PHYSICAL REVIEW B
90
, 165409 (2014)
Tunable large resonant absorption in a midinfrared graphene Salisbury screen
Min Seok Jang (
),
1
,
2
,
*
Victor W. Brar (
),
2
,
3
,
*
Michelle C. Sherrott,
2
Josue J. Lopez,
2
Laura Kim (
),
2
Seyoon Kim (
),
2
Mansoo Choi (
),
1
,
4
and Harry A. Atwater
2
,
3
,
†
1
Global Frontier Center for Multiscale Energy Systems, Seoul National University, Seoul 151-744, Republic of Korea
2
Thomas J. Watson Laboratory of Applied Physics, California Institute of Technology, Pasadena, California 91125, USA
3
Kavli Nanoscience Institute, California Institute of Technology, Pasadena, California 91125, USA
4
Division of WCU Multiscale Mechanical Design, School of Mechanical and Aerospace Engineering, Seoul National University,
Seoul 151-742, Republic of Korea
(Received 24 March 2014; published 8 October 2014)
The optical absorption properties of periodically patterned graphene plasmonic resonators are studied
experimentally as the graphene sheet is placed near a metallic reflector. By varying the size and carrier density
of the graphene, the parameters for achieving a surface impedance closely matched to free-space (
Z
0
=
377
)
are determined and shown to result in 24.5% total optical absorption in the graphene sheet. Theoretical analysis
shows that complete absorption is achievable with higher doping or lower loss. This geometry, known as a
Salisbury screen, provides an efficient means of light coupling to the highly confined graphene plasmonic modes
for future optoelectronic applications.
DOI:
10.1103/PhysRevB.90.165409
PACS number(s): 78
.
67
.
Wj
,
73
.
20
.
Mf
,
78
.
20
.
Jq
The ability to interact strongly with light is important for a
material to be useful in optics-based applications. Monolayer
graphene exhibits a number of interesting optical phenomena,
including a novel photothermoelectric effect [
1
,
2
], strong non-
linear behavior [
3
,
4
], and the potential for ultrafast photode-
tection [
5
]. However, the absolute magnitude of these effects is
limited by the amount of light absorbed by the graphene sheet,
which is typically 2.3% at infrared and optical frequencies
[
6
,
7
]—a small value that reflects the single atom thickness of
graphene. To increase the overall graphene-light interaction,
many novel light scattering and absorption geometries have
recently been developed. These include coupling graphene
to resonant metal structures [
8
–
13
] or optical cavities where
the electromagnetic fields are enhanced [
14
–
16
], or draping
graphene over optical waveguides to effectively increase the
overall optical path length along the graphene [
17
,
18
]. While
those methods rely on enhancing interband absorption pro-
cesses, graphene can also be patterned and doped so as to excite
plasmonic modes that display strong resonant absorption
in the terahertz to midinfrared regime [
19
–
23
]. Graphene
plasmonic modes are highly sensitive to their environment,
and they have been shown to display large absorption when
embedded in liquid salts [
19
,
24
] or by sandwiching dopants
between several graphene layers [
23
]. However, plasmonically
active metallic and semiconductor structures can achieve
near-perfect absorption of radiation at specified frequencies
using a resonant interference absorption method [
25
–
29
]. The
electromagnetic design of these structures derives in part from
the original Salisbury screen design, but with the original
resistive sheet replaced by an array of resonant metal structures
used to achieve a low surface impedance at optical frequencies.
It has recently been proposed that similar devices could be
possible using graphene to achieve perfect absorption from
terahertz to midinfrared [
30
,
31
]. Such a device would offer an
efficient manner of coupling micron-scale free-space light into
*
These authors contributed equally to the work.
†
haa@caltech.edu
nanoscale plasmonic modes, and it would allow for electronic
control of that coupling process. In this paper, we design and
demonstrate a photonic heterostructure based on that principle,
using tunable graphene nanoresonators placed a fixed distance
away from a metallic reflector to drive a dramatic increase in
optical absorption into the graphene.
A schematic of our device is shown in Fig.
1(a)
. A graphene
sheet grown using chemical vapor deposition on copper foil
is placed on a 1-
μ
m-thick, low stress silicon nitride (SiN
x
)
membrane with 200 nm of Au deposited on the opposite side,
which is used as both a reflector and a back-gate electrode.
Nanoresonators with widths ranging from 20 to 60 nm are then
patterned over 70
×
70
μ
m
2
areas into the graphene using
100 keV electron beam lithography (see Sec. I in Supplemental
Material) [
32
]. An atomic force microscopy (AFM) image
of the resulting graphene nanoresonators is shown in the
inset of Fig.
1(b)
. The device was placed under a Fourier
transform infrared (FTIR) microscope operating in reflection
mode, with the incoming light polarized perpendicular to the
resonators in order to maximize the excitation of the resonant
plasmon modes [
20
,
22
]. The carrier density of the graphene
sheet was varied
in situ
by applying a voltage across the SiN
x
between the gold and the graphene, and the resulting changes in
resistance were continuously monitored using source and drain
electrodes connected to the graphene sheet
[
Fig.
1(b)
]
.The
carrier density of the graphene nanoresonators was determined
from experimentally measured resonant peak frequencies (see
Secs. II and III in the Supplemental Material [
32
]).
The total absorption in the device—which includes absorp-
tionintheSiN
x
and the graphene resonators—is determined
from the difference in the reflected light from the nanores-
onator arrays and an adjacent gold mirror. For undoped and
highly doped 40 nm nanoresonators, the total absorption is
shown in Fig.
2(a)
, revealing large absorption at frequencies
below 1200 cm
−
1
, as well as an absorption peak that varies
strongly with doping at 1400 cm
−
1
and a peak near 3500 cm
−
1
that varies weakly with doping. In order to distill absorption
features in the graphene from the environment (i.e., SiN
x
and Au back reflector), we plot the difference in absorption
1098-0121/2014/90(16)/165409(5)
165409-1
©2014 American Physical Society
MINSEOKJANG
et al.
PHYSICAL REVIEW B
90
, 165409 (2014)
FIG. 1. (Color online) (a) Schematic device structure of graphene
Salisbury screen. The inset illustrates the device with the optical
waves at the resonance condition. (b) dc resistance of graphene as a
function of the gate voltage. The inset is an AFM image of 40 nm
nanoresonators.
between the undoped and doped nanoresonators, as shown
in Fig.
2(b)
for 40 nm nanoresonators. This normalization
removes the low frequency feature below 1200 cm
−
1
, which
is due to the broad optical phonon absorption in the SiN
x
and
is independent of graphene doping. The absorption feature
at 1400 cm
−
1
, however, shows a dramatic dependence on
the graphene sheet carrier density, with absorption into the
graphene nanoresonators varying from near 0% to 24.5% as
the carrier density is raised to 1
.
42
×
10
13
cm
−
2
. Because the
absorption increases with carrier density, we associate it with
resonant absorption in the confined plasmons of the nanores-
onators [
19
–
22
,
33
]. In Fig.
2(b)
, we also see that absorption
at 3500 cm
−
1
exhibits an opposite trend relative to the lower
energy peak, with graphene-related absorption decreasing with
higher carrier density. This higher energy feature is due to
interband graphene absorption, where electronic transitions
are Pauli blocked by state filling at higher carrier densities
[
34
]. For spectra taken from the bare, gate-tunable graphene
surface, this effect leads to
8% absorption, i.e., roughly
twice the intensity observed from patterned areas. Finally,
in Fig.
2(c)
, we investigated the graphene nanoresonator
absorption as the resonator width is varied from 20 to 60 nm at
fixed carrier density. This figure shows that the lower energy,
plasmonic absorption peak has a strong frequency and intensity
dependence on resonator width, with the maximum absorption
occurring in the 40 nm ribbons.
The carrier density dependent plasmonic dispersion of this
system is shown in Fig.
3(a)
. The observed resonance fre-
quency varies from 1150 to 1800 cm
−
1
, monotonically increas-
ing with larger carrier densities and smaller resonator widths.
The plasmon energy asymptotically approaches
∼
1050 cm
−
1
due to a polar phonon in the SiN
x
that strongly reduces the
dielectric function of the substrate at that energy [
35
]. This
coupling between the substrate polar phonon and the graphene
plasmon has also been previously observed in back-gated SiO
2
devices [
20
,
22
,
36
]. In Fig.
3(b)
, we plot the intensity of the
plasmonic absorption as a function of frequency at varying
10
20
30
40
(%)
0. (CNP)
1.42
carrier density
(
10
13
cm
-2
)
0
5
10
15
20
25
5
(%)
0.32
0.66
0.95
1.42
carrier density
(
10
13
cm
-2
)
1.42 (bare)
1000
2000
3000
4000
0
5
10
15
20
25
0
1
2
3
Frequency (cm
-1
)
(%)
20 nm
30 nm
40 nm
50 nm
60 nm
(a)
(b)
(c)
FIG. 2. (Color online) (a) The total absorption in the device
for undoped (red dashed) and hole doped (blue solid) 40 nm
nanoresonators. (b) The change in absorption with respect to the ab-
sorption at the charge neutral point (CNP) in 40-nm-wide graphene
nanoresonators at various doping levels. The solid black curve
represents the absorption difference of bare (unpatterned) graphene.
(c) Width dependence of the absorption difference with the carrier
concentration of 1
.
42
×
10
13
cm
−
2
. The resonator width varies from
20 to 60 nm. The dashed curve shows the theoretical intensity of the
surface parallel electric field at the SiN
x
surface when graphene is
absent.
carrier densities, revealing that for all carrier densities, the
maximum in absorption always occurs at 1400 cm
−
1
.
The experimental behavior observed in Figs.
2
and
3
has some similarities with graphene plasmonic resonators
patterned on back-gated SiO
2
devices; however, there are some
significant differences. Most notably, the absolute absorption
observed in this device is one order of magnitude greater than
what has previously been observed. Furthermore, the maxi-
mum absorption in this device always occurs near 1400 cm
−
1
,
in contrast to previous graphene plasmonic devices, where
lower frequency resonances showed greater intensity due to
fewer loss pathways and better
k
-vector matching between the
graphene plasmons and free-space light [
20
,
22
]. These new
absorption features can be understood by considering the role
of the gold reflector. At 1400 cm
−
1
, the optical path length
of the SiN
x
is
λ/
4
n
, and the gold reflector creates a standing
wave between the incident and reflected light that maximizes
the electric field on the SiN
x
surface. As a consequence, when
the graphene nanoresonators are tuned to absorb at 1400 cm
−
1
,
a double resonance condition is met, and the dissipation
165409-2