Supporting Information
DualGated Active Metasurface at 1550 nm with Wide
(>300°)
Phase Tunability
Ghazaleh Kafaie Shirmanesh
†
, Ruzan Sokhoyan
†
, Ragip A. Pala
†
, and Harry A. Atwater
†‡*
†
Thomas J. Watson Laboratory of Applied Physics and
‡
Kavli Nanoscience Institute, California
Institute of Technology, Pasadena, California 91125
, United States
* Corresponding author. E2mail: haa@caltech.edu
1. Electrostatic simulations to extract ITO propert
ies
To accurately calculate the optical response of met
asurfaces under applied bias, we couple the
device physics simulations (Device Lumerical) with
finite difference time domain optical
simulations (FDTD Lumerical). Our electrostatics ca
lculations model the spatial distribution of
charge carriers in the ITO layer embedded in the me
tasurface. In our device physics calculations, we
assume that the work function of Al is 4.3 eV. We a
lso assume that the effective electron mass of
ITO is
m
*=0.35
m
e
, electron mobility of ITO is 25 cm
2
V
21
s
21
, where
m
e
is the free electron mass.
Since our ITO is degenerately doped, we assume that
holes do not significantly contribute to the
observed physical processes. In Device Lumerical so
ftware, we input that the effective mass of
holes is 1×
m
e
, and the hole mobility is 1 cm
2
V
21
s
21
. In our simulations, the bandgap of ITO is set to
2.8 eV,
1
and the electron affinity of ITO is chosen as 4.8
eV. The assumed DC permittivity of ITO is
9.3.
2
Once we have identified the spatial distribution o
f charge under different applied biases, we
then relate the calculated carrier density to the c
omplex dielectric permittivity of ITO
ε
ITO
by using
the Drude model:
⁄
. The plasma frequency
is given by the
following expression
/
∗
. Here,
N
ITO
is the carrier concentration of ITO, which
we extract from the device physics calculations,
e
is the electron charge,
is the DC permittivity of
vacuum,
is the damping constant,
is a fitting constant,
is the angular frequency, which is
related to the wavelength
λ
as
λ=
2πc
/ω
, where
c
is the speed of light in vacuum. When performing
optical simulations, we assume that
m
*=0.35
m
e
,
1.8 10
, and
=3.9. Figure S1a, b show
the imaginary part of the permittivity of ITO for C
ase I and Case II, respectively.
V
0
[V]
a
b
Im (
ε
ITO
)
I
λ
λλ
λ
=1550 nm
I
λ
λλ
λ
=1550 nm
Im
(
ε
ITO
)
V
0
[V]
Figure S1.
Calculated imaginary part of dielectric permittivit
y of a 5 nm2thick ITO film embedded in our
dual2gated metasurface as a function of position an
d applied bias for (a) Case I and (b) Case II. Here
, 0
nm corresponds to the bottom ITO/HAOL interface, an
d 5 nm corresponds to the top ITO/HAOL
interface.
2. Fabrication and characterization of HfO
2
/Al
2
O
3
nanolaminate
We fabricate HfO
2
2Al
2
O
3
nanolaminate films by using atomic layer depositio
n (ALD). We perform
the deposition at 150
o
C by using thermal recipe in our ALD tool (Fiji G2
Plasma Enhanced Atomic
Layer Deposition System). We use tetrakis (ethylmet
hylamino) hafnium, [(CH
3
)(C
2
H
5
)N]
4
Hf, as a
precursor for Hf, while we use trimethyl aluminum,
Al(CH
3
)
3
, as a precursor for Al. During our
ALD process we use water as an oxidant. To fabricat
e HAOL, we adopt two growth periods with
each period consisting of 10 cycles of Al
2
O
3
and 30 cycles of HfO
2
. Immediately after the
deposition, we perform rapid thermal annealing (RTA
) in nitrogen atmosphere. The RTA is
performed for 30 seconds at a temperature of 600
o
C. Previous research has shown that the RTA
treatment causes the diffusion of Al atoms (from Al
2
O
3
layer) into HfO
2
, resulting in formation of
Al2Hf2O bonds.
3
To determine the growth per cycle rates of Al
2
O
3
and HfO
2
films, we fabricate the Al
2
O
3
and HfO
2
control samples on Si substrates. We use 2×10 cycle
s to grow Al
2
O
3
and 2×30 cycles to grow HfO
2
.
We perform transmission electron microscopy (TEM) t
o identify the thicknesses of the grown
samples. The thicknesses of the fabricated Al
2
O
3
, HfO
2
, and HAOL films are 1.54 nm, 7.67 nm, and
9.46 nm, respectively (Figure S2). As shown in Figu
re 1f and Figures S2a, b, while Al
2
O
3
and HfO
2
layers are amorphous, the HAOL layer is partially c
rystallized after RTA treatment. The TEM
images indicate that, as expected, there is a thin
native oxide layer formed on Si substrates. To
enable electrical characterization of the dielectri
c films, we sputter Al top electrodes while using
shadow masks. The continuous Al bottom electrodes a
re deposited by using electron beam
evaporation. To identify the DC permittivities of t
he films, we use the capacitance–voltage (C–V)
measurements of the fabricated metal–oxide–semicond
uctor MOS capacitors at 100 kHz. The DC
permittivities of the fabricated Al
2
O
3
, HfO
2
, and HAOL films are 10.5, 17.8, and 22, respective
ly.
By using current–voltage (I–V) measurements perform
ed on metal2oxide2metal (MIM) structures,
we identify that the breakdown fields of the fabric
ated Al
2
O
3
, HfO
2
, and HAOL films are 7.4
MV/cm, 3.1 MV/cm, and 7.2 MV/cm, respectively.
Figure S2.
TEM images of (a) the Al
2
O
3
control sample deposited via 20 ALD cycles and (b)
the HfO
2
control sample deposited via 60 ALD cycles. Scale b
ar is 2 nm.
3. Fabrication and characterization of indium tin o
xide (ITO)
We deposit our ITO films via room2temperature RF sp
uttering. The deposition pressure is 3 mTorr
while the applied RF power is 48W. We strike the pl
asma by using Ar gas with the flow rate of 20
sccm. We vary the argon/oxygen gas (Ar/O
2
:90/10) flow rate to achieve different carrier
concentrations of ITO.
4, 5
In order to characterize the deposited ITO films,
we perform Hall
measurements and spectroscopic ellipsometry.
6
To this end, we sputter ITO films on quartz and
silicon substrates by changing the Ar+O
2
flow rate while keeping other parameters constant.
We
identify that the deposition rate of ITO is about 1
.11 nm/minute, when the mentioned deposition
parameters are used. Thus, we sputter ITO for 4.5 m
inutes to obtain 5 nm2thick ITO films. Then we
perform Hall measurements on the films deposited on
quartz substrates, and we perform
spectroscopic ellipsometry on the films deposited o
n silicon substrates. After obtaining the charge
carrier concentration
!"#
and electron mobility
μ
of the ITO films from Hall measurements, and
using the relation
$
%&
'()
*
+
∗
+
,
&
'()
%
-
.
, we obtain the complex permittivity of the ITO fil
ms via an
ellipsometry fit to a single Drude function
!"#
/
0
-
/
-
1234
. Here,
Γ
is the damping constant, and
6
is the plasma frequency, which is related to the c
harge carrier density
!"#
via
6
7
&
'()
8
-
9
:
+
∗
.
Here,
,
;
, and
∗
are the electron charge, the dielectric permittivi
ty of vacuum, and the effective
electron mass, respectively. The high2frequency per
mittivity
ε
∞
, damping rate
, and electron
effective mass
∗
are determined via fitting the Drude model to the
measured ellipsometry data.
Thus, the dielectric permittivity of ITO
!"#
is related to the plasma frequency
ω
6
via Drude model,
and the plasma frequency itself depends on the carr
ier concentration of ITO
!"#
. This fact is the
key reason why the optical response of the metasurf
ace is modulated under applied bias. The
electrical and optical constants obtained from Hall
measurements and spectroscopic ellipsometry are
listed in Table S1.
1.93 nm
7.67 nm
1.54 nm
1.05 nm
a
b
Si
Si
Al
2
O
3
HfO
2
Native oxide
Native oxide
When fabricating our dual2gated metasurface (see Su
pporting Information S5 for fabrication steps),
we deposit HAOL on top of ITO. Since the HAOL layer
needs to be RTA2treated at 600
o
C for 30
seconds, we need to take into account the effect of
the RTA treatment on properties of ITO. To
investigate this effect, we fabricate two identical
ITO samples and perform RTA treatment at 600
o
C
for 30 seconds on one of the samples. We do Hall me
asurements and ellipsometry on both samples
and compare the results. As seen in Table S1, the f
itted parameters are in good agreement with the
expected final thicknesses of the films and literat
ure values for the constants (
0.1185 eV,
∗
0.35
8
and
;
3.9), which we use to define the dielectric permitt
ivity of ITO in our
simulations.
5, 7, 8
We consider the bulk charge carrier concentration
of ITO to be
!"#
3×10
20
cm
?@
which draws parallel to the plasma frequency of
1.0874 eV.
Table S1.
Electrical and Optical parameters obtained from Ha
ll measurements and spectroscopic
ellipsometry for the ITO films deposited using diff
erent Ar+O
2
flows rates.
Ar+O
2
flow
rate
[sccm]
Fitted
thickness
[nm] as
deposited
Fitted
thickness
[nm] after
RTA
A
as
deposited
A
after
RTA
B
C
[eV]
as
deposited
B
C
[eV]
after
RTA
D
[eV] as
deposited
D
[eV]
after
RTA
0
4.3637
4.3137 6.0853
5.8447 1.8516 1.924 0.16245
0.14188
0.4
5.3566
5.1242 6.4603
5.402 1.9679 1.2989 0.1409
2 0.12521
0.5
5.2988
5.3237 5.1834
4.8832 1.4075 0.94404
0.16379 0.12981
0.6
4.0852
6.4846 5.338
5.0306 1.4496 1.0185 0.1508
1 0.11095
0.7
5.5826
5.3170 5.9536
4.689±0
1.7932 0.86608
0.13828 0.1543
0.8
5.4923
5.8453 5.6552
5.1296 1.5872 1.1351 0.143
84 0.14262
0.9
5.6060
5.5593 5.1672
5.6363 1.285
1.4352 0.1418
7 0.13105
1
6.2157
6.0063 5.5049
5.4699 1.4416 1.2529 0.13843
0.12189
When fabricating our metasurface we deposit ITO at
Ar+O
2
flow rates of 0.6 sccm. In this case,
the plasma frequency and the charge carrier concent
ration of ITO after RTA treatment is
1.0185 eV and
2.6319×10
20
cm
?@
, respectively. Note that after depositing top gate
dielectric
on ITO, the carrier concentration of ITO is expecte
d to increase due to the leakage of oxygen from
the ITO layer into the dielectric that occurs durin
g ALD process.
9
As a result, we expect the
carrier concentration of ITO in our final device to
be slightly higher than the values we obtain via
Hall measurements.
It should be noted that the smoothness of the ITO f
ilms incorporated into our metasurface is of
great importance. To this end, we deposited differe
nt test samples using different sputtering
powers and temperatures. To investigate the roughne
ss of the deposited ITO films, we performed
atomic forced microscopy (AFM) on our ITO samples a
nd figured out that sputtering at room
temperature with applied RF power of 48 W would res
ult in the ITO films with an average root
mean square
(RMS) roughness of ~0.3 nm. This was expected since
having rough and porous
ITO films would not let us perform Hall measurement
s and/or fit the ellipsometry results with
Drude model.
4. Full wave simulation of tunable metasurface
The simulated reflectance spectra for different app
lied biases for Case I and Case II are presented in
Figures S3a, b. Figures S3c, d show the spectra of
relative reflectance change at different applied
voltages for Case I and Case II, respectively. The
insets of Figures S3c, d show the relative
reflectance change as a function of applied voltage
at a fixed wavelength of 1550 nm.
Figure S3.
Reflectance spectrum for different applied biases
for (a) Case I and (b) Case II. The relative
reflectance change spectrum for different applied v
oltages for (c) Case I and (d) Case II. The insets
show the
relative reflectance change as a function of voltag
e at a wavelength of 1550 nm.
Figure S4 shows the spatial distribution of the
z
2component of the electric field
E
z
inside the
dielectric spacer of the metasurface, which consist
s of HAOL/ITO/HAOL planar layers. The
spatial distribution of
E
z
is calculated at a wavelength of
λ
=1550 nm. Figures S4a2c correspond to
the bias application configuration that is referred
to as Case I (Figure 2a), while Figures S4d2f
correspond to the bias application configuration re
ferred to as Case II (Figure 2d). Both in Case I
and Case II the assumed values of the applied bias
are
V
0
=
−
6.5 V,
V
0
=0 V and,
V
0
=6.5 V (for
definition of
V
0
see Figure 2). As seen in Figures S4c, d and f, th
ere is a strong field enhancement
at the interfaces of ITO and HAOL. Figure S4 also s
hows that the z component of the electric field
E
z
around right and left edges of the antenna are anti
parallel to each other
.
V
0
[V]
∆R/R
0
[%]
V
0
[V]
∆R/R
0
[%]
a
b
c
d
λ
λλ
λ
=1550 nm
λ
λλ
λ
=1550 nm
Figure S4.
Close2up image of the spatial distribution of the
z component of electric field in the
HAOL/ITO/HAOL region for (a) V
0
= −6.5 V, (b) V
0
= 0, and (c) V
0
= +6.5 V in Case I, (d) V
0
= −6.5 V, (e)
V
0
= 0 V, and (f) V
0
= +6.5 V in Case II.
Figure S5 plots the spatial distribution of the abs
olute value of the magnetic field for our dual2
gated metasurfaces. Figures S5a2c correspond to Cas
e I, while Figures S5d2f correspond to Case II.
Both in Case I and Case II we assume the following
values of applied bias voltages:
V
0
=
−
6.5 V,
V
0
=0 V and,
V
0
=6.5 V. As seen in this figure, the magnetic field
is localized in the gap region
between the Al antenna and the back reflector. This
proves the existence of a magnetic dipole
resonance. One can also notice that the strength of
the magnetic dipole is strongly altered by
changing the applied voltage.
Figure S5.
Spatial distribution of the magnitude of the magne
tic field for (a)
V
0
= −6.5 V in Case I, (b)
V
0
= 0
V in Case I, (c)
V
0
= +6.5 V in Case I, (d)
V
0
= −6.5 V in Case II, (e)
V
0
= 0 V in Case II, and (f)
V
0
= +6.5 V in
Case II. The dashed lines specify the boundaries be
tween the back reflector and bottom HAOL, the botto
m
HAOL and the ITO layer, the ITO layer and the top H
AOL, and, finally, the dashed lines outline the pat
ch
antenna.
E
z
[
V
/m]
V
a
= V
b
=
−
6.5 V
V
a
= V
b
= 0 V
V
a
= V
b
= +6.5 V
V
a
=
−
V
b
=
−
6.5 V
V
a
=
−
V
b
= 0 V
V
a
=
−
V
b
= 6.5 V
a
b
c
d
e
f
Bottom gate
dielectric
Top gate dielectric
ITO
Bottom gate
dielectric
Top gate dielectric
ITO
| H | [A/m]
V
a
= V
b
=
−
6.5 V
V
a
= V
b
= 0 V
V
a
= V
b
= 6.5 V
V
a
=
−
V
b
=
−
6.5 V
V
a
=
−
V
b
= 0 V
V
a
=
−
V
b
= 6.5 V
a
b
c
d
e
f
5. Fabrication of doublegated tunable metasurface
In order to fabricate our gate2tunable metasurface,
we first perform RCA1 cleaning
(H
2
O:NH
4
OH:H
2
O
2
= 5:1:1) of silicon substrates. Then by using e2bea
m evaporation, we deposit
an 80 nm2thick aluminum back reflector. On top of t
he Al back reflector we deposit a 9.5 nm2
thick HAOL by using ALD, as described in Supporting
Information Part 2. We then deposit a 5
nm2thick ITO layer on top of the HAOL gate dielectr
ic by using RF magnetron sputtering in
Ar/O
2
plasma environment. Once we have sputtered the ITO
layer, we deposit another 9.5 nm2
thick HAOL layer. Afterwards, we spin e2beam resist
on our Si/Al/HAOL/ITO/HAOL planar
sample and pattern Al fishbone antenna arrays and c
ontact pads via standard e2beam lithography.
After developing the e2beam2exposed sample, we depo
sit Al by using e2beam evaporation. We
obtain our fishbone dual2gated metasurface after pe
rforming lift2off process. Figure S6
summarizes the described fabrication steps of our t
unable metasurface.
Figure S6.
Schematic representation of fabrication steps of ou
r dual2gated tunable metasurface.
6. Reflectance measurements
Figure S7 shows the experimental setup that we use
for reflectance measurements. In our
reflectance measurement setup, the metasurface is i
lluminated by a broadband laser. The laser
beam impinges on the metasurface after passing thro
ugh an optical chopper, a polarizer, a
50/50 non2polarizing beam splitter, and a 20x objec
tive lens. We use a white light source, a
flipping mirror and a CCD camera to make sure that
the laser beam is positioned at the center
of the metasurface. The light reflected from the me
tasurface is then guided to a Ge detector by
the 50/50 beam splitter. The reflectance is obtaine
d via
EFGHIJKH
L
%
N
100
O
P,QRSTUVRW,
?O
XRWYZU[T\]
O
U,V,U,\W,
?O
XRWYZU[T\]
(S1)
where
E
+8^_`abc_d8
and
E
b8c8b8ed8
are the raw reflectance values obtained while illu
minating
the metasurface and the Al back reflector, respecti
vely.
E
f_dghbiaej
is the background
reflectance in the absence of incident laser beam.
E
3
beam evaporation of Al
back3plane
Atomic layer deposition of
HAOL
E
3
beam evaporation of Al
antenna array
Atomic layer deposition
of
HAOL
Sputtering ITO
Metal lift3off
RCA1 treated Si substrate
Spinning resist
E
3
beam exposure and
development
Figure S7.
Optical setup for measuring the reflectance spectru
m of our metasurfaces.
7. Phase shift measurements
We measure the phase shift of the light reflected f
rom the tunable metasurface by using the
optical setup shown in Figure S8. In this setup, th
e beam from a NIR tunable laser is directed
towards the sample via a polarizer, a 50/50 non2pol
arizing beam splitter and then a 20x
objective lens. The beam is positioned at the edge
of the metasurface by using a white light
source and a camera. As a result, the incoming beam
is partly reflected from the metasurface
and partly from the HAOL/ITO/HAOL/Al planar heteros
tructure. The light reflected from the
sample is guided to the camera by using a 50/50 bea
m splitter. On the other hand, the laser
beam itself is directed to the camera by using mirr
ors, where it serves as a reference beam. The
camera records two regions of interference fringes
i) the fringes formed via interference of the
light reflected from the metasurface and the refere
nce beam, and ii) the fringes formed via
interference of the light reflected from the HAOL/I
TO/HAOL/Al planar heterostructure and
the reference beam.
FM
Detector
20x Objective
Metasurface
NIR
Laser
Monochromator
OC
P
M
M
BS
OC:
optical chopper
M: mirror
P: polarizer
BS: beam splitter
White light source
CCD camera
Figure S8.
Optical setup used for measuring the phase shift of
the light reflected from tunable
metasurfaces.
To analyze the phase shift of the light reflected f
rom our tunable metasurface, we
process the images captured by the camera under dif
ferent applied biases. In these images, we
select one spatial cross section from the metasurfa
ce interference fringes and another one –
from the reference fringes area. The intensity valu
es at the cross sections are then interpreted
as the curves which are then smoothened by a moving
average (MA) filter. Figure S9a shows
the interference fringes recorded by the NIR camera
for different applied bias voltages. Figure
S9b depicts the intensity data extracted from the r
eference fringes (indicated by
R
) and
metasurface fringes (indicated by
M
). Figure S9b also plots MA2filtered curves for bot
h
reference and metasurface fringes. Figure S9c shows
the sinusoidal functions fitted to the two
mentioned fringe regions. Considering the offset be
tween these two sinusoidal functions, one
can calculate the phase shift for each applied bias
via
klJm mlFI
n
:
, where
op
is the
distance between the two fixed peaks of sinusoidal
functions that correspond to the
metasurface fringes and reference fringes, and
p
is the period of the sinusoidal wave.
M
M
M
M
IR
Camera
NIR
Laser
White
3
light
Source
Multimode
Fiber
BS
BS
20x
Objective
L
L
Metasurface
P
M: mirror
P: polarizer
BS: beam splitter
L: lens
Figure S9.
(a) The interference fringes captured by NIR camera
. The dashed lines labeled
R
and
M
show the reference and metasurface fringe cross sec
tions, respectively. (b) Extracted intensity data f
rom
reference and metasurface fringe cross sections and
their MA smoothened curves. (c) Fitted sinusoidal
waves for reference and metasurface fringe cross se
ctions. Here,
op
is the distance between the two
3
6.5 V
3
3
.5 V
0
V
3.5
V
6.5
V
R
M
R
M
R
M
R
M
R
M
b
a
c
op
p