Metallic nanostructures can be used to manipulate light on the subwavelength scale to create tailored optical material properties. Next to electric responses, artificial optical magnetism is of particular interest but difficult to achieve at visible wavelengths. DNA-self-assembly has proved to serve as a viable method to template plasmonic materials with nanometer precision and to produce large quantities of metallic objects with high yields. We present here the fabrication of self-assembled ring-shaped plasmonic metamolecules that are composed of four to eight single metal nanoparticles with full stoichiometric and geometric control. Scattering spectra of single rings as well as absorption spectra of solutions containing the metamolecules are used to examine the unique plasmonic features, which are compared to computational simulations. We demonstrate that the electric and magnetic plasmon resonance modes strongly correlate with the exact shape of the structures. In particular, our computations reveal the magnetic plasmons only for particle rings of broken symmetries, which is consistent with our experimental data. We stress the feasibility of DNA self-assembly as a method to create bulk plasmonic materials and metamolecules that may be applied as building blocks in plasmonic devices.
Metallic nanostructures can be used to manipulate light on the subwavelength scale to create tailored optical material properties. Next to electric responses, artificial optical magnetism is of particular interest but difficult to achieve at visible wavelengths. DNA-self-assembly has proved to serve as a viable method to template plasmonic materials with nanometer precision and to produce large quantities of metallic objects with high yields. We present here the fabrication of self-assembled ring-shaped plasmonic metamolecules that are composed of four to eight single metal nanoparticles with full stoichiometric and geometric control. Scattering spectra of single rings as well as absorption spectra of solutions containing the metamolecules are used to examine the unique plasmonic features, which are compared to computational simulations. We demonstrate that the electric and magnetic plasmon resonance modes strongly correlate with the exact shape of the structures. In particular, our computations reveal the magnetic plasmons only for particle rings of broken symmetries, which is consistent with our experimental data. We stress the feasibility of DNA self-assembly as a method to create bulk plasmonic materials and metamolecules that may be applied as building blocks in plasmonic devices.
Entities:
Keywords:
DNA origami; nanoparticles; plasmonic metamaterials; self-assembly
A negative refractive index metamaterial requires both electric permittivity
and magnetic permeability to be negative. While metals naturally exhibit
negative real values for their permittivity at visible frequencies,
negative permeability has to be created through an “artificial
magnetic” response.[1,2] The well-known concept
to achieve this goal in the THz domain is to create electric and magnetic
resonances in split ring resonators,[3] which
has been realized in many examples and led to the development of gradient
index lenses[4,5] and switching devices for terahertz
radiation.[6] This and similar approaches
that are based on electric leads, however, are hard to transfer into
the visible frequency domain due to the changing conduction properties
of metals at high frequencies and the resulting saturation effect
of the magnetic response.[7] As an alternative
route to artificial magnetic responses, it has been proposed that
metal nanoparticles that are brought together in a ring geometry can
support a circulating displacement current induced by plasmonic resonances.[8−10] Similar to the conduction current within split ring resonators the
plasmonic interaction can lead to the emergence of magnetic resonances
at visible or near-infrared frequencies. Recent implementations of
such ring resonators also experimentally demonstrated the appearance
of magnetic resonances and even magnetic-based optical Fano resonance.[11] Studies that investigated size- and shape-controlled
nanoscale ring arrangements of plasmonic particles used either atomic
force microscope nanomanipulation[11] or
lithography[12,13] to achieve their goal. Other
approaches are based on the use of dielectric spheres as substrates
for the random attachment or growth of plasmonic nanoparticles[14−16] or nanoshell clusters.[17,18] To take advantage of
the intriguing concept to use plasmonic particle devices as building
blocks for large scale metamaterials it is necessary to reach full
control over design and manufacturing of subwavelength structures
as well as to be able to produce large numbers of plasmonic structures
in a parallel manner.Here we report a strategy to overcome
these challenges by using the DNA origami method[19−21] to fabricate
versatile templates for metamolecules on a length scale much smaller
than the wavelength of visible light. Structural DNA nanotechnology
allows the assembly of trillions of identical nanoscale three-dimensional
objects at once in a simple, solution-based reaction.[22,23] A DNA origami structure is built from one long scaffolding DNA single-strand
(here: 8634 nucleotides) that is assembled into the desired two- or
three-dimensional shape via the annealing with ∼250 short “staple”
oligonucleotides. Because of their inherent and defined sequence addressability,
DNA structures allow nanometer precise positioning of objects such
as metallic nanoparticles through the incorporation of DNA handle
sequences at virtually any site on the structures.[24−29]For our goal to create nanoparticle ring resonators of defined
geometries, this bottom-up fabrication method was employed to assemble
several nanoring geometries consisting of a defined number of metal
nanoparticles that were all equal or of various sizes and materials.
Using the software cadnano (cadnano.org)[30] and CanDo[31,32] we designed a ring-shaped DNA-origami
structure of 14 parallelly arranged DNA double-helices of 200 nm length
that were bent by design[33] into a full
circle by insertion and deletion of bases at selected sites (see Supporting Information for design details and
experimental protocols S1–S4). The resulting structure has
a diameter of 62 nm and a ring cross-section of ∼10 nm. To
serve as a versatile template for all desirable ring-shaped particle
geometries, the DNA origami structure offers repeating units of DNA
handles that can act as specific attachment sites for DNA-functionalized
nanoparticles, overall 27 evenly spaced sites. It is thus possible
to create a large diversity of particle ring geometries. Figure 1 illustrates the self-assembling process and a variety
of implemented particle arrangements, including symmetric and asymmetric
site-specific organization of particles of different sizes and materials.
In the first experimental step, the DNA origami template is annealed
by heating up and slowly cooling down a solution containing the scaffold
strand and a set of staple oligonucleotides including all sequences
that are needed to form the desired attachment sites for particle
binding. After purification of the folded DNA origami template via
gel electrophoresis, DNA-functionalized gold (AuNP) and silver (AgNP)
nanoparticles of the desired size were hybridized to the attachment
sites. Here particles of 10–40 nm in diameter were used but
in principle also larger and smaller particles could be employed.[26] The AuNPs are covered with 15 nucleotide (nt)
long single-stranded DNA (ssDNA) sequences to bind them to the origami
template as well as to stabilize them against high salt concentrations
that are needed during the annealing process. A second gel electrophoresis
step is performed to purify the now assembled metamolecules from the
excess of unbound AuNPs and little amounts of aggregated structures
(Supporting Information Figure S5). Note
that the presence of an excess of nanoparticles during the assembly
largely inhibits the aggregation of the metamolecules. We achieved
yields of correctly arranged particle rings of up to 73%, (statistics
and additional TEM images can be found in the Supporting Information Figures S6–S9). By choosing
unique handle sequences at specific attachment sites and by introducing
the complementary functionalized particle species, various particle
types can be attached at defined positions in one step (Figure 1d) (see Supporting Information for particle functionalization and purification).
Figure 1
DNA origami–nanoparticle
ring structures. Plasmonic rings were built by arranging nanoparticles
on ring-shaped DNA origami structures. (a) Self-assembly of DNA origami
nanoparticle rings: an 8 kb-long DNA scaffold was annealed with ∼250
synthetic staple strands (each ∼40 bases long) to create a
ring-shaped 14-helix bundle template structure (14HB-template). In
the next step, DNA-functionalized AuNPs were attached to defined positions
on the 14HB-template via hybridization of the handle sequences protruding
from the origami ring and the complementary DNA strands on the AuNPs.
(b) Upper panel: schematic illustration of the 14HB-template with
four identical attachment sites to create a symmetric four particle
ring. Lower panel: Wide-field transmission electron microscopy (TEM)
micrograph of plasmonic ring structures consisting of four 40 nm AuNPs.
Scale bar: 400 nm. Inset: Zoom-in. Scale bar: 40 nm. (c) Upper panels:
Schematic illustration and TEM images of 14HB-templates with five
40 nm AuNPs (left), six 30 nm AuNPs (middle), and eight 20 nm AuNPs
(right) attached. Scale bars: 40 nm. Lower panels: Eight 20 nm AuNPs
attached in a ring of dimers (left), eight 10 nm AuNPs attached in
a ring of equal spacing (middle), and ten 10 nm AuNPs in a C-shaped
geometry. Scale bars: 40 nm. (d) Schematic illustration and TEM images
of asymmetric ring structure assembly where the attachment sites offer
two different sequences (illustrated as red and blue dots). Six (left)
and four (middle) AuNPs with two different diameters (30 and 20 nm,
and 40 and 20 nm, respectively) are shown, as well as three AuNP and
one AgNP (right) with diameters of 40 and 20 nm, respectively.
DNA origami–nanoparticle
ring structures. Plasmonic rings were built by arranging nanoparticles
on ring-shaped DNA origami structures. (a) Self-assembly of DNA origami
nanoparticle rings: an 8 kb-long DNA scaffold was annealed with ∼250
synthetic staple strands (each ∼40 bases long) to create a
ring-shaped 14-helix bundle template structure (14HB-template). In
the next step, DNA-functionalized AuNPs were attached to defined positions
on the 14HB-template via hybridization of the handle sequences protruding
from the origami ring and the complementary DNA strands on the AuNPs.
(b) Upper panel: schematic illustration of the 14HB-template with
four identical attachment sites to create a symmetric four particle
ring. Lower panel: Wide-field transmission electron microscopy (TEM)
micrograph of plasmonic ring structures consisting of four 40 nm AuNPs.
Scale bar: 400 nm. Inset: Zoom-in. Scale bar: 40 nm. (c) Upper panels:
Schematic illustration and TEM images of 14HB-templates with five
40 nm AuNPs (left), six 30 nm AuNPs (middle), and eight 20 nm AuNPs
(right) attached. Scale bars: 40 nm. Lower panels: Eight 20 nm AuNPs
attached in a ring of dimers (left), eight 10 nm AuNPs attached in
a ring of equal spacing (middle), and ten 10 nm AuNPs in a C-shaped
geometry. Scale bars: 40 nm. (d) Schematic illustration and TEM images
of asymmetric ring structure assembly where the attachment sites offer
two different sequences (illustrated as red and blue dots). Six (left)
and four (middle) AuNPs with two different diameters (30 and 20 nm,
and 40 and 20 nm, respectively) are shown, as well as three AuNP and
one AgNP (right) with diameters of 40 and 20 nm, respectively.Obviously, also smaller and larger
ring diameters could be implemented by varying the template DNA structure.
The origami template chosen in this study provides a good balance
between structural stiffness (cross section) of the ring template
and large enough diameter to accommodate the attachment of four or
more 30 and 40 nm AuNPs with small gaps between the particles. Importantly,
we found in our experiments that small interparticle distances are
essential to support circulating current modes along the rings. Because
of their comparatively simple spectra, we focus in the following on
four-particle metamolecules.The plasmonic properties of the
assembled individual ring structures were characterized by dark-field
scattering spectroscopy. In our experimental setup the unpolarized
incident light hits the glass substrate at ∼23° from all
sides and the scattered light is collected above the sample and perpendicular
to the glass surface (Figure 2a). First, numerical
COMSOL simulations conforming to our experimental geometry were performed
to understand the scattering behavior of the particle rings. In our
simulations, we used the empirical dielectric constant of gold taken
from Johnson and Christy.[34] Figure 2b shows the simulated scattering spectra for symmetric
planar rings consisting of four 40 nm AuNPs. The gap size between
the particles was varied between 7 and 1 nm. As observed in previous
studies,[35−37] the peak position of the plasmonic resonance shifts
to longer wavelengths for smaller gap sizes. Additionally, at gap
sizes smaller than 4 nm a second peak becomes distinguishable in the
scattering spectra whereas the absorption simulation shows three peaks.
We analyzed the three modes in a surface charge intensity plot for
the case of 1 nm gaps and found them to originate from dipolar, multipolar,
and magnetic interactions (Figure 2c). Interestingly,
the rings with perfect symmetry exhibit three modes in the computed
absorption spectra, whereas the computed scattering spectra show two
modes only and do not show the magnetic mode. However, the simulations
executed with slightly asymmetric rings (Figure 2b) do show three distinguishable peaks in both absorption and scattering
as here the symmetry breaking induces coupling between the far-field
scattered light and the magnetic modes.
Figure 2
Experimental setup and
simulated scattering and absorption spectra of symmetric and asymmetric
four-particle ring structures. (a) Experimental dark-field setup and
the corresponding model for the numerical simulations (inset). (b)
The scattering spectrum of a symmetric ring structure consisting of
four 40 nm AuNPs strongly depends on the gap size Δ between
the particles. For distances smaller than 4 nm a second resonance
peak appears in the scattering spectra. (c) The symmetric ring with
a gap size of 1 nm shows two peaks in the scattering simulation but
three peaks in the absorption whereas the asymmetric ring shows three
peaks both in scattering and absorption. The origins of these peaks
are dipolar, multipolar, and magnetic interactions.
Experimental setup and
simulated scattering and absorption spectra of symmetric and asymmetric
four-particle ring structures. (a) Experimental dark-field setup and
the corresponding model for the numerical simulations (inset). (b)
The scattering spectrum of a symmetric ring structure consisting of
four 40 nm AuNPs strongly depends on the gap size Δ between
the particles. For distances smaller than 4 nm a second resonance
peak appears in the scattering spectra. (c) The symmetric ring with
a gap size of 1 nm shows two peaks in the scattering simulation but
three peaks in the absorption whereas the asymmetric ring shows three
peaks both in scattering and absorption. The origins of these peaks
are dipolar, multipolar, and magnetic interactions.To investigate the scattering spectra of the plasmonic
ring structures experimentally they were immobilized on a substrate,
analyzed with a dark-field spectrometer, and then imaged with scanning electron microscopy
(SEM) to determine the exact position of every single AuNP in each
ring. For the measurements a droplet of solution containing the ring
structures was deposited on a plasma-cleaned quartz glass. After 10
min, the droplet was flushed away by water and the surface was quickly
dried with nitrogen. As the particle rings are slightly distorted
during the adsorption and the subsequent drying process, small deviations
from the designed particle geometries are observed in the SEM images
and these spatial variations are in turn reflected in the disparity
of the recorded spectra (Figure 3a). To further
investigate the influence of the surrounding medium we performed scattering
measurements of ring structures in air (red lines in Figure 3a) but also immersed in a buffered solution (1×
TBE, 11 mM MgCl2, black lines in Figure 3a).
Figure 3
Scattering spectra of individual ring structures immobilized on
glass substrates. The spectral shape is strongly dependent on the
exact particle ring geometry. (a) Each two scattering spectra were
collected from the same structures immersed in buffer (black curves)
and after drying in air (red curves). Insets: corresponding SEM images
of each ring structure. Scale bar: 100 nm. (b) The spectra in air
were simulated with the AuNP positions as determined from SEM data.
The spectra in water were then fitted by moving the positions of each
AuNP 4 nm radially outward.
Scattering spectra of individual ring structures immobilized on
glass substrates. The spectral shape is strongly dependent on the
exact particle ring geometry. (a) Each two scattering spectra were
collected from the same structures immersed in buffer (black curves)
and after drying in air (red curves). Insets: corresponding SEM images
of each ring structure. Scale bar: 100 nm. (b) The spectra in air
were simulated with the AuNP positions as determined from SEM data.
The spectra in water were then fitted by moving the positions of each
AuNP 4 nm radially outward.For structures immersed in solution only one dominant broad
resonance at around 560 nm is observed while the spectra taken in
air exhibit multiple distinct peaks, as well as a red shift of the
dominant resonances. The change of the effective permittivity could
explain this red shift[38,39] but not the appearance of the
additional resonances. These peaks are a strong indication of a reduction
of the interparticle distance,[35,37] which is expected to
occur during the drying process, as dehydration will reduce the electrostatic
and steric repulsion between the ssDNAs that coat the AuNPs.[38] To study this conformational change of only
a few nanometers and to support the experimental observations we performed
numerical simulations. The positions of all AuNPs in the dried rings
were determined from the SEM images with an accuracy of 1 nm and served
as precondition for the simulations. We found an average surface-to-surface
distance of 2.4 nm ±1.9 nm (Supporting Information Figure S7) and strikingly, the simulated scattering spectra match
the measured spectra of the individual dried structures remarkably
well (red lines in Figure 3b). It was also
possible to simulate the shape of the experimental spectra collected
from the structures immersed in buffer. Therefore, not only the effective
permittivity was adapted but also the position of each AuNP had to
be moved 4 nm radially outward, which validates the assumption of
a shift in the position of the AuNPs during the drying process.Generally, already small deviations in the positions of the AuNPs
compared to a perfect symmetric ring structure cause shifts in the
occurring resonant peaks and the emergence of new resonances (cf.
simulated spectra in Figure 2b,c and observed
and simulated spectra in Figure 3a,b). To analyze
the origin of the measured resonances additional numerical simulations
were performed for the ring structure shown in Figure 4a. The simulations show three distinguishable peaks in the
scattering spectra (Figure 4b). Two peaks (at
550 and 600 nm) were already visible in the symmetric ring case (Figure 2c) and belong to the multipolar and dipolar resonances.
The third peak appearing in the computed scattering spectra at 720
nm has its origin in a magnetic plasmon resonance and appears in the
scattering configuration only for four-particle rings with broken
symmetry. In such asymmetric rings, the magnetic plasmons become active
in the scattering because the broken symmetry opens a coupling channel
between the magnetic mode and the far-field photons propagating along
the z-axis. Figure 4c shows
the surface charge distributions for all peaks as well as the electric
and magnetic field map corresponding to the magnetic resonance at
720 nm. Importantly, the experiments confirm that the response of
plasmonic ring structures is very sensitive to their exact geometries
and that electric and magnetic resonances can be supported at optical
frequencies in our DNA-assembled nanoparticle ring resonators (additional
spectra can be found in the Supporting Information Figures S10 and S11).
Figure 4
Analysis of single scattering spectrum. (a)
Scattering spectra of a structure exhibiting multiple modes. Inset:
corresponding SEM image. Scale bar: 50 nm. (b) Simulation of the scattering
spectra for different polarizations of the incident light corresponding
to the particle ring presented in panel a. Three distinguishable peaks
are visible. (c) Simulation of the surface charge distribution for
all peaks as well as the electric and magnetic field corresponding
to the peak at 720 nm.
Analysis of single scattering spectrum. (a)
Scattering spectra of a structure exhibiting multiple modes. Inset:
corresponding SEM image. Scale bar: 50 nm. (b) Simulation of the scattering
spectra for different polarizations of the incident light corresponding
to the particle ring presented in panel a. Three distinguishable peaks
are visible. (c) Simulation of the surface charge distribution for
all peaks as well as the electric and magnetic field corresponding
to the peak at 720 nm.One of the great advantages of DNA-based self-assembly is
the possibility to manufacture large numbers of identical objects
dispersed in solution, which gives access to the use of the DNA-assembled
metamolecules as building blocks in metafluids and to scale up the
production.[21,25] To demonstrate that our DNA origami
ring structures also perform in liquids, we carried out absorption
measurement of bulk solutions containing plasmonic rings with four
40 nm AuNPs. Indeed, the observed spectral peak is shifted from 530
nm for freely dispersed and DNA-coated 40 nm gold nanoparticles to
540 nm for the DNA-assembled particle rings (Figure 5a). The experimental result is in agreement with simulations,
which assume an interparticle distance of 10 nm and predict a shift
from 530 to 542 nm (Figure 5b).
Figure 5
Optical characterization
of the bulk plasmonic ring structures in solution. The absorption
spectrum of plasmonic ring structures consisting of four 40 nm AuNPs
was measured in bulk solution. (a) Measured absorption spectra of
40 nm AuNPs covered with ssDNA (black curve) and the ring structure
with four attached 40 nm AuNPs in solution (red curve). (b) Corresponding
finite difference time domain (FDTD) simulations of 40 nm AuNP in
solution (black curve) and a symmetric ring structure (red curve)
with a gap of 10 nm between the AuNP.
Optical characterization
of the bulk plasmonic ring structures in solution. The absorption
spectrum of plasmonic ring structures consisting of four 40 nm AuNPs
was measured in bulk solution. (a) Measured absorption spectra of
40 nm AuNPs covered with ssDNA (black curve) and the ring structure
with four attached 40 nm AuNPs in solution (red curve). (b) Corresponding
finite difference time domain (FDTD) simulations of 40 nm AuNP in
solution (black curve) and a symmetric ring structure (red curve)
with a gap of 10 nm between the AuNP.Overall, we successfully implemented DNA origami structures
as templates for the synthesis of a wide variety of ring geometries
containing nanoparticles of selectable size and type. We were able
to construct various light-manipulating ring structures with interparticle
gaps of only a few nanometers and we demonstrated their ability to
support strong plasmonic resonances in the visible frequency domain
both attached to a substrate and dispersed in solution. These resonances
include dipolar, multipolar, and magnetic modes and are very sensitive
to the ring symmetry. Our DNA-based self-assembly approach offers
the opportunity to build metamolecules in large numbers with nanometer
resolution and dispersed in solutions. This opens the route to a new
class of material such as metasprays or metamolecules incorporated
in a solid host, which could then be used in nanophotonic and sensing
applications.
Authors: Anton Kuzyk; Robert Schreiber; Zhiyuan Fan; Günther Pardatscher; Eva-Maria Roller; Alexander Högele; Friedrich C Simmel; Alexander O Govorov; Tim Liedl Journal: Nature Date: 2012-03-14 Impact factor: 49.962
Authors: Maryam Zahedian; Xinlei Huang; Irina B Tsvetkova; Vincent M Rotello; William L Schaich; Bogdan Dragnea Journal: J Phys Chem B Date: 2016-05-11 Impact factor: 2.991
Authors: Eva-Maria Roller; Lucas V Besteiro; Claudia Pupp; Larousse Khosravi Khorashad; Alexander O Govorov; Tim Liedl Journal: Nat Phys Date: 2017-05-15 Impact factor: 20.034