A Facile Route to Bespoke Macro- and Mesoporous Block Copolymer Microparticles

We report a facile and versatile strategy for the bespoke fabrication of macroand mesoporous block copolymer microparticles. A clean synthetic route, RAFT dispersion polymerisation in supercritical carbon dioxide (scCO2), is used to generate block copolymer microparticles. Selective swelling/deswelling is then applied to induce controlled morphology transitions and to trap the resulting porous state. The pore sizes are controllable over a large size range (~20 – 200 nm) by varying the length of the swellable block. Through a systematic approach we then demonstrate that the shape of the pores can also be tailored from isolated spheres through to interconnected/bicontinuous channels by varying the ratio of the two blocks. This process is shown to be applicable to a range of poly(methyl methacrylate) (PMMA)based block copolymer systems, including PMMA-b-poly(4-vinyl pyridine), PMMAb-poly(dimethyl acrylamide) (DMA) and PMMA-bpoly(dimethylaminoethylmethacrylate) (DMAEMA). In each case, the second minority block (e.g. P4VP, etc.) was selectively swollen with an alcohol to induce an order-toorder morphology transition and then quenched rapidly by the non-solvent hexane. This not only takes place on the order of hours, but is also freely scalable for the production of grams of material and beyond in a single step following polymerisation.


Introduction
Porous polymer particles have attracted significant attention because of their potential applicability to mainstream applications such as ion-exchange resins, 1, 2 catalytic supports, 3-7 chromatography column materials, 8,9 and as solid supports for organic synthesis. 10,11 Additionally, there are potentially important emerging applications as templates for materials synthesis, 12,13 scaffolds for tissue engineering, [14][15][16] or as microsensors 17,18 and microreactors. [19][20][21][22] The key problem is that there is no simple scalable route to synthesise nanoporous microparticulate polymer materials with tunable dimensions across both length scales.
Block copolymers (BCPs) have been extensively exploited for the preparation of porous materials with well-defined nanopores. [23][24][25][26][27][28] Owing to the thermodynamic incompatibility of homopolymer chains, microphase separation occurs to yield periodic domains typically on the scale of 10 -100 nm. These are well understood and give characteristic morphologies such as spheres, cylinders, gyroids, and lamellae. [29][30][31][32] The traditional approach to create porous BCPs is to etch away one block to create voids. 33 Nanoporous polymers generated in this way can be exploited as templates for the formation of other nanostructured materials, for example via sol-gel methods. 25 The templated materials published so far have pore dimensions ranging from a few to 50 nm and up until now this has largely been achieved in 2D thin films of up to 500 nm in thickness. However, the process can be tedious and only a very limited range of BCPs with labile blocks have been devised for such purposes; for example, BCPs containing poly(methyl methacrylate) (PMMA), 34,35 polylactide, 36,37 and polymers with double bonds, e.g., polybutadiene or polyisoprene, 38,39 which can be etched away by shortwave UV light exposure, hydrolysis and ozonolysis respectively.
A more recent and simpler approach uses solvent swelling/deswelling and evaporation to generate porosity without the need for sample degradation. The first reports were made on two dimensional (2D) micellar thin films 40 comprised of various amphiphilic BCPs such as PS-b-PAA [41][42][43][44] and polystyrene-b-poly(x-vinyl pyridine) (PS-b-PxVP, x = 2 [45][46][47][48] or 4 [49][50][51] ). The polar solvents quickly penetrate the thin layers of the minority blocks, causing the micelle cores to undergo swelling that is confined within the continuous phase, while the majority blocks remain in a glassy state. Upon drying, the swollen core-forming chains collapse on the surface of the non-swollen matrix and leave pores centred around the original location of the micelles, hence this method is called "confined swelling-induced pore generation". 44,52 The same swelling induced porosity strategy has also been successfully utilised for so called one-dimensional (1D) BCP materials such as nanorods and nanofibers. 46,[53][54][55][56][57] An additional template of anodic aluminium oxide (AAO) was used to produce nanorods (ca. 500 nm) that were then subjected to swelling in a selective solvent, such as ethanol for PS-b-P2VP. Compared to 2D thin films, which are confined to a solid substrate on one-side, the nanorods/nanofibers show morphology reconstruction in all directions. 46,54 Moreover, the pores are formed with an internal homogeneous wall/shell structure from the collapsed chains of the minority domains of the BCP. The pore structures were found to depend on the swelling agent, swelling temperature, swelling time, and BCP architecture. 46,58,59 The pore sizes were all < 50 nm and the film thickness was limited to about 500 nm. Furthermore, Mei et al. were the first to demonstrate that AAO templates can also be used to produce nanoparticles with various internal nano-scale morphologies, including pores and layers, by varying the length of the swellable block. 60 Porosity was then introduced into these structures via the same solvent swelling approach.
Another related approach for creating porous nanoparticles is to use BCP solutions to create micelles via precipitation and then to apply a swelling/deswelling process where appropriate. This approach has also been shown to impart control of the pore size, shape and/or the overall particle diameter by varying the properties of the block copolymer used. 61,62 The limitation of this route is that the nanoparticle synthesis requires a slow and controlled precipitation and then evaporation from very dilute solutions (~1 wt. %), a process that cannot easily be scaled to the gram or kilogram scale production of material.
Previously, we reported RAFT controlled dispersion polymerisation in scCO2 as a unique route to create multigram quantities of microparticulate block copolymers. [63][64][65] Moreover, we have also demonstrated that the sizes of the microparticles can be controlled reproducibly from 300 nm through to 5.3 m. 66 In this paper we exploit these novel BCP structures to create unique nanoporous microparticles from various systems including PMMA-b-P4VP, PMMA-b-PDMA and PMMA-b-PDMAEMA with controllable porosity over an unprecedented size range (10 nm to ca. 200 nm). Our methodology exploits synthesis in scCO2 followed by a rapid alcohol based swelling induced morphology reconstruction of the minor block to create voids. To the best of our knowledge, this is the first time nanoporous microparticles with completely bespoke physico-chemical properties have been produced by a method that is realistically scalable.

Synthesis of BCP microparticles via RAFT controlled dispersion in scCO2
The synthesis of BCP microparticles followed earlier published procedures. 63 pressure in the autoclave was first increased to ~6 MPa and the autoclave was then heated to 60 °C whilst the contents were stirred mechanically. 68 The final pressure and temperature were gradually adjusted to ~27 MPa and 65 °C over a period of 15 min.
The reaction was carried out for 18 hours to achieve full conversion of the MMA. The second monomer 4VP (5 g) and additional AIBN (6.3 mg) were purged with N2 for 15 min and added to the autoclave via a HPLC pump at 1 ml/min. The polymerisation of 4VP was then allowed to proceed for 20 h before the autoclave was first cooled down to 25 °C and then depressurised. The product was observed to be an off-white fine free flowing dry powder and was collected for analysis.

Selective swelling
To introduce porosity to the as-synthesised BCP microparticles, the original BCP microparticles (50 mg) were fully dispersed in alcohol (3 ml) in a glass vial (3.5 ml) and shaken 3 times by hand over a two-hour period. The particles were left standing until they settled to the bottom of the vial. The upper alcohol layer was removed and then hexane (3 ml) was poured into the vial and the particles were fully rinsed by vigorous shaking. The particles were then allowed to resettle before removing the hexane layer after 2 h. The hexane rinse was then repeated three times to completely remove all of the alcohol and then dried in a vacuum oven at 25 °C for > 2 h before further analysis.

Analysis
The block copolymers were characterised by 1 H NMR in CDCl3 on a Bruker AV3400 (400 MHz) spectrometer. Gel Permeation Chromatography (GPC) analysis was carried out on an Optilab-rEX (Wyatt) in a mixture of chloroform/ethanol/triethylamine (90/9/1 by volume) for PMMA-b-P4VP at a flow rate of 0.5 mL/min and 25 °C. GPC Columns were composed of K-G, a K-805L and a HT-803 column (Shodex) and calibrated with PMMA narrow standards.
For SEM, the samples were mounted on an aluminium stub and sputter-coated with platinum or iridium prior to imaging on a JEOL 7100F FEG-SEM at accelerating voltage of 2 kV or a FEI Quanta 650 ESEM at 10 kV. The average pore width based on SEM was measured by counting over 100 pores using the commercial software package NanoMeasurer 1.2.5.
Thin sections for cross-sectional TEM were obtained from the original powder BCP microparticle samples by embedding them in an epoxy resin (Agar 100) cured at 55 °C for 2 days. Thin sections (~80 nm) were cut by ultra-microtome using a diamond knife (Leica Diatome Ultra 45°) and were placed on copper TEM grids (Agar). The porous microparticle samples were similarly mounted but were cut by cryo-ultra-microtome.
The samples were first placed on the surface of a drop of sucrose solution (5 %) on a sample stick following by freezing in liquid nitrogen. The frozen microparticles were then ultra-microtomed into thin sections of ~100 nm at -60 °C using a glass knife. These cross-sections on the copper grids were then imaged using either a JEOL 2100Plus TEM at 200 kV or FEI Tecnai BioTwin-12 TEM at 100 kV at room temperature.
PMMA-b-P4VP samples were stained with I2 vapour for ∼2 h where stated to enhance the domain contrast in the images. P4VP preferentially absorbs I2 and appears as dark regions in bright field TEM. The domain sizes in the TEM were measured by counting over 100 domains using the software package NanoMeasurer 1.2.5.
The porosity and surface area of the prepared samples were characterised with a Micromeritics ASAP 2420 instrument by N2 adsorption at 77 K. Prior to the measurements the samples were first degassed at 70 °C for 16 h. The surface area was calculated by the Brunauer-Emmett-Teller (BET) method using the N2 adsorption isotherm data within the relative pressure ranging from 0.05 to 0.3. The density functional theory (DFT) method, which is described in detail elsewhere, 69,70 was used to extract the pore size distribution from the adsorption branch using the Micromeritics software. When both mesopores and macropores exist in a material it is necessary to assess the pore width distribution relative to both pore volume and surface area; mesopores contribute to surface area, whilst macropores significantly increase pore volume.

Tilt TEM and Tomography
Conventional bright-field TEM of these particles is limited by two factors, first, that the image is a 2-dimensional projection and second, that the sample has to be thin enough to allow at least partial transmission of the electron beam. The complexity of a fully porous 3-dimensional structure is not readily comprehended in a 2-dimensional projection. However, many of the microparticles could be seen to be partially electron transparent even at their thickest points. Since this allows information to be obtained from the entire volume, and the contrast mechanism of conventional bright-field TEM images in these samples is predominantly proportional to volume, tomographic reconstruction from a tilt series was therefore possible.
For an accurate reconstruction, the sample was deposited onto a lacey carbon support TEM grid pre-prepared with 10 nm gold fiducial markers to aid computer alignment. A total of 120 images were manually acquired at 1 degree steps in a single axis tilt series, using a Gatan 916 room temperature tomography holder in a JEOL 2100Plus equipped with a Gatan US1000 CCD camera at 200 kV. Post-acquisition alignment and reconstruction was performed using the IMOD software (http://bio3d.colorado.edu/imod/) utilising the WBP and SIRT reconstructions. RAFT controlled dispersion polymerisation in scCO2. After swelling with ethanol, a selective solvent for P4VP, and rinsing with the non-solvent hexane the particle morphology was clearly porous (Fig. 1a). The SEM image shows open surface pores that are partially interconnected into sub surface cylindrical pores and a deeper pore network, with an average pore width of dw = 112.7 nm. The subsurface porosity is demonstrated by SEM images at higher magnification (Fig. 1b) where much smaller pores with dw = 20.3 nm are visible below. TEM images (Fig. 1c) indicate that the porosity persists throughout the entire microparticle. A thin slice (~150 nm) was cut by cryo-ultra-microtome and the TEM images (Fig. 1d) confirm the pores are interconnected and form a bicontinuous porous network. Prior to ethanol swelling the original M-V27.8 microparticles have a phase separated morphology of spherical P4VP domains in a PMMA matrix (Fig. S1c). These internal patterns were formed in situ by polymerisation induced phase separation. 63 The minority P4VP domains show an average size of d = 21.7 nm (measured from TEM).

Results and
A distinctive periphery clearly shows larger P4VP domains with d = 60.9 nm (refer to S2 for all TEM-derived domain size measurements). After the solvent swelling interconnected channels are formed and these have an average internal pore width of dw = 51.1 nm (Fig. 1d), which is twice the original P4VP domain size of d = 21.7 nm and is a result of the P4VP ethanol swelling interaction.
The swelling -deswelling process is best described schematically (Scheme 1). The swelling of these 3D microparticles is somewhat different from 2D thin films, 41

Pore size and shape control
To investigate the effect of the volume fraction of the minority P4VP block on the porosity, we synthesised PMMA-b-P4VP microparticles with similar PMMA chain lengths (ca. MW = ~50 kg/mol) and varied the P4VP chain length to give molar fractions from 9 -45 mol% (  Fig. S1b-d). These peripheral P4VP domains are enlarged from 57 nm to 106.3 nm at 19.9 and 30.9 mol% P4VP respectively (Fig. S1, 3).
It appears that the peripheral surface structure is likely a consequence of progressive monomer exclusion towards the interior of the microparticles. Therefore, at the latter stages of polymerisation the P4VP domains at the periphery are swollen with excess 4VP monomer, leading to the much larger domains as compared with the interior. We did kinetic studies of the copolymerization via a sampling tap connected to the autoclave as described previously and the corresponding samples were microtomed and investigated by TEM. These images corroborate the 4VP domains become larger at the particle surface towards the latter stages of the reaction (Fig. S5). This also likely accounts for the slightly broader than expected MW distributions ca. 1.4 -1.5 ( Table   1). GPC analysis was used to exclude the possibility of P4VP homopolymer contamination for each sample. In each case no obvious shoulders were present, and extensive washing with ethanol (a good solvent for P4VP) showed no distinct mass loss or GPC profile changes (see Fig. S6 for M-V27.8).  The BCP microparticles underwent swelling/deswelling as described and SEM images ( Fig. 2) show the porous microparticles obtained from PMMA-b-P4VP for a range of P4VP molar fractions. At low P4VP fraction (13.3 mol% -sample M-V13.3) the images show isolated spherical pores on the particle surface (Fig. 2a, b). These surface pores are arranged in a near hexagonal array. As the molar fraction of P4VP is increased to 19.9 mol% (sample M-V19.9), the larger surface open pores partially fuse and where 2 to 4 pores merge, they morph into a short cylindrical surface feature (Fig. 2c, d). This is a direct result of the increased swollen P4VP volume exerting increased pressure on the glassy PMMA matrix, leading to rupturing and merging of isolated spherical P4VP domains. A similar pore structure evolution from spheres to channels was previously observed in the swelling of 1D nanorods at longer swelling times. 46,48 Some interesting pore geometries have also been reported previously by changing the BCP composition and hence the swellable block fraction in large compound micelles or nanospheres. 46,[60][61][62] When the minority P4VP block is further increased to 27.8 mol% (sample M-V27.8) more ethanol absorbs and the swollen P4VP volume increases concomitantly (Fig. 2e,   f), giving interconnected open surface pores and a much larger surface pore width, dw = 112.7 nm (compared to M-V19.9, dw = 74.2 nm). Thus, the original partially merged spherical P4VP domains observed in M-V19. 9 have extended to form a bicontinuous morphology fixed into the glassy PMMA scaffold (Fig. 2e). Increasing the P4VP block further to 30.9 mol% (sample M-V30.9) leads to significantly more surface porosity ( Fig. 2g, h) where the surface bicontinuous channels now have an average pore width of dw = 168.8 nm from the SEM. Finally, the sample with a P4VP block ratio of 36.4 mol% (M-V36.4) shows a loss of control and collapse of the surface pores (Fig. S9a,   b); Coalescence and fusion leads to less well-defined pores because there is insufficient glassy PMMA present to stabilise and fix the swelling induced morphology. The SEM images also hint at sub-surface pores of the order dw = 20 -25 nm (Fig. 2); To probe these, the same samples were further characterised by TEM imaging (Fig. 3).
For M-V13.3b, the pores only develop near the surface and not throughout the whole particle (Fig. 3a). At higher P4VP content > 19 mol%, the TEM data clearly demonstrate that the microparticles are highly porous throughout (Fig. 3b-d) and confirm that ethanol diffuses effectively through the glassy PMMA regions, the major component of the BCP microparticles.
We hypothesised that the swelling process for the 3D microparticles is time-dependent.
Wang et. al investigated 1D nanorods of PS-b-P2VP with ethanol and found that varied pore morphologies were achieved by increasing swelling time from 10 min to 66 h. 46,54 However, they began with just a very dilute suspension of nanorods in ethanol and argued that the swelling at 20 °C was tiny. In our study, the BCP microparticles are a dry powder and we have a much greater material bulk. We found that a substantial time ca. >1 h was required just to fully suspend the BCP microparticle powder in ethanol, whereas for full solvent penetration, >12 h was required to fully swell gram quantities of our microparticles.

Tilt TEM Tomography
To further demonstrate the internal porosity, a 3D structural analysis of the pores throughout a whole microparticle was undertaken using tilt TEM imaging with angles of up to 120°. The particles initially displayed a very small shrinkage of 2 -2.5 % in diameter but then quickly stabilised under the conditions of the measurements being made. A certain level of radiation damage is to be expected upon extended exposure of the soft polymers to the electron beam, 71 but it is clear that the lacy carbon film coated on the TEM grid was sufficient at dissipating the charge on the microparticles.
Pore size measurements throughout the sample were made from the tomographic reconstruction in the Z-axis (i.e. parallel with the original zero tilt image) to minimise distortion due to the missing wedge. Eight image slices (Fig. 4) for sample M-V30.9 reveal the pore structure. Reconstructed slices at the surface of the structure (Fig. 4, image 001, 021) reveal large open pores (dw = 70.8 nm). Slices in the middle of the reconstruction (passing though the centre of the microparticle) reveal much smaller pore widths at the particle centre (dw = 24.4 nm, image 141 in Fig. 4). All slices show the nanopore structures, which further confirm that the pores are developed throughout the particle. layers but the bulk appears to be nonporous (Fig. S7a). However, as the P4VP fraction increases it is clear that high quality internal porosity evolves with very clear interconnectivity throughout the entire particle (Fig. S7b-d).
The porous materials were further investigated using nitrogen adsorption isotherms (BET measurement at -196 °C, Fig. 5). The majority of the isotherms were typical type IV, characterised by the dominance of the H1 hysteresis loops observed at high relative pressures (P/P 0 ) region. The adsorption capacity of nitrogen was negligible for virtually all the samples analysed at relative pressures up to 0.8 indicating that the materials are essentially purely mesoporous but containing some macropores. At low P4VP content, M-V13.3 presents a characteristic H2(b) loop, usually associated with "ink-bottle like" pores. But the porosity in this sample is quite low and consistent with the TEM/tilt TEM observations which reveal mesopores only in the sub-surface layers with pore width 21.2 nm (Fig 3a, Fig. S8a).
The other samples show greater porosity and pore size distributions (Fig. 5b, c) indicating that at higher mole ratios of P4VP there are multiple regular peaks of meso-  (Figs. 3, 4).
The presence of micropores in these samples is small compared to the large population of meso-and macropores (Fig. S12). As a result, the specific BET surface areas of these materials are generally limited, ranging from 11 to 91 m 2 /g but with total pore volumes approaching up to 0.69 cm 3 /g. However, what is clear is that the surface area and pore volume both increase significantly with increasing P4VP-content, demonstrating that this is the pore-generating domain during the swelling/de-swelling process.

Pore size control by MW of BCP
The size of the nanodomains formed via phase separation of the block copolymers is determined by the constituent block size, where the domain size is proportional to N x , with N denoting the repeating unit and x < 1. 29,74 Here we demonstrate that RAFT controlled dispersion polymerisation in scCO2 can be utilised to achieve higher molecular weight blocks up to 330 kDa (see Table 1). As expected, these much higher molecular weights lead to much larger domain sizes, with d(P4VP) increasing from 25.5 nm to 60.4 nm as we progress from M-V30.9 (MW = 59400 g/mol) to the much larger M-V36.9 (MW = 153100 g/mol) (the as-synthesised domains are shown in Fig.   S10). The swelling process for M-V36.9 yielded porous microparticles with isolated large pores at the particle surface (average pore width dw = 168.9 nm) (Fig. 6). A TEM/tilt TEM video (Fig. S7e) clearly shows that the pores are interconnected and that these large pores are found at both the particle surface and internally (Fig. 6c). The absence of interparticle fusion for this system when compared with M-V36.4 (50k-25k) is attributed to the significantly reduced solubility of the P4VP block in ethanol as the molecular weight increases. Tomographic images based on the tilt TEM series show internal porosity (Fig. 6d) and demonstrate that the pore size can be controlled and extended to larger macropores of up to ~200 nm by increasing the MW of the blocks, and hence also increasing the domain size and the available volume for ethanol swelling.
Such large macropores cannot easily be achieved by conventional routes such as UV etching, ozonolysis, hydrolysis, or complexation with additives, which generally lead to much smaller mesopores < 50 nm. 25 In those methods the pore size is limited to the domain size of the intrinsic BCP self-assembled nanostructure, which is usually in the range of 10 -100 nm. By contrast, our approach using swelling induced order-to-order morphology transitions and morphology reconstruction by fast de-swelling can facilitate significantly larger features.

Other block copolymer systems
The same selective solvent swelling and fast de-swelling process was applied to a range of other microparticulate block copolymers. The key was to determine the most appropriate selective swelling solvent and de-swelling pair for other systems (Fig. 7) based  (Fig. 7a,b) when swollen with ethanol compared to the original (Fig.   S11g). The particles also developed surface open pores (dw = 26 nm) which appear interconnected. In contrast, swelling M-DMAEMA11.8 with ethanol causes interparticle fusion (Fig. S13). A lower alcohol of methanol was found to significantly reduce the particle fusion for M-DMAEMA11.8 (Fig. 7c,d). Surface open pores developed, but these were much smaller dw = 12 nm. This illustrates that it is key to find the appropriate selective swelling solvent when changing the BCP system, by considering the solubility parameters of the swelling block/swelling solvent to achieve the optimal porous structures. The successful fabrication of nanoporous particles from PMMA-b-PDMA and PMMA-b-PDMAEMA highlights the versatility of our approach and could likely be applied to a very wide range of block copolymer systems to achieve nanoporous soft materials with bespoke surface functionality.

Conclusion
We present a new methodology for creating controlled nanoporous block copolymer microparticles by synthesis in scCO2 followed by a swelling/rapid de-swelling route.
Our S4. GPC traces of the as-synthesised block copolymers listed in Table 1. Data for the reaction aliquots collected after the first PMMA block was polymerised but prior to injecting the second monomer are also included.