Three-dimensional nanoparticle arrays are likely to be the foundation of future optical and electronic materials. A promising way to assemble them is through the transient pairings of complementary DNA strands.
One of the staple concepts of nanotechnology is that of 'growing' useful materials or devices by coaxing a random mixture of microscopic parts to assemble spontaneously into a desired structure. Versatile self-assembly schemes have been demonstrated that use DNA as the primary building material1. In this issue, two research teams, one led by Oleg Gang (Nykypanchuk et al., page 549)2 and the other by Chad Mirkin (Park et al., page 553)3, recount how they have built on the successes with DNA to aid the self-assembly of gold nanoparticles. Their technique should also work for other varieties of technologically exciting nanoparticles.
Progress in achieving the directed self-assembly of nanoparticles had been elusive, owing to one potentially daunting requirement: selective adhesion. Each microscopic part must be engineered so that it sticks only to the others it should abut in the desired final structure. In earlier experiments4, nanoparticles were found to form ordered arrangements when a surrounding solvent was evaporated. In this case, however, the final structures depended sensitively on the particle chemistry and charge.
This is where DNA comes into its own. Particles carrying complementary strands of DNA selectively adhere to each other when the strands 'hybridize' to form the familiar DNA double helix. The final architecture is thus determined not by chemistry or charge, but by the lengths and nucleotide sequences of the DNA strands. That promises a versatile assembly scheme that might be used with particles of nearly any material to fabricate nanocomposites or 'metamaterials'5 with unusual electronic and optical properties. The applications of such materials might include high-efficiency solar panels and lasers, super-resolution microscopes — and even coatings to render objects invisible.
Nykypanchuk et al.2 and Park et al.3 both start by grafting DNA to gold spheres of the order of 10 nanometres in diameter to give two populations of DNA-capped particles, A and B. Each sphere bears several dozen strands, and the ends of the strands on A-type and B-type particles are complementary. This configuration means that spheres of one type will selectively adhere to spheres of the other, but neither type of sphere will adhere to its own kind.
The authors mixed the A and B spheres in water. Under the right conditions, they found that the nanospheres were rapidly guided, as the DNA strands hybridized, to arrange themselves into well-ordered arrays. The resulting crystal had body-centred-cubic crystal symmetry, with A and B spheres taking up alternating locations in the lattice, so that each A sphere was surrounded by eight B spheres and vice versa (Fig. 1). Such a structure — known as a CsCl lattice after crystals of caesium chloride, which take the exact same form — provides the maximum possible number of A–B adhesion contacts.
Both Nykypanchuk et al.2 and Park et al.3 report that crystallization requires the DNA-binding regions to be connected to the gold spheres by flexible spacers, also made of DNA, that are roughly as long as the sphere diameter. Moreover, crystallization happens only at higher temperatures, at which the DNA binding strands are dynamic, continuously forming double helices and dissociating back into single strands.
The DNA in these experiments is being used in a fundamentally different way from its use in earlier DNA self-assembly techniques such as Ned Seeman's 'DNA tile' approach1. There, each constituent tile of the structure was made of interconnected DNA double strands. Each tile had one binding strand dangling from each corner, so that it could mate with neighbouring tiles. The structure of each tile was thus controlled at the molecular scale. The chemical process for attaching DNA strands to nanoparticles2,3, by contrast, is essentially random, and scatters DNA strands over the gold sphere's surface, rather than at just eight nearest-neighbour locations. The exact number of strands varies from sphere to sphere.
Remarkably, ordered arrangements of the nanoparticles can form despite these random variations in their individual structure. The long spacers and the dynamic binding process seem to be crucial. Flexible spacers fluctuating in space ensure that an extended spherical cloud of strands surrounds each core, washing out the random pattern in which the strands are anchored to the particles. When the clouds of complementary neighbouring particles overlap, hybridization forms transient DNA bridges that briefly pull pairs of spheres towards each other.
In essence, the nanoparticles reach out to each other using their long spacer arms, and temporarily 'shake hands' with their complementary DNA strands. The net attractive interaction is proportional to the time-averaged number of bridges between a pair of spheres. This in turn is determined by the degree of overlap between the two DNA clouds6.
Unlike the strictly determined binding that drives other DNA-based assembly techniques, suspensions of nanoparticles with such handshaking interactions mimic7 the phase behaviour of atomic materials, but using engineered interactions. The ordered arrangement of A and B spheres, for example, mirrors the alternating positive and negative ions in a salt crystal, which also have a long-range, spherically symmetrical attraction.
The analogy is not perfect: Park et al.3 report that one of their samples forms a face-centred-cubic, rather than a body-centred-cubic CsCl structure, when incubated at higher temperatures. They argue that this behaviour stems from a competition between the contributions to the system's total free energy of sphere entropy (which favours the more densely packed face-centred-cubic structure) and A–B binding energy (maximized by the CsCl structure).
Realizing the potential of these new materials will certainly require more research to stabilize their structure. The long DNA spacers imply that the resulting nanoparticle array is roughly 90% water, and is probably quite fragile. Still, existing techniques can probably be adapted to fill the gaps with gels or solid ceramic to yield a robust, solid material. Better models of the handshaking interaction will also need to be developed, validated and applied to computing what periodic structure a given DNA sequence will produce.
Even more exciting would be the possibility of attaching DNA to non-spherical nanoparticles — perhaps preferentially to different crystal facets — to create directional bonding and more complex structures. The ultimate dream is the creation of a DNA tool-kit that will make possible the self-assembly of nearly any material reliably at the nanoscale.
Winfree, E., Liu, F., Wenzler, L. A. & Seeman, N. C. Nature 394, 539–544 (1998).
Nykypanchuk, D., Maye, M. M., van der Lelie, D. & Gang, O. Nature 451, 549–552 (2008).
Park, S. Y. et al. Nature 451, 553–556 (2008).
Shevchenko, E. V., Talapin, D. V., Kotov, N. A., O'Brien, S. & Murray, C. B. Nature 439, 55–59 (2006).
Linden, S. et al. Science 306, 1351–1353 (2004).
Biancaniello, P. L., Kim, A. J. & Crocker, J. C. Phys. Rev. Lett. 94, 058302 (2005).
Tkachenko, A. V. Phys. Rev. Lett. 89, 148303 (2002).
About this article
Physical Review E (2018)
Two-dimensional structures formed in a binary system of DNA nanoparticles with a short-range interaction potential
Japanese Journal of Applied Physics (2018)
Journal of the Physical Society of Japan (2017)