Patchy nanoparticles by atomic stencilling

Chemicals

All chemical compounds have been used with out additional purification after buy: gold(III) chloride trihydrate (≥99.9% hint metals foundation, HAuCl₄·3H₂O, Sigma-Aldrich), sodium borohydride (99%, NaBH₄, Sigma-Aldrich), sodium iodide (99.999%, NaI, Sigma-Aldrich), sodium bromide (≥99.99%, NaBr, Sigma-Aldrich), sodium chloride (99.3%, NaCl, Fisher Scientific), potassium iodide (≥99.5%, KI, Sigma-Aldrich), silver nitrate (≥99.0%, AgNO₃, Sigma-Aldrich), sodium tetrachloropalladate(II) (roughly 36%, Na₂PdCl₄, Acros Organics), l-ascorbic acid (BioXtra, ≥99.0%, AA, Sigma-Aldrich), sodium hydroxide (99.99%, NaOH, Sigma-Aldrich), cetyltrimethylammonium chloride (>95%, CTAC, TCI, product quantity: H0082), benzyldimethylhexadecylammonium chloride (≥97.0%, BDAC, Sigma-Aldrich), cetylpyridinium chloride (>98.0%, CPC, TCI, product quantity: H0078), sodium citrate tribasic dihydrate (≥99.0, C₆H₅Na₃O₇, Sigma-Aldrich), PS₁₅₄–b-PAA₅₁ (M_n = 16,000 for the PS block and M_n = 3,700 for the PAA block, M_w/M_n = 1.04, Polymer Source Inc.), PS₁₅₄–b-poly(acrylamide)₄₉ (M_n = 16,000 for the PS block and M_n = 3,500 for the poly(acrylamide) block, M_w/M_n = 1.15, Polymer Source Inc.), PS₁₅₄–b-poly(ethylene oxide)₁₇₀ (M_n = 16,000 for the PS block and M_n = 7,500 for the poly(acrylamide) block, M_w/M_n = 1.09, Polymer Source Inc.), 2-NAT (99%, Sigma-Aldrich), biphenyl-4-thiol (97%, Sigma-Aldrich), DMF (anhydrous, 99.8%, Sigma-Aldrich) and hydrochloric acid (99.999% metals foundation, HCl, Alfa Aesar). Nanopure deionized (DI) water (18.2 MΩ cm at 25 °C) purified by a Milli-Q Advantage A10 system was used. Note that the purity of cetyltrimethylammonium bromide (CTAB) is necessary for gold NP synthesis. We have used three sorts of CTAB: (1) Sigma-Aldrich, for molecular biology, ≥99%, product quantity: H6269; (2) Sigma-Aldrich, BioXtra, ≥99%, product quantity: H9151; and (3) Sigma-Aldrich, BioUltra, ≥99.0%, product quantity: 52365. For the info offered on this work, CTAB (1) and CTAB (2) have been used for the experiments carried out earlier than February 2024. CTAB (3) has been used to breed the experiments afterwards.

Synthesis of core NPs

Gold octahedron, cuboctahedron, rhombic dodecahedron, dice and bipyramid NPs have been synthesized following beforehand reported strategies with slight modifications^33,34 (Supplementary Note 1). Palladium cubes have been synthesized following earlier literature⁵¹ with modifications (Supplementary Note 2.8).

Synthesis of patchy NPs

Patchy NPs have been synthesized in two steps, that are iodide masking and ligand-mediated polymer grafting. Using the patchy octahedra in Fig. 3c for example, for the iodide-masking step, we first degas DI water by purging with N₂ for 1 h to attenuate oxygen content material. This degassed water is used all through the masking step to attenuate the oxidative etching of NPs. A gold octahedra answer was centrifuged at 2,400 × g for 15 min and the ensuing pellet was diluted in 20 mM CTAC with degassed water to attain a closing NP focus of 0.5 optical density (OD) at a most extinction wavelength λ_max (Stock Solution I). Next, 5.75 µl of freshly ready iodide answer (1 ml of 10 mM NaI, 500 µl of 200 mM NaOH and eight.5 ml degassed DI water) was added dropwise to six.9 ml of Stock Solution I underneath gentle vortex and incubated undisturbed for 30 min. After the incubation, 2.1 ml of 40 mM CTAB was added, adopted by three rounds of centrifugation (3,400 × g, 5,300 × g and a pair of,200 × g for 15 min every). The supernatant was eliminated after every centrifugation and a 10-µl pellet was redispersed in 10 ml of 20 mM CTAB after the primary spherical and 1 ml of DI water after the second spherical, respectively. The closing pellet from the third centrifugation, after supernatant removing, was diluted with DI water to succeed in 5.0 OD at λ_max with 0.07 mM CTAB (Stock Solution II). The closing CTAB focus could be diverse relying on in another way formed NPs, as supplied in Supplementary Note 2.

For ligand-mediated polymer grafting, nonetheless utilizing the patchy octahedra in Fig. 3c for example, we sequentially added 817 µl DMF, 5 µl of 2-NAT answer (2 mg ml⁻¹ in DMF), 200 µl of Stock Solution II and 80 µl of PS-b-PAA (8 mg ml⁻¹ in DMF) into an 8-ml vial underneath gentle vortex. The vial was tightly capped with a Teflon-lined cap, sonicated for five s, Parafilm-sealed and heated at 110 °C in an oil tub for two h with out disturbance. After cooling to room temperature within the oil tub, the response combination was transferred to a 1.5-ml microcentrifuge tube and centrifuged 3 times at 3,400 × g, 1,300 × g and 1,000 × g (15 min every) to take away molecular residues. After every centrifugation, 1.49 ml of the supernatant was eliminated and a 10-µl pellet was redispersed in 1.49 ml of water. Following the ultimate centrifugation, the 10-µl pellet was diluted with 90 µl of water for subsequent characterizations and self-assembly experiments. Detailed descriptions of the experimental process of the synthesis of different patchy NPs are supplied in Supplementary Note 2 and Supplementary Tables 2–8.

Similar procedures with modifications could be prolonged to palladium nanocubes (Supplementary Note 2.8), different thiols and block copolymers (Supplementary Notes 2.1 and a pair of.5) and scale-up synthesis (Supplementary Note 2.9).

Large-scale self-assembly of patchy NPs

For self-assembly experiments, a patchy NP answer in a 1.5-ml microcentrifuge tube was left undisturbed in a single day, permitting the particles to sediment and focus (≥30.0 OD at λ_max).

Coffee ring effect-driven meeting on a silicon wafer

A silicon wafer (3 × 3 mm², Ted Pella) was cleaned with water, acetone and isopropanol via sonication for five min every, adopted by 60 s of O₂ plasma remedy utilizing a Harrick Plasma PDC-32 (most RF energy of 18 W) to make the floor hydrophilic. Meanwhile, a 4 × 4-inch² piece of Parafilm was positioned on a TechniCloth (TX609, Texwipe) with a water-filled Petri dish cowl (60 × 15 mm², Falcon) on high. 3.5 µl of a concentrated patchy NP answer was drop-casted onto the wafer on the Parafilm. Immediately after drop-casting, each the silicon wafer and Petri dish have been lined by a big Petri dish (100 × 15 mm², VWR) and gently pressed right down to seal in opposition to the Parafilm, sustaining a moist setting (humidity: 75–80%). A typical drying course of takes 16–24 h at room temperature (Supplementary Fig. 38a).

Capillary force-driven meeting in a glass capillary tube

A ten-µl glass capillary tube (internal diameter: 0.5573 mm, Drummond Scientific Company) was stuffed with 2 M NaOH, incubated for 20 min, rinsed with water and dried to make the internal wall hydrophilic earlier than use. 5 µl of concentrated patchy NP answer was then drawn into the capillary tube, holding the NP answer a number of centimetres away from each tube ends. The capillary tube was suspended inside a loosely capped 15-ml centrifuge tube, which was then positioned in a desiccator till the answer had absolutely dried, forming a visual gold ring across the inside. See the experimental set-up in Supplementary Fig. 38b.

Electron microscopy characterizations

TEM, SEM and STEM characterizations

TEM photographs have been acquired with a JEOL 2100 LaB6 transmission electron microscope operated at 200 kV. SEM photographs have been captured utilizing an FEI Helios NanoLab 600i operated at 2 kV with a beam present of 0.17 nA and a working distance of 4.0 mm. HAADF-scanning transmission electron microscopy (STEM) imaging was performed on a probe aberration-corrected Thermo Fisher Scientific Themis Z scanning transmission electron microscope operated at 300 kV. STEM-EDX mapping was acquired with a Thermo Fisher Scientific ‘Kraken’ Spectra 300 operated at 60 kV outfitted with Dual-X EDX detectors with a group angle of 1.76 sr.

Sample preparation for HAADF-STEM characterization

For NP side imaging and evaluation, options of pristine NPs have been drop-casted on TEM grids, with CTAB focus diminished via three rounds of centrifugation. Approximately 50–200 µl of the pristine NP options in 40 mM CTAB was used per pattern. Each pattern was first diluted with 1 ml of water in 1.5-ml microcentrifuge tubes, adopted by eradicating 990 µl of supernatant after every of the primary two centrifugations. After every centrifugation, a 10-µl pellet was redispersed in 990 µl of water. Following the third centrifugation, a 1.5-µl pellet was drop-casted onto a carbon-coated copper TEM grid (Electron Microscopy Sciences, CF400-Cu) and dried fully. Before imaging, samples have been plasma-cleaned for two min at 15 W underneath a 12-sccm circulate of Ar + O₂ to take away residual CTAB molecules protecting NPs utilizing a PIE Scientific Tergeo-EM plasma cleaner.

HAADF-STEM imaging

HAADF photographs have been acquired with a beam present of roughly 20 pA and semi-convergence angle of 18 mrad. The digital camera size was 115 mm and the gathering angles of the HAADF detector have been 62–200 mrad for picture acquisition. To decrease picture distortion from pattern drift, a number of sequential frames (10–50 frames) with brief dwell time (100–500 ns) have been acquired, which have been used to render drift-corrected frames to reinforce the distinction.

TEM tomography

TEM photographs for tomography reconstruction have been collected by titling the samples from 0° to −60° after which from 0° to +60° in 2° intervals, capturing 61 photographs per pattern. To decrease beam injury, a low electron dose fee of 6–8 e⁻ Å⁻² s⁻¹ was used. To enhance the polymer patch distinction, a defocus of −2,048 nm was used. Image alignment and contrast-transfer perform correction have been carried out utilizing IMOD64 4.9.3 software program⁵². Tomograms have been generated utilizing OpenMBIR with diffuseness of 0.3 and smoothness of 0.2 as reconstruction parameters, which makes use of a model-based iterative algorithm of tomogram reconstruction⁵³. 3D fashions of the tomograms have been visualized utilizing Amira 6.4 from Thermo Fisher Scientific. The tomograms have been denoised utilizing three filters: median (3 × 3 × 3 voxel neighbourhood, iterations 26), Gaussian filter (kernel dimension 9, 3 × 3 × 3 customary deviation) and edge-preserving smoothing (time 25, step 5, distinction 3.5 and sigma 3). Polymer patches and gold NPs have been segmented by greyscale depth thresholding and refined manually in Amira 6.4 software program.

STEM-EDX characterization

Samples for the STEM-EDX evaluation have been ready by drop-casting Stock Solution II on carbon-coated copper TEM grids (Electron Microscopy Sciences, CF400-Cu). The octahedra and cuboctahedra have been iodide-masked. To decrease carbon contamination, the TEM grids have been baked in a single day at 130 °C in excessive vacuum to take away extra CTAB. STEM-EDX maps have been acquired with a 120-pA probe present, a 30-mrad semi-convergence angle and steady raster-scanning with drift correction, utilizing a 2-µs dwell time over roughly 4 h. For iodide-masked cuboctahedra, the NPs have been predominantly oriented on the carbon assist alongside the [001] route. Therefore, the stage was tilted 45° to the [110] zone axis, aligning each the {111} and {100} aspects ‘edge-on’. Elemental mapping was achieved by becoming and quantifying the height intensities above the background utilizing the Cliff–Lorimer methodology⁵⁴. Two high-quality EDX maps of seven cuboctahedra have been chosen and a complete of 1,692 line profiles have been obtained (Supplementary Fig. 25). Similarly, iodide-masked octahedra have been oriented on the carbon assist alongside the [110] route and their ‘edge-on’ {111} aspects have been used for iodide focus evaluation (Extended Data Fig. 2).

TEM picture evaluation

Patch native thickness t
_loc

The t_loc of patches was decided because the diameter of the biggest circle that may match throughout the patch and contains the goal pixel. Patch contours have been manually outlined in ImageJ to outline the patch areas. The t_loc map was then obtained utilizing the built-in ‘Local Thickness’ perform in ImageJ, as additionally detailed in our earlier work²⁴.

Maximum patch thickness t
_m and patch protection fraction f
_cov

The values of t_m and f_cov have been measured utilizing a neural-network-based methodology on TEM photographs⁵⁵. The neural network-based TEM picture segmentation is detailed in Supplementary Note 3 and Supplementary Figs. 32 and 33. After coaching the neural community with manually labelled TEM photographs of numerous patchy NP shapes, a form fingerprint t was extracted from the patch contours. First, the centroid of the gold NP was recognized and rays extending from θ = −180° to θ = 179° at 1° intervals have been drawn from the centroid. The distance that every ray travels throughout the patch area was recorded as a perform of θ. t_m is set by the utmost worth of t and f_cov is recognized because the vary of θ by which t is non-zero. See detailed description in Supplementary Note 3 and Supplementary Figs. 34–36.

Other characterizations

UV–Vis measurements

Ultraviolet–seen (UV–Vis) spectra have been measured utilizing a Scinco S-4100 PDA spectrophotometer with a quartz cuvette (path size = 1 cm, VWR).

XPS characterization

XPS evaluation was carried out utilizing a Kratos Axis Ultra outfitted with a monochromatic Al Kα radiation X-ray supply and an vitality decision of 0.4 eV. Before characterization, 10 µl of every Stock Solution II, ready with varied iodide concentrations, was drop-casted on a silicon wafer cleaned with water, acetone and isopropanol after which absolutely dried. The silicon wafers have been fastened to the pattern bar and transported to the instrument in an hermetic container underneath an Ar ambiance for XPS sign acquisition. Data processing and peak becoming have been performed utilizing the CasaXPS software program. Atomic concentrations inside samples have been calculated by integrating the fitted XPS spectra for all analysed areas and adjusting for atomic relative sensitivity components. See detailed pattern circumstances in Supplementary Table 9.

Raman characterization

Raman characterization was carried out utilizing a Horiba XploRA-nano TERS/TEPL with a 100× goal lens. The Raman samples have been ready following two steps, with out polymer grafting (Supplementary Table 10). First, gold octahedra have been iodide-masked in various [I⁻] after which concentrated to succeed in 20.0 OD at λ_max in 0.07 mM CTAB, following the iodide-masking process described above. Next, 40 µl of 2-NAT (0.2 mg ml⁻¹ in DMF) and 200 µl of the iodide-masked octahedra have been sequentially added into 860 µl of DMF in a vial with gentle vortex. The remaining response situation and washing steps comply with the usual iodide masking process, apart from centrifugation, which was carried out 3 times at 2,900 × g for 15 min and a pair of,400 × g for 7 min twice. After the primary two centrifugations, the pellet was resuspended in 1 ml of DMF. After the ultimate centrifugation, the sediment was redispersed in 20 µl of DMF and the NP focus was adjusted to 13.0 OD at λ_max by including water. A 5-µl aliquot of this NP answer was drop-casted onto a clear glass slide. The glass slide was used after washing with isopropyl alcohol and water after which absolutely dried. Raman measurements have been carried out with a 638-nm excitation wavelength, 1-mW laser energy and three scans for common (90 s publicity time every).

X-ray tomography of large-scale self-assembly

The X-ray tomography pattern was ready via a focused-ion beam milling utilizing the FEI Helios NanoLab 600i. Before milling, platinum was deposited on the self-assembled patchy rhombic dodecahedra on a silicon wafer, with a deposition diameter of three µm and a thickness of 300 nm. Subsequently, the meeting was milled right into a cylinder with a diameter of two.5 µm. The formed cylinder was lifted with an OmniProbe and mounted onto a tungsten needle tip (Supplementary Fig. 50).

X-ray tomography information have been acquired on the laborious X-ray nanoprobe beamline of National Synchrotron Light Source II at Brookhaven National Laboratory. A monochromatic beam at 12 keV was chosen after which targeted by a set of crossed multilayer Laue lenses to supply a nanobeam⁵⁶ with a dimension of roughly 13 nm. 2D fly-scans have been carried out in a grid sample with at the very least 150 × 150 pixels, 50-ms dwell time and 10-nm step dimension at each 2° to gather 91 projections protecting 180°. Far-field diffraction patterns have been analysed with a ptychographic reconstruction algorithm to retrieve each the complex-valued probe and the item features. The acquired fluorescence spectra have been fitted utilizing the software program bundle PyXRF. Individual 2D frames have been coarsely aligned utilizing ImageJ 1.5, MultiStackReg plugin, whereas fantastic alignments have been adjusted with a cross-correlation perform in Tomviz 1.9, adopted by 3D reconstruction. The information have been additional processed by Fourier filtering utilizing the sharp lattice peaks within the reciprocal house picture of the superlattice to take away noise and sharpen the particle positions for subsequent segmentation, utilizing Dragonfly 2020.2 software program⁵⁷.

DFT calculations of gold surfaces with iodide and 2-NAT

All DFT calculations have been carried out utilizing the Vienna Ab initio Simulation Package^58,59,60 with projector augmented waves⁶¹. The generalized gradient approximation by Perdew, Burke and Ernzerhof was used for the exchange-correlation useful⁶². We selected an vitality cut-off of 450 eV as an optimum worth for our plane-wave foundation set. For the sampling of the primary Brillouin zone, Monkhorst–Pack grids have been used⁶³. We additionally included the DFT-D3 methodology of Grimme et al. with the Becke–Johnson damping to explain long-range van der Waals interactions⁶⁴. See particulars in Supplementary Note 4.

Computation and concept modelling of patchy NPs and their assemblies

Patchy NP grafting simulation

A library of particular person patchy NPs was simulated utilizing MD utilizing the HOOMD-blue simulation engine⁶⁵. First, a single anisotropic particle was positioned on the centre of the simulation field. Polymer chains with one grafting finish have been then randomly distributed on the floor of the central anisotropic particle and allowed to freely transfer on the floor to seek out their equilibrium positions. The location of iodide-masked areas predicted from our iodide adsorption concept (see particulars in Supplementary Note 5) have been modelled as spherical beads which might be strongly repulsive to the polymer chains. All chain–chain interactions exhibit Lennard–Jones attraction with one another to seize the PS–PS aggregation in experimental circumstances. All different interactions have been purely repulsive and modelled utilizing the Weeks–Chandler–Andersen repulsive potential. For detailed description, see Supplementary Note 8.

Theoretical prediction of patches

Polymer patch patterns and sizes have been theoretically predicted utilizing MC grafting simulation involving two steps: (1) figuring out the floor distribution of iodide-masked area and (2) inserting polymer chains onto the unmasked (‘free’) floor websites, whereas accounting for chain–chain interactions. We started by setting up a grid of factors on the core NP floor, defining potential websites for iodide or polymer attachment. Using the Metropolis algorithm with floor energies computed from DFT, iodide-masked factors have been positioned on the varied floor websites (Supplementary Note 5). Placing iodide-masked factors on the particle floor makes use of the Metropolis algorithm, guided by floor energies computed from DFT. Iodide-masked factors can not accommodate polymer grafting, leaving the remaining open areas as the one ‘free’ websites for polymer grafting. After an occupancy matrix logs iodide-masked positions, the primary polymer chain grafts on the floor based mostly on its Boltzmann-weighted free vitality as a perform of floor areas. The matrix updates with this chain attachment and ‘free’ floor websites inside a correlation size ξ of any polymer-occupied websites achieve a beneficial chain–chain interplay, ruled by the Flory–Huggins parameter χ (Supplementary Note 6). This total course of repeats for every polymer chain till the goal grafting density is achieved. See Supplementary Note 7 for particulars.

Simulation of patchy NP self-assembly

MC simulations have been carried out to acquire the self-assemblies of patchy NPs, utilizing the HOOMD-blue simulation engine⁶⁵. We first outlined a patchy particle whose patch areas and sizes are commensurate with these measured from the above MD simulation and validated with concept and experiments. Patches have been modelled to exhibit laborious sphere interactions with one another to seize the sturdy PAA–PAA electrostatic repulsions between the outer floor of the patches (Supplementary Fig. 49). The core NP interactions have been modelled utilizing a Kern–Frenkel attraction between the varied floor websites that weren’t lined by polymeric patches. See Supplementary Note 9 for particulars.