Clathrate colloidal crystals

See allHide authors and affiliations

Science  03 Mar 2017:
Vol. 355, Issue 6328, pp. 931-935
DOI: 10.1126/science.aal3919

Turning colloidal gold into clathrates

Clathrates contain extended pore structures that can trap other molecules. Lin et al. created colloidal analogs of clathrates in which bipyramidal gold nanoparticles functionalized with DNA molecules assembled into polyhedral clusters to create open-pore structures (see the Perspective by Samanta and Klajn). These clathrate colloidal crystals exhibit extraordinary structural complexity and substantially broaden both the scope and the possibilities provided by DNA-inspired methodologies.

Science, this issue p. 931; see also p. 912


DNA-programmable assembly has been used to deliberately synthesize hundreds of different colloidal crystals spanning dozens of symmetries, but the complexity of the achieved structures has so far been limited to small unit cells. We assembled DNA-modified triangular bipyramids (~250-nanometer long edge, 177-nanometer short edge) into clathrate architectures. Electron microscopy images revealed that at least three different structures form as large single-domain architectures or as multidomain materials. Ordered assemblies, isostructural to clathrates, were identified with the help of molecular simulations and geometric analysis. These structures are the most sophisticated architectures made via programmable assembly, and their formation can be understood based on the shape of the nanoparticle building blocks and mode of DNA functionalization.

DNA-programmable assembly has emerged as a powerful method for generating colloidal crystals with exquisite control over crystal symmetry, lattice parameter, and in certain cases, macroscopic habit (14). Indeed, more than 500 different crystals spanning 36 different symmetries have been made via this approach and the complementary contact model and associated design rules that govern DNA-programmable colloidal crystal formation (59). The vast majority of such structures studied to date consist of spherical nanoparticle building blocks, isotropically functionalized with DNA, termed programmable atom equivalents (PAEs) (10). With the advent of methods for preparing anisotropic PAEs, DNA-modified structures can be made in which the shape of the nanoparticle core directs the DNA bonding elements in an anisotropic manner (8, 1114). This is making it possible to synthesize colloidal examples of many solid-state structures that naturally occur, as well as several others that have no known mineral equivalent (8, 11, 15, 16). Clathrates—structures consisting of polyhedral cages with large pores that can be used for host-guest chemistry—represent a challenging target for colloidal assembly (1720). Because of the symmetry and complexity of clathrates, it is not clear how such architectures can be made from spherical building blocks, but they may be attainable with anisotropic ones. We explored, via experiment and simulation, how the symmetry of nanoscale gold bipyramids can be used in conjunction with programmable assembly to generate several clathrate architectures.

Clathrates occur as hydrates of a variety of host molecules such as methane and as open fourfold coordinated crystals in the carbon group (2124). The cages are formed from molecular or atomic nodes that adopt discrete bond angles between 100° and 125°. With the development of synthetic methods for forming monodisperse samples of anisotropic particles (25, 26), shapes can be explored that impose directionality on oligonucleotide bonding motifs (8, 11). In this regard, oblate trigonal bipyramids (TBPs) with {110} facets, which direct bonding elements with ~110° angles between them, are ideal structures with which to explore nanoparticle-based colloidal clathrate formation (Fig. 1, A and B, and fig. S1). To prepare building blocks suitable for assembly, we synthesized TBPs (~250-nm long edge, 177-nm short edge) via reported procedures (27) and functionalized them with 28-base hexylthiol-modified DNA (Fig. 1C, red; sequences of all DNA strands are provided in table S1) (28). After purification by means of centrifugation and washing with phosphate-buffered saline, these particles were hybridized with a variety of different linker strands (23 to 228 bases) (Fig. 1C, blue) that vary in length but are all terminated with a self-complementary GCGC sticky end. Before their binding to the particles, the long linkers were introduced to partially complementary strands (Fig. 1C, green) to form a primarily duplex region, except for the four-base self-complementary termini, 10 A bases near the particle that were maintained as single-stranded DNA for flexibility purposes, and additional individual bases (black) at designated locations that served as flexors facilitating assembly (5).

Fig. 1 Self-assembly into clathrate colloidal crystals.

(A) Geometry and scanning electron microscopy (SEM) image of gold TBPs with {110} facets and 109.5° large-edge angle. (B) When placing TBPs at the middle of network edges and rotating them appropriately, all triangular TBP facets align nearly parallel. In the resulting tetrahedral network, polar tips join at network nodes, and equatorial tips join in the center of network cages. (C) Illustration of the DNA linker design. The linker binds to the surface of Au bipyramid through a 28-base hexylthiol-anchor strand (red), which can recognize a linker strand (blue). Duplexer strands (green) hybridize the linker strand to form double-strand segments, except for specific single-base sites (black) and the four-base self-complementary sticky end. DNA length can be tuned by varying the number of duplexed block segments. (D) The TBP tetramer with its DNA shell has rounded edges and vertices. (E) Whole-view SEM image of the self-assembled TBP superlattice. (F and G) TEM images of the superlattice zoomed in on single-domain and multidomain regions, respectively, from a sectioned sample.

We hypothesized that use of long DNA strands that have enough length and flexibility would be important to reduce the strain in the material imposed upon assembly. If we tried to physically assemble TBPs of this size by sticking faces together, structures with relatively large gaps of up to 20° will form (figs. S8 to S10), creating the need for long, flexible DNA bonds, which simultaneously also round the outer DNA shell (Fig. 1D). When the particles were allowed to assemble by maintaining the temperature of the solution below the DNA melting temperature, long DNA linkers enabled the formation of high-quality crystalline structures (Fig. 1E and fig. S2). Closer inspection of transmission electron microscopy (TEM) images suggested that the assemblies were clathrates (Fig. 1F). In addition to large-area single-domain crystals, we observed multidomain architectures that contained a mixture of at least two different but related materials (Fig. 1G and fig. S14).

To investigate the effect of the DNA linkers, we tuned their length by changing the number of 40-base duplexed block segments (table S2). For the shortest DNA linker without block segments, TBPs only formed tetrahedral units, and higher-level structures were not found (Fig. 2A). Some local order was observed when we used DNA linkers with one block segment, but substantial distortion was still apparent (Fig. 2B). Multishell order resembling clusters but with defects and cracks occurred when using DNA with two- and three-block segments (Fig. 2, C and D). Clearly identifiable clathrates appeared when using DNA with four-block segments, but the domain size was usually small (Fig. 2E). The highest-quality crystals formed when DNA with five-block segments was used, and single-crystalline clathrate domains as large as tens of micrometers were found in such structures (Fig. 2F). As expected for clathrates, cavities were distributed throughout the sample (Fig. 2, G and H), which were generally guest-free in all of the samples.

Fig. 2 Effect of DNA length on TBP assembly and presence of cavities.

(A to F) SEM images of samples by using TBP particles with DNA bonding elements that contain (A) zero-, (B) one-, (C) two-, (D) three-, (E) four-, and (F) five-block segments paired with duplexer strands. An improvement of the assembly quality with the increase of the block segment number is apparent. (G) Cavities are observed in the middle of the clusters representing clathrate cages. Shown in this TEM image is a thin sectioned sample that contains the middle portion of clusters showing the empty spaces (center of the image) and off-center sectioned clusters, including the upper part of the clusters (top left) (fig. S13H). The particles in this sample have DNA ligands with five block segments. (H) Illustration of a single cluster (identified as cluster C) (Fig. 3D) before and after removing the top and bottom TBPs.

To resolve the complex structure of the clathrates, we first reproduced the assemblies thermodynamically using a discrete element method with implicit solvent (29). We constructed a minimal model in which a single PAE is represented as a rigid TBP core surrounded by a DNA ligand shell (Fig. 3A and fig. S3). The interaction between PAEs is described by an effective pair potential that depends only on the separation distance measured from patches on the TBP core (fig. S4 and table S3). In this simplified representation, the increased rounding of the PAE observed experimentally with increased DNA length is automatically included. We chose the shell thickness, determined by DNA length, to match that used in the experiments.

Fig. 3 Model and simulation of DNA-tethered nanocrystals.

(A) The nanocrystal core (TBP) is surrounded by a wide shell of double-stranded DNA (ds-DNA) that terminates in a narrow shell of single-stranded DNA (ss-DNA). The interaction of PAEs is captured by an effective pair potential consisting of a Weeks-Chandler-Anderson (WCA) repulsion upon shell overlap plus a double-Gaussian model (DGM) attraction representing DNA hybridization of the ss-DNA. (B) Nanocrystals with DNA ligands containing five-block segments (68.7 nm length) cluster together in a simulation snapshot and spontaneously order. Without the DNA shell, local motifs of clathrates II and IV structure are identified. (C) Simulations by using DNA ligands containing eight-block segments (103.2 nm length) show exclusively clathrate II. (D) Clathrates are built by four types of clusters. Particles in sixfold rings are colored in red. (E) Relation of geometric frameworks derived from PAE cluster A. The tips of the cluster form a great dodecahedron. The cluster can be mapped onto the (pentagonal) dodecahedron by connecting polar tips. It can also be mapped onto the icosahedron, which is the dual of the dodecahedron and the Frank-Kasper polyhedron with coordination number 12, by connecting equatorial tips. Connecting PAE centers defines an icosidodecahedron. The same principles can be applied to other PAE cluster types (fig. S5).

At high temperature, the model PAEs remained in a disordered fluid phase, but they aggregated at low temperature because of DNA hybridization and eventually ordered into crystals given sufficient time (fig. S6) (18, 19). A strong tendency for alignment and face-to-face contact to maximize hybridization was observed, which is in agreement with experiment. This tendency brought the two polar tips of the TBPs together as well as the three equatorial tips of the TBPs (Fig. 1B). We identified aspects of crystallographic order that developed spontaneously by visualizing the PAEs (Fig. 3, B and C, fig. S7, and movies S1 and S2). A geometric analysis revealed that the assembled structures contain clusters resembling stellated polyhedra (Fig. 3D). Within a cluster, one of the three equatorial tips of each TBP pointed inward toward the cluster center. The remaining two equatorial tips joined to form the outer (stellated) points of the cluster, and the polar tips formed the inner points of the cluster surface (Fig. 3E).

We found four cluster types, all stellated polyhedra, in the simulation data. Thirty PAEs formed a great dodecahedron (a nonconvex polyhedron, cluster A), and 36, 39, and 42 PAEs arranged into the lower-symmetry configurations clusters B, C, and D, respectively (Fig. 3D). By relying only on the inner points corresponding to polar tips on the cluster surface, clusters A, B, C, and D were mapped onto a pentagonal dodecahedron (cage A, 512), tetrakaidecahedron (cage B, 51262), pentakaidecahedron (cage C, 51263), and hexakaidecahedron (cage D, 51264), respectively (Fig. 3E and fig. S5). These polyhedra are well known as clathrate cages.

The four cluster types can be arranged into three crystals, clathrates I, II, and IV, following literature classification (Fig. 4 and figs. S8 to S10) (30). We briefly discuss each of the three clathrate types by comparing TEM images and our structure model. Among hundreds of images, we selected for discussion those that are aligned with their main crystallographic axes. Clathrate I (also known as CS-I or sI) is a cubic phase that is distinctly characterized by square tiles when projected along a fourfold axis, with hexagonal rings at their vertices (Fig. 4, A to C, and fig. S8). Clathrate IV (HS-IQ) is a hexagonal phase that corresponds to triangle tiles when projected along a sixfold axis, with hexagonal rings at their vertices (Fig. 4, D to F, and fig. S9). Last, clathrate II (CS-II or sII) is another cubic phase that has a fundamental rhombohedral building block. When viewed along a twofold axis, pentagonal rings arrange into rhombs (Fig. 4, G to I, and fig. S10). Additional evidence is provided in the supplementary materials (28), where we characterized images not aligned with main crystallographic axes (figs. S11 to S14). Using this classification, we identified the order observed in our simulations (Fig. 3B) as a mix of clathrates II and IV. For thicker DNA shells, exclusively clathrate II was observed (Fig. 3C).

Fig. 4 Identification of the three basic clathrate crystal structures.

In the experimental data, we observe crystals analogous to (A to C) clathrate I oriented along [100], (D to F) clathrate IV oriented along [0001], and (G to I) clathrate II oriented along [110]. Each row shows the construction of a unit cell. [(A), (D), and (G)] Nonrounded TBPs highlight the local geometry. [(B), (E), and (H)] Connecting TBP polar tips reveals the clathrate cage representation. [(C), (F), and (I)] Comparison of electron microscopy images (left), zoom-ins of the red areas (middle), and TBP cores in the structure model (right). Pentagonal rings and hexagonal rings are indicated as white overlays. Characteristic structural features seen in projection along high-symmetry axis are outlined as orange overlays.

We hypothesize that the relative stability of the clathrate crystals can be estimated from geometric principles and the complementary contact model (5). Clathrate I has the most similar clusters (cluster composition AB3) but also the lowest frequency of clusters with icosahedral symmetry (cluster A). Overall, it requires the least deformation of the DNA shell. Thus, we expect clathrate I to be the dominant phase for thin DNA shells. Clathrate IV has a larger range of cluster sizes (A3B2C2) and provides TBPs more wiggle room, which is an advantage for DNA shells of intermediate thickness. Clathrate II has the largest range of cluster sizes (A2D), including the most open cluster (cluster D). It requires the strongest deformation of the DNA shell. Our simulation results are consistent with the hypothesis that clathrate II is the dominant phase for thick, flexible DNA shells (table S4).

Depending on the application, the clathrate colloidal crystals we observed may be described equally well by using the language of different material geometries (Fig. 3E). For example, the equatorial tips define a tetrahedral [Frank-Kasper (31)] network, and the TBP centroids are located in a fashion analogous to the oxygen positions in the recently discovered chibaite mineral network (32). The cavities in the center of the clusters might have applications for host-guest recognition applications (such as proteins or virus) at the mesoscale. In addition, they may allow access to multiple properties either synchronously or asynchronously, yielding materials behavior not yet achievable in simpler colloidal crystals. Last, other complex crystals predicted by simulation (15) but not yet realized should now be possible through the use of the complementary contact model and judicious combination of particle shape and DNA linker.

Supplementary Materials

Materials and Methods

Figs. S1 to S14

Tables S1 to S4

References (3340)

Movies S1 and S2

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: This work was supported as part of the Center for Bio-Inspired Energy Science, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under award DE-SC0000989. C.A.M. additionally acknowledges support from the Air Force Office of Scientific Research awards FA9550-11-1-0275 and FA9550-12-1-0280. H.L. is grateful to a scholarship from the China Scholarship Council (CSC) under Grant CSC no. 201306310060. M.E. acknowledges funding by Deutsche Forschungsgemeinschaft through the Cluster of Excellence Engineering of Advanced Materials and support from the Interdisciplinary Center for Functional Particle Systems (FPS) and Central Institute for Scientific Computing (ZISC). M.S. acknowledges support from the University of Michigan Rackham Predoctoral Fellowship Program. S.C.G. was partially supported by a Simons Investigator award from the Simons Foundation. Computational resources and services were supported by Advanced Research Computing at the University of Michigan, Ann Arbor. This work made use of the Electron Probe Instrumentation Center (EPIC) facility [Northwestern University’s Atomic and Nanoscale Characterization Experimental Center (NUANCE)], which has received support from the Materials Research Science and Engineering Center program (NSF DMR-1121262) at the Materials Research Center, the International Institute for Nanotechnology (IIN), and the State of Illinois, through the IIN. All results are reported in the main paper and supplementary materials. H.L. synthesized, functionalized, and assembled the particles; H.L. and L.S. synthesized and purified the DNA; H.L. collected ultraviolet-visible data; H.L. and L.S. collected EM data; S.L., M.S., M.E., and S.C.G. developed the simulation model; S.L. and M.S. implemented the model in the DEM module of HOOMD-Blue; S.L. performed the simulations; S.L., M.E., and S.C.G. analyzed and discussed the simulation data; S.L. and M.E. performed the structural identification analysis on the simulation data and on the TEM images supplied by H.L.; H.L., S.L., M.E., S.C.G., and C.A.M. wrote the paper; M.E., S.C.G., and C.A.M. supervised the research; C.A.M. is the developer of the concept of programmable colloidal crystallization and the concept of controlled valency through anisotropic particle functionalization with nucleic acids that led to the formation of the observed clathrate structures. The authors declare no competing financial interests.

Stay Connected to Science

Navigate This Article