1
Supporting Information
Phase Modulation with
E
lectrically
T
unable
Vanadium Dioxide
P
hase
-
C
hange
M
etas
u
r
faces
Yonghwi Kim
,
†
Pin Chieh Wu,
†
Ruzan Sokhoyan,
†
Kelly Mauser
,
†
Rebecca G
l
a
udell,
†
Ghazaleh
K
afaie
Shirmanesh
,
†
and Harry A. Atwater
*
†‡
†
Thomas J. Watson Laboratories of Applied Physics, California Institute of Technology, Pasadena, California 91125,
United States
‡
Kavli Nanoscience Institute, California Institute of Technology,
Pasadena, California 91125, United States
*Corresponding author: Harry A. Atwater (
haa@caltech.edu
)
Contents
S1.
VO
2
thin film growth and d
evice fabrication
S2.
VO
2
thin film characterization: e
lectrical
properties
S3.
VO
2
thin film characterization: o
ptical properties
S4.
Full
-
wave simulation: electric field analysis
S5.
Effective medium approximation
S6.
P
hase shift measurement
optical setup
S7.
T
emporal response measurement
2
S1.
VO
2
thin film growth and d
evice fabrication
Figure S1.
(a) Schematic of VO
2
thin film grown on the planar Au/Al
2
O
3
heterostructure.
The
structure consists of a 40
-
nm
–
thick active VO
2
film, a 50
-
nm
–
thick Al
2
O
3
layer, and an optically
thick (150 nm) Au backpla
ne.
(b) Scanning electron microscope (SEM) image of the
heterostructure structure and (c) the zoomed
-
in SEM image of the heterostructure with the Au
backplane.
The inset shows a further
magnified
SEM image of the continuous VO
2
film on the
heterostructure, which was then patterned by the nanofabrication proc
ess
.
The metasurface was fabricated on a cleaned fused silica substrate. First, a
150
-
nm
–
thick
Au backplane was deposited on
a
20
-
nm
–
thick Ti adhesion layer using electron b
eam deposition.
Then
a
50 nm Al
2
O
3
layer was formed on top of the Au backplane by atomic layer deposition
,
followed by 40 nm of VO
2
formed
by pulsed layer deposition (at 650 °C in a 5
-
mTorr O
2
environment), which result
ed
in a heterostructure as shown in Fi
gure S1a.
S
c
anning electron
microscope
images of the planar heterostructure
are
shown in Figure S1b,c.
We verified that during
the growth processes the surface of the heterostructure was smooth and without cracks that could
lead to the failure of the devic
e’s electrical control due to the leakage current.
1
The metasurface
pattern was written on
the
positive electron
-
beam resist using electron
-
beam lithography, and
the
pattern was developed in a resist developer. A 40
-
nm
–
thick Au
layer
and a 10
-
nm
–
thick Cr layer
were then deposited consecutively on the developed resist using electron beam deposition. This
was followed by a lift
-
off process to form the designed patte
rns. The patterned top metal layer (Cr)
was subsequently used as a hard mask for
the
dry etching of
the
VO
2
layer with a fluorine
-
based
inductively coupled plasma
-
reactive ion etching
(
ICP
-
RIE
)
process
, which
yield
ed
self
-
aligned
metal
-
insulator
-
metal ante
nna arrays. Finally, the Cr mask was removed by
a
brief dry etching
process
,
using a chlorine
-
based ICP
-
RIE process.
3
S2.
VO
2
Thin film characterization: electrical properties
Figure S2.
Electrical resistance hysteresis curves of the VO
2
thin film grown on the planar
Au/Al
2
O
3
heterostructure formed on a fused silica substrate. The resistance was measured based
on the van der Pauw method during heating/cooling cycles
,
where the temperature was controlled
by a thermoelectric element.
We
adopted
van der Pauw method
to measure the
electrical resist
ance
of the
VO
2
film
grown
on
the
planar Au/Al
2
O
3
heterostructure formed on the fused
silica
substrate
(
Figure S1a
)
.
The temperature
-
dependent resistance was recorded by applying heating/cooling
temperature
cycles in the 25
–
100 °C range where the temperature was controlled by a thermoelectric element.
Figure S2 shows the temperature
-
dependent hysteresis of the resistance for
a
40
-
nm
–
thick VO
2
film. We observe
d
a drastic change
in
the resistance as the samples
’
temperatures pass
ed
the
insulator
-
to
-
metal (IMT) transition temperature, which is around 68 °C.
4
S3.
VO
2
Thin film characterization: optical properties
Figure S3.
Temperature
-
dependen
t
reflect
ance
of
a
VO
2
film
during (
a
)
the heating cycle and (b)
the cooling cycle. (c)
H
ysteresis cycle at
=
2.7
μ
m
extracted from
figures
(
a
)
and
(
b
)
, showing a
sharp
and reversible reflectance
across
the
phase transition
in the VO
2
thin film.
In order to characterize the optical propert
ies of the VO
2
films grown on the planar
Au/Al
2
O
3
heterostructure (Figure S1a) across the IMT, we measured the optical reflectance curves
in the near
-
and mid
-
infrared ranges as a function of temperature. Figure S3a and S3b represent
the temperature
-
dependent reflectance modulation in the 27
–
110 °C range
of a 40
-
nm
–
thick VO
2
fi
l
m for heating and cooling cycles, respectively. We observe
d
that the temperature
-
dependent
optical reflectance changes significantly across the IMT. Based on the temperature
-
dependent
reflectance data shown
in
F
igures S3a and S3b,
we represent
in
F
igure S3c the reflectance as a
function of temperature at
λ
= 2.7 μm. We observe
d
the thermal hysteresis of the reflectance
as
similar to the one in
F
igure S2.
Wavelength
(μm
)
Reflectance (%)
Heating
Reflectance (%)
Wavelength (μm)
Cooling
a
b
c
20
40
60
80
100
120
60
70
80
90
100
Reflectance (%)
Temperature (
C)
Heating
Cooling
Reflectance (%)
Temperature (
°C
)
λ
= 2.7 μm
5
S4.
Full
-
wave simulation:
e
lectric field
analysis
Figure S4.
Electric field distribution of the metasurface structure. (a) Spatial distribution of the z
component of the electric field E
z
. (b) Spatial distribution of the x component of the electric field
E
x
.
The l
eft panels of each figure indicate
the
insulating ph
ase resonance (
λ
= 1520 nm), while
the
right panels indicate
the
metallic phase resonance (
λ
=
1620
nm).
We used finite
-
difference time
-
domain methods to simulate the periodic antenna structure
under normal incidence illumination with a transverse magneti
c plane wave. In Figure 2b
show
s
the existence of concentrated magnetic field magnitude |H
y
| in the dielectric gap of antenna
elements as a result of a magnetic dipole resonance. Figure S4a shows the enhanced z component
of the electric field E
z
around
the
right and left edges of the antenna, which
are
antiparallel to each
other. Figure S4b also shows the existence of an antiparallel x component of the electric field E
x
around the top (light blue) and bottom (light yellow) of the dielectric gaps, which i
ndicates an
antiparallel electric field as a result of the magnetic dipole resonance. As seen in the right panels
of Figure S4a and S4b, the electric fields become more localized in the Al
2
O
3
layer as the VO
2
layer undergoes an insulator
-
to
-
metal transitio
n
.
The tendency is
consistent with the change in the
magnetic field magnitude distributions of Figure 2b as a result of perturbation in the magnetic
dipole resonance.
6
S5.
Effective medium approximation
Figure
S5
.
Measured dielectric permittivity of
a
VO
2
thin film on a sapphire substrate at RT and
90
°C, which clearly indicates the phase transition. T
he real part of the dielectric permittivity of
VO
2
(solid line) changes its sign from positive to negative upon phas
e transition
.
To estimate the effective optical constants of
the
VO
2
layer
that were used in
the
full
-
wave
electromagnetic simulation, we employed
the
Bruggeman effective medium approximation. The
effective complex optical constants
휀
̃
퐸
(
휔
)
of a two
-
compon
ent system in
the
Bruggeman model
for generalized ellipsoidal inclusions is given by
푓
휀
̃
푟
(
휔
)
−
휀
̃
퐸
(
휔
)
휀
̃
푟
(
휔
)
+
(
1
−
푞
)
푞
휀
̃
퐸
(
휔
)
+
(
1
−
푓
)
휀
̃
푚
(
휔
)
−
휀
̃
퐸
(
휔
)
휀
̃
푚
(
휔
)
+
(
1
−
푞
)
푞
휀
̃
퐸
(
휔
)
=
0
where
휀
̃
푟
(
휔
)
and
휀
̃
푚
(
휔
)
are
the wavelength
-
dependent
complex optical constants of the
rutile
and
monoclinic
phases
,
respectively
as shown in Figure
S
5
,
푓
is the volume fraction
of the
rutile
phase
,
and
푞
is the
depolarization factor that depends on the
shape of the
inclusions
.
The
depolarization
factor is chosen based on the assumption
that the rutile phase inclusions are spherical at low
concentration, and it continuously increases assuming that the rutile phase inclusions form disk
shape
s
at higher concentrations.
2
Dielectric permittivity
ε
R,
RT
ε
I,
RT
ε
R,
90
°
C
ε
I,
90
°
C
Wavelength
(nm
)
7
S6
.
P
hase shift measurement
optical setup
Figure S6.
Schematic of the optical setup for
the
phase shift measurement.
The l
abels of each
optical component refer to
the
following: M (mirror); I (iris); BS (beam splitter); L (lens); P (linear
polarizer); ND (neutral
-
density filter); FM (flip mirror).
The interference fringes of reflected light
from the metasurface (Figure 4a)
are
measured
by a Michelson
-
type interferometer
,
as shown in Figure S6. A tunable near
-
infrared light source
is directed onto a 50/50 beam splitter after passing
through
a linear polarizer. We used a white
light source and
a visible camera to position the incident beam on the metasurface structures. One
beam
was
focused onto a reference mirror, while the other beam
was
focused at the edge of the
metasurface structures to utilize the planar Au/Al
2
O
3
heterostructure as a phase reference.
B
y using
an
infrared CCD camera
, we recorded interference fringes formed by the interference from the
reference mirror’s beam and by the sample beam from the metasurface.
We post
-
processed the
captured interference im
ages as a function of applied bias to analyze the phase shift of reflected
light
,
as described in the main text.
8
S7.
T
emporal response measurement
Figure S
7
.
Schematic of
the
optical setup for
the
temporal response measurement. Here HD refer
s
to a high
-
speed InGaAs detector.
We measured
the device’s
temporal response using a function generator to modulate the
reflected signal
when
rectangular voltage pulses
were applied
. The reflected signal at 1550 nm was
monitored
using an oscilloscope (Fig
ure S7
)
.
Figure S
8
present
s the results
when we applied 2.5
-
s
–
wide and 50% duty
-
cycle electrical voltage pulses for different peak
-
to
-
peak voltages. First, when
the applied bias
was
3.4 V
pp
(Figure S8
a), there
was
no observable change in the reflected sign
al
,
indicating that the applied bias
was
below the transition threshold (Figure 3c). We observe
d
a
decrease in the reflected signal as we increase
d
the bias to 8.9 V
pp
(Figure S8
b). The reflected
signal
decreased
further and reach
ed
its minimum
when
the bi
as
was decreased
to 9.3
V
pp
(Figure
S
8
c), and it saturate
d
when the bias
was
10.9 V
pp
(
Figure S
8d
)
, indicating that
the
VO
2
had
completely switched to the rutile phase. We note
d
that the monitored non
-
monotonic behavior
corresponds to the steady
-
state responses as shown in Figure 3c. At fixed wavelength (
λ
=
1550
nm), the reflected signal show
s
a non
-
monotonic behavior because of the redshift of the
resonance dip when the applied
bias
is
larger than 8.9 V
pp
.
NIR tunable
laser
Metasurface
O
scilloscope
9
Figure S
8
.
Temporal response of the metasurface for different peak
-
to
-
peak voltages.
The
corresponding peak
-
to
-
peak voltages are
(a) 3.4 V
pp
, (b) 8.9 V
pp
, (c) 9.3 V
pp
,
and (d) 10.9 V
pp
,
respectively.
In the figures, the blue and red lines correspond to reflected signals at
λ
= 1550 nm
and trigger signals
(voltage pulses)
, respectively.
1
2
3
4
5
6
7
8
9
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
6
12
18
Voltage (V)
1
2
3
4
5
6
7
8
9
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
6
12
18
Voltage (V)
1
2
3
4
5
6
7
8
9
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
6
12
18
Voltage (V)
1
2
3
4
5
6
7
8
9
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
6
12
18
Voltage (V)
a
b
c
d
10
Figure S
9
.
Temporal response of the metasurface for different applied voltage pulses. (a)
Reflected signal for
10.9 V
pp
-
high and
2.5
-
s
–
wide
voltage pulses
with
a
50% duty cycle
.
(b)
Reflected signal for
25.7
-
V
pp
–
high and 15
-
ms
–
wide voltage pulses
with
a
7.5% duty cycle
.
The measured
ON
and OFF
switching
time
(complete phase transition)
for the
biases with
10.
9
-
V
pp
–
high, 2.5
-
s
–
wide
,
and 50% duty
-
cycle
voltage pulses
were
≈
2
s
and
≈
700
ms,
respectively
(Figure S9
a)
.
However, the switching speed is not limited by the intrinsic transition
speed of
the
VO
2
,
but mainly by the thermal design of the device and input parameters. We note
d
that the previously reported switching speed
s,
involving thermal processes
,
are sub
-
picosecond for
optical excitation
3
and a fe
w nanosecond
s
for electrical switching.
4
Nevertheless, the switching
speed of the metasurface can be si
gnificantly enhanced by utilizing
more
-
localized Joule heating
,
by
manipulating applied electrical pulses.
5
For example, we observed ON and OFF switching time
s
of
≈
15 ms and
≈
100 ms, respectively, by applying high
-
intensity (25.7 V
pp
) short (15 ms) voltage
pulses with a 7.5% duty cycle
,
as shown in Figu
re S
9
b.
1
2
3
4
5
6
7
8
9
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
12
24
36
Voltage (V)
0.1
0.2
0.3
0.4
0.5
0.6
0
2
4
6
8
Photodiode
Trigger signal
Time (s)
Intensity (a.u.)
0
12
24
36
Voltage (V)
a
b
ON
OFF
11
Reference
s
(1)
Zhu, Z.; Evans, P. G.; Haglund, R. F.; Valentine, J. G.
Nano Lett.
2017
,
17
(8), 4881
–
4885.
(2)
Qazilbash, M. M.; Brehm, M.; Chae, B.
-
G.; Ho, P.
-
C.; Andreev, G. O.; Kim, B.
-
J.; Yun,
S. J.; Balatsky, A. V.; Maple, M. B.; Keilmann, F.; et al.
Science
2007
,
318
(5857), 1750
–
1753.
(3)
Liu, M.; Hwang, H. Y.; Tao, H.; Strikwerda, A. C.; Fan, K.; Keiser, G
. R.; Sternbach, A.
J.; West, K. G.; Kittiwatanakul, S.; Lu, J.; et al.
Nature
2012
,
487
(7407), 345
–
348.
(4)
Markov, P.; Marvel, R. E.; Conley, H. J.; Miller, K. J.; Haglund, R. F.; Weiss, S. M.
ACS
Photonics
2015
,
2
(8)
, 1175
–
1182.
(5)
Liu, L.; Kang, L
.; Mayer, T. S.; Werner, D. H.
Nat. Commun.
2016
,
7
, 13236.