Most morphologies were fabricated from polystyrene (weight-averaged molecular weight Mw ~ 5,798, 192,000, and 230,000, Sigma-Aldrich). Polystyrene was dissolved in the following solvents: tetrahydrofuran (VWR), chloroform (99% anhydrous, Sigma-Aldrich), N,N-dimethylformamide (Fisher Scientific), N,N-dimethylacetamide (Sigma-Aldrich), 1,4-dioxane (99.8% anhydrous, Sigma-Aldrich), and ethyl acetate (99.5%, Acros Organics). Nonsolvents used were deionized water and ethanol (anhydrous, VWR).
Additional polymers used were poly(vinylidenefluoride) (average Mw ~ 530,000, Sigma Aldrich), polyester (Ecdel PCCE 9966 copolyester ether elastomer, Eastman), and chitosan (77% deacetylated, Sigma Aldrich). Solvents for these were dimethylsulfoxide (anhydrous, ≥ 99.9%, Sigma Aldrich), chloroform (anhydrous, ≥ 99.9%, contains 0.5-1.0% ethanol as stabilizer, Sigma Aldrich), and 1.5% acetic acid (hygroscopic, Acros Organics). The nonsolvents were ethanol and methanol (Chromasolv® for HPLC, ≥ 99.9%, Sigma Aldrich).
1.2 Fabrication platforms and production
In order to sample the entire range of fluidic regimes, two fabrication platforms are used which are also described in detail in Roh et al.1 Laminar to weakly turbulent flow is modeled using a simple Couette cell, which is also capable of developing turbulent Taylor vortices at higher rotational speeds. This type of apparatus was chosen since Taylor-Couette flow is well-studied and the parameters are easily controlled. An acrylic baffle was placed at the top of the fluid to reduce boundary effects. The baffle also maintained the same injection position within the apparatus. Polymer solution was injected in the middle of the shearing gap, and calculated shear rates and Re are based on the shear rate in the middle of the gap.
An IKA Magic Lab device (IKA Works Inc., Wilmington, NC, USA) with MK module is used for producing extremely turbulent flows (3,000 – 26,000 rpm). The module has a conical rotor with a stator which axially displaces to adjust the shearing gap from 80 – 650 µm. This device should produce turbulent flows at all operational speeds. Due to serrations on the rotor-stator, it is difficult to precisely estimate the true shear rates and flow characteristics due to a probability of cavitation at the higher shear rates. Our approximation of shear rates, not considering the complex rotor-stator geometry, gives a shear rate range of 39,000 – 157,000 s−1.
Unless otherwise specified, morphologies were fabricated with a Couette cell with n = 0.65 and IKA Magic Lab with 250 µm shearing gap and 20,000 rpm. Polymer solutions used in this study ranged from 0.1 – 20 wt.% in various solvents and primarily were made with PS of 230 kDa molecular weight. The solutions were injected into the nonsolvent shearing medium at approximate rates of 0.05 mL/min and 2.5 mL/min for the laminar and turbulent platforms, respectively. In each fabrication run, the volume of polymer solution injected was maintained below 10 v.% of the amount of nonsolvent medium. After complete fabrication, the particles were collected and washed several times in ethanol before storage.
1.3 Building ternary phase diagrams
Binodal curves were found through cloud point tests. Various concentrations of polymer solution as components of different ternary systems, were titrated with nonsolvent medium with stirring until the solution became opaque and would separate into two phases upon rest. All diagrams were obtained with PS of 230 kDa, except for the binodal curve for nonsolvent system of 20% Dioxane (aq) which was found using PS with molecular weight of 190kDa. The NS:S ratios at the onset of phase separation were calculated based on the amount of nonsolvent added at the point of opacity.
The morphologies were visualized mainly by a bright field and fluorescence microscope (BX-61 Olympus) and field emission scanning electron microscope (FEI Verios 460L). Material preparation for imaging includes dilution in ethanol and sonication (~ 10 min) prior to spin-coating (2000 rpm) onto the imaging substrate, glass slide or silicon wafer, to limit particle entanglement and aggregation. Phase contrast filters were used with bright field microscopy. Samples for SEM imaging were coated with Au-Pd nanoparticles (approximately 5 nm layer) for sufficient electron conduction. Only the chitosan particles in Fig. 5 were not coated.
Image analysis was performed with ImageJ software with SEM images for approximating characteristic dimensions of the morphologies. The measurements for the particles, fibers, and ribbons from Fig. 2a and Supplementary Fig. 2a were conducted manually and the sample sizes (N) for 1, 2.5, 5, 10, and 15 wt.% were all between 17 and 2014. The effective pore diameter measurements for Fig. 3 were done by measuring the average area of the pores then calculating the associated diameter if the pores were circular. Sample size of pore size measurements ranged from 3390 to 9995 while those for thickness measurements ranged from 3 to 9. Terminating branch lengths in Fig. S5a were measured manually and the samples sizes for the fabricated PS SDCs in 5,000, 10,000, 15,000, and 20,000 rpm conditions were 22 < N < 39 for 2.5 wt.% PS/THF and 50 < N < 133 for 5.0 wt.%.
1.5 Viscosity and rheological characterization
The viscosities of polymer solutions, ranging from 0.1 to 7.4 v.%, were measured using a Ubbelohde viscometer (calibration constant = 0.09738 (mm2 s−1 s−1). Sample of PS in THF with concentrations ranging from 0 to 7.4 v.% were prepared. Viscometer was carefully filled with appr. 25 mL of polymer solution. Solution was pulled through the tubes with a macro pipette prior to measurement.
The viscoelastic properties were used to approximate the degree of gelation in particle suspensions. The material, already suspended in ethanol after washing procedures, were resuspended in pure polyethylene glycol 400 (Acros Organics, average M.W. 400, d = 1.1275) to have final weight concentrations of 0.5-1.5 wt. %. PEG 400 was chosen as the suspension medium to show greater gelation propensity using PS SDC particles compared to cellulose acetate (CA) SDCs (CA SDCs were used in Roh et al. at similar particle concentrations and showed more fluid-like behavior due to refractive index matching). We did not see a need to reduce van der Waals forces so the suspension medium did not have matching refractive index with PS. The SDC-EtOH-PEG 400 mixtures were gently mixed at room temperature until all ethanol had evaporated. The samples were evaluated at 25oC using a rheometer (Discovery HR-2, TA instruments) equipped with a sand-blasted plate (40 mm diameter) and gap size of 800 µm. Frequency sweeps were performed at 1% strain for 100 – 0.1 rad s−1, and amplitude sweeps were performed at 1 rad s−1 from 0.01 – 1000% strain.