DNA-mediated assembly of Au bipyramids into anisotropic light emitting kagome superlattices

Abstract

Colloidal crystal engineering with DNA allows one to design diverse superlattices with tunable lattice symmetry, composition, and spacing. Most of these structures follow the complementary contact model, maximizing DNA hybridization on building blocks and producing relatively close-packed lattices. Here, low-symmetry kagome superlattices are assembled from DNA-modified gold bipyramids that can engage only in partial DNA surface matching. The bipyramid dimensions and DNA length can be engineered for two different superlattices with rhombohedral unit cells, including one composed of a periodic stacking of kagome lattices. Enabled by the partial facet alignment, the kagome lattices exhibit lattice distortion, bipyramid twisting, and planar chirality. When conjugated with Cy-5 dyes, the kagome lattices serve as cavities with high-density optical states and large Purcell factors along lateral directions, leading to strong dipole radiation along the z axis and facet-dependent light emission. Such complex optical properties make these materials attractive for lasers, displays, and quantum sensing constructs.

Au bipyramids are assembled into kagome lattices driven by DNA hybridization, leading to anisotropic light emission.

INTRODUCTION

Understanding the assembly of atomic and nanoscale building blocks into crystals is one of the cornerstones of chemistry and materials science as it is central to engineering functional materials with properties by design (1–9). Furthermore, it is well established that targeting a set of defined properties depends on both the compositions of a material and the parameters of crystallinity including spacing, orientation, and habit (10–14). Routes to deliberately engineer atomic, molecular, and nanoscale ordering across length scales have led to a swath of important materials with exotic photonic and mechanical properties (15–17). For example, control over periodicity and symmetry of superlattices enables the design of colloidal crystals with tunable optical diffraction (18), superfluorescence (19, 20), lattice plasmon resonance (21), and negative refraction (17), which are not attainable in individual nanocrystals. As synthetic methods for producing nanoparticles of different compositions, sizes, and crystallinity are developed, colloidal crystal engineering provides access to previously unidentified materials with exotic properties by design (22–25).

DNA-mediated programmable assembly of nanomaterials has emerged as one of the most powerful and versatile ways to control colloidal crystal structures (composition, symmetry, lattice parameter, and habit) (1, 3, 9, 26–33). In colloidal crystal engineering with DNA, collections of base pairs act as physical bonds that are decoupled from the identity of the nanoparticle cores or “atoms” (33–36). Through well-established routes, nanoparticle cores consisting of diverse shapes, compositions, and sizes can be modified with deliberately designed DNA strands to create “programmable atom equivalents” (PAEs) (34, 37, 38). The design of most such colloidal crystals follows the complementary contact model (CCM), which operates under the premise that the thermodynamic structure will be the one maximizing complementary DNA interactions between building blocks (9, 39–43). This model has been used to design and assemble crystals spanning over 90 different symmetries to date (26).

Because the CCM is based on maximizing DNA hybridization among neighboring PAEs, it is difficult to predict with the CCM how non–space-filling shapes will arrange because the particle facets are unlikely to perfectly align. Here, to better understand the scope of structural possibilities, prolate Au pentagonal bipyramids were synthesized as non–space-filling PAEs and then used as building blocks for colloidal crystal engineering with DNA. In all cases, one of two superlattices with rhombohedral symmetry are the final product. For bipyramids with short DNA strands, the superlattice has a simple rhombohedral unit cell with threefold symmetry composed of eight unidirectionally oriented bipyramids. For bipyramids with low aspect ratios and longer DNA strands, the superlattice has a complex rhombohedral supercell composed of 48 bipyramids arranged in eight layers in which the bipyramids twist and cluster into trimers arranged in a distorted kagome lattice with planar chirality, a consequence of partial facet alignment. These low-symmetry kagome superlattice structures have complex lattice plasmon resonance modes dependent on the light incident direction and polarization, making them attractive previously unidentified materials for lasers, displays, and optical sensing constructs. When conjugated with Cy5-molecular dyes, these kagome superlattices exhibit anisotropic light emission where the emitted light depends on the facet of the superlattice. The low-symmetry kagome lattices serve as cavities with high-density optical states and high emission enhancement factor (Purcell factor) along the x and y axes, leading to strong dipole radiation along the z axis and facet-dependent, anisotropic light emission.

RESULTS

Synthesis of kagome superlattices

Au pentagonal bipyramids of aspect ratios ranging from 2.1 to 3.8 were prepared via previously reported procedures (Fig. 1A and fig. S1) and functionalized with thiol-modified DNA (44). After washing with phosphate buffer, these bipyramids were dispersed in saline (0.5 M) and assembled by adding DNA linker strands with self-complementary GCGC sticky ends (table S1 and Fig. 1B). Polyethylene glycol blocks were introduced in the DNA linker strands to enable flexibility and control over DNA length. Under programmable slow cooling (70° to 65°C at 0.1°C/10 min; 65° to 45°C at 0.1°C/20 min; 45° to 25°C at 0.1°C/10 min), the bipyramids assembled into colloidal crystals with a crystal structure depending on bipyramid dimensions and DNA length (fig. S2A). For bipyramids with low aspect ratios (sample number 2, figs. S1 and S2A), three-dimensional (3D) crystals composed of stacked kagome lattices formed (Fig. 1C) with bipyramids adopting different orientations (fig. S2C). For bipyramids of intermediate aspect ratios (samples 3 to 7; figs. S1 and S2A), the assembled superstructures depend on the length of DNA strands: long and short DNA strands leading to kagome and rhombohedral lattices, respectively. A high-magnification scanning electron microscopy (SEM) image shows that the {111} facet has trimers of bipyramids sitting in a triangular lattice, which arranges the individual bipyramids in a kagome lattice (Fig. 1D). Within each trimer, the bipyramids have different orientations, as confirmed by SEM in an exposed side facet (Fig. 1E). In assembling bipyramids with high aspect ratios (bipyramids sample number 8 to 9, figs. S1 and S2A), rhombohedral crystals formed (fig. S2B) with unidirectional bipyramid orientation (Fig. 1, F and G, and fig. S3). To investigate the bipyramid positional order in the kagome lattices, the superstructures were fixed in polymer resins, and 50-nm-thick sections were prepared for transmission electron microscopy (TEM). There we observed successive layers of kagome lattices (45–48) with lattice positions offset laterally from one layer to the next (Fig. 1H and I), leading to flower-like stacking motifs (Fig. 1J). This unique assembly contains diverse bipyramid local motifs in different facets of the resulting colloidal crystals (Fig. 1K).

Fig. 1. DNA-directed assembly of Au bipyramids into kagome superstructures.

(A) Dark-field scanning transmission electron microscopy (STEM) images of Au bipyramids. (B) Schematic illustration of the DNA-functionalized Au bipyramids. (C) SEM image of kagome lattices. (D) SEM image of the top {111} facets of the kagome lattices. (E) Side-view SEM images showing the bipyramid assembly in the crystals. SEM images of the (F) {100} and (G) {110} facets in the rhombohedral superlattices. (H) Cross-sectional TEM image of bipyramids assembled into kagome superlattices. (I) Schematic illustration of a single kagome lattice and its stacking within the superlattice. (J) Cross-sectional TEM image containing two adjacent kagome lattices. (K) SEM images of the diverse assembly motifs of bipyramids in the kagome superlattices.

Establishing thermodynamic stability of the colloidal superlattices based on the CCM

We performed molecular dynamics (MD) simulations to ascertain whether the kagome superlattices are minimum free-energy structures following the CCM and to better elucidate the observed structures. The PAEs are modeled as hard cores with surrounding DNA shells (Fig. 2A and fig. S4) (25), with the DNA modeled as patches on the PAE surface (table S2). The patch-patch distance where DNA hybridization occurs, r₀, represents roughly twice the DNA length. DNA hybridization is modeled by the shifted Gaussian potential used successfully in previous studies (fig. S4) (28). More details about the simulation model and methods can be found in Materials and Methods. We chose a pentagonal bipyramid with an intermediate aspect ratio of 2.6 and simulated its assembly using three different DNA lengths (r₀ : 0.5,0.7,0.9σ, where the circumcircle radius of pentagon of the bipyramids is 1σ). The system was prepared initially with randomly dispersed and randomly oriented PAEs and then equilibrated at a temperature below the hybridization temperature (T/T_m~0.9). In the simulation, the bipyramids assembled into superlattices with kagome (fig. S5) and simple rhombohedral ordering (fig. S6) for long (r₀ = 0.9σ) and short DNA strands (r₀ = 0.7σ and r₀ = 0.5σ), respectively. To confirm the thermodynamically preferred phases at given conditions, we simulated mixed phases of the two superlattices constructed using the same DNA length for each superlattice in a bath of dispersed PAEs. For r₀ = 0.9σ, the kagome superlattice grew at the expense of the simpler rhombohedral superlattice; for r₀ = 0.5σ and r₀ = 0.7σ, the reverse occurred (Materials and Methods and fig. S7, A to D). For both the long- and short-DNA cases, the thermodynamically stable crystal grew at the dissolution of the unstable phase. On the basis of the 3D models of kagome lattice (fig. S7, E and F), we calculated the volume fraction of the bipyramids and DNA shells to be 69% in the kagome lattice. The voids and pore topology are shown in fig. S7 (G and H), in which all the voids are filled, and the volume occupied by bipyramids and DNA shells are removed. The simulation results provide insight on the formation mechanism of the lattices and are consistent with the experimental phases (fig. S2A). When long DNA strands are used as linkers during assembly, the kagome lattices form as the thermodynamically stable structures. However, the rhombohedral phase dominates when short DNA strands are used as linkers. The simulations reveal that the long DNA strands provide more space between particles than the shorter ones, allowing the bipyramids to locally reorganize to form kagome lattices featuring different bipyramid orientations, twisting, and lattice distortions. Meanwhile, short DNA strands provide compact bipyramid packing, leading to rhombohedral lattices with unidirectional bipyramid alignment.

Fig. 2. Molecular dynamic simulation of kagome lattice formation.

(A) Schematic illustration of DNA-functionalized Au bipyramids. (B) MD simulation snapshot showing the self-assembled superlattices from randomly dispersed bipyramids. (C and D) FFT patterns of the assembled crystal calculated from different orientations: along (C) and perpendicular (D) to the kagome lattice stacking direction. The insets are snapshots of the crystal (bottom left) and bond-orientational order diagram (bottom right), respectively. (E) Three kagome lattice layers stacked in the assembled crystals. Zoomed-in images are shown on the bottom. Each sphere represents the center of a bipyramid, and the bonds connecting the nearest neighbors show the tiling with equilateral triangles and shields. (F) Local bipyramid motifs in two adjacent kagome lattices. The bipyramid twisting (black arrows), facet alignment, and lattice distortion are shown in the top, middle, and bottom panels, respectively. (G) Schematic illustration and cross-sectional TEM image showing the relative lattice and lattice distortion in two adjacent kagome lattices. (H) Schematic illustration of the bipyramids twisting through partial facet alignment. (I) Schematic illustration and cross-sectional TEM image showing the alignment of one of pentagonal domain boundaries to the lattice orientation within the kagome lattices. (J) Distribution of the angles between one of domain boundaries and one of lattice orientation in the distorted kagome lattices.

Distorted kagome superlattices and bipyramid twisting from partial facet alignment

We calculated fast Fourier transform (FFT) patterns based on the centroids of the pentagonal bipyramids in the simulated kagome lattice for structure analysis (Fig. 2, C and D). The FFT pattern of the simulated crystals shows a sixfold rotational symmetry, which is consistent with the (triangular) hexagonal arrangement of trimer clusters (Fig. 2C and fig. S8). In the perpendicular direction, the FFT pattern shows a periodic stacking of 2D kagome lattices (Fig. 2D). Additional peaks in the in-plane FFT indicate the kagome lattices are distorted. In contrast to regular kagome lattices described by a tiling of regular triangles and hexagons, our distorted kagome lattice is tiled with regular triangles and hexagonal shields that are equilateral but not equiangular ($\frac{\mathrm{\pi }}{2}$ and $\frac{5\mathrm{\pi }}{6}$, radians; Fig. 2E and figs. S9 and S13). The polar axis of each bipyramid within a trimer points in one of three directions out of the kagome plane (Fig. 2F). Specifically, the polar axes cycle and twist unidirectionally within every trimer, leading to the same planar chirality in each layer (Fig. 2F) (49). The three bipyramids within a trimer twist in opposite directions in adjacent kagome lattices, forming right-handed and left-handed layers. The same triangle-shield tiling is also present in the experimental kagome lattices. We prepared 50-nm-thick slices of the superlattices by ultramicrotomy and characterized the bipyramid positional order under TEM. Within one kagome lattice, the arrangement of bipyramids can be described by a triangle-shield tiling (Fig. 2G), which experimentally ascertains the lattice distortion. An alternative trimer orientation is also observed in the TEM image in agreement with the simulation results (fig. S10).

To explain the trimer twisting and kagome lattice distortion, we further analyzed the local structural motifs. In principle, the construction of a 2D kagome lattice could be achieved by repeating the trimer motif in a plane with hexagonal symmetry (fig. S11A). The bipyramid geometry, though, precludes the formation of trimers with perfect facet alignment among all three bipyramids. The geometry does allow for perfect edge-to-edge alignment for the top or bottom halves of the bipyramids, but such alignment does not allow maximum DNA hybridization and is therefore not preferred by the CCM. However, because the DNA is long enough, both the particle geometry and the CCM permit three bipyramids to form a trimer by bipyramid twisting (Fig. 2H and figs. S11B, S12, and S13), resulting in lattice distortion. To experimentally verify the lattice distortion observed in the simulations, we consider how the crystalline domains within individual bipyramids align relative to the lattice orientation. If the lattices are distorted, we expect to see parallel alignment of any of the five domain boundaries with any of the three lattice directions within a kagome layer (Fig. 2I and fig. S14A). To this end, we carefully examined the cross-sectional TEM images of the kagome lattices and analyzed the angles (ϴ) between bipyramid domain boundaries and kagome lattice directions (fig. S14B). The polar plot in Fig. 2J shows that 90% of the bipyramid domain boundaries are within ±6° relative to the kagome lattice direction. Such a near-normal distribution around 0° suggests parallel alignment of one bipyramid domain boundary to one lattice direction, which experimentally verifies the distorted lattices (fig. S14C).

To investigate 3D superlattice formation, we analyzed 2D sectional TEM images and mapped the relative position of pairs of adjacent kagome layers, where lattice distortion was ignored for simplification. As shown in Fig. 3A, the top and bottom kagome lattices are highlighted in blue and red lattices, respectively. The position offset along the x and y axes between the adjacent lattices is 1/2 and $\sqrt{3}$/6 for one unit periodicity, respectively (Fig. 3B and fig. S15). Figure 3C and fig. S16 show that the first and seventh kagome layers have the same bipyramid positions and orientations, demonstrating six kagome layers in each repeat unit along the z-direction. Such stacking leads to a rhombohedral super unit cell, which has one six-bipyramid cluster occupying each lattice point and contains a total of 48 bipyramids in three different orientations (Fig. 3D). The six-bipyramid cluster is observed in sectional TEM images with side and top view shown in the top and bottom panels in Fig. 3E. Three distinct bonding states are recognized in the electron microscopy images: top parallel bonding, bottom parallel bonding (Fig. 3F), and staggered bonding (Fig. 3G), which are represented by yellow, green, and red sticks, respectively, in a ball-stick model. In this model, we delineate the positional order and the bonding states of the bipyramids within one rhombohedral unit cell (Fig. 3I and fig. S17). We find that there is parallel bonding between bipyramids within one kagome layer and staggered bonding between bipyramids in two adjacent layers (fig. S18).

Fig. 3. Six bipyramids form a cluster unit in the rhombohedral unit cells.

(A) Cross-sectional TEM image of a two-layer kagome superlattice and relative lattice position offset. (B) Stacking of four kagome lattices into 3D superlattices. (C) Snapshot of the simulated 3D superlattice with multilayer stacking. The first and the seventh layers are identical, demonstrating a six-layer repeat pattern. (D) Schematic illustration of the rhombohedral unit cell and the registration of a six-bipyramid cluster at each lattice point. (E) High-angle annular dark-field (HAADF)–STEM images of the side (top image) and top (bottom image) view of the six-bipyramid cluster. (F) SEM (left), HAADF-STEM (middle), and bright field–STEM (right) images of the bipyramids in parallel bonding. (G) SEM (left), HAADF-STEM (middle), and bright field–STEM (right) images of the bipyramids in staggered bonding. (H) Ball-and-stick model of the bonding states of bipyramids in a rhombohedral unit cell.

Anisotropic light-emitting kagome lattices

Because of the bipyramid anisotropy and low lattice symmetry, the kagome lattices have complex plasmon resonance modes. To analyze the plasmon resonances, reflection spectra were experimentally measured and simulated using a finite-difference time-domain simulation tool (17) (fig. S19). Reflection is dependent on light incident directions due to reduced lattice symmetry (Fig. 4A and fig. S20). The crystals for measuring reflection spectra under x- and z-axis incident light are shown in fig. S21 (A and B, respectively). A large cluster made of bipyramids of the same size as those assembling into colloidal crystals demonstrated much lower reflectance in the visible spectrum due to the random scattering of disordered bipyramids (fig. S21, C and D). To resolve the plasmon modes, we plotted the electric fields inside the kagome lattices at different resonance wavelengths. Depending on the light polarization direction relative to kagome lattice orientation, longitudinal (fig. S22A) and transverse modes (fig. S22, B and C) are identified, in which the electric fields are strongly enhanced around the bipyramid tips. For light polarization along the z axis, we observed a longitudinal excitation mode at 652 nm and a weak coupling mode at 1519 nm (fig. S22A). For x-polarized light, we observed lattice plasmon resonances at distinct excitation wavelengths (Fig. 4B). At 512 nm, transverse dipole resonance modes appear for the kagome lattices, producing well-aligned dipoles in each bipyramid tip parallel to light polarization. As the wavelength increases to 764 nm, field enhancement is strongly localized around each bipyramid tip, which was also observed inside the 3D superlattice (fig. S22, E and F).

Fig. 4. The kagome lattice plasmon resonance and facet-dependent light emitting.

(A) Measured and simulated reflection spectra of one single crystal under different light incident directions. (B) Simulation results of the electric field mapping in xy plane under light excitation at 512 and 764 nm. (C) The measured, facet-dependent fluorescence spectra on different facets of the lattices. (D) Optical microscopy (left), fluorescence (middle), and merged image (right) of one single crystal in experiments. (E) Bright-field (left), fluorescence (middle), and merged confocal image (right) of one single crystal. (F) Simulated effective refractive index of the kagome lattices along different directions. (G) Simulated Purcell factor of a dipole located inside the crystal and directed along x (blue line), y (red line), and z (black line) axes. (H) Simulated radiation pattern over xz plane of a random dipole cloud located inside the crystal (blue line) and in the free space (red line). (I) Simulated average radiative and (J) nonradiative decay rates of dipoles located at the top (blue line) and side faces (red line) of the crystal.

We further investigated the light-emitting properties of these kagome lattices by conjugating Cy-5 dyes to the DNA linker strands. The facet-resolved fluorescence spectra revealed a sixfold emission enhancement of the top facets compared to the side facets (Fig. 4C), leading to much brighter red-light emission on the top facets (Fig. 4, D and E). However, the cluster made of randomly oriented and positioned bipyramids exhibited similar light emission along different directions (fig. 23). These two experiments reveal that the facet-dependent, anisotropic light emission is highly correlated with the reduced symmetry and anisotropic lattice plasmon resonance of the kagome lattices. To obtain a deeper understanding of the driving factors, we simulated the effective refractive index of the kagome lattices and emission properties of the dye molecules inside and on the surface of the crystals. The kagome lattices are birefringent because of the presence of two different refractive indices, depending on orientation (along z and along the other two directions; Fig. 4F). For the dye molecules inside the crystal, the kagome lattices act as a cavity, which determines the density and distribution of optical states and results in enhancement or diminution of emission (Purcell effect) and contributes to the shapes that define the radiation pattern. As shown in the simulations (Fig. 4G and fig. S24A), the Purcell factor, which is the emission enhancement factor of the cavity, is larger for in-plane dipoles (along x and y axes) than out-of-plane (along z axis) ones, leading to stronger light emission along the z direction (perpendicular to the dipole axis). The effects of the kagome lattice on the radiation pattern of the dye molecules inside the crystal were also simulated and compared with the emission in free space, where the dye molecules are modeled by a cloud of randomly distributed out-of-phase dipoles. As seen from the simulated radiation patterns over the fundamental planes (Fig. 4H and fig. S24, B to D), the kagome lattices emit light toward the top-down direction, which is responsible for the facet-dependent fluorescence observed in experiments. For the dye molecules on the surfaces, the plasmonic effects on the crystal surface provide competing enhancement and loss channels. Individual contribution of these competing factors is elucidated through simulated radiative and nonradiative decay rates (Fig. 4, I and J). The emitters on the top and side faces exhibit similar radiative and distinct nonradiative decay rates. The nonradiative decay rate (Fig. 4J) at excitation and emission wavelengths is larger for the emitters on the top face, indicating that the enhancement expected from strong field localization around the bipyramid tips (fig. S22) is redeemed by the nonradiative (loss) channels. Together, the data show that the faceted, anisotropic light emission is driven by the crystal structures serving as anisotropic cavities, as opposed to plasmonic field enhancement on the surfaces. The low-symmetry kagome lattices serve as cavities with much higher emission enhancement factors (Purcell factor) along the x- and y-directions, which is due to a higher density of optical states along the two directions and leads to stronger radiation along the z axis and facet-dependent, anisotropic light emission.

DISCUSSION

In summary, this work shows how the design of low-symmetry colloidal crystals and kagome superlattices provides the ability to engineer quantum-lattice systems with complex lattice plasmon resonances and facet-dependent light emission properties. These capabilities and observations make such structures attractive previously unidentified materials for lasers, displays, and optical sensors. The recognition that PAE partial facet alignment can expand what is structurally possible through colloidal crystal engineering with DNA is important and must be considered as a previously unknown design principle that complements the perfect facet alignment typically expected from the CCM (1, 9, 26). Last, by leveraging the tunable parameters of DNA length and sequence along with non–space-filling PAEs, we have paved the way to exotic colloidal crystal architectures not attainable with conventional build blocks and with emerging and previously unidentified optical properties.

MATERIALS AND METHODS

Synthesis of Au bipyramids

Penta-twinned Au nanoparticles were prepared, which was used as small seeds to initiate the growth of pentagonal bipyramids. A 10-ml solution of HAuCl₄ (0.25 mM), cetyltrimethylammonium chloride (CTAC, 50 mM), and sodium citrate (5 mM) was prepared and 0.25 ml of NaHB₄ (25 mM) was freshly prepared and added into the HAuCl₄ solution. The solution color turned from yellow to brownish, indicating the formation of Au seed particles. After being stirred at room temperature for 30 min, the seed solution was aged in an oil bath at 80°C for 90 min. The solution color changed from brown to red. The seed solution was stored at room temperature. The growth of Au bipyramids of different sizes was realized by controlling the seed solution volume that was added to a growth solution. In a water bath at a temperature of 30°C, the growth solution was prepared by mixing the following solutions under magnetic stir: 100 ml of CTAB (100 mM), 5 ml of HAuCl₄ (10 mM), 1 ml of AgNO₃ (10 mM), 2 ml of HCl (2 M), and 0.8 ml of AA (0.1 M) in sequence. Au seed solution was then added into the growth solution. After 2 hours in the water bath, the bipyramids were washed to remove excess chemicals and dispersed in 10 ml of water. To remove impurities, 5-ml of bipyramid solution was mixed with 2-ml CTAC solution (0.1 M) and 1-ml NaCl solution (2 M), which was thoroughly mixed using Vortex and left without agitation at room temperature overnight. The large bipyramids precipitated driven by depletion force and small impurity nanospheres remained in supernatant that was carefully removed using pipettes. This process was repeated twice.

DNA-programmable assembly

The bipyramid nanoparticles were modified with thiolated oligonucleotides (DNA anchor strands). In a typical procedure, 5 nmol of DNA was used for each 1-ml original nanoparticle solution, and 10-ml bipyramid solutions were concentrated in 1-ml for DNA functionalization. The DNA anchor strands were reduced by dithiothreitol (0.1 M in 170 mM phosphate buffer) for 1 hour at room temperature and purified using a nucleic acid purification desalting column. Meanwhile, the bipyramids were purified by washing against water twice to remove excess CTAB. After the second wash, the supernatant was carefully removed and reduced DNA anchor strands were added to the bipyramids. Afterwards, SDS (0.1%) and (0.1 M) phosphate-buffered saline (PBS) was added to a final concentration of 0.01% and 0.01 M, respectively. The mixture was sonicated to fully disperse the bipyramids and shaken overnight (1000 rpm). Then, NaCl salt solutions (2 M) were added to promote the ligand replacement on bipyramid surfaces. The salt concentration gradually increased to 0.05, 0.1, 0.2, 0.3, 0.4, and 0.5 M within a 30-min time interval. For 1-ml bipyramid solution, the volume of 2 M NaCl solutions was 25.6, 27.0, 58.5, 65.4, 73.5, and 83.3 ml added each 30 min. After the mixture was shaken for 24 hours, the mixture was centrifugated, and excess DNA strands were removed by washing in 0.01% SDS twice. The bipyramids were dispersed in 100-μl 0.5 M NaCl solution with 0.01% SDS, 0.01 M PBS, with final Au concentration of about 50 mM. DNA linker strands (10 nmol) were added into the bipyramid solution, which was slowly cooled in a thermal cycler (Life Technologies). The cooling rate for temperature between 70° and 66°C, 66° and 45°C, and 45° and 25°C was 0.1°C/10 min, 0.1°C/20 min, and 0.1°C/10 min, respectively.

Fixing crystals using silica and polymeric resin

After assembly, the crystals were transferred into 1.5-ml centrifugal tubes and 0.5 M NaCl solution was added to a final volume of 1 ml. Ten-microliter trimethylammonium chloride 50% in methanol was added, and after 30 min, 5-μl triethoxysilane was added. The mixtures were shaken at 900 rpm overnight. Free silica was removed by washing the crystals with water three times. The solid crystals were drop-casted on silicon wafers or TEM grids for electron microscopy characterization. To prepare thin sections, the crystals were further embedded in a polymer resin, embed 812 from Electron Microscopy Sciences and sectioned into 100-nm thin films on TEM grids for electron microscopy characterization.

Characterization

SEM images of the colloidal crystals were acquired on a JEOL JSM-7900FLV scanning electron microscope under an accelerating voltage of 10 kV and a backscattered electron detection mode. To acquire TEM images of bipyramids and sectioned colloidal crystals, a JEOL ARM200CF transmission electron microscope was used under an accelerating voltage of 200 kV. Confocal images were taken using a Leica SP8 Confocal optical microscope. The optical reflectance spectra of single colloidal crystal were measured on Nikon TI Inverted Microscope coupled to ANDOR Acton Spectrometer with Newton EMCCD Camera.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S24

Tables S1 and S2

References