-
Towards Infrared Reconfigurable
Plasmonics and Metamaterials
A dissertation presented
by
Kan Yao
to
the Department of Electrical and Computer Engineering
in partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
in the field of
Electrical Engineering
Northeastern University
Boston, Massachusetts
July 2017
-
ii
© Copyright 2017
Kan Yao
All Rights Reserved
-
iii
Abstract
Metamaterials are rationally engineered composite artificial
structures with exotic
properties not easily obtained or altogether unavailable in
nature. Since the proof-of-
concept experiments in the microwave regime in the early 2000s,
the operating
wavelength of metamaterials has been quickly pushed from
terahertz (THz) to mid-/near-
infrared and down to visible, where metamaterials started to
interfere with another
emerging field of plasmonics. The dynamic interplay between them
has enabled various
innovative concepts and applications, which are impossible with
either area alone.
Despite the continuous blending of metamaterials and plasmonics,
there are still
several sub-fields not fully explored and limitations not yet
resolved. For instance, metals
do not support pronounced plasmonic responses at mid-infrared
and far-infrared
frequencies. On the other hand, infrared light has found
widespread applications such as
biomedical sensing, spectroscopy, and surveillance. The lack of
suitable platforms
supporting strong resonances at infrared wavelengths, as a
consequence, naturally leads
to the quest of better alternatives for infrared plasmonics and
metamaterials.
Another restriction is tunability. So far, most metamaterials
and plasmonic devices are
in solid phase and have fixed functionalities once they are
fabricated. Since the optical
properties of metamaterials and plasmonic structures rely on
both the constituents and the
way in which they are shaped and organized, introducing
variables into the system, either
changeable material properties or reconfigurable structures,
will open up new
opportunities for dynamically tunable metamaterials and
plasmonic structures.
-
iv
In this work, aiming at solving both problems mentioned above, I
will present several
solutions to achieve infrared reconfigurable plasmonics and
metamaterials. Simulations
and experiments are conducted respectively for the design and
realization of infrared
optical devices, covering different aspects of this topic. I
will show plasmonic core/shell
particles and their self-assembled clusters as infrared
resonators, ultra-compact
nanoantennas with switchable and unidirectional radiation
properties, and tunable
graphene metasurfaces for infrared light beam shaping. Future
work and preliminary
results on programmable plasmonic metamolecules via magnetic
self-assembly will also
be discussed.
-
v
TABLE OF CONTENTS
Table of Contents
List of Figures
Acknowledgements
Chapter 1 Introduction
.............................................................................................
1
Chapter 2 Plasmonic Core-Shell Particles by Chemical Synthesis
.............................. 10
2.1 Introduction
...........................................................................................................
10
2.2 Synthesis of Plasmonic Core-shell Particles
........................................................... 13
2.3 Assembly of Clusters
..............................................................................................
23
Chapter 3 Self-Assembled Plasmonic Core-Shell Particles as
Infrared Resonators..... 28
3.1 Bright-Field Reflection Measurements
..................................................................
28
3.2 Simulation of Single Particle Scattering on a Gold Substrate
................................ 35
3.3 Simulation of Single Cluster Scattering on a Gold Substrate
................................. 40
3.4 Summary and Outlook
...........................................................................................
44
Chapter 4 Designer Plasmonic Core-Shell Particles by
Microfabrication ..................... 47
4.1 Introduction
...........................................................................................................
47
4.2 Self-Assembly of Polystyrene Particle Monolayers
............................................... 49
4.3 Core-Shell Particles by Microfabrication
................................................................
52
4.4 Conclusions
............................................................................................................
56
Chapter 5 Ultracompact Nanoantennas with Switchable
Directionality ..................... 57
5.1 Introduction
...........................................................................................................
57
5.2 Design of Ultracompact Unidirectional Nanoantennas
......................................... 59
5.3 Analysis and Discussions
........................................................................................
70
-
vi
5.4 Conclusions
............................................................................................................
73
Chapter 6 Tunable Graphene
Metasurfaces..................................................................
75
6.1 Introduction
...........................................................................................................
75
6.2 Infrared Light Beam Shaping by Graphene Metasurfaces
..................................... 78
6.2.1 Anomalous Reflection
....................................................................................
82
6.2.2 Reflective Focusing
Lens.................................................................................
84
6.2.3 Non-diffracting Airy Beams
............................................................................
86
6.3 Infrared Circular Dichroism by Graphene Metamirrors
......................................... 88
6.4 Conclusions
............................................................................................................
90
Chapter 7 Summary and Future work
...........................................................................
92
7.1 Summary
................................................................................................................
92
7.2 Future Work
...................................................................................................
93
6.2.1 Programmable Plasmonic Metamolecules via Magnetic
Self-Assembly ....... 93
6.2.2 Core-Shell Particles for Sensing Applications
................................................. 98
References
..............................................................................................................
100
-
vii
LIST OF FIGURES
Figure 1.1: Schematics of some representative metamaterials. (a)
A lattice of metallic
wires to achieve negative permittivity at low-frequency. (b)
Metallic split ring
resonators (SRRs) to produce artificial magnetism. (c) Metallic
helices as building
blocks for chiral metamaterials.
....................................................................................1
Figure 1.2: (a) Illustration of SPPs and oscillation of surface
charges at a dielectric/metal
interface. (b) The typical electromagnetic field profile of
SPPs, reflecting their
bound, non-radiant nature.
............................................................................................2
Figure 1.3: Reconfigurable metamaterials and metadevices by
material property changes.
(a) Functional devices can be written in or erased from a
chalcogenide glass (GST)
phase-change film by trains of femtosecond laser pulses at
different conditions. (b)
The optical image of two Fresnel zone patterns written by light.
(c) The second
Fresnel zone pattern on the right is erased by light. (d) A
hybrid metamaterial
consisting of a vanadium oxide (VO2) film sandwiched by gold
meshes and a ground
plane. Applying electrical currents will induce the
insulator-to-metal transition of
VO2 and change the reflectance. (e) Top left: Microscopic image
of the metamaterial.
Letters are bare VO2 areas without gold meshes. Rest panels in
clockwise order:
Infrared images at 2.67 μm wavelength with 0 A, 2.03 A, and 2.20
A current,
respectively. (f) An infrared light beam excites tunable plasmon
resonances across
the graphene nanoribbons biased at varying gate voltages. (g)
Plasmon-enhanced
light-matter interactions improve detection sensitivity of
molecules adsorbed on
graphene. (a-c) are adopted from Ref. [13], (d, e) are adopted
from Ref. [14], (f) is
adopted from Ref. [17], and (g) is adopted from Ref. [18],
respectively. ....................4
Figure 1.4: Reconfigurable metamaterials and metamolecules by
finite structural changes.
(a) Split ring resonator (SRR)-bar on a stretchable film.
Mechanical deformation of
the elastomeric substrate changes the gap size and thus the
coupling, which causes
noticeable resonance shift in (b). (c) A metamaterial consisting
of “meander near the
wire” gold nanostructures on silicon nitride strings. (d)
Applying voltages to the
neighboring strings induces an attractive electrostatic force in
the plane, which
reduces the gap size, enabling modulation of transmission and
reflection. (e, f) Two
gold nanorods are hosted on a DNA template to form a chiral
metamolecule
exhibiting tunable chiroptical properties as in (g). (e) The
relative angle between the
nanorods can be actively controlled by adding certain DNA
strands to pair the locks
on the sides of the template. (f) Alternatively, the nanorods
are assembled
perpendicular to each other, and the DNA locks function as
binding sites that allow
the rod on the top (yellow) to “walk” stepwise between the two
ends of the template.
(a, b) are adopted from Ref. [20], (c, d) are adopted from Ref.
[21], (e) is adopted
from Ref. [23], and (f, g) are adopted from Ref. [24].
..................................................6
Figure 2.1: Optical resonators at different wavelengths. (a) A
high-index dielectric (lead
telluride (PbTe)) particle can support a series of Mie resonance
modes in the mid
infrared (MIR) region. (b) High-index dielectric (silicon)
particle clusters exhibit
limited spectral tunability at visible frequencies. (c)
Plasmonic core-shell
nanoparticle clusters show complex spectral features in the near
infrared (NIR)
region. Figures are adopted from Refs. [31], [35], and [39],
respectively. ................12
-
viii
Figure 2.2: Flow chart of the synthesis of SiO2/Au core-shell
particles. (a) Synthesis of
gold nanoparticle seeds (i) and modification of SiO2 particles
(ii). (b) Preparation of
precursor particles, where gold nanoparticles are attached to
the surface of the
modified silica beads. (c) Continuous reduction of gold ions in
the planting solution
causes the nanoparticles to increase in size, coalesce, and
eventually form complete
shells covering the silica cores.
..................................................................................14
Figure 2.3: Typical ultraviolet-visible (UV-Vis) extinction
spectra of the THPC gold
nanoparticle solution. Different curves are collected from
different batches of
samples. All of them exhibit a vague bump close to 500 nm
wavelength, indicating
the sizes of the synthesized nanoparticles are 1-3 nm. Inset
shows the image of an
aged solution, which is dark brown in color.
..............................................................16
Figure 2.4: Scanning Electron Microscope (SEM) images of a 1 μm
(a) and a 2 μm (b)
core-shell particle (monomer) on a gold-coated substrate. The
shell thicknesses are
estimated to be around 30 and 10 nm, respectively.
...................................................19
Figure 2.5: Zoom-in SEM images for shell inspection. (a) The
surface of a 2 μm particle
coated by a complete and smooth gold shell. Occasional cracks
such as pinhole
defects and excess gold nanoparticles are smaller than 40 nm and
have negligible
impact on the infrared optical properties of the core-shell
particles. (b) Perspective
view (~70°) of a 2 μm core-shell particle at a rectangular cut
on its surface. The cut
is made by focused ion beam (FIB). Measurements at different
locations on the
cutting edges suggest a shell thickness of ~10 nm.
....................................................20
Figure 2.6: (a,c) SEM images of a 1 μm (a) and a 2 μm (c)
particle after 5 cycles of gold
reduction with a small amount of planting solution. The gold
nanoparticles increase
in size (~10 nm) but are isolated. (b) Zoom-in image of a 1 μm
particle after 12
cycles of reduction. The grain size increases to 20 nm and
larger, whereas the
nanoparticles still do not coalesce into a complete shell. (d)
Zoom-in images of a 2
μm particle after 5 cycles of reduction with a small amount of
planting solution and
an additional cycle with a large amount (5 times in volume) of
planting solution. A
complete shell is formed whereas the smoothness degrades
slightly. ........................22
Figure 2.7: SEM image of a 1 μm particle after a single cycle of
reduction with a
condensed planting solution. The attached seed nanoparticles
increase significantly
in size but are clearly isolated.
....................................................................................23
Figure 2.8: Self-assembly of core-shell particles. A droplet of
diluted particle suspension
is pipetted on a gold-coated substrate. During total
evaporation, the core-shell
particles are assembled by the capillary force into different
planar structures. ..........24
Figure 2.9: Self-assembled 1 μm core-shell clusters.
Representative geometries include
but are not limited to: (a) dimers, (b-c) trimers, and (d)
chains. The shell thicknesses
are estimated to be about 30 nm. Scale bars are 200 nm.
...........................................24
Figure 2.10: Self-assembled 2 μm core-shell clusters.
Representative configurations
include but are not limited to: (a) dimers, (b) trimers, (c)
quadrumers, (d) chains, and
(e) pentamers. The shell thicknesses are estimated to be about
10 nm. Scale bars are
500 nm.
.......................................................................................................................25
Figure 2.11: Two clusters consisting of particles with
incomplete coating. Some of the
attached nanoparticles drop off during the movement of the
particles, forming a trial
on the substrate. The scale bars are 500 nm in (a) and 1 μm in
(b). ...........................26
-
ix
Figure 3.1: Schematic of the Fourier transform infrared (FTIR)
single cluster
spectroscopy. Infrared light is generated by a globar source.
After passing through a
Michelson interferometer, the incident beam is focused by a 36x
Cassegrain
objective lens on the sample. The reflected light is collected
by the same lens and
directed to an MCT detector. A knife-edge aperture is applied in
the conjugate image
plane to define a small area containing exclusively the
individual clusters of interest,
represented by the rectangle bounded by the red dashed lines.
..................................30
Figure 3.2: Infrared optical properties of different clusters
obtained by single cluster
FTIR spectroscopy. (a) Reflectance spectra of uncoated 2 μm
silica clusters on a gold
substrate. Measured clusters include a monomer (black), a dimer
(red), a symmetric
trimer (blue), and a symmetric heptamer (olive). (b) Reflectance
spectra of 2 μm
core-shell SiO2/Au clusters on a gold substrate. The shell
thickness is about 10 nm.
Measured clusters include a monomer (black), a dimer (red), a
symmetric trimer
(blue), a chain of three (cyan), an asymmetric quadrumer
(purple), an asymmetric
pentamer (orange), and a chain of five (green). Spectra are
shifted vertically for
clarity except for monomers. The numbers given in the figure
legend denote the
number of particles in the cluster.
...............................................................................32
Figure 3.3: (a) Infrared optical properties of 2 μm clusters
consisting of particles with a
defective shell. The sizes of the defects on the shell are
greater than 100 nm. The
existence of large defects enable direct illumination of light
on the silica core,
evidenced by the reflection drop between 8 and 9 μm and the
valley between 9 and
10 μm (arising from the absorption of silica). Meanwhile, the
line shape at shorter
wavelengths also deviates from the measurement results in Figure
2.13(b), indicating
undesired failure of resonant modes excitation. (b, c) SEM
images of defective
clusters measured in (a).
.............................................................................................33
Figure 3.4: Infrared optical properties of 1 μm core-shell
particle clusters on a gold
substrate. The shell thickness is about 30 nm. Measured
configurations include a
monomer (black), a dimer (red), a symmetric trimer (blue), an
asymmetric
quadrumer (purple), a chain of four (dark cyan), and an
asymmetric pentamer
(orange). Spectra are shifted vertically for clarity except for
monomer. All the
measurements were based on single cluster FTIR spectroscopy.
...............................34
Figure 3.5: (a) Comparison of measured and simulated reflectance
spectra of an uncoated
silica monomer and an uncoated silica trimer on a gold
substrate. The very good
agreement between measurements and simulations confirms the
validity of the
simulation settings and the material properties. Slight shifts
between simulations and
measurements may come from the simplified normal incidence.
Because of the
absence of the gold shell and relatively low index of the
particles, no resonance can
be excited, as indicated by the field intensity maps at (b) 2.5
μm and (c) 3.3 μm at
two cross sections cutting through a silica particle on the
symmetry planes. ............36
Figure 3.6: Optical properties of a 2 μm core-shell monomer on a
gold substrate. The
shell thickness is about 10 nm. (a) Comparison of measured (red)
and simulated
(black) reflectance. In the simulation, a y-polarized plane wave
is incident from the
top. Representative modes are denoted by the dashed lines. (b)
Field intensity maps
of the cavity plasmon modes and their correspondence to the
spectral features. Only
one quarter of the particle is shown because of symmetry. The
electric (a11) and
magnetic (b11) dipole modes show typical ring-like magnetic and
electric field
-
x
distributions, respectively, whereas the electric (a21) and
magnetic (b21) quadrupoles
are indicated by the two intensity lobes along the perimeter.
(c) The electric field
intensity distribution at 5.3 μm. The extremely weak field
inside the particle and
large field intensity outside the particle in the y-z plane
(right panel) suggest the
occurrence of the fundamental electric dipole resonance in
response to the y-
polarized incident field. (d) Surface charge density at 5.3 μm
wavelength, which also
manifests the excitation of the electric dipole resonance of the
core-shell particle. ..38
Figure 3.7: Comparison of simulated reflectance spectra for 2 μm
(a) and 1 μm (b)
monomers with different shell thicknesses.
................................................................40
Figure 3.8: Optical properties of a 2 μm core-shell trimer on a
gold substrate. The shell
thickness is about 10 nm. (a) Comparison of measured (red) and
simulated (black)
reflectance. In the simulation, a y-polarized plane wave is
incident from the top.
Representative modes are marked by the dashed lines. (b) Field
intensity maps of the
cavity plasmon modes and their correspondence to the spectral
features. The colored
dashed lines in the central panels indicate the position of the
cutting planes for
observing different mode profiles. Only half of the trimer is
shown because of
symmetry. Compared with Figure 3.6, the electric dipole (a11)
and magnetic
quadrupole (b21) are still conspicuous and almost unshifted. The
electric quadrupole
(a21) becomes visible due to the interparticle coupling, while
the magnetic dipole (b11)
diminishes. (c) The electric field distribution at 7 μm
wavelength. The intense field
in the gap between two particles (left panel) suggests strong
coupling between the
electric dipole modes of the two particles. In contrast, the
third particle (right panel)
is off-resonance. (d) Surface charge density at 7 μm seen from
the dimer side (left
panel) and from the top (right panel). Both confirm the strong
dipole resonances in
the dimer.
....................................................................................................................42
Figure 3.9: Comparison of simulated reflectance spectra for 2 μm
(a) and 1 μm (b)
trimers with different shell thicknesses.
.....................................................................43
Figure 3.10: (a) NIR reflectance spectrum of a 2 μm core-shell
trimer on a gold substrate.
Data is obtained by using a tungsten halogen lamp as the light
source in combination
with a quartz beam splitter and a MCT detector. (b) Field
intensity profiles of the
mode b31. Some of the cavity modes cannot be excited efficiently
due to small shell
thickness and symmetry requirement.
........................................................................45
Figure 4.1: Flow chart of the fabrication of PS/Au core-shell
particles. ..........................48
Figure 4.2: Schematic (a) and experimental setup (b) for
self-assembly of monolayers.
Diluted PS particle suspension is slowly injected to the
air-water interface to form a
closely packed monolayer, which is picked up by a substrate
after draining. ............50
Figure 4.3: Optical images of the fabricated large-scale PS
particle monolayers. Each
chip is around 1.5 cm × 1.5 cm. The samples exhibit different
vivid colors when
observed from different perspectives, indicating the particles
are in a closely packed
pattern. The sizes of the PS particles are 1 μm (left), 2.1 μm
(middle), and 2.8 μm
(right), respectively.
....................................................................................................51
Figure 4.4: Reflectance spectra of 500 nm (a) and 1 μm (b)
closely packed PS particle
monolayers. Colored solid curves are FTIR measurements, while
the black dashed
lines are simulation
results..........................................................................................51
Figure 4.5: Reflectance spectra of 2.1 μm closely packed PS
particle monolayers after
different O2 etching periods. The power is fixed at 60 watts.
....................................53
-
xi
Figure 4.6: PS/Au core-shell particles at different fabrication
stages. (a-b) Top view of a
monolayer of 1 μm (a) and 2.1 μm (b) PS beads treated by 30s
reactive ion etching.
(c-d) After gold deposition, the particles are flipped over by
PDMS stamping to have
their uncoated side facing up. (c) Top view of the transferred 1
μm monolayer. (d) A
second cycle of gold deposition leads to fully coated core-shell
particles. A
perspective view at 45° shows the non-spherical shape caused by
anisotropic etching.54
Figure 4.7: Individual 2.1 μm (a) and 2.8 μm (b) PS/Au
core-shell monomers on a gold
substrate. The orientations are random after releasing.
..............................................55
Figure 5.1: Schematic of the nanoantenna consisting of a gold
nanoparticle trimer. The
third particle (diameter D2 = 120 nm) and the dimer (diameter D1
= 100 nm) are
slightly different in size, forming an isosceles triangle. An
electric dipole (ED)
emitter is positioned at the center of the dimer and is oriented
along the y-axis. The
dimer gap size is fixed to be g1 = 5 nm, while the larger
particle can be displaced
along the x-axis. The displacement h is counted relative to the
equilibrium position
(top), where the monomer-dimer gaps (g2) are also 5 nm. By
moving the larger
particle closer to the dimer (h0,
bottom), the radiation direction is switched to −x or backward
direction. .................60
Figure 5.2: Results for the forward-radiation configuration. The
larger particle is moved
4 nm towards (+x direction, h = −4 nm) the dimer from the
equilibrium position. (a)
Spectra of the total radiated power (green) and of the ED (blue)
and MD (red)
amplitude, respectively. (b, c) Surface charge density (left)
and electric field
distribution in vector format (right) at (b) 845 nm and (c) 720
nm wavelengths,
where the magnetic and electric resonances take place
respectively. The minimum
side lobe level of −22 dB for forward radiation is observed at
857 nm, where the ED
and MD have approximately equal amplitude and are nearly in
phase. (d) 3D and (e)
2D radiation patterns at 857 nm.
.................................................................................63
Figure 5.3: Evolution of the amplitude (a) and phase (b) of the
dipole modes. The
numbers in the legend denote the displacement h of the larger
particle. ....................64
Figure 5.4: Results for the backward-radiation configuration.
The larger particle is
moved 4 nm away (−x direction) from the dimer counting from the
equilibrium
position. (a) Spectra of the radiated power (green) and of the
ED (blue) and MD (red)
amplitude, respectively. (b) Surface charge density (left) and
electric field
distribution (right) at 710 nm, where the ED and MD mode are
both on resonance. (c)
Same as (b) but at 780 nm, where the minimum side lobe level of
−4.2 dB for
backward radiation is achieved. (d) 3D and (e) 2D radiation
patterns at 780 nm. .....66
Figure 5.5: Minimum side lobe level that can be achieved with
different trimer
configurations (given by displacement h). Circled dots
correspond to the results
reported in Figures 5.2 and 5.4. Note that for different
configurations, the best
directionality occurs at different wavelengths.
...........................................................67
Figure 5.6: Switching of forward-/backward radiation at the same
wavelength with a
reconfigurable nanoantenna (a) or at different wavelengths with
a fixed structure (b).68
Figure 5.7: Spectra of the Purcell factor Fp (red), emission
enhancement f (blue), and
quantum yield η (green) for (a) forward-radiation configuration
when h = −4 nm, and
(b) backward-radiation configuration when h = 4 nm,
respectively. ..........................69
-
xii
Figure 5.8: Radiation patterns reconstructed from the ED-MD
interference model. (a)
Forward radiation at 857 nm (left, cf. Figure 5.2(e)). (b)
Backward radiation at 780
nm (right, cf. Figure 5.4(e)).
.......................................................................................71
Figure 5.9: Comparison of normalized radiation patterns under
different excitation
conditions. In non-ideal cases, the emitter is either shifted or
tilted. (a) Forward
radiation at 857 nm wavelength with h = −4 nm. (b) Backward
radiation at 780 nm
wavelength with h = 4 nm.
.........................................................................................72
Figure 6.1: (a) Schematic of a metasurface consisting of
subwavelength plasmonic
resonators, which function as an array of local sources with
relative phase and
polarizability difference. Manipulation of wavefront,
polarization and intensity of the
reflected/transmitted light can be achieved by designing the
optical response and
arrangement of the resonators. (b) Generation of orbital angular
momentum by a
metasurface. The left panel shows the SEM image of the
metasurface, which uses
plasmonic nanoantennas to realize 2π phase distribution in the
azimuthal angle
direction. The right panel is the measured spiral patterns
created by the interference
of the vortex beam and a co-propagating Gaussian beam. (b) is
adopted from Ref.
[97].
.............................................................................................................................76
Figure 6.2: Schematic of the reflective graphene metasurface
consisting of
subwavelength graphene ribbons on a dielectric/metal substrate.
(a) By properly
arranging the ribbon widths and choosing the material and
thickness of the spacing
layer, the resultant plasmonic resonance in each ribbon varies
and the reflected IR
light can be controlled in a prescribed manner. (b) The graphene
metasurface forms
an asymmetric Fabry-Perot resonator, which substantially
enhances the interaction
between light and graphene nanostructures on the top.
..............................................78
Figure 6.3: (a) Reflectivity and (b) relative phase of the
reflected light as the functions of
ribbon width (w) and dielectric layer thickness (s). (c)
Reflectivity and phase of the
reflected light with varying ribbon widths, while the thickness
of the dielectric layer
is fixed to be s = 4.37 μm. The inset shows in cross-section the
field distribution of
the plasmonic resonance at a graphene ribbon when w = 1.035 μm.
Hereafter unless
specified, the frequency of the incident light is 12.32 THz, the
periodicity is 3 μm,
the refractive index of the dielectric layer is 1.4, and the
Fermi energy of graphene is
0.64 eV.
.......................................................................................................................81
Figure 6.4: Demonstration of anomalous reflection with (a)
normal incidence, and (b)
oblique incidence at an angle of 15.0°. Hereafter in this
chapter, unless specified, the
incident field is subtracted from the total field for clarity,
and the amplitude of the
reflected field is normalized to that of the incidence. Top:
Diagrams of rays
indicating the incident and reflected light. Bottom: Electric
field distributions. ........83
Figure 6.5: Schematic of achieving reflective focusing in (a)
bulk optics and (b) by a
metasurface.
................................................................................................................84
Figure 6.6: Demonstration of a reflective focusing lens at three
different frequencies: (a)
12.32 THz, (b) 11.82 THz, and (c) 12.82 THz.
..........................................................85
Figure 6.7: A dynamically tunable reflective focusing lens by
varying the Fermi energy
of graphene. (a) Ef = 0.56 eV and (b) Ef = 0.48 eV.
...................................................85
Figure 6.8: Generation of Airy beams. (a) Ideal phase and
amplitude profile at the
emitting plane to generate an Airy beam. (b) Electric filed
intensity map of an ideal
-
xiii
Airy beam resulting from (a). (c) Simplified Airy function. (d)
An Airy beam
generated by a graphene metasurface following the profile in
(c). ............................87
Figure 6.9: Schematic of a graphene metamirror that strongly
reflects right circularly
polarized (RCP) light with preserved helicity but weakly
reflects left circularly
polarized (LCP) light, giving rise to large differential
absorption or circular
dichroism (CD). The inset illustrates the unit cell consisting
of a pair of “LI”-shaped
graphene patches. The periodicities in the x and y direction are
250 and 120 nm,
respectively. Other geometric parameters are: Lx = 65 nm, Ly =
100 nm, w = 50 nm,
and g = 50 nm. The CaF2 spacing layer thickness t = 1550 nm.
................................89
Figure 6.10: Helicity-selective reflection of the graphene
metamirror with the carrier
mobility μ = 10000 cm2/V/s (a) and μ = 7500 cm
2/V/s (b). The Fermi energy is fixed
to be 0.8 eV. At the resonance frequency, RCP light is largely
reflected with
preserved helicity (green curve), while LCP light is strongly
absorbed showing much
lower reflection (blue curve). The cross-polarization conversion
is weak and
identical for both polarizations (overlapping red solid and
black dashed lines). .......90
Figure 7.1: Magnetic field assisted assembly of linear chains
consisting of 2 μm
polystyrene particles in a ferrofluid. (Images taken by Rasam
Soheilian in the DAPS
Lab)
.............................................................................................................................95
Figure 7.2: Schematic of the magnetic-field assisted
self-assembly of programmable
metamolecules. (a) In bulk fluid, core-shell microparticles are
assembled into certain
geometries by changing the external magnetic fields. (b) By
using predefined
magnetic templates on a substrate, core-shell particles can be
settled by external
magnetic fields with a global order.
...........................................................................95
Figure 7.3: SEM images of fabricated nanorod pair array by
electron beam lithography,
aimed at testing the fabrication limit for potential templates
preparation. .................96
Figure 7.4: (a) Fully assembled cell for FTIR spectroscopy in
aqueous environment.
CaF2 slides and the spacer are sandwiched between the aluminum
nest plates. Core-
shell particle suspension is injected from one of the two ports.
The IR transparent
window (orange box) allows both observation and measurement. (b)
Comparison of
transmission through the cell filled by water (red) and by
ferrofluid EMG 705 (blue),
respectively.
................................................................................................................97
-
xiv
Acknowledgements
At this point of approaching the very end of my graduate study,
a lot of memorable
moments come out in my mind. During the past four and a half
years, I have received so
much invaluable help and support from my colleagues, friends,
and family.
First of all, I would like to thank my advisor, Prof. Yongmin
Liu, for choosing me as
his first student—this really saved me from a hard time
deviating from my dream
career—and for his patient guidance and unreserved support
throughout all these years.
The privilege of being the only audience of his lectures on the
fundamentals of
plasmonics and metamaterials could never happen again to the
later group members, and
either will I forget the wonderful discussions on the latest
literatures. The most important
thing I have learned from him is the ambition to keep chasing
tough but sweeping
research topics, even if they appear impossible at the very
beginning. The lessons on time
management and on establishing collaborations also influenced me
a lot.
I would also like to express my grateful thanks to all the
committee members: Prof.
Matteo Rinaldi, Prof. Randall Erb, and Prof. Nian. X. Sun for
their helpful guidance,
inspiring advice, and kind support. Before starting the PhD I
had no prior knowledge on
nanofabrication or chemistry. It was Matteo’s course that opened
the door of
nanofabrication to me and introduced me how everything works
out. In the collaborative
work on reconfigurable metamaterials, where I even needed to
start from learning using a
pipette, I received numerous suggestions from Randy on assembly,
surface chemistry,
magnetism and so forth. When there were any unplanned requests
on metal deposition,
Nian always kindly gave me permission to use the facilities in
his lab, even they already
-
xv
had a tight schedule. Without the support from any of you, it
would be impossible for me
to present most results in this dissertation.
Practicing research in the multi-disciplinary field is highly
collaborative. Here I
would give some special thanks to the colleagues who had
selflessly shared their time,
thoughts, and expertise to promote my research. I acknowledge Yu
Hui and Fangze Liu
for their supervision on SEM and ELB, and thank Zheng Liu for
spending so much time
with me polishing these skills. I greatly appreciate Liao Chen,
who helped me out with
almost all the troubles in the chemical synthesis, and Shuai
Chen, who assisted me to
overcome the initial hesitation in performing chemistry. It was
a great pleasure to work
with Rasam Soheilian on the project of magnetic self-assembly. I
have bothered him too
much while he was busy. Hope this collaboration could become
fruitful in the near future
and the friendship lasts a lifetime. I am also very grateful to
Qing Hu for her help on
fabrication, to Xuefei Tan for her training on FTIR, and to
Scott McNamara, David
McKee, and Sivasubramanian Somu for their continuous support on
using the facilities.
I thank the rest of the Liu group, including but not limited to
Zubin Li, Hui Jia, Zuojia
Wang, Feng Cheng, Anna Nardella, Xiaokang Bai, Wei Ma, Lin Li,
Zhong Huang, and
Han Gao. And there is even a longer list of external colleagues
and collaborators, such as
Ji Hao, Xinjun Wang, Zhenyun Qian, Ryan Sungho Kang, Ismail
Ebraheim, William
Fowle, Wentao Liang, Abdelkrim Benabbas, etc.
Last and most importantly, my deepest appreciation goes to my
parents. Their love
and understanding supported me not only these years but all my
life so far. This work is
dedicated to them.
-
1
Chapter 1
Introduction
Metamaterials are rationally engineered composite materials with
exotic properties
not easily obtained or altogether unavailable in nature [1].
Since the early 2000s,
metamaterials have emerged as a new frontier of science
involving physics, material
science, optics, engineering, nanotechnology and so forth,
mainly because they provide
unprecedented opportunities to control light-matter interactions
and light propagation in
unusual ways. The exceptional properties of metamaterials arise
from the elegant design
of subwavelength resonators or “meta-atoms” as well as judicious
arrangement of these
building blocks (see Figure 1.1). This is distinctly different
from natural materials, whose
properties are primarily determined by the chemical constituents
and bonds [2].
Figure 1.1: Schematics of some representative metamaterials. (a)
A lattice of metallic
wires to achieve negative permittivity at low-frequency. (b)
Metallic split ring resonators
(SRRs) to produce artificial magnetism. (c) Metallic helices as
building blocks for chiral
metamaterials.
Early proof-of-concept experiments of metamaterial were mainly
done in the
microwave regime [3]. The operating frequency was quickly pushed
higher from
-
2
terahertz (THz) to infrared (IR) and then to visible [4].
Because most metamaterials are
composed of metals, the plasmonic effect of metals plays an
important role in optical
metamaterials.
Plasmonics focuses on the unique properties and applications of
surface plasmon
polaritons (SPPs), the quasiparticles arising from the strong
interaction between light and
free electrons in metals [5-7]. At the interface between a
semi-infinite metal and a semi-
infinite dielectric, SPPs behave as a surface wave propagating
along the interface while
exponentially decaying into both sides, as illustrated in Figure
1.2.
Figure 1.2: (a) Illustration of SPPs and oscillation of surface
charges at a dielectric/metal
interface. (b) The typical electromagnetic field profile of
SPPs, reflecting their bound,
non-radiant nature.
The interplay between metamaterials and plasmonics has led to
tremendous efforts
devoted to the invention and applications of plasmonic
metamaterials in the optical
region [8, 9]. However, a few areas under this research topic
are not fully explored, and
some restrictions limiting its further development are yet to be
overcome. For instance,
infrared light has found widespread applications such as
biomedical sensing,
spectroscopy, and surveillance [10]; while on the other hand,
metals do not support
pronounced plasmonic resonances at mid-infrared (MIR) and
far-infrared (FIR)
frequencies as in the visible and near-infrared (NIR) region.
The lack of suitable
-
3
platforms that can provide strong light-matter interactions at
infrared wavelengths, as a
consequence, naturally leads to the quest of better alternatives
for infrared plasmonics
and metamaterials.
Another restriction is the tunability. Up to date, most
metamaterials and plasmonic
devices are in solid phase and have fixed functionalities once
they are fabricated. Since
the optical properties of metamaterials and plasmonic
nanostructures rely on both the
constituents and the way in which they are shaped and arranged,
introducing variables
into the system, either changeable material properties or
reconfigurable architectures, will
open up new opportunities for dynamically tunable metamaterials
and plasmonics [11].
The recent progress towards this target is exciting. Figure 1.3
summarizes representative
attempts that employ materials with tunable dielectric
properties. To this end, phase-
change materials are among the most promising candidates [12,
13]. It has been shown
that the refractive-index-changing phase transition of
chalcogenide glass (GST) can be
induced optically by trains of femtosecond laser pulses [13].
Hence, despite the
achievable refractive index is limited, it is possible to write
arbitrary patterns on the GST
film to perform functionality such as lensing or erase them,
both with light (Figure 1.3(a-
c)). Vanadium oxide (VO2) is also popular as a phase-change
material [14, 15]. Figure
1.3(d) and (e) shows a design of metamaterial absorber, which
utilizes electrical currents
to induce Joule heating and in turn the phase transition of the
VO2 film [14]. Because the
infrared reflection of the devices is sensitive to the index
change, applying different
currents allows turn-on/off and continuous contrast tuning of
the infrared image. Another
important class of materials that has tunable dielectric
properties is the semiconductor. In
particular, two-dimensional (2D) materials such as graphene can
be dynamically tuned on
-
4
Figure 1.3: Reconfigurable metamaterials and metadevices by
material property changes.
(a) Functional devices can be written in or erased from a
chalcogenide glass (GST)
phase-change film by trains of femtosecond laser pulses at
different conditions. (b) The
optical image of two Fresnel zone patterns written by light. (c)
The second Fresnel zone
pattern on the right is erased by light. (d) A hybrid
metamaterial consisting of a vanadium
oxide (VO2) film sandwiched by gold meshes and a ground plane.
Applying electrical
currents will induce the insulator-to-metal transition of VO2
and change the reflectance.
(e) Top left: Microscopic image of the metamaterial. Letters are
bare VO2 areas without
gold meshes. Rest panels in clockwise order: Infrared images at
2.67 μm wavelength with
0 A, 2.03 A, and 2.20 A current, respectively. (f) An infrared
light beam excites tunable
plasmon resonances across the graphene nanoribbons biased at
varying gate voltages. (g)
Plasmon-enhanced light-matter interactions improve detection
sensitivity of molecules
adsorbed on graphene. (a-c) are adopted from Ref. [13], (d, e)
are adopted from Ref. [14],
(f) is adopted from Ref. [17], and (g) is adopted from Ref.
[18], respectively.
-
5
their dielectric properties via electrostatic gating [16].
Figure 1.3(f) and (g) shows such
an example, in which graphene is patterned into nanoribbons to
provide a platform for
plasmon-enhanced MIR biosensing [17-19].
The other side of the solution to reconfigurable metamaterials
is altering the structure.
As the coupling between the constituent parts is critical in
determining the global optical
properties, introducing finite structural changes is sufficient
to induce notable variations
in the spectra and field distributions. In Figure 1.4(a)-(d),
two prototypes of gap-tuning
metamaterials are presented. The first one is fabricated on an
elastomeric stamp; thus
stretching the substrate will enlarge the gap size and shift the
resonance [20]. In the
second example, similar in-plane mechanical deformation can be
induced by applying
electrical voltages to the supporting strings [21]. The
associated attractive forces are in
the range of a few nanonewtons, providing a very fine control on
the delicate structures.
It will not be too ambitious to imagine that the ultimate
reconfigurable metamaterials
or metamolecules should possess the capability to actively
transform between distinct
structures under a regulated control. This reorganization of
architecture will significantly
extend the bounds of the material properties available so far.
The path to the Holy Grail is
rather rocky, and the attempts with self-assembly techniques
appear to be most promising.
Figure 1.4(e) and (f) illustrates two types of reconfigurable
chiral metamolecules enabled
by the DNA stapling [22], where upon the addition of
specifically designed DNA strands,
the relative position of the paired nanorods varies under
control [23, 24]. The resulting
states of the metamolecules correspond to different chirality,
which can be readily
distinguished in the spectra in Figure 1.4(g). Although the
structural changes are still
finite, topologically the metamolecules at each state are
different.
-
6
Figure 1.4: Reconfigurable metamaterials and metamolecules by
finite structural changes.
(a) Split ring resonator (SRR)-bar on a stretchable film.
Mechanical deformation of the
elastomeric substrate changes the gap size and thus the
coupling, which causes noticeable
resonance shift in (b). (c) A metamaterial consisting of
“meander near the wire” gold
nanostructures on silicon nitride strings. (d) Applying voltages
to the neighboring strings
induces an attractive electrostatic force in the plane, which
reduces the gap size, enabling
modulation of transmission and reflection. (e, f) Two gold
nanorods are hosted on a DNA
template to form a chiral metamolecule exhibiting tunable
chiroptical properties as in (g).
(e) The relative angle between the nanorods can be actively
controlled by adding certain
DNA strands to pair the locks on the sides of the template. (f)
Alternatively, the nanorods
are assembled perpendicular to each other, and the DNA locks
function as binding sites
that allow the rod on the top (yellow) to “walk” stepwise
between the two ends of the
template. (a, b) are adopted from Ref. [20], (c, d) are adopted
from Ref. [21], (e) is
adopted from Ref. [23], and (f, g) are adopted from Ref.
[24].
-
7
In this work, I explore the possible solutions for achieving
infrared reconfigurable
plasmonics and metamaterials. Simulations and experiments are
both conducted to study
the different aspects of this topic, from fundamentals to
applications, from near-infrared
to far-infrared, and from static material property tuning to
finite structure altering and
eventually towards the ultimate architecture programming. The
contents of this work are
organized in the following order:
The first two consecutive chapters, namely Chapters 2 and 3,
explore in depth the
optical properties of SiO2/Au core-shell particles and their
self-assembled clusters for the
application of mid-infrared resonators. In Chapter 2, based on
wet chemistry, high quality
core-shell particles are synthesized and the conditions leading
to different surface
morphologies are discussed. Self-assembly is then carried out to
obtain clusters of varied
configurations.
The optical characterization of the core-shell particles and
associated numerical study
are presented in Chapter 3. Infrared spectroscopy is performed
to measure the optical
properties of the particles at the single cluster level. It is
found that depending on the
number and configuration of the constituent particles, each
cluster exhibits a unique
spectrum line shape at the infrared, especially mid-infrared
wavelengths. Full-wave
simulations are conducted to reproduce the measured spectra. In
addition to the excellent
agreement, clear correspondence is established between the
spectral features and various
resonance modes. The systematic study in this chapter
undoubtedly shows that plasmonic
core-shell particle can be tailored to achieve tunable
resonances.
Chapter 4 discusses an alternative method to make core-shell
particles. Based on
microfabrication techniques, polystyrene/Au core-shell particles
are produced with a high
-
8
yield. Self-assembled monolayers (SAMs) of polystyrene (PS)
particles are realized first
as a template. Plasma etching is applied to reshape the
particles. Mediated by a transfer
process, the dielectric cores are coated by two cycles of
deposition, resulting in another
type of designer infrared resonators.
In Chapter 5, an ultra-compact plasmonic nanoantenna with
switchable directionality
is proposed and systematically studied by simulations and
analyses. The antenna
comprises an asymmetric gold nanoparticle trimer and is fed by a
nano emitter in the gap.
When the so-called “Kerker conditions” are satisfied, the
antenna shows superior
radiation directionality in the near-infrared region. Moreover,
by displacing one of the
three particles for less than 10 nm, the radiated light beam can
be switched to the
opposite direction. Analyses based on a dipole interference
mdoel are performed to
understand the evolution of the radiation patterns. Techniques
for potential experimental
demonstrations are also discussed.
In Chapter 6, several designs of tunable flat lenses based on
graphene metasurfaces
are proposed to steer far-infrared light. Practical
functionalities, such as anomalous
reflection, reflective focusing, generation of non-diffracting
beams and strong chiroptical
effects are demonstrated by full-wave simulations. Because the
dielectric properties of
graphene can be tuned via chemical doping or electrostatic
gating dynamically, the
possibility of tuning the focal length of a reflective metalens
is also numerically proven
by changing the Fermi energy. These results expand the
applications of graphene-based
devices towards complete beam shaping in the infrared
region.
-
9
Finally, following a brief summary, an outlook and some
preliminary results on the
ongoing study of programmable plasmonic metamolecules via
magnetic self-assembly
and their sensing applications are presented in Chapter 7.
-
10
Chapter 2
Plasmonic Core-Shell Particles by Chemical
Synthesis
2.1 Introduction
The use of subwavelength structures to achieve exotic control of
electromagnetic
fields has stimulated the blossom of metamaterials, plasmonics,
and nanophotonics over
the past decades [9]. The cornerstone of these fields lies on
the fact that micro- or nano-
scale building blocks can resonantly interact with light waves
in a predefined manner by
programmatically designing their structures and arrangements.
Metallic structures play a
crucial role in building resonators operating in the visible and
NIR region, predominantly
because of their capability to support SPPs. The collective
oscillation of free electrons in
metals can couple efficiently with light, which ensures strong
light-matter interactions
necessary for further construction and engineering of
metamaterials. However, using
metallic elements as building blocks for metamaterials is
accompanied by several
limitations, among which the most influential one is the loss.
Ohmic losses in metals exist
across the entire visible spectrum and remain considerable in
the infrared region, whereas
the excitation efficiency of surface plasmons decreases rapidly
as the wavelengths
deviating from the visible. As a consequence, metals are not
ideal platforms for NIR
plasmonic devices and are seldom used to implement resonators at
MIR wavelengths.
-
11
With the urgent quest of infrared optical applications, suitable
building blocks for
infrared optics are in a pressing need [25, 26].
Recently, high-refractive-index semiconductors such as silicon,
germanium, and lead
telluride are proposed as alternative platforms for infrared
metamaterials and photonic
devices [27-31]. As shown in Figure 2.1(a), due to the large
refractive index and low-loss
properties, individual subwavelength particles can interact with
infrared light strongly via
a series of Mie resonances. In addition, the opportunity of
tuning the dielectric constants
by doping [30] or by thermal effects [31] makes these materials
very attractive compared
with conventional dielectrics. However, the fabrication of
regular and uniform micro- or
nano-particles is challenging for high index semiconductors.
Laser ablation can create
small amounts of particles from various substrates but lacks
control of the particle size
and roundedness (see insets of Figure 2.1(a) and (b)) [31-33].
Chemically synthesized
amorphous silicon particles show improved size uniformity, but
they are still far from
being identical, hindering the acquisition of strictly symmetric
clusters [34]. Furthermore,
compared with their metallic counterparts, dielectric particles
generate less intense local
fields and possess weaker coupling when they are tightly packed
[35]. The former issue
limits the light-matter interaction strength, which is critical
in numerous applications such
as sensing and spectroscopy; the latter restricts the spectral
tunability of the cluster,
because introducing additional particles into an existing
cluster will not cause significant
changes to its optical properties [33]. In Figure 2.1(b),
although the number, size, and
configuration of the particles vary a lot, the reflection curves
only show limited changes
in the line shape. Therefore, the task of seeking suitable
building blocks for infrared
-
12
plasmonics and metamaterials with notable spectral tunability is
still unsolved by using
dielectrics alone.
Figure 2.1: Optical resonators at different wavelengths. (a) A
high-index dielectric (lead
telluride (PbTe)) particle can support a series of Mie resonance
modes in the mid infrared
(MIR) region. (b) High-index dielectric (silicon) particle
clusters exhibit limited spectral
tunability at visible frequencies. (c) Plasmonic core-shell
nanoparticle clusters show
complex spectral features in the near infrared (NIR) region.
Figures are adopted from
Refs. [31], [35], and [39], respectively.
Core-shell particles consisting of a dielectric core and a
metallic shell offer a
promising solution. Such core-shell structures can support
sophisticated, hybridized
plasmonic modes as well as cavity modes [36-38]. When they are
closely packed into
clusters, strong interparticle coupling arises, resulting in
complex near-field distribution
and far-field scattering patterns [39, 40]. It has been
demonstrated that magnetic
resonance and Fano resonance can be excited in core-shell
nanoparticle clusters at large
incidence angles (Figure 2.1(c)). However, the sizes of
nanoparticles determine that these
resonances occur only in the NIR region, and cavity modes can
hardly be excited or
-
13
utilized. In this chapter, the chemical synthesis of
micron-sized plasmonic core-shell
SiO2/Au particles (SiO2/Au denotes silica core with gold shell)
is discussed in detail. The
resulting particles are uniformly coated and almost identical,
enabling the preparation of
highly symmetric core-shell clusters by subsequent self-assembly
process. The infrared
optical characterization of individual clusters and the
theoretical study on their optical
properties will be discussed in the next chapter.
2.2 Synthesis of Plasmonic Core-Shell Particles
The chemical synthesis of SiO2/Au core-shell particles involves
a series of steps. The
general flow is illustrated in Figure 2.2. For all the
experiments, the chemicals are
purchased from Sigma-Aldrich without further modification.
Although silica nano- and
micro-spheres can also be synthesized in the lab following the
standard procedure
developed by Stöber et al. [41], commercially available products
are used for better size
and shape control. Because the main focus of this work is to
explore the optical properties
of plasmonic core-shell particles as infrared resonators, the
associated silica core particles
are chosen to be 1 and 2 μm in size. Smaller particles resonate
in the visible region rather
than in the infrared, whereas larger particles may not fulfill
the requirement for potential
applications for infrared metamaterials. Other necessary
chemicals include: chloroauric
acid (gold(III) chloride or HAuCl4·3H2O, 99%),
tetrakis(hydroxymethyl)phosphonium
chloride (THPC, 80% aqueous solution), sodium hydroxide (NaOH),
potassium
carbonate (K2CO3), sodium chloride (NaCl),
3-aminopropylthiethoxysilane (APTES,
98%), formaldehyde (H2CO, 37%), and 200 proof ethanol. This
protocol is adapted from
an earlier work by Brinson et al. [42, 43]. Mainly caused by the
large sizes of the
particles used in this study, the amount of reactants in many
key steps needs to be
-
14
modified, which nevertheless does not follow the scaling law
simply. Meanwhile, how
the way of introducing gold ions and reducing agent will affect
the shell completeness
and surface morphology is also investigated, which appears less
sensitive for small
particles.
Figure 2.2: Flow chart of the synthesis of SiO2/Au core-shell
particles. (a) Synthesis of
gold nanoparticle seeds (i) and modification of SiO2 particles
(ii). (b) Preparation of
precursor particles, where gold nanoparticles are attached to
the surface of the modified
silica beads. (c) Continuous reduction of gold ions in the
planting solution causes the
nanoparticles to increase in size, coalesce, and eventually form
complete shells covering
the silica cores.
During a single synthesis cycle, aging takes quite a large
portion of time. Therefore,
despite several sections and associated preparations are
parallel in the flow chart, careful
considerations in arranging the sequence of experiments can be
very helpful to shorten
the synthesis period. In the following, the experimental details
will be introduced in an
optimized order by assuming all the chemicals are in the
as-arrived state.
First, 1 g of gold(III) chloride powder is dissolved in
deionized water (referred to DI
water or simply water herein) to produce a 1% (by weight)
chloroauric acid solution. The
-
15
dissolving and subsequent storage should both be in a glass
container, and the final
solution is stored in the dark for at least 48 hours before
use.
Second, the silica particles are functionalized with APTES.
After diluting the as-
purchased suspension to 1 wt%, the treatment starts by washing
the particles thoroughly
three times via centrifugation for 30 minutes, complete removal
of the supernatant, and
redispersion in water under strong sonication. These operations
are intended to detach
undesired stabilization groups on the particle surface.
Following the last washing cycle,
the particles are redispersed in ethanol and washed for
additional three times. After that,
the particles are resuspended in ethanol, placed in a
polypropylene bottle and stirred
rapidly. The volume of ethanol is equal to the diluted 1 wt%
aqueous suspension before
washing (4 mL in my experiments). The modification is carried
out in a glovebox filled
with nitrogen. Fresh APTES is added to the suspension with
continuous stirring for 10
minutes. The amount of APTES can be calculated by:
30[μL] / 4[mL] 126[nm]/ ,APTESV V D (2.1)
where V is the volume of suspension and D is the diameter of the
particles. The amount
of APTES from Equation (2.1) is about five times the amount
needed to form a silane
monolayer (~0.7 nm thick) covalently attached to the silica
surface. The actual thickness
of the silane film varies from 1 to 3 nm, corresponding to
several monolayers. However,
it is also noticed that in my experiments, successful synthesis
of high quality core-shells
is achieved even with wild excess of APTES. The impact of the
silane film thickness on
the properties of core-shell particles may need further study.
The suspension is sealed and
stirred slowly overnight. Subsequent operations can be done in
the open lab environment.
The suspension is boiled in a borosilicate container for 2
hours. Total evaporation is
-
16
prevented by continuous addition of ethanol. Care should be
taken to prevent bumping.
The particles are then washed three times in ethanol by
centrifugation, followed by
redispersion in a proper volume of ethanol (2 mL in my
experiments). The final
suspension is stored at 4 °C and has a long shelf life of one
year or longer. After
functionalization, the surface of silica particles is coated by
amino groups, which allows
attachment of gold nanoparticles to be prepared in the next
step.
Gold nanoparticles are synthesized following the procedures
developed by Duff et al.
[44]. 1.2 mL of THPC is diluted to 100 mL, and 1 M NaOH solution
is prepared by
dissolving a NaOH pellet in a certain amount of water. Under
rapid stirring to a strong
vortex, 1.2 mL of the 1 M NaOH solution is added into 180 mL of
water in a glass beaker,
Figure 2.3: Typical ultraviolet-visible (UV-Vis) extinction
spectra of the THPC gold
nanoparticle solution. Different curves are collected from
different batches of samples.
All of them exhibit a vague bump close to 500 nm wavelength,
indicating the sizes of the
synthesized nanoparticles are 1-3 nm. Inset shows the image of
an aged solution, which is
dark brown in color.
followed by addition of 4 mL of diluted THPC. After 5 minutes of
continuous stirring,
6.795 mL of 1% HAuCl4 solution is poured into the beaker in a
quick motion. Within a
second, the solution turns from colorless to medium brown,
indicating preliminary
-
17
success of the reaction. After 10 minutes of stirring, the
resulting solution is aged at 4 °C
in the dark for 14 days, during which its color changes to dark
brown gradually. Because
the optical extinction is very sensitive to the nanoparticle
size, any deviation of color
from above may signify potential failure, which will cause quick
settlement of the
precursor particles in the later step. Figure 2.3 compares the
ultraviolet-visible (UV-Vis)
extinction spectra from three batches of samples. As can be
seen, although the curves
offset slightly from each other, they follow the same trend
consistently and exhibit a
weak plasmon feature: a broad and vague bump close to 500 nm
wavelength. According
to transmission electron microscopy (TEM) inspection, the mean
size of the gold
nanoparticles is between 1 and 3 nm. Larger particles giving
stronger plasmonic
responses will cause a red-shifted and more distinct extinction
peak, rendering the
solution purpler in color. The THPC gold solution has a shelf
life of several months and
can be used reliably at least four months after synthesis
according to my experiments.
Gold planting solution is prepared next by adding 300 μL of 1
wt% HAuCl4 solution
into 20 mL of 1.8 mM K2CO3 aqueous solution. The latter is
prepared by dissolving 50
mg of K2CO3 powders in 200 mL of water. The final planting
solution has a pale yellow
color and should be aged at 4 °C in the dark for 2-3 days before
use, while its shelf life
has not been tested.
Once the THPC gold solution is ready for use (sufficiently
aged), the preparation of
precursor particles can be performed. This is the last step
before the final core-shell
formation and is the trickiest one. The precursor particles
consist of a functionalized
silica core and uniformly distributed gold nanoparticles
attached on its surface. The
attachment is naturally driven by the electrostatic force, since
the amino groups on silica
-
18
are positively charged whereas the gold nanoparticles are
negatively charged. On the
other hand, the Coulombic repulsion between nanoparticles will
limit the achievable
surface coverage. Radloff compared two methods to increase the
ionic strength of the
solution and thereby decrease the repulsive force for better
coverage [45]. I adopt the
more convenient one in my experiments, which requires only
addition of an inert salt
(NaCl) solution. To do this, 1 M NaCl aqueous solution is
prepared by dissolving 0.584 g
NaCl powder in 10 mL of water. Then, 40 mL of aged THPC gold
solution is sonicated
for 1 minute, followed by addition of 4 mL of the NaCl solution.
300 μL of the
functionalized silica particle ethanol suspension is added
immediately after that. The final
solution is stored in the dark after additional sonication for
2-3 minutes. Two findings
different from or not mentioned in previous literatures are
found in my experiments. First,
although highly concentrated NaCl can effectively increase the
ionic strength of the
solution, a threshold seems to exist for a certain size of the
silica beads. According to my
trials, 2 M NaCl already has a high risk to cause aggregation
and settlement of the 2 μm
particles in a few hours. Therefore, 1 M concentration is used
for all my experiments.
Second, the reliable period for the solution to reach
equilibrium is about 72 hours, much
longer than the reported values (12-20 hours). This is probably
because of the larger
particle sizes. The overall surface coverage of the silica
particles is estimated to be
around 25-30% according to literature. However, direct evidence
from TEM is not
practical due to the large particle size. The resulting
precursor particles are washed 2-3
times in water and then redispersed in 5 mL of water. The
solution is stored at 4 °C and
has a shelf life of about 3 months. Beyond this period
aggregation may occur.
-
19
Figure 2.4: Scanning Electron Microscope (SEM) images of a 1 μm
(a) and a 2 μm (b)
core-shell particle (monomer) on a gold-coated substrate. The
shell thicknesses are
estimated to be around 30 and 10 nm, respectively.
The final step of synthesis is the core-shell formation. With
the help of a reducing
agent, the attached THPC gold nanoparticles function as “seeds”
or nucleation sites, upon
which the gold ions in the planting solution are continuously
reduced. Depending on the
ratio of gold ion to precursor particle, core-shells with
different surface morphologies can
be readily synthesized under control. The first example is
plasmonic core-shell particles
with complete coating, as shown in Figure 2.4. Typically, 200 μL
of precursor particle
suspension is placed in a glass vial, followed by addition of 1
mL of planting solution
under rapid stirring. 2.2 μL of formaldehyde is then added into
the mixture in a quick
motion. With continuous stirring, the color of the solution
turns from light brown
gradually to light purple or pink within 10-15 minutes. The
ultimate color does not
correspond to the plasmonic response of the micron-sized
core-shell particles but is
determined by that of the excess THPC gold nanoparticles, which
are unavoidable in the
precursor particle solution. Upon successful synthesis, the
silica particles are uniformly
coated by a gold shell, while the surface is quite smooth,
showing only occasional cracks
such as pinhole defects and attached gold nanoparticles. From
high magnification
-
20
scanning electron microscope (SEM) images (e.g. Figure 2.5(a)),
it is clearly seen that
these imperfections are not larger than 40 nm, much smaller than
the particles and the
infrared wavelengths. Therefore, their influence on the optical
properties will be
negligible.
Figure 2.5: Zoom-in SEM images for shell inspection. (a) The
surface of a 2 μm particle
coated by a complete and smooth gold shell. Occasional cracks
such as pinhole defects
and excess gold nanoparticles are smaller than 40 nm and have
negligible impact on the
infrared optical properties of the core-shell particles. (b)
Perspective view (~70°) of a 2
μm core-shell particle at a rectangular cut on its surface. The
cut is made by focused ion
beam (FIB). Measurements at different locations on the cutting
edges suggest a shell
thickness of ~10 nm.
-
21
The determination of the exact shell thickness, however, becomes
difficult. Typically
for nanoparticles that allow high energy electrons to penetrate
through, transmission
electron microscopy (TEM) is a standard technique for shell
inspection. In the present
study, the large particle size impedes the use of TEM while the
tiny pinhole defects can
only provide a rough estimate of the shell thickness. In Figure
2.5(b), by making a
through cut on the surface by focused ion beam (FIB) milling,
the shell thickness is
measured at different locations of the cutting edge. Although
the poor conductivity of the
silica core hinders a sharp imaging contrast at the gold-silica
interface, the measurements
give direct evidence of the shell thickness, which is about 10
nm.
The way how gold ions are introduced into the reaction is
critical in determining the
surface morphology. As the second example, it is found that if
the planting solution is
added by a small amount at one time and for multiple times, the
attached THPC gold
nanoparticles will only increase in size but cannot coalesce
into a complete shell. Figure
2.6(a) and (c) shows the SEM images of a 1 μm and a 2 μm
particle, respectively.
Compared with the protocol used to obtain core-shells in Figure
2.4, the volumes of the
precursor particle suspension and of the planting solution are
reversed. The reaction is
conducted for 5 cycles, and particles are washed carefully after
each cycle to remove
excess formaldehyde. From the SEM images, it can be seen that in
both cases, the growth
of gold nanoparticles are obvious. However, despite the high
surface coverage, the grains
are isolated and do not form a continuous monolayer. When the
same process is repeated
for 12 cycles, interestingly, the grain size increases
significantly, as shown in Figure
2.6(c), but the shell is still not compete, exhibiting
widespread discontinuities and large
roughness. Up to 20 cycles of reduction have been tested.
Nevertheless, the particles
-
22
aggregate easily during later cycles of reaction, frustrating
further attempts on shell
formation. In comparison, in Figure 2.6(d), when a large amount
of planting solution is
added to react with the particles shown in Figure 2.6(c), a
complete shell is obtained,
although the surface smoothness degrades slightly. Therefore, it
can be concluded that the
amount of gold ions introduced to a single reaction is critical
to the shell formation.
Figure 2.6: (a,c) SEM images of a 1 μm (a) and a 2 μm (c)
particle after 5 cycles of gold
reduction with a small amount of planting solution. The gold
nanoparticles increase in
size (~10 nm) but are isolated. (b) Zoom-in image of a 1 μm
particle after 12 cycles of
reduction. The grain size increases to 20 nm and larger, whereas
the nanoparticles still do
not coalesce into a complete shell. (d) Zoom-in images of a 2 μm
particle after 5 cycles of
reduction with a small amount of planting solution and an
additional cycle with a large
amount (5 times in volume) of planting solution. A complete
shell is formed whereas the
smoothness degrades slightly.
The concentration of gold ions also plays an important role. In
a separate trial, a
condensed planting solution is used. To make a fair comparison
to the result in Figure
2.6(a), the concentration is increased by 5 times while the
volume of planting solution is
-
23
unchanged (200 μL), and the reduction is conducted for only one
cycle. The image of
resulting particles is shown in Figure 2.7, where the morphology
further differs from
previous results. The increase of gold ion concentration causes
abrupt growth of the seed
nanoparticles but does not promote the coalescence. The grains
are clearly isolated,
forming satellite nanostructures on the silica surface.
Figure 2.7: SEM image of a 1 μm particle after a single cycle of
reduction with a
condensed planting solution. The attached seed nanoparticles
increase significantly in
size but are clearly isolated.
In this section, the synthesis of SiO2/Au core-shell particles
is discussed in detail.
Three different morphologies can be obtained by changing the
ratio of the reactants and
by varying the concentration of the planting solution. Silica
particles coated by a
complete shell (Figure 2.4), a defective shell (Figure 2.6(d)),
a discontinuous shell
(Figure 2.6(a-c)), and satellite nanostructures (Figure 2.7)
have different optical
properties and applications, which will be addressed in later
chapters.
2.3 Assembly of Clusters
The self-assembly process is then conducted in a relatively
straightforward manner
[39, 46]. A droplet of diluted core-shell particle suspension is
pipetted on a gold-coated
-
24
substrate and dried slowly at room temperature. During total
evaporation, the particles are
assembled by the capillary force randomly into small clusters of
different planar
geometries. Figure 2.8 illustrates the schematic. Although any
insoluble substrate can be
used, gold-coated high reflective mirrors are favored in the
subsequent infrared optical
characterization. Polished silicon substrate, TEM grids,
infrared transparent windows
made of Calcium Fluoride (CaF2) are also possible options.
Figure 2.8: Self-assembly of core-shell particles. A droplet of
diluted particle suspension
is pipetted on a gold-coated substrate. During total
evaporation, the core-shell particles
are assembled by the capillary force into different planar
structures.
Figure 2.9: Self-assembled 1 μm core-shell clusters.
Representative geometries include
but are not limited to: (a) dimers, (b-c) trimers, and (d)
chains. The shell thicknesses are
estimated to be about 30 nm. Scale bars are 200 nm.
-
25
Figure 2.10: Self-assembled 2 μm core-shell clusters.
Representative configurations
include but are not limited to: (a) dimers, (b) trimers, (c)
quadrumers, (d) chains, and (e)
pentamers. The shell thicknesses are estimated to be about 10
nm. Scale bars are 500 nm.
Figures 2.9 and 2.10 show the SEM images of representative
clusters consisting of 1
and 2 μm core-shell particles, respectively. In addition to
single particles (monomers),
other regular configurations include dimers, trimers,
quadrumers, chains of different
lengths, and by chance pentamers, heptamers or larger ensembles.
The interparticle
spacing within a cluster is tiny compared with the particle
sizes, not larger than 30 nm
estimated from the SEM images. On the other hand, the assembled
clusters are relatively
isolated from each other. The typical distance between
neighboring clusters ranges from
10 to some hundreds micrometers. As a result, the scattering
from one cluster will not be
affected much by nearby particles, which allows us to
characterize the optical properties
of individual clusters.
During the assembly process there are two phenomena worthy of
attention. First, the
probability of forming simple and symmetric structures, such as
dimers, trimers, and
-
26
quadrumers, are much higher than that of forming large and
complex structures.
Meanwhile, it is found that large clusters tend to appear near
the edge of the droplet,
accompanied by more excess nanoparticles, as shown in Figure
2.10(d) and (e). This is a
consequence of the kinetics of evaporation. To make the clusters
cleaner, either thorough
washing after synthesis or moving the droplet during evaporation
could be helpful.
Figure 2.11: Two clusters consisting of particles with
incomplete coating. Some of the
attached nanoparticles drop off during the movement of the
particles, forming a trial on
the substrate. The scale bars are 500 nm in (a) and 1 μm in
(b).
The second phenomenon is about the bonding strength between the
silica core and the
gold shell. In Figure 2.6(a), (b) and (d), several defects sized
over 100 nm can be seen.
However, it is not very clear if they are caused by incomplete
functionalization—this
cannot be fixed in later synthesis steps—or by centrifugation.
The reason is distinguished
-
27
by a revised assembly process, during which the droplet is
gently blown to move a short
distance on the substrate. Figure 2.11 shows the resulting
clusters formed by 2 μm
particles with incomplete coating, which are synthesized by
using small amounts of
plating solution as discussed above. From the images,
unexpectedly, one can see dropped
nanoparticles from the silica surface. Moreover, the
distribution of these nanoparticles is
not random but follows clear trials, indicating that the
exfoliation occurs during the
movement of the silica cores. In an extreme case in Figure
2.11(b), numerous
nanoparticles drop off from a particle in a heptamer, likely
during rolling, leaving a belt-
like uncoated area on the silica surface. Similar defects are
only observed for particles
with incomplete coating or from multi-step reduction, whereas
for complete shells, no
visible damage is found after centrifugation, sonication, and
assembly. Defective shells
can diminish spectral features of the particles and clusters and
thus have negative impacts
on their optical properties. These results provide useful
information to avoid defects in
the synthesis and assembly processes.
-
28
Chapter 3
Self-Assembled Plasmonic Core-Shell Particles
as Infrared Resonators
The successful synthesis of high quality core-shell particles
and assembly of
symmetric clusters provide a set of targets that is ideal for
optical characterization.
Following the fabrication processes above, in this chapter, the
infrared optical properties
of individual core-shell particles and self-assembled clusters
are systematically studied by
both experiments and simulations. It is shown that core-shell
particle clusters exhibit
unique optical properties in the infrared, especially MIR
region. Depending on the
number and configuration of the constituent particles, the
reflection spectra from
individual ensembles show distinct line shapes. In other words,
the optical properties of
the clusters are sensitive to structural and geometric changes,
indicating a promising
toolbox for infrared plasmonics, metamaterials and associated
applications.
3.1 Bright-Field Reflection Measurements
Fourier transform infrared (FTIR) spectroscopy is employed to
characterize the
optical properties of individual clusters with different numbers
of particles and varying
configurations. The measurements are conducted using the Bruker
spectrometer Vertex
70 combined with a microscope Hyperion 1000. Figure 3.1
illustrates the schematic of
the setup for reflection measurement. Infrared light is
generated by a globar source (for
MIR) or a tungsten halogen lamp (for NIR). After passing through
a Michelson
-
29
interferometer, the light beam is focused on the sample surface
by a 36x Cassegrain
objective (NA = 0.5), which shapes the incident beam to be
donut-like in cross section,
corresponding to incident angles ranging from ~18 to 30°. In the
present study, the
reflected light is collected by the same objective and directed
to a Mercury-Cadmium-
Telluride (MCT) detector. In the case of transmission
measurement, two identical
Cassegrain objectives are needed. The incident light is launched
by one lens below the
sample stage, and the transmitted light is collected by the
other on the top. In order to
measure the signal from a single cluster, a knife-edge aperture
in the conjugate image
plane is carefully adjusted to define a rectangular area
containing exclusively the cluster
of interest. Depending on the signal-to-noise ratio (SNR) as
well as the number and
configuration of the particles in the clusters, the size, shape,
and orientation of the
aperture varies, roughly from 100 to 400 μm2. Focusing should be
carefully tuned to
maximize the signal intensity. For each measurement at MIR (NIR)
wavelengths, the
sample is scanned for at least 2500 times with a 4 cm-1
(8 cm-1
) spectral resolution and a
proper amplification. This scheme can effectively suppress
noises and improve SNR. A
reference measurement in a nearby sample-free area is carried
out immediately after each
sample measurement, while all the settings on focusing,
knife-edge aperture size, number
of scans and spectral resolution remain unchanged. Reflectance
is then calculated by
normalizing the sample spectrum to the reference spectrum. All
the measurements are
based on unpolarized illumination.
Uncoated silica particles and particles with incomplete coating
(see Figures 2.6(c) and
2.11) are first measured. Because bare silica particles will
aggregate into a monolayer
easily on the substrate, extremely low concentration is used to
obtain isolated monomers.
-
30
Figure 3.1: Schematic of the Fourier transform infrared (FTIR)
single cluster
spectroscopy. Infrared light is generated by a globar source.
After passing through a
Michelson interferometer, the incident beam is focused by a 36x
Cassegrain objective
lens on the sample. The reflected light is collected by the same
lens and directed to an
MCT detector. A knife-edge aperture is applied in the conjugate
image plane to define a
small area containing exclusively the individual clusters of
interest, represented by the
rectangle bounded by the red dashed lines.
Surprisingly, it is found that the infrared optical properties
of an uncoated silica particle
and of a particle with incomplete coating are nearly identical.
Although the gold
nanoparticles cover about half of the silica surface, their
sizes are around 10 nm, more
than 150 times smaller than the infrared wavelengths (1.5 μm and
longer). Therefore, the
existence of the non-coalesced gold nanoparticles on silica
beads will only change the
optical properties in the visible region but will not affect the
infrared optical response.
This allows one to investigate the infrared optical properties
of uncoated silica particle
clusters by measuring their counterparts with incomplete
coating, which are easy to
assemble. In the following, these two types of particles will
not be distinguished, and
both of them will be referred to uncoated silica particles.
-
31
Figure 3.2(a) compares the reflectance spectra of four different
2 μm silica bead
clusters on a gold substrate. With the number of constituent
particles increasing from 1
up to 7, the measured clusters include a monomer, a dimer, a
symmetric trimer, and a
symmetric heptamer. As one can see, in addition to the
absorption from carbon dioxide
(CO2) in air at 4.3 μm and from silica at 8-10 μm, a broad
reflection dip occurs at 1.5-4
μm, which is caused mainly by light scattering. However, instead
of exhibiting any
noticeable shifts, all the spectral features (peaks and dips)
line up very well in the vertical
direction and simply become more evident with the increasing
number of particles. This
is because the refractive index of silica is relatively low,
less than 1.5 below 8 μm and not
significantly greater than the surrounding medium (air). When
the particle size is
comparable to the wavelengths, no