8/3/2019 Dmytro Nykypanchuk, Mathew M. Maye, Daniel van der Lelie and Oleg Gang- DNA-guided crystallization of colloida

1/4

LETTERS

DNA-guided crystallization of colloidal nanoparticlesDmytro Nykypanchuk1*, Mathew M. Maye1*, Daniel van der Lelie2 & Oleg Gang1

Many nanometre-sized building blocks will readily assemble intomacroscopic structures. If the process is accompanied by effectivecontrol over the interactions between the blocks and all entropiceffects1,2, then the resultant structures will be ordered with a pre-cision hard to achieve with other fabrication methods. But itremains challenging to use self-assembly to design systemscomprised of different types of building blocksto realize novelmagnetic, plasmonic and photonic metamaterials35, for example.A conceptually simple idea for overcoming this problem is theuse of encodable interactions between building blocks; thiscan in principle be straightforwardly implemented usingbiomolecules610. Strategies that use DNA programmability tocontrol the placement of nanoparticles in one and two dimensionshave indeed been demonstrated1113. However, our theoreticalunderstanding of howto extend this approach to three dimensionsis limited14,15, and most experiments have yielded amorphousaggregates1619 and only occasionally crystallites of close-packedmicrometre-sized particles9,10. Here, we report the formation ofthree-dimensional crystalline assemblies of gold nanoparticlesmediated by interactions between complementary DNA moleculesattached to the nanoparticles surface. We find that the nanopar-ticle crystals form reversibly during heating and cooling cycles.Moreover, the body-centred-cubic lattice structure is temper-ature-tuneable and structurally open, with particles occupying

only 4% of the unit cell volume. We expect that our DNA-mediated crystallization approach, and the insight into DNAdesign requirements it has provided, will facilitate both the cre-ation of new classes of ordered multicomponent metamaterialsand the exploration of the phase behaviour of hybrid systems withaddressable interactions.

Theoretical predictions indicate15 that transitions to three-dimen-sional (3D) ordered phases from commonly observed disorderedstates1619 in a DNA-guided particle assembly can occur for particularshapes and ranges of interparticle interaction potentials, which aredefined by the interplay of attraction and repulsion energies, Eaand Er. The phase behaviour can be parameterized using Ea/Er, ande5 dr/dp, which represents a relative range of repulsive interactionsdr with respect to particle size dp (ref. 15). These parameters in DNA

particle assembly systems can be conveniently controlled in a varietyof ways, including thedesign of individual DNA, DNAshell structureand solution ionic strengths2023. Experimentally, DNA-inducedparticle crystallization into random hexagonal close-packed crystalswas observed near surfaces for single-component micrometre-sizedparticle systems with short-range interactions (eR0)9,10. Particlecrystallization with long-range interaction potentials (e< 1) forwhich diverse non-close-packed structures are expected15 (such asDNA-induced crystallization of nanoparticles) has not beenachieved. Here we vary the length of the linking DNA molecules inthe model DNAnanoparticle system, providing a straightforwardmeans of systematically changing the repulsive component of theinterparticle potential.

Figure 1a and b shows a schematic illustration of the assemblysystem used to measure the effect of DNA structure on assemblylong-range order, as studied under a variety of thermal conditions.In each assembly system, a set of DNA-capped gold nanoparticles(denoted A and B) with different DNA structures (SupplementaryTable 1) were allowed to assemble by DNA hybridization into meso-scale aggregates (Fig. 1b, c). The complementary outer recognitionsequences of the DNA capping provided the driving force for A and Bparticle assembly.The length of therecognition sequence, Na, sets thescale of adhesion (per hybridized linker), Ea / Na, to be from

,30 kT at room temperature (,25 uC)to,0 kT at the DNA meltingtemperature Tm. Hence, Ea is approximately constant for all studiedsystems at any given temperature, with van der Waals interactions(,0.5 kT) contributing insignificantly20,24. In a brush regime, thelength N of DNA and the flexibility of the non-complementaryinternal spacer allowed us to tune the range, dr / N

3/5, of repulsiveinteraction and its strength Er / N

3/5/(N3/52 cNa), where c isdefined by persistence length and molecule surface density and isconstant for all studied systems20,25. For single-strandedDNA, experi-mentally relevant separations between particles, sufficient DNAsurface densities and suitable salt concentrations, an estimated mag-nitudeofEr can reach several kT perchain

20. Thus, the use ofmultiplesystems with constant Ea (that is, recognition sequence), and varieddr (that is, spacer lengths), enabled effective interparticle potential

tuning, providing the environment required to achieve crystallinemorphologies of nanoparticle assemblies via the thermal pathwayshown in Fig. 1.

To monitor directly the in situphase behaviour of the systems, weused synchrotron-based small-angle X-ray scattering (SAXS). Theinternal structure of the nanoparticle assemblies was investigatedunder different thermal conditions, including: assembly at roomtemperature; heating the assembly to pre-melting temperatures(Tpm) and DNA melting temperatures (Tm); and cooling the assem-bly below Tm. Figure 2 illustrates the observed scattering changesalong this thermal pathway towards crystallization. The peak posi-tions and relative heights in the SAXS patterns and in the extractedstructure factors S(q) (see Methods and Supplementary Discussionand Supplementary Fig 1 and 2) reveal insights into the structure of

the assemblies, while the number of peaks and their widths reflect thedegree of ordering within the structure.

From the thermal cycling described, we found that systems IV andV, which have the longest flexible spacers (35 and 50 bases respec-tively), showed spontaneous crystalline organization with remark-able degrees of long-range order. In contrast, systems with shorterspacers (systems IIII) or more rigid spacers (see SupplementaryDiscussion and Supplementary Fig. 1) remained amorphous uponcooling. For the crystallizing system IV (Fig. 2), the ordered structureappears after melting and subsequent cooling below Tm at ,60 uC;this is signified by the emergence of several sharp diffraction peaks inthe presence of strong diffuse scattering attributed to a large contri-bution by the nanoparticles form factor. This suggests the presence

*These authors contributed equally to this work.

1Center for Functional Nanomaterials, 2Biology Department, Brookhaven National Laboratory, Upton, New York 11973, USA.

Vol 451 | 31 January 2008 | doi:10.1038/nature06560

549

NaturePublishingGroup2008

http://www.nature.com/doifinder/10.1038/nature06560http://www.nature.com/naturehttp://www.nature.com/naturehttp://www.nature.com/doifinder/10.1038/nature06560

8/3/2019 Dmytro Nykypanchuk, Mathew M. Maye, Daniel van der Lelie and Oleg Gang- DNA-guided crystallization of colloida

2/4

of nuclei of thenewly forming phase in coexistence with unassembledparticles. Upon further cooling to 5957 uC, we observed completecrystallization of the sample, indicated by the dramatic reduction ofparticle form factor and the presence of sharp circular patterns char-acteristic of un-oriented polycrystalline samples (that is, powderscattering). This crystalline formation occurred within only a fewminutes, unaffected by the cooling rate available in our set-up (upto 1 uCmin2

1). The observed crystallization is the result of specificDNADNA interactions, as confirmed by multiple control experi-

ments with noncomplementary DNA-capped nanoparticles or withuncapped particles, none of which exhibit assembly. In addition, theDNA specificity of the assembled systems is manifested in meltingprofiles obtained with both ultravioletvisible spectrophotometry(Supplementary Table S2) and SAXS for all systems.

The SAXS patterns in Fig. 3 reveals seven and ten orders ofresolution-limited Braggs peaks for systems IV and V, respectively,demonstrating their crystalline 3D structures, remarkable degrees of

long-range ordering, and crystallite sizes of at least ,0.5mm, as esti-matedfrom thescattering correlationlength,j< 2p/Dq(ref.26), whereDqis the resolution-corrected (Dqres< 0.0015A

21) full-width at half-maximum (FWHM) of the first diffraction peak. Once formed, thesecrystalline structures were indeed reversible, as confirmed by multipleassemblydisassembly cycles, without a noticeable loss of orderingquality or changes in system behaviour or Tm. Analysis of the peakposition ratios reveal q

x/q15!1:!2:!3:!4:!5:!6:!7, which correspond

to the Im93m space group, a body-centred cubic (b.c.c.) structure, as

shown in Fig. 3a. The S(q) peak height also qualitatively follows therelativeintensities predicted forb.c.c. arrangement.The observed b.c.c.structure meets the requirement for optimizing interaction energies inthe studied binary AB system15 (with AA and BB interactions beingmainly repulsive), by having only particle B in the coordination shellof A and vice versa, thus forming CsCl-type superlattices.

The measured lattice parameters afor the observed b.c.c. structureare ,35nm at 30 uC and ,42.4 nm at 28 uC for systems IV and V,

+

DNA-capped nanoparticles AmorphousDNA-linked aggregates

Reorganizedpre-melted aggregates

Melting-induceddisassembly

Assembly T Tpm T> Tm T< Tm

Re-assembled aggregatesof varied long-range order

or

Crystalline(systems IVV)

Amorphous(systems IIII)

dc

a

b

10 nm

10 m

100 nm Synchrotron X-ray

gnirettacS

Sample

Buffersolution

Assembledaggregates

T-control

T= 56 C

T= 71 C

T= 28 C

S S

15-bp linker

3-b spacer

A B

System I

BS S

15-bp linker

15-b spacerSystem II

0.2

0.1

0.00 0.02 0.04q (1)

0.06 0.08 0.1000

1

2

30 40 50T(C)

60 70

A S S

15-bp linker

35-b spacer 15-b spacer

B

System III

A S S

15-bp linker

35-b spacer

B

System IV

A S S

15-bp linker

50-b spacer

B

System V

A

A525nm

S(q)

Figure 1 | Schematic of experimental design. a, The assembly system ofDNA-capped nanoparticles, the aggregates of which show a series ofstructural changes under a variety of thermal conditions. b, DNA linkagesbetween nanoparticles (one interparticle linkage is shown for clarity, not toscale) with recognition sequences for the A (blue) and B (red) sets of DNAcapping. bp, base pairs. b, bases. s, thiol termination of DNA.c, Representative transmission (top) and scanning (middle, bottom)

electron microscopy images of nanoparticles before (top) and after (middle,bottom) assembly at room temperature. d, Typical example of experimentalmeasurements that reveal a correlation between the ultravioletvisiblemelting profile of theaggregate andits internal structureas probedbyin situSAXS measurements at room temperature, pre-melting temperature, andabove the disassembly/melting temperature.

LETTERS NATURE | Vol 451 | 31 January 2008

550

NaturePublishingGroup2008

8/3/2019 Dmytro Nykypanchuk, Mathew M. Maye, Daniel van der Lelie and Oleg Gang- DNA-guided crystallization of colloida

3/4

respectively. At these a values, the particles are d1IV5 18.9 nm and

d1V5 24.2 nm apart, in the first coordination shell of the obtained

crystals, for systems IV and V, respectively, as calculated from a(d15 !3/2a2dp). These values are close to the equilibrium linkerdimensions, estimated from the scaling argument27 for the DNA

capping environment to be 18 and 23 nm. These a values indicatethat the crystalline structures are remarkably open, in which nano-particles occupy only,24%of the volume and the DNAsoccupyanadditional 45%. Thus, more than ,90% of the assembled structurevolume is occupied by solvent molecules, which is far higher than thetypical void space in packed hard spheres in a b.c.c. orientation(,32%). Such open framework of a superlattice makes the structurevulnerable to collapse uponsolvent removal, whichprevents accuratemorphology visualization by electron microscopy, but at the same

time, makes the crystals highly accessible for modifications, mole-cular transport and storage. In addition, a was found to be highlysensitive to temperature and is primarily defined by thermal pro-perties of DNA. A measured thermal expansion coefficient a