separation of symmetric randomly crosslinked networks50 at χN = 19.2). Further increases in pore size should also be possible, limited by the miscibility of high molecular weight polymers in the solvent, and the efficiency of the end-linking chemistry used. To generate anisotropic interconnected porous materials, an RECN is uniaxially stretched to a strain of ε = ΔL/L = 1.5 at 120 °C, which is above the glass transition temperature Tg of both PLA (38 °C) and PS (95 °C), and then quenched to room temperature to preserve the oriented structure. After etching of PLA, the morphology of the resulting anisotropic PS monolith is studied by SAXS and SEM. As shown in Figure 5a, before stretching, the 2D SAXS pattern of PS/PLA network



⟨cos2 θ ⟩ =

∫0 (Iq(θ) cos2 θ|sin θ|) dθ 2π

∫0 (Iq(θ)|sin θ|) dθ where a value of P2 = −0.5 would indicate perfect orientation of the normal vectors to the domain interfaces perpendicular to the stretching direction.52 In this case, P2 is −0.21, indicating a modest degree of orientation parallel to the stretching direction, although as shown in Figure S13, the orientation can be continuously tuned with the applied strain and increased in magnitude to at least P2 = −0.37 for ε = 2.5. Notably, the characteristic domain size along the perpendicular direction (d⊥ = 29 nm, Figure S14) remains almost the same as that for unstretched PS/PLA networks (d = 30 nm), as predicted,36 while d∥ is increased to 45 nm. This suggests that the stretchinduced orientation does not merely reflect an affine deformation but rather reflects a change in the self-assembled morphology in response to stress applied to the network.



CONCLUSION We have developed a robust route to highly percolating and anisotropic porous materials from randomly end-linked copolymer networks. The presence of random cross-links frustrates ordered structures, thereby providing a pathway to bicontinuous materials over a broad composition range spanning ∼30 vol % in PLA content. After etching the degradable PLA component, monoliths are generated with narrowly distributed pore sizes that can be tuned by varying the molecular weight of the network strands. Anisotropy can be introduced, while preserving bicontinuity, simply by prestretching of the cross-linked network above the glass transition

Figure 5. Anisotropic porous monolith (S11KL10K-55) formed by stretching at 120 °C, followed by cooling to room temperature and then etching of PLA domains. (a) 2D-SAXS patterns before and after stretching and etching. (b) SEM image of anisotropic PS monolith (with 1 nm Au coating). (c) Scattering intensity along the azimuthal direction for anisotropic PS monolith (red: after stretching and etching) and isotropic PS/PLA network (blue: before stretching). F

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

filling analysis, porosity, and pore size distribution calculation (using the BoneJ plugin49,56).

temperature of both network strands, followed by quenching and selective removal of PLA domains. With this PS/PLA model system, we have opened a simple means to generate nanoporous materials with well-controlled pore sizes and anisotropies. Although at this moment further practical application is limited by the brittleness of the porous PS materials obtained, other systems with better mechanical properties are likely to be accessible, given the generality of the method with respect to polymer chemistry, offering promise for a variety of applications, including filtration, insulation, and catalysis.





ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.macromol.7b00007. SAXS, FTIR, TEM, SEM, NMR, GPC, and N 2 adsorption characterizations of PS/PLA networks and PS monoliths (PDF)



EXPERIMENTAL METHODS

Dihydroxyl-terminated polystyrene (PS) and poly(D,L-lactic acid) (PLA) samples, tetraisocyanatosilane, anhydrous tetrahydrofuran, and methanol were purchased from Polymer Source, Advanced Polymer Materials, Gelest, Fisher Scientific, and Sigma-Aldrich, respectively, and used without further purification. Functionalities of PS and PLA provided by the suppliers are 1.9 and 2.0, respectively. Taking sample S11KL10K-55 as an example, 0.177 mL (0.012 mmol of OH) of PS solution (in THF with polymer concentration c = 0.38 g/mL) and 0.204 mL (0.017 mmol) of PLA solution (in THF with c = 0.40 g/mL) were mixed together in 0.120 mL of THF in a glovebag purged with dry N2. After complete mixing, 1.4 μL (0.041 mmol of NCO) of tetraisocyanatosilane (NCO/OH = 1.4) was added to the solution and then immediately shaken gently by hand. The vial was capped to prevent solvent evaporation, and the reaction was allowed to proceed for 24 h. The gel was dried under N2 flow for 1 day and then immersed into THF for 1 day to extract the gel fraction, followed by drying under N2 flow for 1 day and in a vacuum chamber for 1 day. The dry sample was then immersed in a 2 M solution of NaOH in water/methanol (6/4, v/v) for a defined length of time, followed by washing with water/methanol (6/4, v/v) at least three times. The sample was finally dried under reduced pressure for 1 day. 1 H NMR spectra were measured at 500 MHz using a Bruker 500 Ascend spectrometer and chloroform-d as the solvent. Gel permeation chromatography (GPC) was performed using an Agilent 1260 series system. THF was used as the eluent at a flow rate of 1.0 mL/min. Differential scanning calorimetry (DSC) measurements were conducted using a TA Instruments DSC Q200, and samples were analyzed under a heating rate of 5 °C/min under a flow of nitrogen (50 mL/ min). Fourier transform infrared spectroscopy (FTIR) was measured using a PerkinElmer Spectrum 100. Small-angle X-ray scattering (SAXS) measurements were performed using a GANESHA 300 XL SAXS, and samples were prepared with thickness around 2 mm. For transmission electron microscopy (TEM) analysis, ultrathin sections were cut on a Leica Ultracut UCT microtome operating at −70 °C, stained with RuO4 for 10 min, and imaged using a JEOL 2000FX operating at an accelerating voltage of 200 kV. Scanning electron microscope (SEM) measurements were conducted using a Magellan 400 FESEM in immersion mode. Samples were coated with 1 nm of Au using a Cressington Sputter Coater 108. For N2 adsorption isotherm analysis, a sample tube was filled with dried sample (40−70 mg) and thoroughly dried under vacuum at 60 °C for 1 day and then was measured with a Micromeritics Tristar II at 77.3 K. A thin porous sample for TEM tomography was prepared using a Leica Ultracut UCT cryo-ultramicrotome at −70 °C. A drop of basic solution (2 M solution of NaOH in water/methanol) was placed on the TEM grid to remove PLA, followed by washing with water/methanol. A tilt series from −60° to +60° at 1° intervals was collected using a JEOL-2200FS EFTEM transmission electron microscope operated at 200 kV. To increase contrast, zero-loss filtering was applied employing a slit aperture of ΔE = 20 eV after the energy filter. Reconstructions were performed using the Etomo part of the IMOD software53 applying a simple back projection algorithm. Volume rendering was done using UCSF Chimera54 through applying a 3D Gaussian filter and choosing a threshold to delineate pores from PS walls by matching the porosity to the value measured by N2 adsorption. ImageJ55 was used for flood

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected] (R.C.H.). ORCID

Ryan C. Hayward: 0000-0001-6483-2234 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the BASF North American Center for Research on Advanced Materials (NORA) with additional support from the Department of Energy, Basic Energy Sciences, through Grant DE-SC0016208 (tomographic analysis). The authors thank Louis Raboin, Dr. Sekar Thirunavukkaras, and Zhenpeng Li (Case Western) for assistance with transmission electron microscopy, scanning electron microscopy, X-ray scattering, and N 2 adsorption isotherm measurements, respectively, and Dr. Marc Schroeder and Dr. Rene Arbter at BASF for helpful discussions.



REFERENCES

(1) Lu, G. Q.; Zhao, X. S. Nanoporous Materials: Science and Engineering; Imperial College Press: London, 2004. (2) Jenkins, S. B. Nanoporous Materials: Types, Properties, and Uses; Nova Science Publishers: 2010. (3) Hsueh, H. Y.; Yao, C. T.; Ho, R. M. Well-ordered nanohybrids and nanoporous materials from gyroid block copolymer templates. Chem. Soc. Rev. 2015, 44, 1974−2018. (4) Urbas, A. M.; Maldovan, M.; DeRege, P.; Thomas, E. L. Bicontinuous cubic block copolymer photonic crystals. Adv. Mater. 2002, 14, 1850−1853. (5) Chan, V. Z. H.; Hoffman, J.; Lee, V. Y.; Iatrou, H.; Avgeropoulos, A.; Hadjichristidis, N.; Miller, R. D.; Thomas, E. L. Ordered bicontinuous nanoporous and nanorelief ceramic films from self assembling polymer precursors. Science 1999, 286, 1716−1719. (6) Ndoni, S.; Vigild, M. E.; Berg, R. H. Nanoporous materials with spherical and gyroid cavities created by quantitative etching of polydimethylsiloxane in polystyrene-polydimethylsiloxane block copolymers. J. Am. Chem. Soc. 2003, 125, 13366−13367. (7) Pitet, L. M.; Amendt, M. A.; Hillmyer, M. A. Nanoporous linear polyethylene from a block polymer precursor. J. Am. Chem. Soc. 2010, 132, 8230−8231. (8) Bates, F. S.; Fredrickson, G. H. Block copolymers-designer soft materials. Phys. Today 1999, 52, 32−38. (9) Mulder, J. Basic Principles of Membrane Technology; Springer Science and Business Media: 2012. (10) Guillen, G. R.; Pan, Y.; Li, M.; Hoek, E. M. Preparation and characterization of membranes formed by nonsolvent induced phase separation: a review. Ind. Eng. Chem. Res. 2011, 50, 3798−3817. (11) Li, L.; Shen, X.; Hong, S. W.; Hayward, R. C.; Russell, T. P. Fabrication of Co-continuous Nanostructured and Porous Polymer G

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules Membranes: Spinodal Decomposition of Homopolymer and Random Copolymer Blends. Angew. Chem., Int. Ed. 2012, 51, 4089−4094. (12) Sai, H.; Tan, K. W.; Hur, K.; Asenath-Smith, E.; Hovden, R.; Jiang, Y.; Riccio, M.; Muller, D. A.; Elser, V.; Estroff, L. A.; Gruner, S. M.; Wiesner, U. Hierarchical porous polymer scaffolds from block copolymers. Science 2013, 341, 530−534. (13) Dorin, R. M.; Sai, H.; Wiesner, U. Hierarchically porous materials from block copolymers. Chem. Mater. 2014, 26, 339−347. (14) Samitsu, S.; Zhang, R.; Peng, X.; Krishnan, M. R.; Fujii, Y.; Ichinose, I. Flash freezing route to mesoporous polymer nanofibre networks. Nat. Commun. 2013, 4, 2653. (15) Nakanishi, K.; Tanaka, N. Sol−gel with phase separation. Hierarchically porous materials optimized for high-performance liquid chromatography separations. Acc. Chem. Res. 2007, 40, 863−873. (16) Nakanishi, K. Pore structure control of silica gels based on phase separation. J. Porous Mater. 1997, 4, 67−112. (17) Chen, Y.; Yuan, D.; Xu, C. Dynamically vulcanized biobased polylactide/natural rubber blend material with continuous cross-linked rubber phase. ACS Appl. Mater. Interfaces 2014, 6, 3811−3816. (18) Seo, M.; Hillmyer, M. A. Reticulated nanoporous polymers by controlled polymerization-induced microphase separation. Science 2012, 336, 1422−1425. (19) Saba, S. A.; Mousavi, M. P.; Bühlmann, P.; Hillmyer, M. A. Hierarchically Porous Polymer Monoliths by Combining Controlled Macro-and Microphase Separation. J. Am. Chem. Soc. 2015, 137, 8896−8899. (20) Tsui, B. Y.; Yang, C. C.; Fang, K. L. Anisotropic thermal conductivity of nanoporous silica film. IEEE Trans. Electron Devices 2004, 51, 20−27. (21) Burresi, M.; Cortese, L.; Pattelli, L.; Kolle, M.; Vukusic, P.; Wiersma, D. S.; Steiner, U.; Vignolini, S. Bright-white beetle scales optimize multiple scattering of light. Sci. Rep. 2015, 4, 7271. (22) Malekmohammadi, S.; Zobeiry, N.; Gereke, T.; Tressou, B.; Vaziri, R. A comprehensive multi-scale analytical modelling framework for predicting the mechanical properties of strand-based composites. Wood Sci. Technol. 2015, 49, 59−81. (23) Erdodi, G.; Kennedy, J. P. Amphiphilic conetworks: definition, synthesis, applications. Prog. Polym. Sci. 2006, 31, 1−18. (24) Sperling, L. H.; Mishra, V. B. A. T. The current status of interpenetrating polymer networks. Polym. Adv. Technol. 1996, 7, 197−208. (25) Lee, J. C. Polymerization-induced phase separation. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1999, 60, 1930. (26) Seo, M.; Amendt, M. A.; Hillmyer, M. A. Cross-linked nanoporous materials from reactive and multifunctional block polymers. Macromolecules 2011, 44, 9310−9318. (27) Iván, B.; Almdal, K.; Mortensen, K.; Johannsen, I.; Kops, J. Synthesis, characterization, and structural investigations of poly (ethyl acrylate)-l-polyisobutylene bicomponent conetwork. Macromolecules 2001, 34, 1579−1585. (28) Bruns, N.; Tiller, J. C. Amphiphilic network as nanoreactor for enzymes in organic solvents. Nano Lett. 2005, 5, 45−48. (29) Dinu, M. V.; Perju, M. M.; Drăgan, E. S. Porous SemiInterpenetrating Hydrogel Networks Based on Dextran and Polyacrylamide With Superfast Responsiveness. Macromol. Chem. Phys. 2011, 212, 240−251. (30) Walker, C. N.; Bryson, K. C.; Hayward, R. C.; Tew, G. N. Wide Bicontinuous Compositional Windows from Co-Networks Made with Telechelic Macromonomers. ACS Nano 2014, 8, 12376−12385. (31) Cui, J.; Lackey, M. A.; Madkour, A. E.; Saffer, E. M.; Griffin, D. M.; Bhatia, S. R.; Crosby, A. J.; Tew, G. N. Synthetically Simple, Highly Resilient Hydrogels. Biomacromolecules 2012, 13, 584−588. (32) Erdodi, G.; Kennedy, J. P. Ideal tetrafunctional amphiphilic PEG/PDMS conetworks by a dual-purpose extender/crosslinker. I. Synthesis. J. Polym. Sci., Part A: Polym. Chem. 2005, 43, 4953−4964. (33) Erdödi, G.; Iván, B. Novel amphiphilic conetworks composed of telechelic poly (ethylene oxide) and three-arm star polyisobutylene. Chem. Mater. 2004, 16, 959−962.

(34) Walker, C. N.; Versek, C.; Touminen, M.; Tew, G. N. Tunable Networks from Thiol-ene Chemistry for Lithium Ion Conduction. ACS Macro Lett. 2012, 1, 737−741. (35) de Gennes, P. G. Effect of cross-links on a mixture of polymers. J. Phys., Lett. 1979, 40, 69−72. (36) Panyukov, S.; Rubinstein, M. Stress-induced ordering in microphase-separated multicomponent networks. Macromolecules 1996, 29, 8220−8230. (37) Sakurai, S.; Aida, S.; Okamoto, S.; Ono, T.; Imaizumi, K.; Nomura, S. Preferential Orientation of Lamellar Microdomains Induced by Uniaxial Stretching of Cross-Linked Polystyrene-b lockpolybutadiene-b lock-polystyrene Triblock Copolymer. Macromolecules 2001, 34, 3672−3678. (38) Zalusky, A. S.; Olayo-Valles, R.; Wolf, J. H.; Hillmyer, M. A. Ordered nanoporous polymers from polystyrene-polylactide block copolymers. J. Am. Chem. Soc. 2002, 124, 12761−12773. (39) Elwell, M. J.; Ryan, A. J.; Gruenbauer, H. J.; Van Lieshout, H. C. In-situ studies of structure development during the reactive processing of model flexible polyurethane foam systems using FT-IR spectroscopy, synchrotron SAXS, and rheology. Macromolecules 1996, 29, 2960−2968. (40) Price, S. C.; Ren, X.; Jackson, A. C.; Ye, Y.; Elabd, Y. A.; Beyer, F. L. Bicontinuous alkaline fuel cell membranes from strongly selfsegregating block copolymers. Macromolecules 2013, 46, 7332−7340. (41) Bates, F. S.; Wiltzius, P. Spinodal decomposition of a symmetric critical mixture of deuterated and protonated polymer. J. Chem. Phys. 1989, 91, 3258−3274. (42) Vonk, C. G.; Billman, J. F.; Kaler, E. W. Small angle scattering of bicontinuous structures in microemulsions. J. Chem. Phys. 1988, 88, 3970−3975. (43) Hashimoto, T.; Jinnai, H.; Nishikawa, Y.; Koga, T.; Takenaka, M. Sponge-like structures and their Gaussian curvatures in polymer mixtures and microemulsions. Prog. Colloid Polym. Sci. 1997, 106, 118−126. (44) Jinnai, H.; Nishikawa, Y.; Koga, T.; Hashimoto, T. Real-Space Studies on Interface in a Phase-Separated Polymer Blend by Laser Scanning Confocal Microscopy. In Interfacial Aspects of Multicomponent Polymer Materials; Springer: New York, 1997; pp 53−61. (45) Bates, F. S.; Wiltzius, P. Spinodal decomposition of a symmetric critical mixture of deuterated and protonated polymer. J. Chem. Phys. 1989, 91, 3258−3274. (46) Barrett, E. P.; Joyner, L. G.; Halenda, P. P. The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms. J. Am. Chem. Soc. 1951, 73, 373−380. (47) Brunauer, S.; Emmett, P. H.; Teller, E. Adsorption of gases in multimolecular layers. J. Am. Chem. Soc. 1938, 60, 309−319. (48) Hillmyer, M. A. Nanoporous materials from block copolymer precursors. In Block Copolymers II; Springer: Berlin, 2005; pp 137− 181. (49) Dougherty, R.; Kunzelmann, K. H. Computing local thickness of 3D structures with ImageJ. Microsc. Microanal. 2007, 13, 1678−1679. (50) Read, D. J.; Brereton, M. G.; McLeish, T. C. B. Theory of the order-disorder phase transition in cross-linked polymer blends. J. Phys. II 1995, 5, 1679−1705. (51) Sakurai, S.; Aida, S.; Okamoto, S.; Sakurai, K.; Nomura, S. Mechanism of Thermally Induced Morphological Reorganization and Lamellar Orientation from the Herringbone Structure in Cross-Linked Polystyrene-block-polybutadiene-b lock-polystyrene Triblock Copolymers. Macromolecules 2003, 36, 1930−1939. (52) Pester, C. W.; Schmidt, K.; Ruppel, M.; Schoberth, H. G.; Bö ker, A. Electric-Field-Induced Order−Order Transition from Hexagonally Perforated Lamellae to Lamellae. Macromolecules 2015, 48, 6206−6213. (53) Kremer, J. R.; Mastronarde, D. N.; McIntosh, J. R. Computer visualization of three-dimensional image data using IMOD. J. Struct. Biol. 1996, 116, 71−76. (54) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF Chimeraa H

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules visualization system for exploratory research and analysis. J. Comput. Chem. 2004, 25, 1605−1612. (55) Schneider, C. A.; Rasband, W. S.; Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 2012, 9, 671−675. (56) Doube, M.; Klosowski, M. M.; Arganda-Carreras, I.; Cordelières, F. P.; Dougherty, R. P.; Jackson, J. S.; Schmid, B.; Hutchinson, J. R.; Shefelbine, S. J. BoneJ: free and extensible bone image analysis in ImageJ. Bone 2010, 47, 1076−1079.

I

DOI: 10.1021/acs.macromol.7b00007 Macromolecules XXXX, XXX, XXX−XXX