See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232031906

Photonic surface states in plasmonic crystals of metallic nanoshells

Article in Physical Review B · September 2011

DOI: 10.1103/PhysRevB.84.115455

CITATIONS READS

16 432

4 authors:

Christos Tserkezis Nikolaos Stefanou

University of Southern Denmark National and Kapodistrian University of Athens
181 PUBLICATIONS 2,228 CITATIONS 192 PUBLICATIONS 6,096 CITATIONS

SEE PROFILE SEE PROFILE

G. Gkantzounis N. Papanikolaou
University of Surrey National Center for Scientific Research Demokritos
43 PUBLICATIONS 711 CITATIONS 143 PUBLICATIONS 5,713 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Christos Tserkezis on 12 May 2018.

The user has requested enhancement of the downloaded file.

 PHYSICAL REVIEW B 84, 115455 (2011)

Photonic surface states in plasmonic crystals of metallic nanoshells

C. Tserkezis,1 N. Stefanou,1 G. Gantzounis,2 and N. Papanikolaou2

1
Section of Solid State Physics, University of Athens, Panepistimioupolis, GR-157 84 Athens, Greece
2
Institute of Microelectronics, NCSR “Demokritos,” GR-153 10 Athens, Greece
(Received 13 July 2011; published 27 September 2011)
We report on the occurrence and properties of photonic surface states in fcc crystals of metallic nanoshells,
by means of full-electrodynamic calculations using the layer-multiple-scattering method, properly extended.
Detailed dispersion diagrams of the surface states associated with the (001) and (111) surfaces are calculated for
such semi-infinite crystals and corresponding finite slabs, and convergence by increasing the slab thickness is
discussed. It is shown that these states can be tuned over a broad frequency range by varying the shell thickness and
can be characterized, along high-symmetry directions, according to their symmetry. Absorption in the metallic
material limits the propagation length which can, however, be as long as several tens of lattice constants for
low-loss metals and relatively broad bands.

DOI: 10.1103/PhysRevB.84.115455 PACS number(s): 42.70.Qs, 42.25.Bs, 73.20.Mf, 78.67.Pt

I. INTRODUCTION each metallic cylinder coated with a dielectric layer. The metal
was considered as perfect conductor and also modeled with
The possibility of electromagnetic (EM) modes localized
a Drude-like dielectric function that takes strongly negative
at the surface of a semi-infinite photonic crystal was initially
values in the frequency region of interest. Starting from a
theoretically investigated by Meade et al.,1 who argued that
crystal of dielectric cylinders that supports photonic surface
such modes must always exist for some termination of any
states and inserting a thin metallic cylinder at the center of
surface of the crystal, and was subsequently experimentally
each dielectric cylinder, it was shown that the surface states
demonstrated.2 Motivated by these works, several numerical
gradually disappear by increasing the radius of the metallic
studies of surface states in two-dimensional (2D) and three-
cylinders. To the best of our knowledge, there is no study
dimensional (3D) photonic crystals, based mostly on the
of surface states in 3D metallodielectric photonic crystals as
supercell approach, have been reported,3–9 and the dependence yet. Here we show, by means of rigorous full-electrodynamic
of such states on the surface termination has been explored.10,11 calculations, that fcc crystals of metallic nanoshells, which
Interest in surface waves has been reignited in recent years, also can be easily fabricated using self-assembly methods,30–32 do
in relation to spoof plasmons in corrugated perfect-conductor support surface EM modes. It is also worth noting that, while
surfaces,12,13 and to the potential appearance of forward and surface states are usually introduced in a photonic crystal by
backward Tamm and Shockley photonic states,14–16 while either terminating a surface at an appropriate point1,10 or by
several applications exploiting surface states in 2D and 3D adding a surface defect layer,33,34 in our case the surface is
photonic crystals have been proposed. When appearing in well defined and no defect layer is needed.
dielectric photonic crystals, surface waves can be considered
as efficient replacements of surface plasmons in metallic films,
II. METHOD OF CALCULATION
thanks to their much lower absorptive losses.17,18 Moreover,
surface modes can be excited using diffraction gratings and Our calculations are based on the full-electrodynamic layer-
can be engineered to collimate light exiting a photonic crystal multiple-scattering (LMS) method,35–37 which is ideally suited
waveguide.19–24 In addition, they can be of particular interest for the study of surface modes in metallodielectric crystals.
in the field of photonic metamaterials, since they can transfer The method can describe not only a finite slab of a given
the evanescent components of an incident wave field to the crystal, but also the corresponding semi-infinite crystal exactly
other side of a negatively refracting slab, thus enabling for and not by appealing to the commonly employed supercell
subwavelength imaging,25–27 and can be engineered to have approximation. Moreover, since the method solves Maxwell’s
opposite phase and group velocities.28 Recently, Ishizaki and equations in the frequency domain, it can treat dispersive and
Noda experimentally observed surface states in 3D photonic dissipative materials like metals in a straightforward manner.
crystals as well,29 and established a new route for photon The concept of the surface is inherent in the LMS method,
manipulation. because the structure under consideration is built as a sequence
Although surface states in 2D—and to a lesser extent of successive layers of scatterers arranged with the same 2D
3D—dielectric photonic crystals have been adequately studied periodicity parallel to a given crystallographic plane, which is
in the past two decades, this is not the case for metallodielectric taken to be the x-y plane. The properties of the individual
crystals. Zhang et al.5 reported that such states could not be scatterers enter only through the corresponding scattering
easily encountered in 2D crystals of metallic rods due to T matrix which, for homogeneous spherical particles, is
the expulsion of the EM field from the surface because of given by the closed-form solutions of the Mie scattering
the presence of the metallic material, though they did not problem,36 while for spheres consisting of a number of
exclude such a possibility. They considered two different kinds concentric, homogeneous spherical shells, it can be calculated
of metallodielectric crystals in a square lattice with cylindrical by an efficient recursive algorithm.38 At a first step, in-plane
scatterers, which are either pure metallic or of layered type with multiple scattering is evaluated in a spherical-wave basis

1098-0121/2011/84(11)/115455(5) 115455-1 ©2011 American Physical Society

TSERKEZIS, STEFANOU, GANTZOUNIS, AND PAPANIKOLAOU PHYSICAL REVIEW B 84, 115455 (2011)

using the single-scatterer T matrix and appropriate propagator By definition, each eigenmode propagates through the crystal
functions. Subsequently, interlayer scattering is calculated without changing its state and, on the other hand, for a
in a plane-wave basis defined by the reciprocal vectors g semi-infinite crystal, there is no rear surface to reflect the
of the given 2D lattice and the linear polarization state forward into backward Bloch waves. Therefore, the appro-
p = 1,2 of the plane wave beam, for given angular frequency, priate boundary condition for the scattering problem of an
ω, and reduced in the surface Brillouin zone (SBZ) wave EM wave incident on a semi-infinite photonic crystal from
vector, k , which are conserved quantities in the scattering the homogeneous host material that extends to infinity is
process. The waves transmitted through and reflected from V− = 0,42 and then Eq. (2) yields
a single layer are obtained by appropriate transmission and
reflection matrices: QIgp;g p , QIII E−
0 =F
−+ ++ −1 +
[F ] E0 ≡ R∞ E+
0. (3)
gp;g p for incidence from the
IV II On the other hand, the condition for the existence of surface
left (z < 0) and Qgp;g p , Qgp;g p for incidence from the right
(z > 0), respectively.36,37 The Q matrices of slabs consisting modes is to have nonzero forward Bloch modes (V+ = 0) in
of successive layers with the same 2D periodicity are obtained the absence of incoming field (E+
0 = 0) and then Eq. (2) gives
9

by properly combining the corresponding matrices of the

F++ V+ ≡ E+
0 = 0, (4)
component layers. The ratio of the transmitted or reflected
energy flux to the energy flux associated with the incident wave which is satisfied when
defines the transmittance, T , or reflectance, R, respectively, of
det[F++ ] = 0. (5)
a given slab. The possible eigenmodes of the slab are obtained
by requiring the existence of a wave field localized within the
slab in the absence of incident wave. Dividing the slab into III. RESULTS AND DISCUSSION
a left and a right part, with corresponding reflection matrices We consider an fcc crystal of composite spherical nanopar-
QIIL and QIII
R , this requirement leads to the secular equation
39
ticles consisting of a silica core (εsilica = 2.13; μsilica = 1) of
  radius R, coated with a concentric metallic shell of thickness
R = 0.
det I − QIIL QIII (1) D, so that S = R + D is the total radius of the particle. We
For a 3D crystal consisting of an infinite periodic sequence assume that the metallic material is described by the simple,
of layers stacked along the z direction, Bloch’s theorem leads yet effective, Drude relative permittivity43
to an eigenvalue problem of either the scattering or the transfer ωp2
matrix,40 which provides the z component of the Bloch wave εm = 1 − , (6)
ω(ω + iτ −1 )
vector, kz , for given ω and k . The solutions kz (ω,k ), looked
upon as functions of real ω, define, for each k , lines in the where ωp is the bulk plasma frequency and τ is the relaxation
complex kz plane. These so-called real-frequency lines, taken time of the conduction-band electrons that accounts for
together, constitute the complex band structure of the infinite dissipative losses, and relative permeability μm = 1. For
crystal associated with the given crystallographic plane. A convenience, we use throughout this paper ωp as the frequency
line of given k may be real (in the sense that kz is real) over unit and c/ωp as the length unit (considering a typical value
certain frequency regions, and be complex (in the sense that of 10 eV for h̄ωp , c/ωp corresponds to about 20 nm). Metallic
kz is complex) for ω outside these regions. It turns out that, for nanoshells are easily tunable resonant units; plasmons of
given k and ω, out of the solutions kz (ω,k ), none or, at best, the outer and inner surfaces of the shell interact with each
a few are real and the corresponding eigenvectors represent other, giving rise to coupled resonant modes, one below the
propagating modes of the EM field in the given infinite lower (particle-like) and one above the higher (cavity-like)
crystal. The remaining solutions kz (ω,k ) are complex and plasmon modes.44,45 The interaction and the resulting level
the corresponding eigenvectors represent evanescent waves. shifts increase as the overlap between the corresponding wave
These have an amplitude which increases exponentially in the fields becomes larger, i.e., by reducing the shell thickness, and
positive or negative z direction and, unlike propagating waves, is more pronounced for the dipole plasmon modes because of
do not exist as physical entities in the infinite crystal. However, their relatively larger spatial extent.46 We take R = 0.7c/ωp
they are an essential part of the physical solutions of the EM and D = 0.3c/ωp , in which case the eigenfrequency of the
field in the case of a surface or a slab of finite thickness. A fundamental dipole particle-like plasmon mode (0.37ωp ) is
region of frequency where propagating waves do not exist, for significantly lower than that of the corresponding homoge-
given k , constitutes a frequency gap of the EM field for the neous metallic sphere of radius c/ωp (0.52ωp ). The fcc √ crystal
given k . If, over a frequency region, no propagating wave under consideration has a lattice constant a = 2.1 2c/ωp ,
exists whatever the value of k , then this region constitutes an which corresponds to nearly touching particles, and we are
absolute frequency gap. looking for surface states localized at its (001) and (111)
The reflection matrix of the corresponding semi-infinite surfaces. Therefore, in the framework of the LMS method, we
crystal, R∞ , is obtained from the set of eigenvectors of the have to build the crystal in two different ways: as a sequence of
transfer matrix by properly constructing a matrix, F, which successive (001) and (111) layers of nanoshells, with nearest-
projects the space of forward and backward Bloch eigenmodes, neighbor distance a0 = 2.1c/ωp , as shown schematically in
V+ and V− , onto the original plane-wave basis, as follows:40,41 Fig. 1. Successive (001) and√(111) layers are separated by a
 +   ++   + distance d = a/2 and d = a 3/3, respectively.
E0 F F+− V In the left diagrams of Figs. 2 and 3, we depict the
− = −+ −− . (2)
E0 F F V− photonic band structure of the given crystal along the X

115455-2
PHOTONIC SURFACE STATES IN PLASMONIC CRYSTALS . . . PHYSICAL REVIEW B 84, 115455 (2011)

FIG. 2. (Color online) Left-hand diagram: photonic band struc-

ture of an fcc crystal of composite spherical particles, consisting of
a silica core and a metallic shell described by the Drude dielecric
function without dissipative losses (core radius R = 0.7c/ωp , shell
thickness D = 0.3c/ωp , and nearest-neighbor distance in the fcc
crystal a0 = 2.1c/ωp ), along the X direction. Thick and thin
lines correspond to doubly degenerate and nondegenerate bands,
respectively. Right-hand diagram: projection of the photonic band
FIG. 1. (001) (left) and (111) (right) surfaces of the fcc lattice structure on the SBZ of the (001) surface along the symmetry lines
(top panel), and the corresponding reduced k zones (bottom panel). shown in the inset. Shaded and blank regions represent frequency
The conventional (bulk) Brillouin zone of the fcc lattice is shown for bands and gaps, respectively. With solid lines in gap regions, we
comparison (shaded decatetrahedron). show the dispersion curves of the surface modes. The dashed lines
denote the light cone in air.
and L directions, respectively, calculated by deliberately
disregarding absorption in the metallic material, taking τ −1 = are even, while those about 0.35ωp and 0.42ωp are odd
0 in Eq. (6), in order to be able to interpret our results in upon reflection with respect to the x-z plane. Similarly, along
an unambiguous manner. At low frequencies, we obtain a the M direction, the modes about 0.35ωp are even upon
linear, doubly degenerate dispersion curve, as expected for reflection with respect to the corresponding vertical plane.
propagation in a homogeneous medium characterized by a Along other directions of the (001) SBZ, as well as along any
frequency-independent effective refractive index. At higher direction of the (111) SBZ, there are no such planes of mirror
frequencies, flat, almost dispersionless bands are formed, symmetry and thus a symmetry characterization of the bands
about ∼0.3ωp and ∼0.4ωp , from the dipole and quadrupole of surface states is not possible.
particle-like plasmon modes of the individual nanoshells, In order to examine the tunability of these surface states,
respectively. Frequency gaps open up due to hybridization we carried out systematic calculations by varying the shell
between the effective-medium-like bands and flat bands of thickness of the individual scatterers, keeping their total radius
the same symmetry,46,47 giving rise to the band diagrams and the nearest-neighbor distance constant. In Fig. 4, we
shown in Figs. 2 and 3. Next to these diagrams, we show display the projection of the band structure on the SBZ of
the projection of the photonic band structure on the SBZ of the (001) surface, and the corresponding bands of surface
the corresponding surface along the symmetry lines shown states, for the limiting cases of very thin metallic shells, with
in the insets. The shaded regions extend over the frequency D = 0.1c/ωp and R = 0.9c/ωp , and homogeneous metallic
bands of the EM field: at any one frequency within a shaded
region, for given k , there exists at least one propagating
EM mode in the infinite crystal. The blank regions represent
frequency gaps for the given k . With solid lines, we show the
dispersion of surface modes, while the dashed lines denote the
light cone. These modes, which originate from interference
effects in the photonic crystal,1,5 lie indeed outside the light
cone and in gap regions, i.e., they are true surface states that
decay exponentially in the crystal and in the outer region.
Interestingly, many of the calculated dispersion curves of
surfaces states have segments with negative slope and can
be associated with backward-propagating surface waves.28
Along high-symmetry directions of the (001) SBZ, the
surfaces states can be classified as odd or even upon reflection
with respect to appropriate vertical planes of mirror symmetry. FIG. 3. (Color online) Same as Fig. 2, for the L direction and
For example, along the X direction, the modes about 0.28ωp the (111) surface of the crystal.

115455-3
TSERKEZIS, STEFANOU, GANTZOUNIS, AND PAPANIKOLAOU PHYSICAL REVIEW B 84, 115455 (2011)

FIG. 4. (Color online) Projection of the photonic band structure FIG. 5. (Color online) (a) Enlarged view of band 1 of Fig. 2
of fcc crystals of (a) composite spherical particles consisting of a silica and the dispersion of corresponding surface modes calculated using
core and a metallic shell described by the Drude dielectric function Eq. (1) for finite slabs consisting of two, four, and eight layers
without dissipative losses (core radius R = 0.9c/ωp ; shell thickness of nanoshells (thin solid, dashed, and dotted lines, respectively).
D = 0.1c/ωp ) and (b) corresponding homogeneous metallic spheres (b) Estimated propagation length of the surface modes under
with radius S = c/ωp , on the SBZ of their (001) surface (nearest- consideration.
neighbor distance in both fcc crystals a0 = 2.1c/ωp ), along the
symmetry lines shown in the inset of Fig. 2. Shaded and blank regions the interaction of these modes becomes smaller, and the results
represent frequency bands and gaps, respectively. With solid lines in
converge to the single dispersion curve that characterizes the
gap regions, we show the dispersion curves of the surface modes,
semi-infinite crystal. This is shown in Fig. 5(a) for band 1 of
while the dashed lines denote the light cone in air.
Fig. 2 and for slabs consisting of two, four, and eight (001)
layers of nanoshells. Interestingly, convergence is faster along
spheres with radius S = c/ωp . It can be seen that, as discussed the XM direction, which indicates that, in this case, the surface
in Ref. 46, variation of the shell thickness results mainly in a states are more strongly localized, as we verified by calculating
frequency shift of the plasmon modes and the corresponding the relevant attenuation length.
band gaps. For example, the position of band 1 shown in Absorptive losses, which have been ignored so far, can
Fig. 2 shifts from ∼0.15ωp for the crystal of thin nanoshells be taken into account by including a small imaginary part
to ∼0.35ωp for the crystal of homogeneous metallic spheres. in the Drude dielectric function. We set τ −1 = 0.002ωp in
Interestingly, while the partial band gaps associated with the Eq. (6), a value which is appropriate for silver. Then Eq. (5)
dipole particle-like plasmon modes retain almost the same is satisfied in the lower complex frequency plane, at ω − iγ ,
form and their maximum width increases by only about 25%, where γ > 0 denotes the inverse of the lifetime of the mode.
those associated with the quadrupole particle-like plasmon The mode lifetime, multiplied by the corresponding group
modes widen drastically for all points of the SBZ as the velocity, vg = dω/dk, along a given direction in the surface,
thickness of the metallic shell increases, and finally an absolute provides an estimate of the propagation length, L, of this mode.
gap is formed. It is also worth noting that, unlike Ref. 5, where In Fig. 5(b), we display the propagation length calculated in
surface states were only encountered for photonic crystals of this way for band 1 of Fig. 2. It can be seen that, even for this
metallic rods coated by a thick dielectric layer, in our case relatively flat band, a propagation length as long as 25a0 can
surface states do appear even for crystals of homogeneous be achieved. Similar calculations for more extended bands of
metallic spheres. The presence of metallic material does not surface states yield propagation lengths as long as 80a0 , which
prohibit the appearance of surface states in a photonic crystal, makes these modes interesting, also in view of photon transport
provided that the dielectric function of the metal is not highly applications. These propagation lengths are of the order of a
negative in the frequency region of interest. Moreover, our few microns, while surface plasmon polaritons on a flat silver
calculations show that the geometry of the crystal is also an surface have propagation lengths in the range 10–100 μm in the
important factor affecting the occurrence of surface states; for visible spectrum, which increase toward 1 mm as one moves
example, such states tend to disappear if we increase the lattice into the 1.5 μm near-infrared telecommunication band.49
constant, i.e., decrease the volume fraction of the nanoshells.
The dispersion curves of the surface states presented in
IV. CONCLUSION
Figs. 2–4 have been calculated for the semi-infinite photonic
crystal according to Eq. (5). In addition, they can be obtained To summarize, we extended the LMS method so as to
by assuming corresponding finite slabs, using Eq. (1), a method provide the surface states of a semi-infinite photonic crystal
which has already been employed in the study of surface and applied it to the specific case of fcc crystals of metallic
states in phononic crystals.48 Obviously, for a finite slab we nanoshells. In contrast to earlier claims that surface states
obtain two dispersion curves, close to each other in frequency, are unlikely in metallodielectric photonic crystals, our study
corresponding to surface modes localized at the two surfaces of reveals the existence of surface states localized at the (001)
the slab. As the number of layers comprising the slab increases, and (111) surfaces of the crystals under consideration, which

115455-4
PHOTONIC SURFACE STATES IN PLASMONIC CRYSTALS . . . PHYSICAL REVIEW B 84, 115455 (2011)

can be tuned over a broad frequency range by varying the ACKNOWLEDGMENTS

shell thickness. We characterized these states according to their
This research has been co-financed by the European Union
symmetry along high-symmetry directions of the SBZ and dis-
cussed the convergence of the associated dispersion curves, as [European Social Fund (ESF)] and Greek national funds
obtained for finite slabs of increasing thickness. Moreover, we through the Operational Program “Education and Lifelong
provided an estimate of the corresponding propagation length, Learning” of the National Strategic Reference Framework
which can be as long as several tens of lattice constants for rel- (NSRF), Research Funding Program: “Heracleitus II. Invest-
atively broad bands and nanoshells made of a low-loss metal. ing in knowledge society through the European Social Fund.”

1 27
R. D. Meade, K. D. Brommer, A. M. Rappe, and J. D. Joannopoulos, R. Chatterjee, N. C. Panoiu, K. Liu, Z. Dios, M. B. Yu, M. T. Doan,
Phys. Rev. B 44, 10961 (1991). L. J. Kaufman, R. M. Osgood, and C. W. Wong, Phys. Rev. Lett.
2
W. M. Robertson, G. Arjavalingam, R. D. Meade, K. D. Brommer, 100, 187401 (2008).
28
A. M. Rappe, and J. D. Joannopoulos, Opt. Lett. 18, 528 (1993). S. Foteinopoulou, M. Kafesaki, E. N. Economou, and C. M.
3
P. Etchegoin and R. T. Phillips, Phys. Rev. B 53, 12674 (1996). Soukoulis, Phys. Rev. B 75, 245116 (2007).
4 29
J. M. Elson and P. Tran, Phys. Rev. B 54, 1711 (1996). K. Ishizaki and S. Noda, Nature (London) 460, 367 (2009).
5 30
X. Zhang, L.-M. Li, Z.-Q. Zhang, and C. T. Chan, Phys. Rev. B 63, C. Graf and A. van Blaaderen, Langmuir 18, 524 (2002).
31
125114 (2001). J. Zhang, X. Chen, W. Hu, J. Shao, W. Lu, and Z. Wang, Solid State
6
M. Qiu and S. He, Phys. Lett. A 282, 85 (2001). Commun. 145, 582 (2008).
7 32
S. Enoch, E. Popov, and N. Bonod, Phys. Rev. B 72, 155101 (2005). S. Uchida, K. Yamamura, and N. Zettsu, Thin Solid Films 518,
8
M. Che and Z.-Y. Li, J. Opt. Soc. Am. A 25, 2177 (2008). 3581 (2010).
9 33
F. J. Lawrence, L. C. Botten, K. B. Dossou, R. C. McPhedran, and B. Wang, W. Dai, A. Fang, L. Zhang, G. Tuttle, Th. Koschny, and
C. M. de Sterke, Phys. Rev. A 82, 053840 (2010). C. M. Soukoulis, Phys. Rev. B 74, 195104 (2006).
10 34
F. Ramos-Mendieta and P. Halevi, Phys. Rev. B 59, 15112 (1999). Y.-C. Hsu and L.-W. Chen, J. Opt. 12, 095709 (2010).
11 35
Y. A. Vlasov, N. Moll, and S. J. McNab, Opt. Lett. 29, 2175 (2004). N. Stefanou and A. Modinos, J. Phys. Condens. Matter 3, 8135
12
J. B. Pendry, L. Martı́n-Moreno, and F. J. Garcia-Vidal, Science (1991).
36
305, 847 (2004). N. Stefanou, V. Yannopapas, and A. Modinos, Comput. Phys.
13
F. J. Garcia-Vidal, L. Martı́n-Moreno, and J. B. Pendry, J. Opt. A: Commun. 113, 49 (1998).
37
Pure Appl. Opt. 7, S97 (2005). N. Stefanou, V. Yannopapas, and A. Modinos, Comput. Phys.
14
N. Malkova and C. Z. Ning, Phys. Rev. B 73, 113113 (2006). Commun. 132, 189 (2000).
15 38
A. Namdar, I. V. Shadrivov, and Y. S. Kivshar, Appl. Phys. Lett. N. Stefanou, C. Tserkezis, and G. Gantzounis, Proc. SPIE 6989,
89, 114104 (2006). 698910 (2008).
16 39
A. P. Vinogradov, A. V. Dorofeenko, A. M. Merzlikin, and A. A. N. Stefanou, G. Gantzounis, and C. Tserkezis, Int. J. Nanotechnol.
Lisyansky, Phys. Usp. 53, 243 (2010). 6, 137 (2009).
17 40
W. M. Robertson and M. S. May, Appl. Phys. Lett. 74, 1800 (1999). C. Tserkezis and N. Stefanou, Phys. Rev. B 81, 115112
18
M. Shinn and W. M. Robertson, Sensors Actuators B 105, 360 (2010).
41
(2005). L. C. Botten, N. A. Nicorovici, R. C. McPhedran, C. Martijn de
19
E. Moreno, F. J. Garcı́a-Vidal, and L. Martı́n-Moreno, Phys. Rev. Sterke, and A. A. Asatryan, Phys. Rev. E 64, 046603 (2001).
42
B 69, 121402(R) (2004). Z.-Y. Li and K.-M. Ho, Phys. Rev. B 68, 155101 (2003).
20 43
P. Kramper, M. Agio, C. M. Soukoulis, A. Birner, F. Müller, N. W. Ashcroft and N. D. Mermin, Solid State Physics (Saunders,
R. B. Wehrspohn, U. Gösele, and V. Sandoghdar, Phys. Rev. Lett. New York, 1976).
44
92, 113903 (2004). E. Prodan, C. Radloff, N. J. Halas, and P. Nordlander, Science 302,
21
E. Moreno, L. Martı́n-Moreno, and F. J. Garcı́a-Vidal, Photon. 419 (2003).
45
Nanostruct. Fundam. Appl. 2, 97 (2004). T. V. Teperik, V. V. Popov, and F. J. Garcı́a de Abajo, Phys. Rev. B
22
A. I. Rahachou and I. V. Zozoulenko, J. Opt. Soc. Am. B 23, 1679 69, 155402 (2004).
46
(2006). C. Tserkezis, G. Gantzounis, and N. Stefanou, J. Phys. Condens.
23
W. Śmigaj, Phys. Rev. B 75, 205430 (2007). Matter 20, 075232 (2008).
24 47
H. Caglayan, I. Bulu, and E. Ozbay, Appl. Phys. Lett. 92, 092114 G. Gantzounis and N. Stefanou, Phys. Rev. B 72, 075107
(2008). (2005).
25 48
C. Luo, S. G. Johnson, J. D. Joannopoulos, and J. B. Pendry, Phys. R. Sainidou, B. Djafari-Rouhani, and J. O. Vasseur, Phys. Rev. B
Rev. B 68, 045115 (2003). 77, 094304 (2008).
26 49
S. Xiao, M. Qiu, Z. Ruan, and S. He, Appl. Phys. Lett. 85, 4269 W. L. Barnes, A. Dereux, and T. W. Ebbesen, Nature (London) 424,
(2004). 824 (2003).

115455-5

View publication stats