Functional Plasmonic Nanocircuits with Low Insertion and
Propagation Losses
Arian Kriesch,
*
,
†
,
‡
Stanley P. Burgos,
‡
Daniel Ploss,
†
Hannes Pfeifer,
†
Harry A. Atwater,
‡
and Ulf Peschel
†
†
Institute of Optics, Information and Photonics, Erlangen Graduate School in Advanced Optical Technologies,
Friedrich-Alexander-University Erlangen-Nuremberg (FAU) and Max Planck Institute for the Science of Light, 91058 Erlangen,
Germany
‡
Kavli Nanoscience Institute, California Institute of Technology, Pasadena, California 91125, United States
*
S
Supporting Information
ABSTRACT:
We experimentally demonstrate plasmonic
nanocircuits operating as subdi
ff
raction directional couplers
optically excited with high e
ffi
ciency from free-space using
optical Yagi-Uda style antennas at
λ
0
= 1550 nm. The optical
Yagi-Uda style antennas are designed to feed channel plasmon
waveguides with high e
ffi
ciency (45% in coupling, 60% total
emission), narrow angular directivity (<40
°
), and low insertion
loss. SPP channel waveguides exhibit propagation lengths as
large as 34
μ
m with adiabatically tuned con
fi
nement and are
integrated with ultracompact (5
×
10
μ
m
2
), highly dispersive
directional couplers, which enable 30 dB discrimination over
Δ
λ
= 200 nm with only 0.3 dB device loss.
KEYWORDS:
SPP, circuit, antenna, waveguide, directional coupler, Yagi-Uda
S
urface plasmon polariton (SPP) waveguides are uniquely
advantaged by their high con
fi
nement, allowing for
subwavelength integration. This is a requirement for integrating
optics with a footprint size that is comparable with electronic
circuits, thus enabling plasmonic-electronic hybrid integration,
a path that has been repeatedly highlighted as a future key
application of plasmonics.
1
−
4
However, high con
fi
nement in
plasmonics usually increases loss due to the larger
fi
eld overlaps
with the metal. The second major obstacle to deep
subwavelength plasmonics is high insertion loss due to the
limited modal
fi
eld overlap of less-con
fi
ned waveguide schemes
like Si integrated photonics
5,6
or optical
fi
bers,
7
thus intrinsi-
cally limiting the performance of hybrid dielectric-plasmonic
circuits.
8
Here, we illustrate how in-circuit-loss can be mitigated by
restricting strong optical con
fi
nement only to components
where it is absolutely essential (Figure 1c), and how insertion-
loss can be addressed by coupling light into plasmonic
nanocircuits via impedance-matched optimized Yagi-Uda
9
style nanoantennas (Figure 1a,d). Using this platform, we
experimentally demonstrate optical directional couplers
10,11
integrated on a micrometer scale that show unusually strong
spectral dispersion, a key prerequisite for integrated wavelength
division multiplexers.
To implement these device concepts, we use SPP channel
waveguides (also called SPP gap waveguides),
12,13
which o
ff
er
maximum con
fi
nement
14
in a narrow rectangular gap etched a
few hundred nanometers into a metal
fi
lm. We note that this
waveguide geometry does not su
ff
er from optical mode cuto
ff
when scaled down. By
fi
lling the air gap (
n
= 1) in the
plasmonic waveguide with the substrate material silica (
n
≈
1.45) (Figure 1b) we make the modal
fi
eld distribution more
symmetric, therefore eliminating an upper modal cuto
ff
that
otherwise prohibits larger waveguide channel widths (>80 nm
for usual Au channel waveguides), therefore limiting the
propagation length to below 10
μ
m. Wherever possible, the
connections from one functional plasmonic unit to the next
must be bridged with low loss plasmonic waveguides. As the
waveguide mode is basically maintained for di
ff
erent channel
widths, easy to fabricate adiabatic waveguide tapers can form
the transition from highly con
fi
ning to low loss sections. We
experimentally demonstrate t
hat the investigated circuits
achieve a propagation length of
L
P/e
=34
μ
m, while they are
still subwavelength with a waveguide gap width of 300 nm and
that they have an e
ff
ective refractive index of
n
eff
≈
1.54 at
λ
0
=
1550 nm with low dispersion.
To reduce insertion loss, each presented nanoplasmonic
circuit utilizes at least two connected Yagi-Uda antennas to
enhance coupling e
ffi
ciency from a focused laser beam and to
achieve narrow directionality (Figure 1a,d). We measured the
antenna and waveguide properties spectrally and derive their
fundamental properties: e
ffi
ciency (45% in coupling, 60% total
emission), directionality (<40
°
), and spectral dispersion.
Received:
July 13, 2013
Revised:
August 14, 2013
Published:
August 20, 2013
Letter
pubs.acs.org/NanoLett
© 2013 American Chemical Society
4539
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
Terms of Use
For our experiments, embedded SPP nanocircuits (Figures
1c and 6b) were fabricated in a two-step process. A 220 nm thin
Au
fi
lm on a thin 160
μ
m thick, polished quartz glass substrate
was patterned by lift-o
ff
with a 100 kV electron beam
lithography system. The designed minimum feature size at
the antenna gap (Figure 1e,h) was 80 nm. Subsequently,
fl
owable spin-on glass (Filmtronics 400F) was spin-coated and
baked to form a dielectric cladding layer. Scanning electron
microscope images of device cross sections, cut with a focused
ion beam (FIB) at di
ff
erent positions, show that the cladding
layer
fi
lls all gaps, covers the whole structure without cavities
(Figure 1f
−
h), and forms a smooth surface that is
∼
320 nm
higher than the underlying structures. The fabricated circuits
consistently deviate less than 15
°
from perfectly rectangular
sidewall angles. To investigate the dependence of the optical
properties of each circuit component on design parameters,
arrays of circuits were fabricated, while systematically varying
waveguide length and bend radii (Figure 1c).
Spectral transmission measurements of each circuit were
taken on an optical far-
fi
eld excitation and imaging setup
15
with
a 4 W supercontinuum light source, spectrally
fi
ltered to
λ
=
1200
−
1850 nm at 5 nm fwhm with an acousto optical tunable
fi
lter (AOTF) (for sketch and details see Supporting
Information). A collimated and linearly polarized laser beam
was focused by an objective (NA = 0.9 dry respectively 1.3 oil
immersion) to a di
ff
raction-limited spot (Ø < 2
μ
m). The
sample was positioned with 5 nm resolution on a stabilized
piezo stage to couple maximum power into the optical antennas
(Figure 1a,d). The whole circuit was simultaneously imaged
with 150
×
or 300
×
magni
fi
cation through the excitation
objective, a polarization
fi
lter and NIR imaging optics on an
infrared InGaAs CCD camera (Xenics XS, 320
×
256 pixels).
With this setup, the linearly polarized emission from output
antennas, turned by 90
°
with respect to the input antenna
(Figure 1c), was imaged with enhanced dynamic range by
suppressing the back-re
fl
ection from the incident beam with a
ratio of up to 10 000. Absolute circuit transmission values were
determined by normalizing with the spectral re
fl
ectivity of a
silica
−
air interface on the same sample (see Supporting
Information for details). With additional Fourier imaging, the
polarization dependent angular emission pattern of the
antennas was analyzed.
For any application, including probing the circuit properties
presented here, light has to be coupled into and out of the
guided mode of the plasmonic waveguide with highest possible
power conversion e
ffi
ciency.
2
Plasmonic antennas are best
suited to ful
fi
ll this task for exciting channel plasmons from
freespace.
16,17
The concept of optical antennas has already been
successfully transferred from w
ell-proven radiofrequency
technology to the optical and near-infrared with a resulting
antenna size in the micrometer range.
18
−
20
Here we apply the
macroscopic design concept of Yagi-Uda antennas
9
in the near-
infrared
21
−
25
and connect them to feed the plasmonic circuits.
For the investigated antennas, the feed element, an
approximately 1
μ
m long, 100 nm wide, split Au rod, is
centrally illuminated with a focused laser beam (Ø < 2
μ
m) that
is polarized parallel to the antenna arms (Figure 1a,d). This
element is driven at its resonance frequency and feeds the 300
nm wide SPP channel waveguide. An additional nonresonant
Yagi-re
fl
ector element is placed in a distance of 600 nm to the
feed element to constructively re
fl
ect the electromagnetic
fi
eld
back into the waveguide, thus enhancing the in-coupling
e
ffi
ciency.
Our whole design was developed using an iterative particle-
swarm-optimization algorithm based on full 3D FDTD
simulations while taking into account the limitations of
fabrication and real optical material parameters as determined
with ellipsometry. As it was found that in-coupling e
ffi
ciency
increases with decreasing size of the antenna gap,
21
we limited
the antenna gap dimension to 80 nm based on fabrication
limitations. A taper was added to connect the antenna with a
300 nm wide low-loss waveguide to minimize impedance
mismatch and back re
fl
ections.
To simplify the experimental analysis of our circuits we
attributed lumped properties to each element (Figure 2a).
1
The
antennas were assigned a characteristic spectral transmission
T
ant
(
λ
)=
P
wg
(
λ
)/
P
freespace
(
λ
) that represents the ratio of power,
which is converted into the waveguide mode for a speci
fi
c
wavelength, a spectral absorption
A
ant
(
λ
) due to ohmic losses
and a re
fl
ectivity
R
ant
(
λ
) at the antenna-waveguide connection
caused by residual impedance mismatch.
1,17
For the wave-
guides, we assigned a length-dependent transmission
T
wg
(
λ
,
L
)
=1e
−
L/L
0
that is quanti
fi
ed in terms of the propagation length
L
0
of their supported SPPs. Hence,
L
0
is separated by multiple
measurements over a systematic variation of the waveguide
Figure 1.
(a) Optical loaded Yagi antennas are optimized to couple a
highly focused, linearly polarized laser beam into waveguides (b) that
are covered with a cladding layer of spin-on-glass. (c) SEM of an array
of multicomponent plasmonic nanocircuits and (d) of a single Yagi
antenna. Cross-section cuts with a focused ion beam device (FIB), (f)
through the 300 nm wide waveguide, (g) the antenna connector, and
(h) the antenna with a minimum feature size of only 80 nm display the
sandwich of a substrate layer of silica, the 220 nm thick metal
structures and the 320 nm thick cladding layer of spin-on-glass, which
completely
fi
lls the metal gaps.
Nano Letters
Letter
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
4540
length
L
. In applying this model to our experiments, we further
assume that in- and out-coupling antennas had similar,
inversion-invariant properties. This is a reasonable approx-
imation as we inject and extract light using the same objective.
Hence, the total system transmission is given by
T
tot
(
λ
)=
T
ant1
(
λ
)
T
wg
(
λ
)
T
ant2
(
λ
)=
T
ant
(
λ
)
2
T
wg
(
λ
).
We experimentally measured the spectral system trans-
mission
T
tot
(
λ
) in the range of
λ
= 1200
−
1850 nm in steps of
5 nm, coupling in and out of the circuit from air or through the
silica substrate. An ensemble of 20 (5
×
5) di
ff
erent plasmonic
nanocircuits was characterized (Figure 1c), repeated on 4
di
ff
erent samples with the same fabrication settings. Each
parameter variation ensemble contained circuits with di
ff
erent
total waveguide lengths (
L
= 12.28
−
42.28
μ
m) and di
ff
erent
radii (
R
=1
−
4
μ
m) for the 90
°
bend that turns the waveguide
mode polarization for the emitting antenna. A systematic
analysis of the bend loss over radius demonstrated that bends
with
R
=3,4
μ
mo
ff
er negligible additional loss compared to
the linear waveguide propagation loss, which is comprehensible
since the modal e
ff
ective wavelength di
ff
erence for those radii is
negligible compared to straight waveguides. Hence, we focused
our analysis to circuits with a 3
μ
m bend radius.
On the basis of this
fi
nding, we
fi
tted the transmissions of the
circuits with an iterative two-dimensional least-squares
fi
t over
the independent variables wavelength and waveguide length.
The antenna transmission was modeled assuming a Lorentzian
spectral broadening response,
T
ant
=
T
ant
max
/[1 + ((
λ
−
λ
0
)/
γ
)
2
].
Under the assumption that dispersive e
ff
ects of the in- and
outcoupling are solely caused by the antenna resonance, we
fi
nd that the peak wavelength
λ
0
and line width
γ
are
independent of the excitation, whereas the maximum antenna
e
ffi
ciency
T
ant
max
is higher for excitation from and emission into
thesubstratethanintoairduetotherefractiveindex
asymmetry of the structure,
n
substrate
≈
1.44 >
n
air
=1.
Figure 2.
(a) The plasmonic circuits are analyzed as a sequence of
black boxes, each with a characteristic, spectral transmission
T
(
λ
),
absorption loss
A
(
λ
), and re
fl
ection coe
ffi
cient
R
(
λ
) toward the
waveguide. (b) Experimentally determined lower bound for the
spectral e
ffi
ciency of the Yagi antennas
T
ant
(
λ
) reaching a spectral peak
value of 60% of intensity to couple the bound mode inside the
waveguide out into air (15%) and silica (45%), respectively, from a
focused beam into the waveguide. The antenna e
ffi
ciency was
determined by
fi
tting measured data with a Lorentzian line shape
and separating loss e
ff
ects. The shaded areas indicate
±
σ
error bands.
Figure 3.
A narrow and strongly linear polarized angular emission directionality of the Yagi antenna at
λ
= 1550 nm (a
−
c) into air and (d
−
f) into the
silica substrate was measured experimentally by Fourier plane imaging of the antenna emission during excitation with (b) an objective of NA = 0.9
and (e) an immersion objective of NA = 1.3. The maximum azimuthal emission cone of 30
°
is in accordance with the 3D FDTD simulation of the
same structure into (c) air and (f) silica.
Nano Letters
Letter
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
4541
From this
fi
t the spectral propagation length
L
0
(
λ
) (Figure
5a, blue curve) and the spectral antenna transmission
T
ant
(
λ
)
were determined over all measurements and circuits (few
circuits with obvious fabrication errors were excluded from the
evaluation). The
fi
t error (standard deviation
σ
), therefore
includes statistical errors of both the measurement and
fabrication (Figure 2b). The determined antenna e
ffi
ciency is
a lower limit, as we might not have eliminated all possible
sources of losses in the circuit. The antenna transmission peaks
at
λ
max
= 1530 nm with a spectral width of
γ
= 244 nm,
corresponding to a fwhm = 488 nm. The spectral peak
e
ffi
ciency of the antennas is
T
ant,air
max
=15
±
1% from air and
T
ant,substr
max
=45
±
2% from the substrate with minimum statistical
error around the peak and increasing error toward the
measurement limits of
λ
= 1200 nm and
λ
= 1850 nm, where
the absolute circuit transmission and the InGaAs camera
e
ffi
ciency decrease.
Keeping the initial assumption that the antenna properties
obey optical inversion symmetry, the total antenna e
ffi
ciency,
giving the total emission into air and silica, can be spectrally
summated to be
T
ant, tot
max
=60
±
3%, the highest power
transmission for a waveguide loaded optical antenna, reported
to date.
Validity of the applied transmission analysis is based on the
hypothesis that the objective collects all light that is emitted
from the Yagi antennas and that a good overlap of the focused
exciting beam and the angular emission directionality is
achieved. To verify this assumption, the antenna was excited
from air (Figure 3a) and from the substrate (Figure 3d), while
the far-
fi
eld emission of the antenna was imaged into the
excitation direction (Figures 3b,e) in the Fourier plane and
compared to the angular emission spectrum obtained from 3D
FDTD simulation with the same geometry parameters and
excitation (Figures 3c,f). In experiment and simulation, the
antenna emits with strong directionality within a polar angle
cone of
Φ
≲
40
°
, which is remarkably narrow
17,23
and
completely covered by the NA of our objectives (NA = 0.9
from air and 1.3 immersion from silica). Hence, radiative losses
seem to play a minor role in our measurements.
However, the guided
fi
eld in a waveguide cannot be directly
measured in the far-
fi
eld with the only limited information
coming from slight scattering of impurities in strongly
overexposed leakage microscopy (see Supporting Information).
Therefore aperture near-
fi
eld scanning optical microscopy
(NSOM) was used to image the antennas and waveguides
(Figure 4a) directly.
The
fi
eld distribution of the focused beam at the antenna, as
predicted by 3D FDTD simulations (Figure 4c), leads to a
distortion of the incident beam in the area between the antenna
feed element and the re
fl
ector (Figure 4b). No propagating
waves are visible along the surface of the sample away from the
antenna, demonstrating the e
ffi
ciency of the Yagi re
fl
ector. We
note that the evanescent
fi
eld of the guided mode inside the
waveguide is still possible to image through the 320 nm thick
cladding layer, which is an advantage for probing the operation
of the nanocircuit dynamics.
Embedded SPP channel waveguides feature a characteristic
propagation length and e
ff
ective refractive index. In particular,
the latter property is important for potential applications as it
determines the phase velocity, but it is usually di
ffi
cult to
measure, as phase information is lost when measuring emitted
intensities. From transmission measurements on circuits with
short waveguides, we found spectral oscillations on the total
transmission with maximu
m amplitude at the antenna
resonance, which correspond to Fabry
−
Pe
́
rot resonances,
allowing us to probe the mode index of the waveguides.
For the shortest waveguides (
L
= 14.28
μ
m), the spectral
oscillations are very distinct (Figure 5b). The lumped circuit
transmission model (Figure 2a) allows us to treat the circuit as
a Fabry
−
Pe
́
rot resonator de
fi
ned by a channel waveguide of
length
L
and mirrors formed by the antennas due to imperfect
antenna impedance-matching. This model was
fi
tted to the
spectral system transmission, taking the already determined
waveguide loss (Figure 5a, blue curve) and the antenna
e
ffi
ciency into account. Hence, the e
ff
ective refractive index of
the waveguide mode (Figure 5a, green curve) was determined
to be
n
eff
≈
1.54 at
λ
0
= 1550 nm with low spectral dispersion, a
value that coincides well with the expected guided mode
e
ff
ective index from FDTD calculations.
Simulated modal
fi
eld distributions (Figure 5a insets)
indicate decreasing con
fi
nement and increasing
fi
eld-overlap
with the dielectric for longer wavelengths,
26
thus explaining the
experimentally observed decrease in propagation loss for these
wavelengths (Figure 5a, blue curve).
We integrated the developed Yagi antennas and waveguides
with optical directional couplers (ODC) in the developed
nanoplasmonic circuit platform, demonstrating the functional
application of subdi
ff
raction plasmonics as ultrashort coupling
length devices. ODCs have been a standard component in
macroscopic integrated and
fi
ber optics for several decades.
11
They allow for de
fi
ned, dispersion-engineered transfer of power
from one waveguide (bar) to a second waveguide (cross) by
evanescent coupling of the
fi
eld of the guided modes, which can
be well described with coupled-mode-theory.
10
The coupling-
ratio
I
cross
/
I
bar
can be tuned with the length of the coupler. In
the investigated nanocircuits, two embedded SPP waveguides
run in parallel for up to 12
μ
m (Figure 6b). The 3D FDTD
simulations of the ODC (Figure 6a) clearly show that with thin
metal
fi
laments between the two waveguides the investigated
geometry features coupling lengths
L
c
for a
fi
rst full power
transfer from bar to cross down to few micrometers.
Feed (Figure 6b, red) and probe antennas (Figure 6b,
green), and an additional fourth antenna for monitoring
internal back-re
fl
ections (Figure 6b, blue) are connected to the
di
ff
erent ports of the ODC. As before, the emission antennas
Figure 4.
(a) SEM of a Yagi antenna that was illuminated with a highly
focused beam through the substrate and (b) scanned with an aperture
near-
fi
eld optical microscope (NSOM) through the 300 nm thin layer
of cladding SiO
2
. Strong enhancement of the spatial near-
fi
eld
distribution is visible around the antenna and the Yagi re
fl
ector. The
re
fl
ector suppresses emission of the antenna away from the waveguide
and couples the electromagnetic wave into the SPP channel waveguide
with only slight emission toward the cladding. (c) Electric-
fi
eld
distribution (
|
E
|
2
) in the structure plane of the optimized antenna
geometry from a full 3D FDTD simulation in the same con
fi
guration
as the measurement.
Nano Letters
Letter
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
4542
are all turned by 90
°
, radiating with a linear polarization
perpendicular to the excitation antenna. The length of the
couplers was varied in an array of optical circuits (Figure 6c)
from
L
=0
−
12
μ
min3
μ
m steps in one direction, while the
nominal width of the
fi
lament that separates the waveguides
was varied from
w
=50
−
90 nm in 10 nm steps in the other.
Even the smallest, 50 nm thin metal
fi
laments (aspect ratio 4.4
for a 220 nm metal
fi
lm) demonstrated good fabrication
fi
delity
over the entire 12
μ
m length of the longest couplers (Figures
6d,e). To enable smooth transitions, the cross waveguide
approaches the bar waveguide in a 90
°
bend with a radius of
R
=3
μ
m, which for the single waveguide circuits showed
negligible bend loss. Fabrication was consistently reproducible
with four di
ff
erent samples.
As for the basic straight waveguide system, the directional
coupler was spectrally probed and modeled with a slightly more
elaborate lumped circuit model (Figure 7a) in which the
previously investigated Yagi antennas are connected to the
directional coupler with a characteristic spectral transmission
T
wg
(
λ
,
L
) and coupling length
L
c
(
λ
). Each circuit was probed at
λ
0
= 1550 nm, while two clearly distinct emission spots from
the cross and bar antenna were separately integrated and the
total system transmission was monitored similar as for the
single waveguide circuits. (Figure 7e, see Supporting
Information for details). We note that the resulting power
ratio is intrinsically robust to variations of the coupling
e
ffi
ciency into the circuit as those variations simply lead to a
linear scaling of the emission from both monitor antennas.
Coupled mode theory predicts a power exchange between
the waveguides in analogy to two weakly coupled damped
harmonic oscillators.
10
An iterative least-squares
fi
t of this
model was applied to the measured cross- and bar- emission,
while being careful to avoid numerical divergences and ensuring
equalweightingofallemissionratios(seeSupporting
Information for details). Following this
fi
t routine, we obtained
a reproducible coupling length of only
L
c
= 2.46
±
0.04
μ
m for
the thinnest
fi
lament width of 50 nm. This
fi
t, as a second free
parameter, determines an additional equivalent length
L
bend
=
Figure 5.
(a) The 300 nm wide and 220 nm deep embedded SPP
channel waveguides o
ff
er remarkably low loss with a propagation
length of
P
0
/e (
λ
= 1550 nm) = 34
μ
m. Propagation length and
dispersion (blue curve) were determined by iterative
fi
tting to the
experimental spectral circuit transmission for series of di
ff
erent
waveguide lengths. The e
ff
ective refractive index of the SPP mode
was determined as
n
eff
(
λ
= 1550 nm) = 1.54 with
fl
at dispersion, by
(b) iterative
fi
tting of a Fabry-Pe
́
rot oscillator model the experimental
spectral transmission (see Figure 3a). The shown pronounced
oscillations (red dots) are caused by slight back-re
fl
ections of the
wave inside the waveguide from the antennas in circuits with 12.28
μ
m
length.
Figure 6.
(a) Two embedded SPP channel waveguides in close vicinity
exchange energy and form an optical directional coupler (ODC) with
a subwavelength length
L
c
= 2.46
μ
m for full transfer of the displayed
power
fl
ow from waveguide bar to cross (in-plane cross section of a
3D FDTD with the experimental geometry). (b) For full spectral
characterization of this passive SPP component, it was integrated into
a nanocircuit and probed with 4 Yagi antennas, one to couple in (red),
two emitting in crossed polarization to observe cross and bar (green),
and one to monitor internal re
fl
ections (blue). (c) The ODC
parameters, length, and waveguide separation were varied systemati-
cally. (d,e) The thinnest separation is formed by an only 50 nm wide,
220 nm deep
fi
lament of Au, running uniformly over a length of up to
12
μ
m. (f) At the beginning and the end of the ODC, the waveguides
approach with a smooth, adiabatic bend that already causes signi
fi
cant
coupling.
Nano Letters
Letter
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
4543
1.7
±
0.07
μ
m of the coupler that is caused by the transition
into and out of the straight coupling region by 90
°
bends
(Figure 6). Hence, even ODCs with a parallel coupler length of
L
=0
μ
m demonstrated signi
fi
cant coupling of the guided
plasmon from the bar waveguide over to the cross waveguide.
Furthermore, direct comparison of experimental and FDTD
results demonstrate that the fabricated
fi
lament widths are
e
ff
ectively a bit smaller than expected, which is reasonable since
the metal side-walls are inclined
∼
15
°
from the vertical, as is
evident from SEM cross-sectional images of the fabricated
device (Figures 1f
−
h). Consequently, ODCs simulated with a
rectangular
fi
lament of 40 nm width (Figure 6a) had the same
coupling length as real couplers consisting of 50 nm angled
fi
laments. A complete ODC device can therefore be integrated
into an unprecedentedly small area,
11,27,28
while maintaining
reasonably low loss (Figure 5a, blue), therefore achieving a
single coupling length system transmission of 0.93, respectively
a loss of
−
0.31 dB.
The dispersion of directional couplers is utilized in many
compact wavelength division multiplexing (WDM) systems
10
as it allows for low-loss wavelength division. Probing the
spectral transmission of the investigated ODC circuits reveals a
pronounced wavelength-dependence.
We characterized di
ff
erent ODCs spectrally in a range of
λ
0
=
1250
−
1850 nm. Starting with
λ
0
= 1550 nm, the ODCs with a
fi
lament width of 70 nm and length of
L
=10
μ
m (Figure 7b)
operating with several full coupling lengths demonstrated a
well-balanced output with
I
cross
/
I
bar
≈
1 (Figure 7e). Then, by
sweeping the wavelength we observed full switching between
cross (Figure 7d) and bar (Figure 7f) in going from short to
long wavelengths, leading to a 30 dB wavelength discrimination
between 1450 and 1650 nm (see Supporting Information for a
video demonstrating spectral switching). The whole device
features a footprint of only
∼
5
×
10
μ
m
2
and could easily be
stacked to form an arrayed WDM, enabling even more narrow
wavelength discrimination within little more space. The size is
extremely compact compared to current dielectric integrated
WDMs requiring millimeter dimensions.
6
The strong dispersion of the coupling is dominated by the
frequency dependent dielectric constant of the Au as the
fi
eld-
penetration into the thin metallic
fi
lament between the
waveguide is deep. Thus, inside the ODC the total metal-
fi
eld overlap is large compared to the single waveguide where
the metal-dispersion does not in
fl
uence the e
ff
ective index of
the mode considerably (Figure 5a, green curve). However, the
low dispersion is a desired property for the waveguides as is the
high dispersion of the couplers for reducing the e
ff
ective size of
the ODC.
In conclusion, we have demonstrated a highly reproducible
design scheme for plasmonic nanocircuitry that combines the
high con
fi
nement and dispersive properties of plasmonics
28
together with the low-loss and high-coupling e
ffi
ciencies of
Yagi-Uda antennas. For the
fi
rst time, loaded optical Yagi
antennas are used to reduce insertion loss and couple light with
a high total e
ffi
ciency of 60% (15% from air, 45% from
substrate) into and out of plasmonic nanocircuits. Furthermore,
this platform is used to demonstrate the operation of embedded
SPP based ODCs with extraordinarily short coupling length.
The ODCs feature low transmission loss and compete well with
other, for example, Si-integrated circuitry components
6,27
in
terms of coupling-length-overloss ratio while being superior in
overall compactness. Distinct spectral switching is observed,
thus allowing down-scaling of wavelength division multiplexing
from the millimeter to the micrometer range. Additional, we
note that other nanoplasmonic circuitry components can easily
be transferred to the Yagi-Uda loaded embedded plasmonic
waveguide platform, leading a path toward highly integrated,
spectrally functional plasmonic chips like resonant-guided wave
networks
29
or on-chip detectors,
12
ideally with no need for
subsequent intersections
5,7,8
between di
ff
erent types of wave-
guides.
■
ASSOCIATED CONTENT
*
S
Supporting Information
Experimental details on the fabrication, the far
fi
eld optical IR
setup and characterization procedures, leakage microscopy
measurements, spectral switching, NSOM measurements,
experimental data analysis and statistics on the antennas, and
directional couplers and numeric details on the FDTD
simulations. This material is available free of charge via the
Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*
E-mail: arian.kriesch@mpl.mpg.de.
Author Contributions
A.K. and S.P.B. conceived the experiments and developed the
device design, A.K. performed numerical simulations and S.P.B.
fabricated the samples. A.K., D.P., H.P., and S.P.B. performed
the optical and FIB/SEM measurements. A.K., S.P.B., U.P., and
H.A.A. analyzed the data and wrote the
fi
rst draft of the
Figure 7.
(a) Schematic diagram and (b) SEM of an ultracompact SPP
circuit with four Yagi antennas (red in, green out, black re
fl
ection
reference) and a directional coupler (coupler length
L
=10
μ
m),
designed for several full coupling cycles at
λ
= 1550 nm. It shows
strong spectral dispersion of the coupling length
L
c
(
λ
). (c) This leads
to full switching of the power from output channel cross to bar
(green), clearly visible in the optical emission images for (d) 1440, (e)
1550, and (f) 1720 nm, resulting in a 30 dB wavelength discrimination
between 1450 and 1650 nm, while back-re
fl
ections (black) are
negligible.
Nano Letters
Letter
dx.doi.org/10.1021/nl402580c
|
NanoLett.
2013, 13, 4539
−
4545
4544