-
QUEENSLAND UNIVERSITY OF TECHNOLOGY
SCHOOL OF PHYSICAL AND CHEMICAL SCIENCES
APPLIED OPTICS AND NANOTECHNOLOGY PROGRAM
Theoretical and Numerical Investigation of Plasmon Nanofocusing
in Metallic Tapered
Rods and Grooves
Submitted by Michael Werner Vogel, Dipl.-Ing. (FH), to the
School of Physical
and Chemical Sciences, Queensland University of Technology, in
partial
fulfilment of the requirements of the degree of Doctor of
Philosophy.
2009
-
ii
-
Key Words
Near field optics, nano-optics, plasmonics, surface plasmons,
localised
surface plasmons, film plasmons, gap plasmons, nanofocusing,
adiabatic
nanofocusing, non-adiabatic nanofocusing, local field
enhancement, metallic
V-groove, metal tapered rod.
iii
-
iv
-
Abstract
Effective focusing of electromagnetic (EM) energy to nanoscale
regions is
one of the major challenges in nano-photonics and plasmonics.
The strong
localization of the optical energy into regions much smaller
than allowed by
the diffraction limit, also called nanofocusing, offers
promising applications in
nano-sensor technology, nanofabrication, near-field optics or
spectroscopy.
One of the most promising solutions to the problem of efficient
nanofocusing
is related to surface plasmon propagation in metallic
structures. Metallic
tapered rods, commonly used as probes in near field microscopy
and
spectroscopy, are of a particular interest. They can provide
very strong EM
field enhancement at the tip due to surface plasmons (SP’s)
propagating
towards the tip of the tapered metal rod. A large number of
studies have
been devoted to the manufacturing process of tapered rods or
tapered fibers
coated by a metal film. On the other hand, structures such as
metallic V-
grooves or metal wedges can also provide strong electric field
enhancements
but manufacturing of these structures is still a challenge. It
has been shown,
however, that the attainable electric field enhancement at the
apex in the V-
groove is higher than at the tip of a metal tapered rod when the
dissipation
level in the metal is strong. Metallic V-grooves also have very
promising
characteristics as plasmonic waveguides.
This thesis will present a thorough theoretical and numerical
investigation
of nanofocusing during plasmon propagation along a metal tapered
rod and
into a metallic V-groove. Optimal structural parameters
including optimal
taper angle, taper length and shape of the taper are determined
in order to
achieve maximum field enhancement factors at the tip of the
nanofocusing
structure.
An analytical investigation of plasmon nanofocusing by metal
tapered rods
is carried out by means of the geometric optics approximation
(GOA), which
is also called adiabatic nanofocusing. However, GOA is
applicable only for
analysing tapered structures with small taper angles and without
considering
a terminating tip structure in order to neglect reflections.
Rigorous numerical
v
-
methods are employed for analysing non-adiabatic nanofocusing,
by tapered
rod and V-grooves with larger taper angles and with a rounded
tip. These
structures cannot be studied by analytical methods due to the
presence of
reflected waves from the taper section, the tip and also from
(artificial)
computational boundaries. A new method is introduced to combine
the
advantages of GOA and rigorous numerical methods in order to
reduce
significantly the use of computational resources and yet achieve
accurate
results for the analysis of large tapered structures, within
reasonable
calculation time.
Detailed comparison between GOA and rigorous numerical methods
will
be carried out in order to find the critical taper angle of the
tapered structures
at which GOA is still applicable. It will be demonstrated that
optimal taper
angles, at which maximum field enhancements occur, coincide with
the
critical angles, at which GOA is still applicable. It will be
shown that the
applicability of GOA can be substantially expanded to include
structures
which could be analysed previously by numerical methods
only.
The influence of the rounded tip, the taper angle and the role
of dissipation
onto the plasmon field distribution along the tapered rod and
near the tip will
be analysed analytically and numerically in detail. It will be
demonstrated that
electric field enhancement factors of up to ~ 2500 within
nanoscale regions
are predicted. These are sufficient, for instance, to detect
single molecules
using surface enhanced Raman spectroscopy (SERS) with the tip of
a
tapered rod, an approach also known as tip enhanced Raman
spectroscopy
or TERS.
The results obtained in this project will be important for
applications for
which strong local field enhancement factors are crucial for the
performance
of devices such as near field microscopes or spectroscopy. The
optimal
design of nanofocusing structures, at which the delivery of
electromagnetic
energy to the nanometer region is most efficient, will lead to
new applications
in near field sensors, near field measuring technology, or
generation of
nanometer sized energy sources. This includes: applications in
tip enhanced
Raman spectroscopy (TERS); manipulation of nanoparticles and
molecules;
vi
-
efficient coupling of optical energy into and out of plasmonic
circuits; second
harmonic generation in non-linear optics; or delivery of energy
to quantum
dots, for instance, for quantum computations.
vii
-
viii
-
Glossary of Terms and Abbreviations
ATR Attenuated Total Reflection
CPU Computer Processing Unit
DLM Drude-Lorentz Model
EELS Electron Energy Loss Spectrum
EIM Effective Index Method
EM Electromagnetic
FDTD Finite-Difference Time-Domain
FEM Finite Element Method
FIB Focused Ion Beam
FTIR Frustrated Total Internal Reflection
GDT Green’s dyadic technique
GOA Geometric Optics Approximation
IMI Insulator-Metal-Insulator
LRSPP Long Range Surface Plasmon-Polariton
LSP Localized Surface Plasmon
MIM Metal-Insulator-Metal
MoM Method of Moments
MOSFET Metal-Oxide Semiconductor Field-Effect Transistor
NFM Near Field Microscope
NSOM/SNOM Near-field Scanning Optical Microscope
PSTM Photon Scanning Tunnelling Microscope
SERS Surface Enhanced Raman Spectroscopy
SP Surface Plasmon
SPP Surface Plasmon Polaritons
SRR Split Ring Resonators
TE Transverse Electric
TEM Transmission Electron Microscope
TERS Tip Enhanced Raman Spectroscopy
TM Transverse Magnetic
ix
-
x
-
List of Publications and Manuscripts
Refereed publications
A1. Michael W. Vogel, Dmitri K. Gramotnev, “Adiabatic
Nano-focusing in metal tapered
rod in the presence of dissipation”, Physics Letters A, 363,
507-511 (2007)
A2. Dmitri K. Gramotnev, David F. P. Pile, Michael W. Vogel,
Xiang Zhang, “Local
Electric Field Enhancement during Nano-focusing of Plasmons by a
Tapered Gap”,
Physics Review B, 103, 034304 (2008).
A3. Dmitri K. Gramotnev, Michael W. Vogel, Mark Stockman,
“Optimized non-adiabatic
nano-focusing of plasmons by tapered metal rods”, Journal of
Applied Physics, 104, 034311(2008)
A4. Michael W. Vogel, Dmitri K. Gramotnev, “Optimization of
plasmon nano-focusing in
tapered metal rods”, Journal of Nanophotonics, Vol. 2, 021852
(21. Nov 2008)
A5. Michael W. Vogel, Dmitri K. Gramotnev, “Shape effects on
adiabatic nano-focusing
of localized plasmons in tapered metal rods”, to be submitted to
Journal of Applied
Physics
Conference publications
1. Michael W. Vogel, Dmitri K Gramotnev. “Does Adiabatic
Nano-Focusing in Metallic
Tips Really Exist?” Australasian Conf. on Optics, Lasers and
Spectroscopy
(ACOLS), Rotorua, New Zealand (2005), Poster.
2. Michael W. Vogel, Dmitri K. Gramotnev. “Adiabatic
Nano-Focusing in Metallic Nano-
Structures in the Presence of Dissipation”, ETOPIM 7, Sydney,
Australia (2006).
Oral Presentation.
3. Michael W. Vogel, Dmitri K. Gramotnev. “Analyses of adiabatic
nano focusing in
metal tips and nano-holes in the presence of dissipation.”, 17th
Australian Institute of
Physics Congress – AOS, Brisbane, Australia (2006).
4. Michael W. Vogel, Dmitri K. Gramotnev, " Adiabatic
Nano-Focusing in Metal Tips in
the Presence of Dissipation ", Surface Plasmon Photonics 3,
p.168, Dijon, France
(2007).
5. Michael W. Vogel, Dmitri K. Gramotnev. “Adiabatic
Nano-Focusing in Metal Tips with
different profiles”, Surface Plasmon Photonics 3, p.168, Dijon,
France (2007).
xi
-
6. Michael W. Vogel, Dmitri K. Gramotnev. “Non-Adiabatic
Nano-Focusing in Metallic
Rods”, Oral Presentation, Surface Plasmon Photonics 3, p.168,
Dijon, France
(2007).
7. Michael W. Vogel, Dmitri K. Gramotnev. “Excitation of Surface
Plasmons in Metal
Tapered Rods using Hollow Waveguides”, 18th Australian Institute
of Physics
Congress – AOS, Adelaide, Australia (2008).
8. Michael W. Vogel, Dmitri K. Gramotnev. “Adiabatic
Nano-Focusing in Metal Tips with
different profiles”, 18th Australian Institute of Physics
Congress – AOS, Adelaide,
Australia (2008)
Invited Talks (Other than Conference)
1. Michael W. Vogel. “Plasmonics Research at QUT”, invited talk
at the Max Planck
Institute for Polymer Research, Mainz, Germany (June 26,
2007).
xii
-
Statement of Original Authorship
The research contained in this thesis has not been previously
submitted to
meet requirements for an award at this or any other higher
education
institution. To the best of my knowledge and belief, the thesis
contains no
material previously published or written by another person
except where due
reference is made.
Signature
Date
xiii
-
xiv
-
Acknowledgement
Writing a PhD thesis is a very solitary process, I am sure that
everybody who
went through such an ordeal understands what I am talking about.
On the
other hand, it is the most gratifying moment when you have
finished your
work. I would never have come so far without the love and
the
encouragement I got (and still get) from Judith.
My gratitude also goes to my supervisors, Dmitri (Благодари
многих) and
Esa (Kiitos paljon!) for their support.
Without the support from the high performance computational
group (HPC)
this project would not have been possible, thanks a lot to Mark
Berry and the
whole team who showed me how to reduce the runtime of my
MATLAB(C)
program from 35 hours to 1.5 micro seconds.
Finally I would like to thank Birgit Alves-Stein and Tanya
Cairns who were
forced to proof read my thesis and compelled to give me valuable
comments.
xv
http://mexiko.pauker.at/pauker/DE_DE/FI/wb/?x=Kiitoshttp://mexiko.pauker.at/pauker/DE_DE/FI/wb/?x=paljon
-
xvi
-
Table of Contents
Key Words
.....................................................................................................
iii
Abstract
...........................................................................................................
v
Glossary of Terms and Abbreviations
............................................................
ix
List of Publications and Manuscripts
..............................................................
xi
Statement of Original Authorship
.................................................................
xiii
Acknowledgement
.........................................................................................
xv
Table of Contents
........................................................................................
xvii
Table of Figures
...........................................................................................
xix
List of Tables
...............................................................................................
xxiii
1. Introduction
..........................................................................................
1
1.1. Research Problem
.....................................................................
1 1.2. Research Background
...............................................................
8 1.3. Aim of the project and linking the papers
................................. 11
2. Theory and Background information
................................................ 17
2.1. Introduction
..............................................................................
17 2.2. Maxwell’s Equations and Wave Equation
................................ 21 2.3. Constitutional
Parameters .......................................................
23
2.3.1. Drude-Lorentz model
..................................................
25 2.3.2. Non-local Effects and Loss
Mechanism ...................... 26
2.4. Surface Plasmons on a Flat Surface
....................................... 28
2.5. Surface plasmons in Three Layer System
............................... 33
2.5.1. MIM-Structure (Metal Gap)
.......................................... 33 2.5.2.
IMI-Structure (Metal Film)
............................................ 36
2.6. Excitation of Surface Plasmons
............................................... 41
2.6.1. Excitation by Electrons
................................................
41 2.6.2. Excitation by Photons
.................................................. 42
2.6.2.1. Otto and Kretschmann Configuration ................
42 2.6.2.2. Grating and Surface Structures
......................... 44 2.6.2.3. Tip of Near Field
Microscopes ........................... 45
xvii
-
2.7. Detection of Surface Plasmons by Near Field
Microscopes .... 47
2.8. Surface Plasmons on Cylindrical Surfaces
............................. 53
2.8.1. Field Structure and Dispersion Relation
...................... 57 2.8.2. Analyses of
the Dispersion Relation ........................... 63
2.9. Nanofocusing in Tapered Rods and Grooves
......................... 66
2.9.1. Methods of Analysis
.................................................... 70
2.9.1.1. Staircase Approximation
................................ 70 2.9.1.2. Geometric Optics
Approximation ................... 72 2.9.1.3. Intensity
Distribution with GOA ...................... 72 2.9.1.4.
Numerical methods ........................................
75
3. Adiabatic Nano-focusing of Plasmons by Metallic Tapered
Rods in the Presence of Dissipation
............................................................
79
4. Local Electric Field Enhancement during Nanofocusing of
Plasmons by a Tapered Gap
...........................................................
87
5. Shape Effects on Adiabatic Nano-focusing of Localised
Plasmons in Tapered Metal Rods
.....................................................................
95
6. Optimized Nonadiabatic Nanofocusing of Plasmons by
Tapered Metal Rods
......................................................................................
129
7. Optimisation of Plasmon Nano-focusing in Tapered Metal
Rods 139
8.
Conclusion........................................................................................
159
Appendix A: Derivation Dispersion Relation: Flat Metal Surface
................ 169
Appendix B: Determination of Incident Wave from Fabry-Perot
Pattern ..... 171
Appendix C: Optimal Length of Tapered Rod
............................................. 173
Appendix D: Determination Intensity Distribution in GOA
........................... 177
References
.................................................................................................
179
xviii
-
Table of Figures
Figure 2.1: 3D-model of a surface plasmon propagating along a
flat metal
surface in the x-direction. Schematically a snap shot of the Hy
distribution
(TM-mode) is shown. The relative permittivity εd are for the
dielectric material
and εm for the metal, respectively. The evanescent waves in
y-direction are
indicated by the dash-dotted line.
................................................................
29
Figure 2.2: Propagation length for SP’s on a flat surface for
gold, silver and
aluminium. The permittivities of Au, Ag and Al were taken from
Palik [129]. 31
Figure 2.3: Dispersion relation of SP’s on a flat Ag surface,
with the
permittivity of Ag modelled by DLM, λp is the corresponding
plasmon
wavelength and εd = 1 is the permittivity of the adjacent
dielectric material.
See text for description of the curves.
..........................................................
32
Figure 2.4: MIM structure with (a) anti-symmetric and (b)
symmetric
magnetic field distribution with respect to the middle planeThe
field
distribution is also indicated by the dash-dotted black line at
an arbitrary
cross section.The metal gap width h is assumed to be 500 nm.
.................. 34
Figure 2.5: Dispersion relation qx(h), h being the gap width, of
a MIM (Ag-
Vacuum-Ag) plasmon waveguide for symmetric magnetic field Hy
(anti-
symmetric surface charge distribution) and anti-symmetric
magnetic field Hy
(symmetric charge distribution). The dash-dotted line indicates
the threshold
wave number; below the value qx / k0 = 1 no gap plasmons exist.
............... 36
Figure 2.6: 3D image of SP’s propagating along a metal film (IMI
structure)
with (a) anti-symmetric and (b) symmetric magnetic field
distribution with
respect to the middle plane. The field distribution is also
indicated by the
dash-dotted black line at an arbitrary cross section. The grey
box indicates
the metal film which is aligned along the xy-plane.
...................................... 37
Figure 2.7: Dispersion relation qx(h), h being the film
thickness, of a IMI
(Vacuum-Ag-Vacuum) plasmon waveguide for symmetric magnetic
field Hy
xix
-
(anti-symmetric surface charge distribution) and anti-symmetric
magnetic
field Hy (symmetric charge distribution).
.......................................................
38
Figure 2.8: Otto-Raether configuration: A laser beam is coupled
into a prism
at the critical angle θc. Due to total internal reflection an
evanescent wave
penetrates the metal surface and triggers surface plasmon
propagation. The
reflected beam is measured by a photodiode. In the diagram the
photo
current i vs. Angle of incident θ is shown schematically. The
dip indicates an
energy transfer to excite SP’s at the critical angle θc.
................................... 43
Figure 2.9: Kretschmann geometry: Similarly to Otto-Raether
configuration a
laser beam is coupled into the prism at the critical angle θc.
The evanescent
wave penetrates through the thin metal film, attached at the
base of the
prism, and triggers SP propagation at the outward surface of the
metal. As in
Figure 2.8, the dip in the photocurrent indicates the transfer
of energy to the
SP’s at the critical angle θc.
..........................................................................
43
Figure 2.10: Excitation of SP’s by means of a 1D diffraction
grating. .......... 44
Figure 2.11: Basic approach in near field microscopy
(NFOM/NSOM). The
yellow area indicates the near field region, or the evanescent
field. ............. 49
Figure 2.12: 3 D Model of a metal cylinder or wire with
permittivity εm
surrounded by a dielectric εd. The cylinder axis is aligned
parallel to the z-
axis of a cylindrical coordinate system. The surface plasmon
propagation is
assumed along the cylindrical surface in z-direction.
.................................... 57
Figure 2.13: Dispersion relation for the fundamental (n = 0) and
the first two
hybrid plasmon modes HE1 and HE2 (corresponding to n=1 and n=2)
on a
silver cylinder (εm = -16.102+ 1.076i at λvac = 632.8 nm)
surrounded by air (εd
= 1). The circles indicate the common radius r = 500 nm at which
the electric
field distribution is determined (see text below and Figure
2.14). ................ 59
Figure 2.14: Normalized electric field E distribution for a
metallic cylinder
with a radius r = 500 nm for a) TM-plasmons b) HE1- and c)
HE2-plasmons.
The parameters correspond to those in Fig. 2.13.
........................................ 61
xx
-
Table 1: EM field components of TM-plasmons in cylindrical
coordinates. . 62
Figure 2.15: Modified Bessel functions of the first (I) and
second (K) kind of
the zero and first order.
................................................................................
64
Figure 2.16: Wave number of TM-plasmons propagating along
tapered gold
and a silver wire. Curve 1 corresponds to silver and curve 2 to
gold ........... 65
Figure 2.17: Tapered rod design with parameters. The rounded tip
is
considered only within non-adiabatic or rigorous numerical
calculations.
However, the same structures can be considered for adiabatic
nanofocusing,
one has to replace the half sphere by an infinitely long wire
with radius equal
the exit radius to avoid any reflections.
........................................................
66
Figure 2.18: Staircase approximation of a tapered structure. The
cone, the
wedge or the V-groove (shown as cross section along the symmetry
axis z) is
decomposed into a consecutive row of basic structures with
thickness t and
with sufficiently small distances dz between adjacent
structures. ................ 71
Figure 2.19: Intensity distribution for gold tapered rod with
different taper
angles β=2α, initial radius r1 = 600 nm and exit radius r2 = 5
nm. ................ 73
Figure 2.20: Adiabatic parameter versus distance from the tip of
a metal
tapered rod for different taper angles with the same parameters
as in Figure
2.19. The gray shadowed area indicates the region of δ at which
GOA is
increasingly breached.
.................................................................................
74
Figure 2.21: Phase velocity versus distance from the tip of a
metal tapered
rod for different taper angles. The parameters correspond to
those in Figure
2.19.
.............................................................................................................
75
Figure 2.22: COMSOL model setup. Magnification of the tip area
a) node
distribution for FEM and b) Plot normalized electric field as
surface plot,
magnetic field H as contour plot. Taper angle β = 20o and tip
radius rtip = 5
nm, c) schematics of the computational window
.......................................... 76
xxi
-
Figure A.1: Electric field distribution at the interface of the
metal and the
dielectric. The different curves stem from the phase shift, see
text. ........... 171
Figure B.1: Tapered rod with rounded tip
.................................................. 173
xxii
-
List of Tables
Table 1: EM field components of TM-plasmons in cylindrical
coordinates. . 62
xxiii
-
xxiv
-
1. Introduction
1.1. Research Problem
Currently, microelectronics and photonics are among the
fundamental
technologies on which modern science and engineering is based.
Around 25
years ago, the US government initiated a study to identify and
define
important scientific and technological projects for the near
future, also called
great challenges [1]. Among the issues identified are: long term
weather
prediction; evaluating the effects of climate change;
deciphering and
cataloguing the human DNA (human genome project); analyses of
air
turbulences for prediction of tornadoes; computational ocean
science for
tsunami warning; and speech recognition. Due to the complexity
of the
problems, the study of these challenges relies almost entirely
on high-
performance computation. In order to increase computational
power, the data
processing speed has to be increased by either further
miniaturizing
electronic and optical devices or increasing the processing
speed, or clock
rate.
A modern single CPU (Computer Processing Unit), commonly used
in
personal computers or other microelectronic devices, contains
more than 200
million basic transistors, typically metal-oxide semiconductor
field-effect
transistors or MOSFET’s, within an area of ~ 800 mm2 [2, 3].
Although this
enormous number of transistors, packed into such small areas,
clearly
highlights the advantage of microelectronics, there is one
problem related to
the space between the n-junction (electrons) and the p-junction
(holes) in
MOSFET’s. This distance becomes so small that, quantum
mechanically,
electrons could tunnel across the potential barrier between
them, and yield
an unwanted electron current, also known as the tunneling
current [2].
In addition to tunneling currents, parasitic inductance due to
the
heterostructure of a basic transistor, is also present [2, 4].
The insulating
oxide layer has to be very thin, less than 1 nm, in order to
reduce the
inductance [2]. This measurement, however, increases the tunnel
current,
1
-
because the decrease of the width of the potential wall would
lead to an
increase of the probability that electrons penetrate the barrier
[2]. Other
limiting factors for miniaturization are the interaction of
electrons with an
interface as well as with other electrons, the practical problem
of granularity
of semiconductor dopants, not to mention the challenges related
to thermal
effects [2, 5].
Increasing the clock rate by keeping the characteristic sizes of
basic field
transistors constant brings along the problem of the transit
time for an
electron from the source to the hole. This characteristic
transit time,
determined by structure and material used would be much larger
than one
period of the oscillation, which limits the operational
frequency in
microelectronics to the GHz region (~109 Hz) [4].
In summary, all these problems, as illustrated above, are among
the
reasons why currently the computer industry is shifting towards
parallel
processing and multiprocessor techniques.
Photonic devices, on the other hand, operate at much higher
frequencies,
theoretically increasing the speed of information processing in
(hypothetical)
optical computers by a factor of ~ 105 or more, compared to
conventional
computers based on microelectronics. Photonic devices, such as
optical
fibers, couplers and switches have already been successfully
incorporated
into telecommunication and computer networks [6]. In fact,
photonic elements
today are the backbone of the worldwide communication system
because
they are insensitive to electromagnetic (EM) fields, experience
less heat
production and therefore lower losses during operation, and
possess
transmission bandwidths at almost the speed of light because
they operate at
optical frequencies (~1014 Hz) [6].
The advantages of photonic devices over microelectronic devices
arise
from the fact that the carriers of information are photons
rather than electrons
[6, 7]. Unfortunately, photonic devices are much larger than
their
microelectronic counterparts (by at least 10 times) because of
the diffraction
limit, theoretically limiting the characteristic minimum sizes
of photonic
2
-
devices (such as the core diameter of an optical fiber) to
around half the
wavelength [6-9]. How the diffraction limit effects ordinary
photonic devices
can be demonstrated by two well known examples, namely a
conventional
microscope and an optical fiber used, for instance, for long
signal
transmissions.
The resolution δ of a conventional microscope is determined by
Abbè’s
criterion [10]
θλδsinn2
=
(1.1)
where λ is the operational wavelength, n is the refractive index
and θ is the
aperture angle. The term n sinθ is also known as the numerical
aperture or
NA [6, 8, 9, 11, 12]. According to (1.1) the resolution can be
improved by
increasing the numerical aperture or decreasing the wavelength.
Increasing
the numerical aperture can be achieved by employing special
objectives to
increase the aperture angle [11, 12], or by applying immersion
oil to increase
the refractive index [9, 12]. However, as previously mentioned,
the best
resolution achievable with conventional optical microscopes is
in the range of
half the operational wavelength. It is mainly the lack of
materials available
(natural or synthetic) with a high enough refractive index n
that inhibits the
achievements nanometer resolution with common optical
instruments.
The transmission efficiency of a multimode optical fiber
decreases rapidly
with decreasing core diameter until a cut-off is reached. The
cut-off diameter
for a multimode fiber, at which no propagating waveguide mode is
supported,
is typically in the region of tens of micrometers at optical
frequencies [13, 14].
For a single (fundamental) mode fiber, which has no cut-off
diameter, the
electric field also becomes less and less confined in the fiber
with decreasing
core diameter, with the result that the evanescent field (due to
total reflection
at the interface core-cladding) spreads out further from the
cladding into the
surrounding area and potentially interferes with other optical
devices nearby.
Therefore to prevent cross talk the optical cables have to be
coated with an
3
-
extra layer or be separated by a larger distance with the result
of further
increase of the typical dimensions of the device [13, 14]. It is
evident that
photonics based on conventional optical techniques faces a
fundamental
miniaturization problem and it seems imperative to develop new
methods and
other technologies to further reduce the characteristic size of
optical devices
and components [15, 16].
On the other hand, microelectronics has already reached a high
degree of
miniaturization, but faces the problem of manufacturing at the
nanometer
scale, not to mention the intrinsic problems related to heat or
interference
with external EM fields [17]. An enormous synergy effect could
be obtained
with respect to increasing computational power by combining the
advantages
of microelectronics (high degree of miniaturization) with
photonics (operating
at high frequency). It should be mentioned though that the first
steps toward
this goal have been already completed by combining silicon (a
typical
material used in microelectronics) with photonics and
investigating the
possibility of chip-to-chip or even inter-chip communication
[15, 18, 19].
However, the essential problem that remains is the size mismatch
between
photonic and microelectronic devices [15, 16].
One possible and promising solution is to employ EM surface
waves on
metallic interfaces, called surface plasmons (SP’s) or just
plasmons. Surface
plasmon research, also known as plasmonics, is a new branch
of
nanophotonics or nano-optics [7, 19, 20]. Plasmonic components
and
devices are based on the propagation of SP’s on artificially
created metallic
structures and may be one of the future approaches to resolve
the problem of
how to merge microelectronics and photonics. This is because
they combine
the benefits of very strong electromagnetic field localization
at the boundary
of the metal and the dielectric within the operation at optical
frequencies. The
strong localisation of SP’s is due to the opposite signs of the
real part of the
dielectric permittivities of both media in contact. At optical
frequencies the
dielectric permittivity of metals, such as silver and gold or
aluminium, is
negative which renders them as promising materials for future
plasmonic
devices.
4
-
Plasmonic Devices
Surface plasmons (SP’s) are collective and coherent oscillations
of the
conduction electrons coupled to photons at the surface of a
metal. Once
excited they propagate along the interface between the metal and
the
adjacent dielectric [7, 11, 16, 17, 19-34]. One of the unique
properties of
SP’s, as mentioned in the previous section, is the strong
localization at the
interface, which renders SP’s very sensitive to any change of
the surface
condition. This sensitivity has been exploited in sensors in a
large variety of
applications in chemistry, medicine and biology [35-37]. SP’s
are also ideal
candidates to design real “flat” optics, that is, where the
propagation of light is
confined to the surface of a metal. This could pave the way to
the ability to
shrink photonic devices and components down to
sub-wavelength
dimensions [17, 24, 26, 33, 38-40].
Merging plasmonics with microelectronics is a major long-term
goal in
science and engineering due to the huge potential benefits, as
indicated in
the previous section. Hence, SP based circuits have to perform
the same
tasks as their electronic counterparts [15, 41]. For instance,
plasmonic
circuits have to be able to a) support the propagation of SP’s
around sharp
corners and bends without significant losses b) split one
plasmon stream into
two and vice versa c) couple and switch and d) focus SP’s into
nanometer
regions, to name just few possible options. A plasmonic circuit
must also be
capable of coupling light into and recover light from the
plasmonic circuit with
high coupling and decoupling efficiency as well as
interconnecting with
common microelectronic circuits [15, 22, 41]. However, owing to
the
presence of the metal in all potential plasmonic devices,
dissipation is one of
the intrinsic problems of plasmonics which limits the
propagation distances
and hence limits miniaturisation of plasmon devices [20, 22, 28,
33, 42].
It should also be mentioned that SP’s not only exist on metals
but also on
doped semiconductors [43]. However, this is beyond the scope of
this thesis,
for an detailed overview of SP’s on semiconductors see, for
instance, the
extensive review from Kushwaha [43].
5
-
Over the last two decades several metallic structures to guide
SP’s, also
called plasmonic waveguides, have been suggested, including a
thin metal
film sandwiched between two dielectrics [44-46], metal stripes
embedded in
dielectrics [47-49], nano-chains [50-59], a V-groove in a metal
[60-66], a
metal wedge [67, 68], or metal wires [69-75]. In order to
compare the
characteristics (and performance) of different plasmonic
waveguides and
devices, criteria have to be defined, such as confinement of the
electric field
within or close to the waveguide, propagation distance and also
the
possibility of single mode operation. Other more practical
figures of merit are
the efficiency of transmission around sharp bends and corners,
the tolerance
to imperfections or even the complexity of the manufacturing
process for
plasmon waveguides [76-78].
6
-
Localized Surface Plasmons and Nanofocusing
SP’s are not only bound to flat surfaces but can also propagate
along
interfaces of curved surfaces, such as metal cylindrical wires.
They can even
be confined to tiny metal structures, called nanoparticles. In
this case, the
conducting electrons in the nanoparticles do not propagate but
oscillate in
resonance with the external EM field. The first rigorous
analytical study of
surface waves on metallic spheres was carried out by G. Mie
(1869-1957)
[79], as he studied scattering of an EM wave by small metallic
spheres. SP’s
excited on nanoparticles (sometimes called Mie-resonance [17,
80]) are also
known as localized surface plasmons (LSP’s), to distinguish them
from
ordinary SP’s. In a sense they can be considered as
non-propagating SP’s.
The dispersion relation of LSP’s depends strongly on the form
and shape of
the nanoparticles and the surrounding environment, which renders
LSP’s
very sensitive to any changes of the environmental condition.
This sensitivity
has been exploited, for instance, in bio sensors [23, 33, 81].
Their resonance
spectrum also exhibits strong peaks, indicating that LSP’s can
only be
excited at certain frequencies [23, 33]. One has only to match
the resonance
frequency in order to excite LSP’s whereas to excite SP’s both
the resonance
frequency and the wave number have to match [23, 33].
The strong localization effect, due to the resonance effect of
the EM field
in close proximity of the nanoparticles, has been utilized in
research areas
such as non-linear optics where the high energy density near
the
nanoparticles can trigger, for instance, second harmonic
generation [82, 83].
Another important application of LSP’s is related to Raman
spectroscopy
where the Raman signal is strongly enhanced due to the presence
of
roughness of the metal surface, a metal tip (also known as tip
enhanced
Raman spectroscopy or TERS) [84, 85] or nanoparticles close to
the probe
being examined [85-89], hence the name surface enhanced
Raman
spectroscopy (SERS). The overall SERS enhancement factor is
proportional
to the fourth power of the local electric field [82, 85]. A
local electric field
enhancement factor of ~1000 would enhance the SERS signal by a
factor of
~1012 which would be sufficient to detect single molecules [87,
90].
7
-
The SP’s, propagating along metal tapered rods or other
confining metal
structures, such as a wedge or a V-groove, may experience a
decrease of
both the phase and group velocity. In that event the plasmon
wave number
increases. This is equivalent to the transformation of
(propagating) SP’s to
(non-propagating) LSP’s and the optical energy may become
concentrated at
the tip of the structure, or focused, when the effect of
dissipation is
sufficiently weak [91-95]. The effect of the concentration of
plasmon energy
down to the nanometer region, much smaller than allowed by
diffraction limit,
is known in the literature as superfocusing [91, 94, 96, 97] or
nanofocusing
[92, 93, 95, 98-100]. Nanofocusing by a solid metal tapered rod
has been
studied analytically [91, 95, 96], numerically [98, 101] and
verified
experimentally [102]. Other metallic structures have also been
investigated
including nanoparticles [53, 54], metal wedges [67, 93, 97, 103]
or a metallic
V-groove [92, 97, 99, 104].
1.2. Research Background
As indicated in the preceding section, efficient delivery of
optical energy to
the nanoscale region is crucial for the performance of sensors
employed in
nanophotonics and spectroscopy. However, the generation of
nanometer
sized energy sources is a considerable challenge. Several ways
have been
suggested to overcome this difficulty including a laser beam
focused close to
the tip region [105], an approach also known as tip enhanced
microscopy
(see chapter 2.6). Unfortunately it suffers from poor
efficiency, and since the
laser beam heats up the tip, damage to the probe or even the
sample can
occur [12, 106]. Another approach consists of guiding light
through a tapered
optical fiber coated with a metal film and terminated by an
aperture.
However, this design essentially experiences the problems of low
throughput
and emergence of heat [11, 27, 107-112]. On the other hand,
tapered rods
are probably the most common component in near field optics,
compared to
more complex structures capable of nanofocusing, including
metallic wedges,
metallic V-grooves and gaps. They are made of either solid
metals or
dielectrics, uncoated or coated with a metal film. The
manufacturing process
8
-
is well understood and high precision tapered rods and tips of
any shape can
be produced in great quantities [11, 113].
SP propagation along metal tapered structures like a metal rod
or a
metallic V-groove may overcome the problems related to efficient
delivery of
optical energy to the nanoscale region. As mentioned
previously,
Babadjanyan et al. [91] and Stockman [95] demonstrated that the
SP’s
propagating towards the tip of a metallic tapered rod may lead
to a strong EM
field enhancement (nanofocusing) when dissipation is
sufficiently weak.
Babadjanyan and colleagues analysed plasmon focusing by a
tapered rod by
means of the wave equation in spherical coordinates in
combination with
separation of variables [91]. This could only be done under very
restrictive
conditions which are applicable only in close proximity of the
tip, as pointed
out by Kurihara et al. [96]. Stockman [95] studied plasmon
nanofocusing by a
metal tapered rod on the base of the geometric optics
approximation (GOA)
or adiabatic theory, also referred to in quantum mechanics as
WKB-method.
GOA is applicable only under the condition that the tapered
structure does
not vary significantly within one wavelength. Stockman [95]
also
demonstrated that the applicability condition of GOA includes
the whole
tapered rod including the tip itself, when the taper angle is
sufficiently small.
The effect of nanofocusing under GOA is also called adiabatic
nanofocusing
[95, 114]. As demonstrated, the attainable plasmon field
enhancement
factors were in the range of ~ 1000 for a silver tapered rod
with a taper angle
α ~ 2.3O and taper length ~2.5 μm. However, the effect of
dissipation and
taper angle on the efficiency of nanofocusing has not been
discussed. No
attempt had been made to determine the limits of the
applicability condition of
GOA.
Adiabatic nanofocusing during plasmon propagation by a V-groove
in
metal has been first reported by Gramotnev [92]. He demonstrated
within
GOA that the magnetic field at the tip of the groove remains
finite even in the
absence of dissipation. The analysis of the dependence of
dissipation and
taper angle on the maximum local field enhancement has also been
carried
out [92]. In addition, a comparison study of adiabatic
nanofocusing and
9
-
numerical has been conducted by Pile and Gramotnev [99].
However, the
field enhancement factors have been determined for the magnetic
field only.
No comparison study with other structures, such as a tapered rod
or a metal
wedge has been conducted.
Issa and Guckenberger [98] studied numerically nanofocusing
(also called
non-adiabatic nanofocusing) by metal tapered rods attached to a
long
straight cylindrical wire and terminated by a spherical tip of
radius rtip = 5 nm.
The length of the wire was chosen in such a way in order to
avoid any
interference from reflections originating from the initial
computational
boundary and the tip region. Optimal taper angles at which the
plasmon field
enhancement is maximal were found, which did not depend on the
radius of
the cylindrical wire [98]. It has been demonstrated that the
plasmon field
enhancement increases with increasing tip length. Although an
optimal angle
has been found, no attempt had been made to determine other
structural
parameters such as an optimal taper length. The dependence of
the tip
radius on maximum field enhancement has not been determined and
no
detailed analysis of the field structure near the tip has been
conducted.
In summary, no thorough investigation into the effects of
different levels of
dissipation, taper angles and shape variations on plasmon
nanofocusing and
plasmon field distribution by metal tapered rods has been
conducted. The
influence of the tip geometry on the maximal achievable field
enhancement
factors as well as the field structure around the tip has not
been considered.
No detailed analysis has been carried out to determine the
limits of the
applicability condition of the adiabatic approximation (GOA).
For a metallic
V-groove no electric field enhancement factors have been
determined and no
comparison with a metal tapered rod has been done.
10
-
1.3. Aim of the project and linking the papers
As emphasized in the preceding section, the propagation of SP’s
along
tapered metallic structures may lead to a strong field
enhancement at the tip
of the structure which can be used to deliver electromagnetic
energy to the
nanometer region. The strong field enhancement, due to the
strong
localisation of the plasmon field near the tip, is crucial and
determines, for
instance, the resolution of near field microscopes [11, 12,
115]. It is also of
particular importance in SERS or TERS, because the sensitivity
of SERS,
along with the distance sample to probe, depends mainly on the
strength of
the local electric field [82, 85].
The overall goal of this PhD project was to analyse in detail
plasmon
nanofocusing by metal tapered rods and metallic V-grooves and
determine,
theoretically and numerically, structural conditions, such as
optimal taper
angles, taper length or ideal shape, under which maximum plasmon
field
enhancement (nano-focusing) can be achieved.
With these topics in mind, the project was divided into two
parts. In the first
part, the adiabatic approximation (GOA) has been employed to
study
analytically plasmon nanofocusing by a tapered rod and a
metallic V-groove.
However, this approach is applicable, as emphasized in the
previous section,
only for small taper angles; hence the effect of reflection from
the tapered
section and from the tip is not taken into account.
In particular, the specific topics investigated in the first
part were:
• Analysis of the effects of different levels of dissipation and
taper
angles on the efficiency of plasmon nanofocusing by metal
tapered
rods. Determination and interpretation of the different
characteristics of the intensity distribution (chapter 3).
Methods and outcomes:
The optimal parameters, such as optimal taper lengths and
taper
angles at which the plasmon field enhancement at the tip is
maximal, were determined and discussed. The different
plasmon
11
-
intensity distributions along the tapered rod, between two
fixed
radii, were analysed in detail by means of GOA. It will be
shown
that three distinctive intensity distribution patterns exist,
owing to
the presence of two critical taper angles, or equivalently, two
critical
dissipation levels.
• Detailed analysis of the electric field enhancement in a
metallic V-
groove. Comparison study of plasmon field enhancement during
nanofocusing by a tapered rod and a V-groove in a metal
(chapter
4). Comparison of the results achieved by GOA and rigorous
numerical methods.
Methods and outcomes:
Following recent reports on adiabatic nanofocusing by a
metallic
V-groove [92, 99], a comparison study with a metal tapered rod
was
conducted, to determine the characteristics of the EM field in
the
V-groove and the plasmon field enhancement. It will be shown
that
the attainable field enhancement at the tip of a V-groove in
the
presence of strong dissipation in the metal is higher than
compared
to a tapered rod having the same structural parameters, such
as
taper angle or taper length. A comparison study of the
results
achieved by GOA and rigorous numerical methods (FEM) has
been
carried out to determine the limits of GOA for plasmon
nanofocusing by a metallic V-groove.
• Analyses of the shape variation of the tapered rod on
nanofocusing
(chapter 5).
Methods and outcomes:
A detailed analysis, based on GOA, on the effects of shape
deviations from the ideal conical geometry on the plasmon
field
enhancement at the tip as well as the intensity distribution
along the
rods, is presented. Two basic structures were considered in
detail,
a convex (curved outwards) and a concave (curved inwards)
tapered rod. It will be demonstrated that, in general, the
attainable
12
-
plasmon field enhancement by a concave tapered rod is
significantly higher compared to a convex tapered rod or even
a
conical tapered rod.
In the second part, non-adiabatic nanofocusing by a tapered
rod
terminated by a spherical tip and with large taper angles is
studied in detail.
Owing to the presence of the a) tapered section and b) a rounded
and closed
tip, parts of the plasmon wave are reflected back and interfere
with the
forward propagating (towards the tip) plasmon wave. This leads
to potentially
complex interference patterns, also called Fabry-Perot
resonance, which
cannot be treated by analytical methods. Therefore, in order to
study
nanofocusing accurately, rigorous numerical methods have to be
employed.
In practice, many different kinds of numerical techniques for
analysing
electromagnetic fields have been applied, such as finite
difference time
domain (FDTD), methods of moments (MOM) and finite element
methods
(FEM). Within this thesis, however, the numerical simulations
for tapered
rods have been carried out exclusively by means of FEM. To this
end, a
commercial software package COMSOL© and in-house programs,
mainly
developed in MATLAB©, have been employed.
As mentioned in the previous section, Issa and Guckenberger [98]
studied
numerically (with COMSOL©) non-adiabatic nanofocusing by
considering
tapered rods attached to long cylindrical wires. Optimal taper
angles at which
the plasmon field enhancement is maximal were found, which did
not depend
on the radius of the cylindrical wire. However, the dependence
of
nanofocusing on the tip radius has not been taken into account.
It has been
demonstrated, though, that the plasmon field enhancement
increases with
increasing tip length. This result was found for short tapered
rods only and so
the analysis was limited to a small section near the tip. It is
not clear if the
increase would continue with further increase in the taper
length. Of course
the field enhancement will not increase infinitely but it is of
interest whether a
correlation between optimal taper angle and optimal taper length
exists.
In response to these questions, the topics in the second part
were:
13
-
• Determination of the optimal taper angle and an optimal
taper
length at which the plasmon field enhancement is maximal
(chapter
6 and chapter 7).
Methods and outcomes:
Non-adiabatic nanofocusing by metal tapered rods was studied
by
means of FEM. The existence of an optimal taper angle, as
concluded by [98, 101], and an optimal tip length at which
the
plasmon field enhancement at the tip is maximal was found.
An
equation to determine the optimal taper length, based on
energy
conservation consideration, has been derived. The plasmon
field
enhancement is determined by the ratio of normalized electric
field
amplitude at the top of the rounded tip to the normalized
electric
field at the launching point. To this end, a new approach
was
introduced, which combines GOA and numerical analysis in
order
to study large tapered structures within reasonable time and
reducing significantly the use of computational resources. A
new
method was also introduced to extract the contribution of
the
forward propagating plasmon wave from the interference
pattern
due to Fabry-Perot resonance.
• Analysing the dependency of the tip radius on plasmon
field
enhancement and determining the characteristics of the EM field
in
the proximity of the spherical tip region (chapter 6 and chapter
7).
Methods and outcomes:
The achievable plasmon field enhancement was calculated for
different metals and different tip radii. The attainable plasmon
field
enhancement strongly depends on the radius of the rounded
tip.
The maximum field enhancement factor, as predicted in the
numerical calculations, are in the range of ~2500, which is
sufficient
for detecting single molecules with SERS. The field
distribution
along the rounded tip, the effect of the tip radius and the
presence
of different metals on to the field enhancement factors has
been
studied in detail.
14
-
• Investigation of the validity of the adiabatic theory for
nanofocusing
by a metal tapered rod. Determining the accurate
applicability
condition for adiabatic nanofocusing by a tapered rod (chapter
6
and chapter 7).
Methods and outcomes:
In chapter 4, for a V-groove in a metal, the critical taper
angle, at
which the plasmon field enhancement calculated by means of
the
adiabatic theory starts to deviate from the results obtained
by
rigorous numerical methods, is in the range of 12-14o, for gold
at
wavelength λvac = 632.8 nm. No such comparison had been done
for tapered rods. Therefore, a detailed investigation of the
applicability condition for different metals and different
wavelengths
was carried out. Surprisingly, the critical angle was found to
be
much larger than for the groove in the metal. This increases
the
number of possible applications of the adiabatic theory in
the
context of plasmon nanofocusing by metal tapered rods. As
another
key result, it has been shown that the applicability condition
for
adiabatic nanofocusing by metal tapered rods, as applied in
[95], is
too strict. As a consequence, a new modified applicability
condition
for adiabatic nanofocusing by metal tapered rods has been
suggested.
15
-
16
-
2. Theory and Background information
2.1. Introduction
Robert W. Wood (1868-1955) was the first scientist who
accurately described
the effects of surface plasmons (SP’s) as he examined the
spectra of an
incandescent lamp with a new metallic grating. In 1902, he wrote
in his report
[116]: “When the light is incident on the opposite side of the
normal from the
spectrum we find the red and orange extremely brilliant up to a
certain wave-
length, where the intensity suddenly drops almost to zero, the
fall occurring,
as I have said, within a range not greater than the distance
between the D
lines. A change of wave-length of 1/1000 is then sufficient to
cause the
illumination in the spectrum to change from a maximum to a
minimum. The
theory of the diffraction-grating, as it stands at the present
time, appeared to
me to be wholly inadequate to explain this most extraordinary
distribution of
light, and I accordingly endeavoured to find out if possible the
necessary
modifications which must be introduced.”
Wood expected a smooth and slow change in the intensity
distribution but
instead he observed sharp and narrow bright and dark bands. He
also
noticed that this unexpected behaviour occurs only when the
incident wave is
s-polarized, that is, when the only magnetic field component of
the incident
wave is parallel to the grating. Woods could not explain the
results he
obtained within the framework of the existing theory and
described them as
anomalies. Ever since, this phenomena is referred to as Wood’s
anomalies
[28, 117, 118]. Fano [117] and later Hessel and Oliner [118]
were among the
first to develop a theory to explain Wood’s anomalies by
coupling of light into
EM surface waves, mediated by the metallic grating
structure.
According to the free electron model, a metal consists of a
lattice formed
by positive ions and electrons from the valence band which move
through the
body , only weakly bound to the ion cores, similar to a gas [25,
119-121].
The coherent oscillations of the valence (or conducting)
electrons are called
bulk (volume) plasmons or SP’s, depending on whether the
plasmon
17
-
oscillation takes place inside or at the surface of the metal.
Bulk plasmons
had been first scrutinized in detail in the early 1950’s by
Pines and Bohm as
they studied the effect of electrons passing through the metal
[122-125]. In
1957, SP’s were first predicted by Ritchie [21] to explain
unexpected low
frequency losses (below the bulk plasmon frequency ωp, see also
section
2.3.1) during a study of the angle-energy distribution of
electrons passing
through thin metal films. Two years later, SP’s had been
verified
experimentally by Powell and Swan [126] by measuring the
energy
distribution of electrons passing through an aluminium film.
In the literature, SP’s are sometimes called surface plasmon
polaritons
(SPP) which reflects the hybrid nature of being a mixture of
plasma
oscillations and photons, but essentially SP’s and SPP’s are
identical [28,
119]. The EM field, associated with SP’s, which propagates along
the
interface between the metal and the dielectric, is highly
localized and decays
rapidly into both media, like a surface wave [16, 20, 24, 26,
28, 29, 33, 42].
The EM surface wave has both, longitudinal and transversal
components,
whereas surface plasmons are purely longitudinal surface charge
density
waves [119]. Because of their wavelike propagation, SP’s can be
described
by Maxwell’s equations and the EM wave equation. It should be
mentioned,
however, that in the beginning of plasmonics research most of
the results
including the dispersion relation or the response of metals to
the external
applied field, were derived by applying Bloch’s equations,
because of the free
electron model used to describe the metal [21, 127, 128].
Maxwell’s equations are introduced in section 2.2, followed by a
brief
discussion of the constitutive relations in section 2.3, in
context of the
interaction of light and matter in regions smaller than allowed
by diffraction
limit. At this length scale, metallic structures, such as
nanoparticles or
nanorods, have an impact on the permittivity ε, permeability μ
or conductivity
δ. Within section 2.3, the Drude-Lorentz model is introduced,
a
representation of the electric response of metals to an applied
EM field which
is commonly applied in plasmonics and near-field optics,
followed by a brief
discussion about loss mechanisms.
18
-
In section 2.4, a detailed analysis of the propagation of SP’s
on a flat
metal surface is presented. Starting from Helmholtz’s equation
and applying
the appropriate boundary conditions, the dispersion relation for
the plasmon
wave number q(ω), with ω being the angular frequency, is derived
and the
main characteristics of the solutions are discussed. As a major
result, it will
be demonstrated in section 2.4, that only TM-plasmons can
propagate along
the metal surface. In the optical and near infrared frequency
regime the real
part of the complex relative permittivity of most of the metals,
such as silver,
gold, aluminium and copper, takes on negative values which is
the basic
requirement for the existence of SP’s [129].
SP propagation on a metal film bound by a dielectric on each
side; also
called IMI (insulator-metal-insulator) and within a metal gap
filled with a
dielectric, called MIM (metal-insulator-metal) is studied in
section 2.5. In
these configurations coupling of SP’s for sufficiently small
film or gap widths,
give rise to additional plasmon modes, called film and gap
plasmons. As it
will be shown, these gap (film) plasmons have either a symmetric
or an anti-
symmetric EM field distribution across the gap (film) with
respect to the
middle plane. The dispersion relations of all possible modes are
analysed
and the results in relation with nanofocusing are discussed.
In section 2.6 an overview of the different excitation methods
of SP’s is
presented, followed by an outline of the tools and sensors used
to detect and
visualise SP’s, such as fluorescent molecules or near field
microscopes, in
section 2.7.
SP propagation along a metal cylindrical wire is studied in
detail in section
2.8. Starting from the wave equation the dispersion relation
q(r) is derived,
with r being the local radius, followed by a detailed
investigation of the most
important solutions (modes). Contrary to a flat metal surface,
cylindrical
structures not only support fundamental TM-plasmon modes but
higher order
modes (hybrid modes) as well. However, all modes except of
the
fundamental TM-mode experience a cut-off when the diameter
decreases. As
it will be shown in section 2.9, TM-plasmons mode can be
employed for
nanofocusing.
19
-
Flat metal surfaces, IMI, MIM and cylinders constitute the
building blocks
for the theoretical and numerical analyses of adiabatic
nanofocusing of SP’s
by a tapered rod and a metallic V-groove. A detailed analysis
of
nanofocusing by those structures, including analytical and
numerical
methods applied within this thesis, is presented in section 2.9
which will
conclude the literature review.
20
-
2.2. Maxwell’s Equations and Wave Equation
SP’s are EM surface waves which propagate along the interface of
a metal-
dielectric system. They can be analysed analytically by means of
the
Maxwell’s equations. The amplitude of the SP is maximal at the
interface and
decays exponentially into both media; thus they are strongly
localised at the
interface at optical frequencies [16, 20, 24, 26, 28, 29, 33,
42].
In SI-units, which are adopted throughout this thesis, the
equations for the
electric field strength E 1 (or simply electric field) and the
magnetic field
strength H (magnetic field) are given in differential forms
as
t∂∂
−=×∇ E B (2.1)
and
t∂
∂+=×∇
DJH (2.2)
where B is the magnetic flux density, D the electric
displacement and J is the
density of the free currents [130]. The magnetic field H and the
magnetic flux
density B, as well as the electric field E and the displacement
D are related
by the constitutive equations
HB r0μtμ= (2.3)
ED r0εt
ε= (2.4)
Where μr and εr denote the relative permeability and
permittivity tensor,
respectively. µ0 2 and ε0 3 are the permeability and
permittivity in vacuum,
respectively. Metals considered in this thesis are isotropic,
linear and
1 Vectors are denoted by bold letters
2 ε0 ≈ 8.854·10-12 F/m
3 μ0≈ 4π·10-7 H/m
21
-
homogeneous; hence equations (2.3) and (2.4) are simple
proportional
relations. The constitutional parameters, however, may still be
dependent on
the applied frequency but for the sake of simple notation, the
arguments are
not shown.
By combining (2.1), (2.2) and (2.4) the wave equation for the
electric field E
can be derived as
tr00 ∂
∂−=×∇×∇ E εεμ E . (2.5)
A plane wave solution has the form
(2.6) )t(ie rk0EE
⋅−= ω
where k denotes the wave vector, ω the angular frequency and E0
the
amplitude of the electric field, the wave equation (2.5) can be
expressed as
(2.7) ( ) EkEEkkEkk ),(kk r202 ωε=−⋅=××
with k0 = ω/c with c being the speed of light. In the last step
the vector
relation x x A = ( ·A) - 2A has been applied. For transversal
waves
(k⋅E = 0), (2.7) is reduced to the well known Helmholtz
Equation
0k 22 =+∇ EE . (2.8)
For longitudinal waves (k⋅E = kE) the wave equation for any
arbitrary electric
field is reduced to
(2.9) 0),(0),(k rr2
0 =⇒= kkE ωεωε
Equation (2.10) implies that at the zeros of εr(ω,k) only,
longitudinal waves
can propagate in the metal.
22
-
2.3. Constitutional Parameters
The relative permittivity εr in (2.5) describes the response of
matter to an
applied external electric field E, likewise the relative
permeability μr in (2.4) describes the magnetic response to the
applied magnetic field H. In the high-
frequency regime of optics and plasmonics, however, the magnetic
response
can be neglected for two reasons. Firstly, the magnetic domains
are too large
and therefore too inert to follow the rapidly varying EM field
[114]. Secondly,
as mentioned in the introduction of this chapter, noble metals
such as Gold
and Silver are isotropic and diamagnetic by nature; whereas
aluminium (as
another important metal) is a paramagnetic material [25]. Hence,
the relative
permeability μr reduces to 1 in the optical frequency range for
all metals
considered in this thesis. Nevertheless it is worth noting that
Ishikawa and
colleagues have shown theoretically that the relative
permeability μr can
indeed take on negative values at optical frequencies for
artificial nano-sized
structures, composed of split ring resonators (SRR) [131]. It
should also be
mentioned that external magnetic fields have been applied in the
context to
plasmonics, for instance, to study the effect on surface plasmon
propagation
in semiconductors, also called magnetoplasmons. An extensive
overview of
this topic, which is beyond the scope of this thesis, has been
given, for
instance, by Kushwaha [43].
The electrical properties of metals, in contrast to the magnetic
properties,
play a crucial role in the response to optical waves. This is
because electrons
can react almost instantly to any changes of an external EM
field, especially
in metals where the valence electrons are nearly free to move
throughout the
whole lattice structure.
The characteristic sizes of metallic structures studied in
nano-photonics
and plasmonics are in the nanoscale region. At this length scale
the effective
constitutive parameters become size dependent. This is the
subject of a
whole new and rapidly growing scientific research area dealing
with left-hand
materials and metamaterials [132]. As a simple definition,
metamaterials are
designed materials that gain their EM properties not only by the
material
23
-
properties they are composed of, but also by the structural
composition at the
nanoscale. The potential applications of metamaterials are
staggering. Just
recently, for instance, it has been demonstrated that an object
surrounded by
a cylindrical shell, composed of SRR’s, is rendered invisible
[133]. The
incoming EM wave is routed around the shell and no scattering
occurs from
the object inside [133]. Another interesting example is the
possibility of
designing a perfect lens in the form of a simple slab made of
negative
refractive index material [134], where both the permittivity and
the
permeability are simultaneously negative [135], a concept
originally
envisioned by Veselago [136].
Photonic crystals are another class of artificial material,
composed of
periodically built up dielectrics or metal-dielectrics in 1D-,
2D- or 3D-structure
[20, 137-140], though, the 2D- and 3D-structure are still
difficult to engineer
[140]. EM waves can propagate in photonic crystals with designed
gaps or
along designed defects at the surface [138, 140] and it has also
been
demonstrated, for instance, that light can be guided around
bends with low
losses [22, 141]. However, the propagation of EM waves along
those
manufactured defects is still subject to the diffraction limit,
therefore photonic
crystals are not applicable for the purpose of nanofocusing or
sub
wavelength guiding. Plasmon nanofocusing by structures made of
artificially
designed materials, however, due of its complexity, is beyond
the scope of
this PhD-project and cannot be considered.
Over the past century several models to describe the dielectric
properties
of metals been developed. One of the first models, the
Drude-Lorentz model
(DLM), is still commonly applied in plasmonics and
nanophotonics, because
at optical frequencies it describes sufficiently well the
dielectric response of
noble metals, despite the simple equation [11, 21, 28, 33, 142].
Noble metals
play an important role in plasmonics, because the dielectric
permittivity of
gold or silver, for instance, is negative at frequencies within
the visible
spectrum. As a consequence it is possible to excite SP’s at
optical
frequencies. This will be the subject of the next section.
24
-
2.3.1. Drude-Lorentz model
Around 100 years ago, Paul Drude (1863-1906) developed a
simple
description of the dielectric response of a metal, based on the
free electron
model, in order to explain the high electric conduction and
thermal
conductivity of metals [143]. Later improved by Hendrik Lorentz
(1853-1928),
the Drude-Lorentz Model (DLM) essentially treats the electrons
as a gas
inside the metal, only weakly bound to the positive ions, which
are fixed in a
rigid lattice structure. Starting from a simple model of an
electron exerted to
an applied external electric field with a restoring force
originating from the
positive background, the relative permittivity εr(ω) can be
expressed as
γωω
ωωε
i1)( 2
pr −
−=2
(2.10)
where ωp is the bulk plasma frequency, and γ is the damping
frequency,
which accounts for scattering of the electrons by other
electrons or the lattice
[11, 25, 33, 42, 114, 119, 120, 142]. For a weakly damped system
(γ 0)
the relative permittivity takes on a particularly simple
form:
2p
r 1)(2
ωω
ωε −= (2.11)
According to (2.11), a metal becomes transparent (wave
propagates insides
the metal) when the applied frequency ω is larger than the bulk
plasma
frequency ωp. The metal becomes impenetrable (the metal acts as
a
reflector) when ω becomes larger than ωp, as the wave is damped
inside the
metal and no wave propagation occurs. When ω is equal to the
bulk plasma
frequency ωp, or equivalently εr becomes zero, however,
longitudinal EM
waves can propagate inside the metal, as mentioned in the
previous section.
The bulk plasma frequency can be intuitively understood by
considering
the following scenario: In equilibrium, the free electron
density is uniform
across the whole metal surface, with a fixed and positively
charged
background. A slight disturbance causes an inhomogeneous surface
charge
distribution which generates a restoring force. However, due to
the non
25
-
uniform charge distribution an electric field is generated in
which the
electrons gain additional momentum. As a result the electrons
move towards
the equilibrium position but overshoot due to inertia of the
electrons, hence
an oscillation of the electron gas with a natural frequency ωp
follows, damped
by internal frictions, collisions and other mechanisms
[144].
The DLM predicts sufficiently well the dielectric response of
noble metals
such as gold or silver to an applied EM field at optical and
near infrared
frequencies [42]. However, for completeness it should be
mentioned, that the
DLM does not take quantum mechanical phenomena, such as inter-
and
intra-band transitions, into account. These effects have an
impact on the
dielectric response, and in particular on dissipation of the
metal [25, 119-
121].
2.3.2. Non-local Effects and Loss Mechanism
The complex problem of dissipation and dielectric response of
metals has
been addressed by many researchers [25, 114, 120, 122-125, 127,
128, 130,
145, 146]. Landau damping (damping of longitudinal waves in
plasma), inter-
and intra-band transitions and the anomalous skin effect are
examples of
non-local loss mechanisms which have to be considered when the
plasmon
wavelength becomes very short. This is especially important when
the SP’s
approaches the tip of a tapered metallic structure. The wave
number
increases anomalously near the tip, hence both the phase and
group velocity
decrease and the plasmon wavelength shortens [91, 92, 94, 95].
This will be
demonstrated in section 2.9, when nanofocusing by metal tapered
rods and
metallic V-grooves is discussed.
When the size of the sample or structure becomes comparable with
the
mean-free path of an electron, spatial dispersion has to be
taken into
account. In this case the permittivity ε becomes size and wave
vector
dependent [114]. In conjunction with nanofocusing by tapered
rods,
Stockman [95] discussed qualitatively the effect of spatial
dispersion,
whereas Ruppin [100] demonstrated, analytically and numerically,
that spatial
dispersion has to be considered only when the local diameter of
a tapered
26
-
rod becomes much smaller than the plasmon wavelength. To this
end, he
derived a modified dispersion relation q(r), owing to the fact
that at the
interface an interaction of the transverse surface wave with the
longitudinal
bulk plasmons takes place. The intensity of the SP field at a
diameter
of ~1 nm, calculated without spatial dispersion, would be more
than twice as
large [100]. However, this result has been achieved only for a
silver tapered
rod of one plasmon wavelength (λvac = 632.8 nm).
Within this PhD thesis, the response of a metal to an applied EM
field, as
well as all loss mechanisms mentioned above, are represented by
the
complex dielectric permittivity, as published by Palik [129].
The real part of
the permittivity describes the propagation of SP’s and the
imaginary part is
linked to dissipation. A minimum diameter of 4 nm as a lower
limit in the
investigations of nanofocusing in metallic structures within
this thesis has
also been applied in order to avoid any effects from spatial
dispersion, as
suggested by Stockman [95].
27
-
2.4. Surface Plasmons on a Flat Surface
A single flat metal surface is probably the most basic metal
structure to
describe SP propagation. Nevertheless, some important results
can be
derived with this simple configuration. Within this section, it
is assumed that
the interface, which separates the metal and the dielectric, is
placed on the
x-z plane of a Cartesian coordinate system. It is also assumed
that the metal
occupies the region y < 0 and the SP’s propagate along the
x-axis. It can be
demonstrated that only TM-plasmons can propagate along the
surface [28,
33, 42]. This is due to the fact that the relative permeability
μr is close to unity
for most natural metals at optical frequencies, for details see
Appendix A.
There is alternative explanation derived from the boundary
condition
⋅− nDD md = δ)( (2.12)
where Dd and Dm are the dielectric displacement in the
dielectric and
metal, respectively and n is the normal vector at the interface.
As a
consequence of (2.12) only the components of E in the plane of
the incident
wave can induce longitudinal surface charge density oscillations
in the
direction of propagation (x-axis) [29]. The only magnetic field
component left
is Hy, hence only TM-plasmon can propagate along the surface of
the metal.
The EM field associated with TM-plasmons, also decay
exponentially away
from the interface into both media, see also Figure 2.1.
Therefore, the following solution for Hy can be assumed:
( ))xxqt(j −ω
( )
metal,0yyexpeAH my −⋅⋅= α (2.13b)
where A and B are amplitudes to be determined by the boundary
conditions,
αm and αd are the reciprocal (positive and real) penetration
depths into the
28
-
metal and the dielectric, respectively. The TM-plasmon wave
number qx is
complex because of the presence of a metal with the
permittivity
εm = εm′ + iεm′′. Consequently, the amplitude of the SP’s
decreases with
increasing propagation distance and eventually dissipates. The
penetration
depth is defined as the distance from the interface
metal-dielectric at which
the amplitude of Hy is reduced by e-1:
md
20
2x
md kq εα −= (2.14)
Figure 2.1: 3D-model of a surface plasmon propagating along a
flat metal surface in the x-direction. Schematically a snap shot of
the Hy distribution (TM-mode) is shown. The relative permittivity
εd are for the dielectric material and εm for the metal,
respectively. The evanescent waves in y-direction are indicated by
the dash-dotted line.
The penetration depths and the permittivities are also related
by (see
Appendix A for derivation):
d
m'
m
d
εα
εα
=− (2.15)
where εd and εm′ are the real permittivities of the dielectric
material and the
metal, respectively. The permittivity of the dielectric is
assumed lossless;
hence εd is real throughout this thesis. From (2.14) it can also
be concluded
29
-
that the penetration depth of the plasmon into the metal is much
smaller than
into the dielectric, as indicated in Figure 2.1.
The real part of the permittivity of the metal, εm′, must take
on negative
values to satisfy (2.15), because the penetration depths, by
definition, are
positive real numbers. The real part of the complex wave number
can then
be written as [20, 26, 28, 29, 42]
''k'qmd
md0x εε
εε+
= . (2.16)
In order to obtain a positive wave number the condition for the
permittivity is
│εm│> εd. The imaginary part of the plasmon wave number
is
2m
m2
3
md
md0x '
'''
'2k''q
εε
εεεε
⎟⎟⎠
⎞⎜⎜⎝
⎛+
= . (2.17)
The propagation distance of SP’s along the surface is limited,
owing to losses
in the metal, which is represented by the imaginary part of the
permittivity
εm′′. The propagation length L is defined as [28, 33]
1−x )''q2(L = . (2.18)
Inserting (2.17) into (2.18) yields
'')'('
k1L
m
2m
2
d
md
0 εε
εεε
⎟⎟⎠
⎞⎜⎜⎝
⎛ +=
13
. (2.19)
The propagation distance L on a flat metal surface for gold,
silver and
aluminium as determined by (2.19) is depicted in Figure 2.2. For
optical wavelengths in the range of 500-700 nm, silver has the
largest propagation
30
-
distance followed by aluminium and gold. The strongly increased
propagation
distance towards the infrared region is of importance for the
applicability and
possible implementation of plasmonic devices in
telecommunication
technology [66, 147, 148].
Figure 2.2: Propagation length for SP’s on a flat surface for
gold, silver and aluminium. The permittivities of Au, Ag and Al
were taken from Palik [129].
Returning to dispersion relation (2.16), the solutions for the
real part of the
plasmon wave number qx for TM-plasmons on a surface of silver,
with the
permittivity of the metal described by DLM (2.12) surrounded by
vacuum is
shown in Figure 2.3. For small wave numbers the dispersion curve
of the
TM-plasmons (Figure 2.3, curve 1) asymptotically approaches the
dispersion curve of the bulk wave (light-line) in the adjacent
dielectric (Figure 2.3, curve 3). For large wave numbers it
approaches the SP condition ωsp=ωp/(εd+1)1/2,
which also corresponds to the condition εd = -εm(ω) (Figure 2.3,
curve 2). One important result is that curve 3 does not intersect
curve 1, which
represents the dispersion relation of the bulk wave in the
adjacent dielectric.
Consequently, light waves cannot excite SP’s on a flat metal
surface or SP’s
cannot radiate into light waves without appropriate devices,
hence the name
non-radiative SP. This is a consequence of Snell’s law, which
imposes the
condition that the component of the wave vector parallel at the
interface is
continuous. In other words, the wave vector of the surface
plasmon and the
wave vector of the bulk wave do not match at the interface; once
the SP’s are
coupled into a smooth metal surface, they are trapped [20].
31
-
Figure 2.3: Dispersion relation of SP’s on a flat Ag surface,
with
the permittivity of Ag modelled by DLM, λp is the
corresponding
plasmon wavelength and εd = 1 is the permittivity of the
adjacent
dielectric material. See text for description of the curves.
Between the bulk plasma frequency ωp and the surface plasmon
frequency ωsp, no solution corresponding to SP’s exist. This can
also be
shown by substituting (2.11) into (2.16). This frequency gap is
also known as
the plasmon gap [42]. For applied frequencies higher than the
bulk frequency
ωp another solution exist, at which the plasmon wave radiates in
the metal
(Figure 2.3, curve 4) [28, 42]. However, these modes are not
confined to the interface and are of no interest for this
project.
32
-
2.5. Surface plasmons in Three Layer System
An important extension of the simple metal surface is a three
layer system
sometimes also called heterostructure [46], where each of the
layers has an
infinite extension in two dimensions. Two basic heterostructures
can be
distinguished, a dielectric gap in a metal, or MIM
(metal-insulator-metal)
system and a metal film surrounded by two dielectrics, or IMI
(insulator-
metal-insulator) system. The method of analyses for SP’s is
essentially the
same as shown in the previous chapter for a single flat surface.
However,
because of the additional interface the dispersion relation
becomes more
complex, owing to the fact that SP’s on each interface couple
and form new
additional modes when the thickness of the metal film or the gap
width
becomes sufficiently small. The SP’s on each interface can be
excited in
phase and in anti-phase by means of, for instance, the
Kretschmann
configuration, see also section 2.6.2. The new plasmon modes are
also
called film plasmons (for IMI) or gap plasmons (for MIM) [92,
93].
In this section it is assumed that the structure extends
infinitely in the
x- and y- direction and the plasmon propagation is along the
x-axis. In view of
the fact that only TM-plasmons can propagate, the distinction of
the different
plasmon modes is determined by the distribution of the sole
magnetic field
component Hy across the gap or metal film [92, 93].
2.5.1. MIM-Structure (Metal Gap)
Two infinite metal planes separated by a gap filled with a
dielectric or vacuum
in which a film plasmon propagates constitute a MIM-
(metal-insulator-metal)
or simply a slot waveguide [149]. Economou [44] analysed in
detail, for the
first time, the dispersion relation q(ω) by means of the DLM.
Dionne and
colleagues [149] extended the investigation to SP’s propagation
in a Ag-
SiO2-Ag p