Gold nanocrystals with DNA-directed morphologies

Precise control over the structure of metal nanomaterials is important for developing advanced nanobiotechnology. Assembly methods of nanoparticles into structured blocks have been widely demonstrated recently. However, synthesis of nanocrystals with controlled, three-dimensional structures remains challenging. Here we show a directed crystallization of gold by a single DNA molecular regulator in a sequence-independent manner and its applications in three-dimensional topological controls of crystalline nanostructures. We anchor DNA onto gold nanoseed with various alignments to form gold nanocrystals with defined topologies. Some topologies are asymmetric including pushpin-, star- and biconcave disk-like structures, as well as more complex jellyfish- and flower-like structures. The approach of employing DNA enables the solution-based synthesis of nanocrystals with controlled, three-dimensional structures in a desired direction, and expands the current tools available for designing and synthesizing feature-rich nanomaterials for future translational biotechnology.


Supplementary Figure 4 | In situ investigations of the DNA-directed crystallisation
of Au in a sequence-independent manner. a, Detailed configuration of the nanoplasmonic sensing system and sample data obtained from the system. The profiles of a single nanoparticle are obtained using the integrated system consisting of a 100-W tungsten lamp, a dark-field condenser, a microfluidic device, a high-precision stage controller, an oil-immersion objective, a colour camera, a Rayleigh light scattering spectrograph, a highly sensitive CCD camera, and a data analysis unit. The in situ images were taken by dark-field microscopy (dark background) and CCD camera (blue

Supplementary Note 1 | Comparison of DNA-mediated approaches in nanomaterial synthesis and advantages of DNA-directed nanostructures.
Major differences between the previously reported DNA-templated metallisation and the current DNA-directed crystallisation are in synthesis procedures, mechanisms, and final products. It was previously shown that the synthesis of colloidal NCs in controlled shapes was limited only to symmetric ones with identical surface facets (polyhedral structures;

DNA-templated metallisation
control resolution in a scale of 10 nm) 1,2 . The synthesis of irregularly shaped nanocrystals was empirical rather than scientific, without principles to follow 3 . The controllability is hampered by lacking organic molecules that are able to specifically coat targeted crystal facets and thereby produce nanomaterials with desired structures. A strategy used in a recent study to cast asymmetric shapes using pre-defined molds was quite complex with a low casting yield 4 . We synthesised asymmetric nanocrystals in a desired direction (e.g. a branched form of nanocrystals; named "nanostars") with a relatively high morphological yield. Nanostars generally exhibit stronger local electric field than polyhedral and irregularly shaped nanocrystals because the nano-branches provide natural focusing of S20 electromagnetic field. This is a phenomenon similar to the fact that nanorods typically exhibit higher local electromagnetic field than nanospheres. Compared to nanorods, nanostars offer more branches that can support optical hot spots. For biosensing or diagnostics applications, more hot spots per nanocrystal naturally lead to higher sensitivity. Compared to conventionally synthesised nanostars with uncontrollable branches produced using surfactants [5][6][7][8][9][10][11][12][13] , the DNA-directed nanostars developed in the current study offer a significant advantage to facilely tune the local electromagnetic field by their DNA-directed branches (e.g., UV-vis spectra shown in this study).

AuNCs.
We further conducted optical simulations using the commercial software COMSOL to understand the plasmon behaviour of the star-shaped AuNCs with one branch ( AuNC-1 is modelled as a spherical core with one tapered cylindrical branch ( Supplementary Fig. 14a,b). The extinction, scattering, and absorption cross-section for AuNC-2 exhibits two resonances: one transverse resonance concentrated at the core and one longitudinal resonance with hot spots at the ends of the branches according to the S22 simulations ( Supplementary Fig. 15). The transverse resonance is located at 520 nm and the longitudinal resonance at 810 nm. The simulation results faithfully reproduced the experimental spectra, which also showed two peaks (528.5 nm and 779.5 nm). The transverse resonance peak position agrees almost perfectly. The experimental longitudinal resonance is located at a shorter wavelength than the simulation. We attribute this difference to the slight overestimation of branch length in simulation. Reduction of the branch length by a few nm can make the peak position agree well.
AuNC-3 has less obvious symmetry and thus the simulations had to be done for a number of different orientations to identify all resonances supported by this structure.
This modelling study revealed that AuNC-3 exhibits three resonances in all ( Supplementary Fig. 16). One is the now familiar 520 nm resonance concentrated at the core ( Supplementary Fig. 16c,d). The other two are resonances delocalised across the structure with hot spots at the tips of the three branches ( Supplementary Fig. 16e). One of these two exhibits equal intensities at the tips of the three branches ( Supplementary Fig.   16f) with a peak wavelength of 645 nm. The other has uneven distribution of electric field over the three branches ( Supplementary Fig. 16g,h) and this resonance is located at 762 nm. The simulation results once again agree well with the experimental spectrum, which also show three peaks at 537 nm, 613.5 nm, and 749 nm.
For nanostars with four and five branches, their lack of symmetry requires a large number of simulations for multiple orientations. Since the simulations for the one-, twoand three-branched structures have already established the plasmon hybridizations between the core and the branches in agreement with experiments, we did not conduct further simulations for the four-and five-branched structures.