Controlling Internal Pore Sizes in Bicontinuous Polymeric Nanospheres**

Complex polymeric nanospheres were formed in water from comb-like amphiphilic block copolymers. Their internal morphology was determined by three-dimensional cryo-electron tomographic analysis. Varying the polymer molecular weight (MW) and the hydrophilic block weight content allowed for fine control over the internal structure. Construction of a partial phase diagram allowed us to determine the criteria for the formation of bicontinuous polymer nanosphere (BPN), namely for copolymers with MW of up to 17 kDa and hydrophilic weight fractions of ≤0.25; and varying the organic solvent to water ratio used in their preparation allowed for control over nanosphere diameters from 70 to 460 nm. Significantly, altering the block copolymer hydrophilic–hydrophobic balance enabled control of the internal pore diameter of the BPNs from 10 to 19 nm.


Dialysis
Dialyses were conducted in miliQ water using a 5ml QuixSep Dialyzer TM (Orange Scientific) covered with a dialysis membrane (Medicell International Ltd., 36/32" diameter) with a cut off at approximately 12.0-14.0 kDa. Water (HPLC Grade, Fisher Scientific) and THF (Analytical grade, Fisher Scientific) were filtered using Sartorius Minisart 0.45 µm filters before use.
In a typical procedure, 10 mg of block copolymer was dissolved in 4 ml THF and left stirring in an oil bath at 35 °C. 6 ml water (HPLC Grade, filtered) was added drop-wise to the stirring solution over 90 mins. The solution was transferred to the dialyzer, sealed with the dialysis membrane and immersed in 4 L of stirring 35 °C miliQ water for 24 h to displace the THF. During this time, the water was replaced twice.

Optical and Size Measurements
Dry transmission electron microscopy (TEM) was carried out using a JEOL JEM (200-FX) machine, operating at 120 kV. 20 μl of the dialyzed sample was deposited onto a carboncoated copper grid (200 mesh) left for 30 s and removed via suction. The grid was then stained with a solution of 5 % uranyl acetate and 1 % acetic acid. 20 μl of this solution was deposited onto the grid and removed after 5 s. Excess solution was dabbed away using filter paper. Particle size measurements were carried out using an electron density profile tool, and averages were determined from 50 particles. Dynamic light scattering measurements were carried out on a Malvern High Performance Particle Sizer (Nano Zetasizer HPPS HPP5001) with a laser at a wavelength of 633 nm. 1 ml of the dialyzed solution was taken, filtered using a 1.2 μm filter and placed in a clean cuvette. The desired temperature was set (4 °C) and the sample left at this temperature for 15 mins before the runs were conducted. The particle sizes, size distributions and polydispersity for each aggregate solution were analyzed. At each temperature, ten size readings were obtained and the quoted size values are an average of these results.

Cryo TEM and Cryo ET sample preparation
CryoTEM samples for 2D and 3D imaging were prepared on 200 mesh Cu TEM grids containing a hydrophilized R2/2 Quantifoil carbon support film (Quantifoil Micro Tools GmbH, Jena, Germany). The carbon support film was hydrophilized by surface plasma treatment (glow discharge) in a Cressington 208 carbon coater directly prior to use. In general, 27 μl of the dialyzed solution equilibrated to the desired temperature (4 °C), was mixed with 3 μl of a colloidal gold solution (4 °C, 5 nm diameter, PAG coated). Subsequently, 3 μl of this mixture was applied to the hydrophilized TEM grid, blotted and vitrified in an automated vitrification robot (FEI Vitrobot TM Mark III) by plunging into liquid ethane. This process was performed with the environmental chamber of the Vitrobot conditioned to 100 % humidity and 4 °C to prevent temperature and drying artefacts.

CryoTEM and Cryo ET imaging
Samples were imaged in two dimensions and three dimensions (cryo electron tomography) in a FEI CryoTitan operated at 300 kV and equipped with a field emission gun (FEG). For imaging low-dose mode and underfocus conditions were used (-5 μm to -50 μm depending on the magnification) with a 70 μm objective aperture inserted in the beam path. Images were recorded zero-loss filtered with a post column Gatan Energy Filter (20 eV energy window) using a 2k x 2k Gatan slow scan CCD camera. Tomographic tilt series acquisition was performed with Inspect 3D software (FEI Company). Alignment and reconstruction was carried out in IMOD [2] using SIRT with 10 iterations. Subsequently data were denoised by nonlinear anisotropic diffusion prior to visualization. [3] Typical electron tomography acquisition conditions: Temperature at vitrification: 4 °C Typical angular sampling: 0° to +64 and to -64° at 1° or 2° increments Magnification: 19000 × or 24000 × Defocus: -5 to -10 μm Total electron dose: 60 to 100 e -.Å -2 Pixel size at 19000 x : 0.463 nm for images; 0.93 nm voxel size in the corresponding tomogram Pixel size at 24000 x : 0.370 nm for images; 0.74 nm voxel size in the corresponding tomogram

Segmentation
Segmentation of the reconstructed tomograms was conducted with several steps to reduce noise and specifically segment out the features that were observed and interpreted visually from the slices of the reconstructed tomograms. Noise reduction was implemented using a nonlinear anisotropic diffusion (NAD) filter in IMOD. [2][3] The rest of the segmentation was carried out in Avizo® Fire, a 3D analysis software. A mask was generated by manually tracking the edges of the sphere and subsequent interpolation over a few slices. This mask was then applied to the tomogram to select only the sphere; the background surrounding the sphere was made transparent. Subsequently, the dark regions in the sphere were segmented out (PODMA) by using a tophat filter and threshold to segment out the local structures, whilst ignoring the largescale intensity variations. The threshold was chosen by visually judging the best match of the segmentation with the slices of the tomogram. Optionally, in some volumes a morphological opening, closing or erosion was done to highlight certain features that were observed in the slices of the tomogram. The figures show a volume rendering combined with an isosurface and an orthoslice. The volume rendering in yellow is of the PODMA phase, whilst the water/PEO phase is transparent. The isosurface was generated on the border of the PODMA phase and the water/PEO phase. The orthoslice was taken from the NAD filtered reconstruction and overlaid with the local segmentation in yellow. A full summary of the segmentations including the morphological operations and the ball sizes that were used as structuring elements in pixels are given in the table below.

Preparation of Ethylene Glycol Methyl 2-Bromo 2-Methyl Propanoate (PEO Macrointiator)
A modification of a literature method was used [4] and was typically carried out as follows: a solution of PEGME (M n ca. 2.0 kDa) (10 g, 5 mmol) in 50ml dichloromethane was added dropwise to a stirred mixture of 2-bromoisobutyryl bromide (1.24 ml, 10 mmol), triethylamine (1.4 ml, 10 mmol), and dimethylamino pyridine (1.22 g, 10 mmol) in dichloromethane at 0 °C for 1 h under nitrogen. The solution was stirred for a further 18 h at room temperature. A third of the reaction solvent was evaporated off and the resultant yellow precipitate was filtered off. The remaining solution was made up to 100 ml with dichloromethane and transferred to a separating funnel. The solution was washed several times with a saturated sodium bicarbonate solution and then with a 10 % hydrochloric acid solution. The organic layer was collected and dried over anhydrous magnesium sulphate for an hour, filtered and the dichloromethane evaporated off. Finally, the resultant yellow solid was left under vacuum at 50 °C for two hours. Yields ranged from 50-95 %. The structure was confirmed using 1 H, 13 C NMR and Fourier Transform Infrared (FTIR) spectroscopy, and the molecular weight parameters determined using size exclusion chromatography (SEC).

Synthesis of PEO-PODMA block copolymers via ATRP
A modified literature synthesis was conducted as follows: [1a] The PEO macroinitiator (0.84 g, 0.44 mmol) was placed in a Schlenk tube along with a magnetic stirrer, and dissolved in 2 ml of a solution of xylene:IPA (9:1 vol/vol). To this solution were added octadecyl methacrylate (also dissolved in 2 ml xylene:IPA (9:1)) (2.5 g, 7.4 mmol), N-(n-octyl)-2pyridyl(methanime) (0.19 g, 0.88 mmol) and Cu I Br (0.06 g, 0.44 mmol). The Schlenk tube was then sealed and the solution degassed with nitrogen for 30 mins. The reaction mixture was stirred at 95 °C for 12 h under N 2 . The reaction mixture was then exposed to air, diluted in THF and run through an alumina column. The polymer was isolated by precipitation from the THF solution into methanol and yields ranged from 40-90 %. The structure was confirmed using 1 H, 13 C NMR and FTIR spectroscopy, and molecular weight parameters determined using SEC.

CryoTEM and DLS Measurements
All results are acquired from 0.1 wt% polymer dispersions unless otherwise stated. CryoTEM and DLS were conducted at 4 °C. In DLS figure legends, I ave , N ave and V ave are the intensity, number and volume size average curves respectively. Black particles in cryoTEM micrographs are gold nanoparticles used as fiducial markers for electron tomography. All scale bars represent 100 nm.             Figure S14: Galleries of select z-slices of 3D reconstructions of tilt series of aggregates from PEO-PODMA block copolymers with varied PEO contents (numbers stated in each gallery below), revealing the internal structure. All scale bars represent 100 nm. Figure S15: Segmentations of the reconstructed tomograms of different morphologies, with corresponding cuts through the volume (from Figure 2 in manuscript). Figure S16: Segmentations of the reconstructed tomograms of bicontinuous nanospheres from block copolymers with different PEO weight fractions (f), with corresponding cuts through the volume (from Figure 3 in manuscript).

Domain size measurements
Measurements for the internal PODMA and hydrated PEO regions were obtained by analysing the 3D reconstructions of the aggregates using in-house MATLAB scripts. The PODMA wall thickness was defined as the shortest distance between two neighbouring bright aqueous channels when browsing up and down through the xy-slices of the tomogram. Measurements were obtained from an average of at least 20 points in the PODMA wall. Equivalently, pore size measurements were performed by finding the widest diameter of bright aqueous regions when browsing up and down through the xy-slices of the tomogram. An illustration of the process is provided below in Figure S15.