of 28
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the
Version of Record
. Please cite this article as
doi:
10.1002/adma.202104863
.
This article is protected by copyright. All rights reserved.
Electrically Tunable and
Dramatically Enhanced
Valley
-
Polarized Emission of Monolayer WS
2
at
Room Temperature
with
Plasmonic Archimedes Spiral Nanostructures
Wei
-
Hsiang Lin
*
, Pin Chieh
Wu, Hamidreza Akbari, George. R. Rossman, Nai
-
Chang Yeh* and Harry A.
Atwater*
Wei
-
Hsiang Lin, Hamidreza Akbari, Prof. Harry A. Atwater
Department of Applied Physics, California Institute of Technology, Pasadena, CA 91125, USA
E
-
mail:
haa@caltech.edu
and
whlin@alumni.caltech.edu
Prof. P
in Chieh Wu
Department of Photonics, National Cheng Kung University, Tainan 70101, Taiwan
Prof. George. R. Rossman
Department of Geological and Planetary Sciences, California Institu
te of Technology, Pasadena, CA
91125, USA
Prof. Nai
-
Chang Yeh
Department of Physics, California Institute of Technology, Pasadena, CA 91125, USA
E
-
mail: ncyeh@caltech.edu
Keywords:
transition metal dichalcogenides, exciton
-
plasmon interaction, valleytroni
cs, circular
dichroism, chiral plasmonic metasurface
Abstract
Monolayer transition metal dichalcogenides
(TMDs)
have intrinsic valley degrees of freedom, making
them
appealing for exploiting valleytronic application
s
in information storage and processing. WS
2
monolayer possesses two inequivalent valleys in the Brillouin zone, each valley coupling selectively
with
a
circular
polarization of
light. The degree of valley polarization (DVP) under the excitation of
circular
ly polarized light
(CPL)
is a parameter that determines the purity of valley polarized
photoluminescence (PL) of monolayer WS
2
.
Here, we report
efficient tailoring of
valley
-
polarized PL
from
monolayer WS
2
at room temperature (RT) through surface plasmon
-
e
xciton interaction
s
with
plasmonic Archimedes spiral (PAS) nanostructures. The DVP of WS
2
at RT
can
be enhanced
from <
5%
to 40% and 50%
by using 2 turns (2T) and 4 turns (4T) of PAS
, respectively. Further enhancement
and control of excitonic valley polari
zation
is
demonstrated by
electrostatic
al
ly
doping
monolayer
WS
2
.
For
CPL
on WS
2
-
2TPAS heterostructure
s
,
the
40% valley polarization is
enhanced
to 70% by
modulating the carrier doping via a backgate
, which
may be attributed to the screening of
momentum
-
dependent long
-
range electron
-
hole exchange interactions. The
manifestation
of
electrical tunibil
i
ty in the valley
-
polarized emission from WS
2
-
PAS heterostructures
presents a
new
strategy
towards
harness
ing
v
alley excitons for application in ultrathin valleytronic devices.
This article is protected by copyright. All rights reserved.
2
Introduction
Monolayer
2H
-
phase transition metal dichalcogenides (
TMDs
)
MX
2
are direct band gap
semiconductors
that
consist of a
n atomic layer
of transition metal atoms (M) sandwiched between
a
top and
a
bottom
layer of
halogen atoms (X)
that
are arranged in the
ir
respective hexagonal lattice
structure
.
[1,2]
The band structures of monolayer TMDs
[3
7]
consist
of two inequivalent K (
K) valleys
in the
hexagonal
Brillouin zone.
T
he
strong spin
-
orbit coupling and broken
inversion symmetry
in
monolayer
2H
-
TMDs
result in
a
large energy splitting between the top spin
-
up (spin
-
down) and
bottom sp
in
-
down (spin
-
up)
valence bands
in the K (
K) valley
.
[8
13]
This
spin
-
valley coupling
gives
rise to
valley
-
dependent optical selection rules
:
[14
16]
Circularly
polarized light with positive helicity
(
+
) couples to the K va
lley and
that of
the negative helicity (
) couples to the
K valley.
It is
therefore possible to selectively populate and manipulate the different valleys (K or
K) by means of
circularly polarized light
(CPL).
[11,15]
T
he spin and valley degrees of freedom
(DOF)
in these
monolayer
TMDs materials provide an important platform
for exploring new condensed
matter
physics
and
for
application
s
in spintronic
and valleytronic
devices
.
Noting that the control of
circularly polarized
emission of light
is also a
key step towards developing functional optical devices
for
applications
in
d
isplay
technologies
[17,18]
and
quantum
information
processing
/
communication
,
[19
21]
manipulation of the valley
DOF
is one of the promising approaches
to
achieving
practical information processing devices that are more
energy
efficient
than
existing
charge
-
based
technologies
.
Although a
ccessing different valley
DOF
through CP
L
has been demonstrated in several TMDs,
including
molybdenum disulfide (MoS
2
)
,
[8
10]
molybdenum diselenide (MoSe
2
)
,
[22]
tungsten disulfide
(WS
2
)
,
[13,23]
and tungsten
diselenide (WSe
2
),
[24
27]
it has been challeng
ing
to manipulate the valley
polarization and the resulting photoluminescence (PL) of monolayer 2H
-
TMDs at room temperature
(RT)
because of the weak light
-
matter interaction and substantial defects
[
13]
in these atomically thin
This article is protected by copyright. All rights reserved.
3
layers. Various approaches, such as applying an in
-
plan electric field
,
[24,28]
an out
-
of
-
plan magnetic
field
,
[29
33]
or a localized magnetic field
,
[34]
have been proposed in an attempt to enhance the
circular dichroism (CD) of P
L in
these
TMDs.
An alternative approach to enha
n
cing the CD of PL in TMDs
is
by means of chiral plasmonic
metasurfaces
.
[34
43]
This approach is based on the strong optical chirality in metallic chiral
metasurfaces
and the enhancement of light
-
matter interaction through localized surface plasmon
resonance (LSPR) modes. Specifically, the confined LSPR modes can strongly interact with the carriers
within the monolayer TMD material
to
form a compact, fuctional nanophotonic devices. Futhermore,
the geometry
-
dependent LSPR modes of chiral plasmonic me
tasurface provide wavelength tunability
for the wide emission range of different TMDs. Thus, implementing chiral plasmonic metasurfaces
[44
54]
on
2H
-
TMDs is likely a
n efficient
approach to manipulating the CD of PL in these
monolayer
materials.
In this study, a metal
-
dielectric
-
metal (MDM) structure
that
consist
s
of monolayer WS
2
and
plasmonic Archimedes spiral (WS
2
-
PAS)
nanostructures
is designed to investigate the monolayer WS
2
exciton
-
plasmon
interaction and to manipulate the valley
-
polarized PL in WS
2
-
PAS heterostructures
by spectroscopic measurements and numerical simulations. The device configuration
included
a
monolayer WS
2
single crystal grown by chemical vapor deposition (CVD) and
transf
erred to a
substrate of 60 nm
-
thick SiO
2
on top of a 150 nm
-
thick gold reflection layer. A 60 nm
-
thick gold film
was
subsequently deposited on the WS
2
/SiO
2
/Au structure, and then
either righthanded (RH) or
lefthanded (LH)
spiral gold rings
were
patterned o
n the gold film by E
-
beam lithography.
To
demonstrate the crystallinity of
our monolayer
WS
2
samples,
we performed Raman spectroscopic
studies of the
1
and
2
1
modes using a 514.3 nm laser (2.41 eV) as the excitation source.
Additionally, X
-
ray phot
oelectron spectroscopy (XPS) was conducted to examine the chemical
This article is protected by copyright. All rights reserved.
4
composition and valence states of monolayer WS
2
,
as detailed
previously.
[13]
Representative Raman
and XPS characterization results of a monolayer WS
2
are
given
in Figures S1 and S2
,
respectively
.
The magnitude of the degree of valley polarization (DVP) in as
-
grown monolayer WS
2
wa
s < 5 %
at RT
, whereas those
of the DVP in monolayer WS
2
-
2T
RH
PAS
(
with
2 turns of PAS)
and WS
2
-
4T
RH
PAS
(
with
4 turns of PAS)
under the excitation of right
-
ha
nded circularly polarized (RCP) light
were
found
to
be
40% and 50%, respectively.
Here the DVP value (
P
DVP
) is defined by the following expression:
DVP
=
(
+
)
(
)
(
+
)
+
(
)
,
(1)
where
I
(
σ
+
) and
I
(
σ
) denote the right
-
and left
-
handed
circular polarization
-
resolved PL intensity,
respectively.
In contrast, the corresponding values of the DVP under the excitation of left
-
hand
ed
circularly polari
z
ed (LCP) light decrease
d
to
~
5 % in both monolayer
WS
2
-
2T
RH
PAS and WS
2
-
4T
RH
PAS heterostructur
es. Additionally, the DVP
values could
be further enhanced by applying a
back
-
gated voltage
V
G
ate
from 40% for
V
Gate
= 0
up
to 70% for
V
Gate
=
2
0 V. This finding suggests that
modulating the carrier doping level can enhance the valley polarization by
screening the long
-
range
electron
-
hole exchange interactions
,
thus reducing the momentum
-
dependent inter
-
valley
scattering.
The capabil
i
ty of enhancing and tuning the degrees of valley polarization in monolayer
2H
-
TMDs
at room temperature
by plasmonic chir
al metasurfaces
and electrical gating
offers
new
oppo
r
tunities toward
s
developing valley
-
dependent optoelectronic devices for energy
-
efficient
information processing.
Our proposed
underlying mechanism for tailoring the valley
polarized PL
of monolayer 2H
-
WS
2
by PAS
and gated voltage
is schematically
shown
in Figures 1a
-
1c
: In WS
2
monolayer, excitons in the
K valley couple with the right handed
(RH)
circularly polarized light (
+
), whereas excitons in
the
K
valley couple with the left handed
(LH)
circularly polarized light (
). However, after
+
light
excitation, the exciton population pumped in the K valley could be scattered to the
K valley due to
This article is protected by copyright. All rights reserved.
5
phonon
-
and defect
-
i
nduced inter
-
valley scattering. Therefore, the DVP of PL was usually low at
RT
as show in Figure 1a. On the other hand, when WS
2
monolayer interacts with the RHPAS structures
in the near
field range, optical absorption is enhanced, which contributes to more
+
excitons under
right handed CPL in the WS
2
-
RHPAS structure. In the K
v
alley, the
+
excitons interact with the right
handed (RH) chiral near
field of the
surface plasmon polaritons (
SPP
)
of PAS
, leading to giant optical
activit
ies
that contribute to enhanced
the
PL intensity in the far field,
as shown in Figure 1
b. In
contr
ast, the
exciton emission generated from
K valley is suppressed by the interaction with the
RH chiral
near
-
field.
The interplay of exciton generation, exciton emission, and inter
-
valley scattering
in each valley contributes to far
field measurements o
f the valley
polarized PL spectra. As
schematically shown in Figure 1
a, the DVP of the PL emission depends on the decay rates of the
excitons at K (
) valley,
K valley (
), and
the
inter
-
valley scattering rate between the two
valleys. Therefore,
the DVP can be enhanced through increasing the decay rates of the valley
-
polarized excitons or decreasing the inter
-
valley scattering rate in two valleys by the chiral Purcell
effect. Although the presence of RHPAS structures contributes to increasing the
decay rate of the
+
excitons, the inter
-
valley scattering is still significant so that the DVP enhancement is limited. The
latter situation may be mitigated by applying a gate voltage between the WS
2
and the gold mirror,
as
shown in Figure 1
c, so that th
e excess carrier doping helps screen the momentum
-
dependent long
-
range electron
-
hole exchange interaction, thereby reducing the inter
-
valley scattering and leading to
significantly enhanced DVP in gated WS
2
-
RHPAS
hetero
structures.
Results and Discussion
C
oupling TMDs with chiral plasmonic metasurfaces can enhace the light
-
matter interaction
through the confined localized surface plasmon and can be a promising approach to enhance the CD
of PL in 2H
-
TMDs.
Figure
2
a
is a
schematic
illustration
of the metal
-
di
electric
-
metal (MDM) structure
that consists of a monolayer WS
2
and plasmonic Archimedes spiral nanostructures (WS
2
-
PAS) under
This article is protected by copyright. All rights reserved.
6
the excitation of
CPL
. The WS
2
monolayer was grown by chemical vapor deposition (CVD) method
and then transferred onto a substrate consisting of a 60 nm
-
thick SiO
2
layer on top of a 150 nm
-
thick
gold mirror layer. The Archimedes spiral rings were patterned by E
-
beam lithography on a 50
nm
-
thick Au film deposited on the
top of the
WS
2
monolayer
.
Figure
2
b shows the PL intensity
differences between the as
-
grown monolayer WS
2
and the
MDM structure of a monolayer WS
2
integrated with right
-
handed
plasmonic
Archimedes spiral
(RHPAS)
nano
s
tru
ctures. Under the excitation of right
-
hand circularly polarized light (RCP, red), the
PL intensity of the
WS
2
-
2T
PAS
heterostructure
was
enhanced by more than 10 times relative to that
of the as
-
grown WS
2
(green). In contrast, the PL intensity
was
barely en
hanced under the excitation
of left
-
handed circularly polarized light (LCP, blue).
The strong polarization
contrast is a good
indicator for exciton
-
plasmon coupling,
and
the
quantitative strength can be characterized by the
intensity
enhancement factor (EF). The
enhancement factor
(EF)
is defined as EF
= I
wi
/I
wo
, where I
wi
(I
wo
) is the maximum intensity of WS
2
with (without)
the
Archimedes spiral nano
s
tructures.
Figure
2
b
shows
the EF
is larger than 10 between
the
RCP emi
s
sion
of the
W
S
2
-
2T
PAS
and
that of
the as
-
grown
WS
2
,
whereas the
EF
is nearly 1 between
the
LCP emission
of
the
WS
2
-
2T
PAS
and
that of
the as
-
grown WS
2
.
This spectroscopic phenomenon
suggests that coupling
WS
2
with PAS nanostructures is
an efficient way for tailoring the exciton
-
plasmon interaction
through
actively
controlling
the
polarization of the incident light.
The inset
of Figure
2
b is a
scanning electron microscope (SEM)
image of
arrays of
WS
2
-
2T
PAS
heterostructure
,
where
the scale bar
represents
5
m
.
To understand the enhancement of PL intensity and valley polarization, we note that i
n the
saturation regime
,
the
EF
consists of
contributions from three processes
:
an enhanced absorption
rate (
α
),
an
enhanced light extracti
on
factor
(
ε
), and
a
total rate enhancement factor
(
on
off
)
from the Purcell effect
that
includes
both
radiative and
nonradiative rates as well as metal
lic
loss
es.
Thus, we have
EF
=
훼휀
(
on
off
)
, where
on
is the
total decay rate of WS
2
coupled to
the
PAS
This article is protected by copyright. All rights reserved.
7
nanostructures
,
and
off
is the total decay rate of
the as
-
grown
WS
2.
The total rate enhancement
factor
can be directly determined by the
time
-
resolved PL measurements
.
As shown i
n Figure S
3, the
extract
ed
decay time of
the as
-
grown
WS
2
was
1
.
5
ns (
off
=
1
off
), whereas that for
the
WS
2
-
2TPAS
wa
s
0
.
3
ns
(
on
=
1
on
)
. The
WS
2
-
2TPAS
heterostructure
s
displayed
a significantly faster
decay
rate
and
the total rate enhancement
was found to be
(
on
off
)
~
5
(
~
500%).
Therefore, the
underlying Purcell factor (
F
p
)
could be est
i
mated from the relation
=
(
+
)
(
on
off
1
)
off
1
, as
detailed in
Supporting
Information
S2
.
Here
and
are the radiative and metallic
-
loss decay
rates, respectively, and
off
(
+
푁푅
)
represents the quantum efficiency of the as
-
grown
WS
2
. From
t
he measured
value
(
on
off
)
~
5
and using
(
+
)
1
in the limit of a large
and
taking
off
=
0
.
1
,
[24
27]
we obtained
an underlying
Purcell factor
40
.
T
he coupling factor
(
1
+
)
thus estimated
was around 98%
,
which
implied
that
nearly all the
PL
emission
was
coupled to the
MDM nano
-
cavity mode.
The exciton
-
plasmon interaction in WS
2
-
PAS
nano
structures is dependent on the incident laser
power, the
PL intensity of
as
-
grown
WS
2
, and the specific designs of the Archimedes spiral
nanostructures. The laser power dependence of PL, under 514 nm RCP laser excitation in WS
2
-
2TPAS
with RH spiral nanostructures,
was revealed
in Figure
2
c. The PL intensity of
the WS
2
-
2TRHPAS
was
found to increase rapidly with the laser power from 1.9 μW to 16.2 μW, and the PL peak position
also exhibit
ed
a slight blue shift with increasing laser power up to 5.5 μW
,
and then
remained
invariant from 5.5 μW to 16.2 μW. Figure
2
d
p
resented
a log
-
log plot of the power
-
dependent PL
intensity of WS
2
with (red, triangle) and without (blue, inversed triangle) the plasmonic coupling
effect. The relationship between PL intensity (
I
PL
) and laser power (
P
laser
)
could
be approximately
express
ed by the relation
I
PL
= (
P
laser
)
m
, where
m
denoted
a constant. For both WS
2
with and without
spiral structures, the corresponding
m
values
were
fitted to
be
~ 1.37 for laser powers ranging from
1.9 μW to 16.2 μW, and a maximum enhancement factor
ε
reached
>
10 at 9.1 μW.
This article is protected by copyright. All rights reserved.
8
V
alley polarization
in monolayer TMDs
could
be induced by
CPL
excitations and detected by PL
measurements at room temperature.
Figure
3
shows the
comparison
of the valley p
o
larized PL
spectra between an as
-
grown monolayer WS
2
and a WS
2
-
2T
RH
PAS
heterostructure
under the
excitations of RCP and LCP light. The wavelength of the excitation laser
was
514 nm (
E
pump
= 2.41 eV),
which
was
far away the neutral exciton resonant energy (
E
X
= 2 eV). A linear polarizer and a quarter
-
wave plate (400
-
800
nm)
were
placed after the laser to obtain circularly polarized light. The
polarized PL signal
was
resolved by a broadband polarizer and
a
quarter
-
wave plate placed before
the spectrometer.
+
+
+
+
+
+
+
This article is protected by copyright. All rights reserved.
9
+
To better understand the exciton
-
plasmon interactions, we further investigated the effect of
different plasmonic structures on the valley
-
polarized PL spectra.
Figure
S
4
a
show the
SEM image of
2
-
turn RHPAS, and the
structural
parameters
of the PAS
are
given
in Figure
S4b
.
The spiral structure
has an inner radius
r
0
= 200 nm, width = 50 nm, and pitch = 310 nm so that the outer radius
r
=
r
0
+
N
(
R
Plasmon
/2), where
N
is the i
nteger number of turns and
R
Plasmon
= 620 nm is the resonance
wavelength of the surface plasmon polariton, which coincide
s
with the neutral A
-
exciton emission
wavelength in the PL spectrum of monolayer WS
2
.
These structure parameters
were
chosen by the
COM
SOL
simulation results (Figure
s
S4
and
S5
)
so that
strong field enhancement
could
be
achieved
at the ring center under the RCP excitation.
Moreover, the pitch of the spiral must match the
wavelength of the surface plasmon polariton
(SPP)
.
Overall, p
roper
choices of the structural
parameters are
essential to optimize the exciton
-
plasmon interaction in the PAS structures.
Figure
S4c
show
s
the near
-
field
PL
intensity
in the
x
-
y plane
under
RCP excitation.
When a
2
-
turn
RH
PAS
interacted
with the RCP light, t
he surface plasmon polaritons (SPPs)
were
generated by the
RH spiral structures and coherently propagated toward to the ring center
to form
a
constructive
focus point. The excitons thus generated (RCP emission) interact
ed
constructively with the SPP field,
which result
ed
in enhanced PL intensity at the center of the RH spiral structure.
Figure
S4e
shows
the corresponding phase map of
the
2
-
turn
RH
PAS
in the
x
-
y plane
under
RCP excitation
s
.
The phase
at the center of the
spiral ring
revealed a
locally unifor
maly phase distribution as
shown by
the white
dotted circle in Figure
S4e
.
Thus
, the
constructive
interaction of exciton
s
and
SPP
field
s
resulted in
significant
PL enhancement under
the
RCP excitation. In contrast, the LCP
excitation
induced
weak
E
-
field distribution without focusing
effects, as
shown in Figure
S4d
. At the center of the spiral ring the
phase
varied
from
π to π
as shown in Figure
S4f
. The SPPs propagate
d
as before with
the
LCP
This article is protected by copyright. All rights reserved.
10
excitation, but the phase
could not
compensate
to yield
constructive, focusing effects at the center
of the spiral structure
.
Interestingly, t
he E
-
field and phase distribution
were
also
dependent on
the number of
turns of
the
PAS
, and the
dependence
of the exciton
-
pla
s
mon interaction on the number
of turns
i
s
manifested
in Figure
4
and Figure
S5
.
In
Figure
4
a
-
4
d
, we show
the SEM images of 1
-
, 2
-
, 3
-
and 4
-
turn
PAS
arrays on
a
monolayer WS
2
single crystal.
The corresponding PL intensity maps
under RCP
excitation
are shown in Figure
4
e
-
4
h
,
where it is evident that the PL intensity
became
dramatically
enhanced when the WS
2
monolayer
coupled
with the 2
-
turn and 4
-
turn
RH
spiral rings. In contrast,
the intensity enhancement almost
disappeared
in the case of 1
-
turn and 3
-
turn
RH
spiral rings.
Th
e
corresponding
E
-
field distribution
s
are shown in Figure
s
S5a
-
S5d
, where we
found
that the focusing
effects only
happen
e
d
for the RHPAS structures with an
even
number of
turn
s
.
This finding may
be
explained
in terms of
the phase compensation
effects
induced
by the
outer and inner spiral coupling
at subwavelength
s
. The phase distribution of the PAS heterostructure demonstrate
d
a
strong
dependence on the number
of turns in the spiral structure,
as shown in Figure
S5e
-
S5h
.
For
PAS with
an
even number
of
turns
, the phase at the center of the spiral ring
revealed a
locally uniform phase
distribution
,
as
manifested by
the white dotted circle Figure
S5f
and
S5h
. In contrast
, the phase
changed
from
π to π for
PAS with an
odd number
of turns
.
In addition to t
he PL intensity,
spatial
ly
resolved DVP
maps
further
reveal
ed
that the DVP
wa
s
also dependent on the number of turns in the PAS structure, as shown in Figure
s
4
i
-
4
l
and Figures
S6
-
S9
for
1
-
, 2
-
, 3
-
and 4
-
turn
PAS
arrays on
a
monolayer WS
2
.
For
a
RH
PAS with an odd number of
turns, the DVP values
were
lower than 10%.
In contrast
, for
a RHPAS with an
even number of turns,
the DVP values were
much enhanced,
r
anging from 35% to 55% for 2T
-
RHPAS
and from 40% to 55%
for 4T
-
RHPAS.
Such
strongly enhanced
valley
polariz
ation associated with the far
-
field
PL mapping
of
This article is protected by copyright. All rights reserved.
11
the WS
2
-
PAS
heterostructure
may
be attributed to the
consequences of the
near
-
field interaction
between the valley exciton
s
in
WS
2
a
nd
the surface pla
s
mon modes
in the PAS
.
We further investig
ated the effect of the PAS resonance wavelength (
R
Plasmon
) on the
enhancement of the DVP by performing PL measurements on
WS
2
-
2TRHPAS
heterostructures with
three different resonance wavelengths
R
Plasmon
= 310 nm (2.42×10
14
Hz), 620 nm (4.84×10
14
Hz) and
930 nm (7.26×10
14
Hz), and the corresponding PL spectra and the DVP values are shown in Figure
S10
.
These results revealed
the best room temperature DVP enhancement
occurred at
R
Plasmon
=
620
nm, which coincided with the neutral A
-
exciton emission
wavelength.
N
eutral exciton
s
are
the natural low
-
energy excitations of
a charge
-
neutral
semiconductor,
whereas trions
are
only formed in the presence of excess
charge. Therefore, the intensity of trion
emissions is generally
dependent on
the amount of exc
ess charge in the semiconductor. For this
reason, trion emissions
were
usually not
found
in
the PL spectra of
our CVD
-
grown monolayer WS
2
samples unless a back gate voltage
was
applied.
For the gated samples, the PL spectra typically
exhibit
ed
additional emissions at 30
60 meV below
the neutral excitonic line, which
may be
attributed to the emission from
negatively
charged
trions
(X
). Thus, by simply varying the applied
back
gate
voltage, we
were able to
control the ratio between neutral
excit
on
and
charged
trion
emissions.
Figure
5
a
shows
a
schematic
of
a back
-
gated FET device based on
WS
2
-
2TRHPAS
and
an
SEM
image of
the top view of such a
device.
For a given gate voltage,
there
were
two well
-
defined
PL
spectral
components associated with the
emission bands of the neutral excitons (X) and the
negativ
ely
charge
d
trions
(X
).
We found that
the emission near 620 nm (2 eV) from neutral excitons
(X) was dominant around the charge neutrality point at
V
Gate
= 0.
The optimal line
shape
fitting
for
the
spectral contributions
was found to be
a mixed Gaussian
Lorentzian function
,
as shown in
Supporting Information
Figure
S11
for WS
2
and
Figure
S1
2
for
WS
2
-
2TRHPAS, respectively
.
The PL
This article is protected by copyright. All rights reserved.
1
2
spectral evolution of these two emission bands with the gate voltage is
illustrated in
Figure
s
S1
3
and
S1
4
for
as
-
grown
1L
-
WS
2
,
and
in
Figure
s
5
b
-
5
c
for
WS
2
-
2TRHPAS
.
We note that the trion
-
to
-
exciton
intensity ratio
s
of both WS
2
-
2TRHPAS and 1L
-
WS
2
exhibited
similar
dependence on the gate voltage,
as shown in Figure 5c and Figure
S1
3
c
, respectively.
Figure
S13a
shows a color map of the PL spectra from a
back
-
gated field effect device
of as
-
grown 1L
-
WS
2
with the
back
-
gate voltages (
V
Gate
) from −25 to 25 V at RT.
For negative gate voltages
(
V
Gate
<
0), both X and X
features
were
visible, and the integrated
intensity of the X
-
peak gradually
decreased
while that of the
X
-
peak increase when |
V
Gate
|
decreases. In contrast, for positive g
ate
voltage (
V
Gate
> 0
), the integrated
intensity of the X
-
peak gradually
decreased
and the X
-
peak
became
dominant with increasing
V
Gate
due to electrostatic doping, which
led
to a single
X
-
peak
profile
for sufficiently high
carrier densities
, implying that the Fermi level
moved
much closer to the
energy of the
bound
trion
states than to that of the neutral exciton states.
Helicity resolved PL spectra of an as
-
grown gated WS
2
sample are shown in Figures
S1
4
b
and
S1
4
c
for
V
Gate
=
2 V and +5
V, respectively. When the WS
2
sample
was
negatively gated at
V
Gate
=
2
V
, both X and X
emissions reveal
ed
relatively low DVP of 4.2% and 5.8%, respectively.
Similarly, at a
positive gate voltage (
V
Gate
= 5V
), both X and X
emissions also reveal
ed
low DVP
of 4.7% and 5.1%,
respectively
. These low DVP values impl
ied
that the
significant inter
-
valley scattering
was insensitive
to the gate voltage in the as
-
grown
WS
2
sample
.
In contrast to the
small
effect of the gate voltage
on the DVP of the as
-
grown WS
2
sample, t
he
gate voltage
dep
e
ndent
DVP
became
significantly different in the case of
1L
-
WS
2
-
2TRH
PAS
heterostructure
s
.
Figure
5
b
-
5
c show the
PL
spectral evolution of X and X
emissions from the
1L
-
WS
2
-
2TRHPAS sample with the gate voltage.
Additionally, pol
arization
-
resolved PL spectra of the
1L
-
WS
2
-
2TRHPAS sample under
+
exc
itations are shown in Figures
5
d
-
f
for
V
Gate
=
0
,
10 V and
20 V,
respectively. For
V
Gate
= 0, the photon
-
excited excitons at the K valley
were
more populated under
This article is protected by copyright. All rights reserved.
13
+
excitations, which
corresponded
to the valley
-
polarized state in pristine WS
2
and WS
2
-
2TRHPAS.
By increasing the electron density via
de
creasing the gate voltage from 0 to
10 V and
20 V, the
difference between the
+
and
components of the PL spectra
became
more significant
,
implying
increased valley polarization of the neutral excitons. Specifically, we
found
that the values of DVP for
V
Gate
= 0,
10 V and
20 V
were
40%
,
50%
and
70%
, respectively, suggesting significantly enhanced
valley polarization as the applied bias
wa
s detuned from the charge neutral point.
The corr
e
sponding
RT
spatial
ly
resolved DVP map
and PL maps
of
a
WS
2
-
2TRHPAS
device at
V
Gate
=
20 V
are
show
n
in
Figure
s
5
g
and Figure S15, respectively.
To gain
further
insights into this behavior, we perform
ed
gate
-
dependent transport
measurements
, using a scheme where a positive bias
induced
hole
-
doping and a negative bias
introduced
electron
-
doping.
We observe
d
typical
n
-
type transport behavior with on/off current
ratios
greater than 10
4
at room temperature, as shown in Figure
S1
6
.
The doped carrier density
n
under gate voltage
V
G
ate
was
estimated from the geom
e
tric capacitance and back
-
gate voltage using
the
following
r
elation
=
(
0
0
x
)
(
Gate
CNP
)
,
(2)
where
=
3
.
9
is the dielectric constant of SiO
2
,
0
x
is the thickness of SiO
2
,
and
0
is the vacuum
permissivity
. As shown in Figure
S16
, the charge neutral point (CNP)
was
observed to be at 2
V
so
that
the n
-
type carrier concerntration
could be
esimated
by using Eq. (2), which yielded
7.19
10
11
cm
-
2
,
4.31
10
12
cm
-
2
, and
7.91
10
12
cm
-
2
for 0
V,
10
V and
20
V,
respectively.
Given a carrier
concentration
n
, the
corresponding Fermi
-
level position
could
be es
ti
mated by using
the following
expression for semiconductors:
=
ln
(
2
1
)
,
(3)
This article is protected by copyright. All rights reserved.
14
where
is the Fermi
-
level energy,
is the
conduction band edge,
2D
=
v
ћ
2
is the density
of the states,
v
is the valley degeneracy factor
, and
=
0
.
34
0
. Therefore, the Fermi
-
level
position
was
estimated to be 0.3
eV, 0.25
eV and 0.23
eV lower than the conduction band minimum
for 0
V,
10
V and
20
V
, respectively.
As sho
wn in
Figure
s
5d
5f
,
the
degree of
valley polarization
steadily
increased
with
the carrier density
at RT, which
may be attributed
to the noticeable
screening effect by the doped carrier
s
.
Specifically, the mechanism for the enhancement of valley polarization by electrostatic doping
in
the WS
2
-
2TRHPAS heterostructure
may be attributed to carrier doping
-
induced suppression on the
inter
-
valley relaxation process. The inter
-
valley relaxation pr
ocess of bright excitons is dominated by
the long range electron
-
hole (e
-
h) exchange interaction, and the long
-
range e
-
h exchange
interactions can be efficiently screened by adding 2D electrons/holes with electrostatic doping.
[55]
The screen length is det
ermined by the inverse of the Thomas
-
Fermi wave vector, which is given by
the following expression:
[55]
푇퐹
(
,
)
=
푇퐹
0
[
1
(
)
]
,
(4)
where
푇퐹
0
=
s
v
2
(
4
휋휀
ћ
2
)
is the zero temperature Thomas
-
Fermi wave vector,
s
(
v
)
is
the degeneracy for spins (valleys),
is the effective electron or hole mass, and
ε
is the dielectric
constant. The Fermi energy
measured from the bottom of the conduction band (to the top of the
valance band) is defined by
=
2
휋푛
ћ
2
(
s
v
)
, where
n
is the doped electron (hole) density.
Therefore,
푇퐹
increases rapidly with increasing
n
according to Eq. (4).
In the strong scattering limit,
the inter
-
valley scattering rate
(
v
)
1
due to e
-
h exchange interaction may be approximated by
the
relation
(
v
)
1
(
푇퐹
)
2
.
[55]
Therefore, the inter
-
valley scattering rate
(
v
)
1
is strongly
suppressed by carrier doping. In contrast, the intra
-
valley relaxation time
0
is much less affected by
carrier doping, as supported by the stable linewidths a
nd integrated intensities upon doping
.
[56]
Our
This article is protected by copyright. All rights reserved.
15
work aims to tailor the degree of valley polarization by manipulating the inter
-
valley scattering rate
(
v
)
1
via tuning the carrier doping. Noting that the valley polarization
P
DVP
[57]
is given by
퐷푉푃
=
0
1
+
2
(
0
v
)
,
(5)
where
P
0
is the ideal valley polarization
, we find that
the suppression of
(
v
)
1
by electrostatic
doping leads to the enhancement of
P
DVP
, which agrees well with our experimental observations.
Conclusion
In summary, we have de
monstrated that ultra
-
compact
plasmonic Archimedes spiral (PAS)
nanostructures can efficiently tailor the valley
-
polarized PL of monolayer WS
2
at room temperature.
The chiral
-
SPP focusing fields from the PAS nanostructures can significantly amplify the exc
iton/trion
emissions in the WS
2
monolayer with specific chiral emission due to the interaction of WS
2
excitons/trions with the plasmons of the PAS nanostructures. The plasmon
-
exciton coupling is
shown to provide control of the valley
-
polarized PL through
manipulating
the imbalance of valley
population, leading to significant PL modulations through valley
-
dependent control of the excitonic
emissions at room temperature. In particular, we have demonstrated enhancement of the degrees
of valley polarization (D
VP) at room temperature from
~
3% for the as
-
grown 1L
-
WS
2
to ~ 40% for
the 1L
-
WS
2
-
2TRHPAS
heterostructures
and
to ~ 50% for the
1L
-
WS
2
-
4T RHPAS heterostructures.
Furthermore, by appl
y
ing a back
-
gate bias voltage (
V
Gate
), the valley polarization is found to
be
further enhanced from 40% to 70% for
V
Gate
=
2
0 V, which implies that
the
excess
carrier doping can
enhance the valley polarization by screening of the momentum
-
dependent long
-
range electron
-
hole
exchange interaction and therefore reducing the inter
-
v
alley scattering.
The methodology described
in this
work
provides a promising platform to manipulate the valley degrees of freedom in 2H
-
TMDs
efficiently at room temperature, paving ways for future applications of
opto
-
valleytronic
/spintronic
devices based
on the
se
2D materials.
This article is protected by copyright. All rights reserved.
16
Experimental Section
Synthesis of WS
2
: Monolayer WS
2
was grown
by
using APCVD as
reported
previously. 95 mg WO
3
precursor mixed with 5 mg KI was placed in a quartz boat containing the SiO
2
/Si substrates
that were
set face
-
down
directly above the W source precursor, and the quartz boat was then positioned at
the center of the furnace. A second boat containing 100 mg S was placed upstream at 18 cm away
from the W source. Next, the system was pump down to 3×10
-
2
torr to eliminate a
ir and moisture.
After the system reached the base pressure, the Ar/H
2
(80/40 sccm) carrier gas was introduced until
atmospheric pressure was achieved. The furnace was then heated up with a ramp rate of 35
C/min
to the growth temperatures (750 to 850
C).
The S component melted at 150
C was sent into the
furnace at the growth temperature to grow WS
2
. The sample growth procedure proceeded for 10
minutes, after which the furnace was directly opened to room temperature to stop the reaction
immediately.
Transfer of mono
-
and multi
-
layer WS
2
: Polystyrene (PS) was used as the supplementary film to peel
off the WS
2
crystals from sapphire substrates. PS (M.W. 192000)
dispersed in toluene solution (20
mg/mL) was spin
-
coated on
top of the WS
2
sample at a speed
of 3000 rpm. The edge of the PS film
was scribed with a blade, and then the sample of PS/WS
2
/Si
-
substrate was inserted into water
slowly. The PS/WS
2
film became peeled off naturally in water, and was subsequently rinsed with
deionized
-
water thrice before i
t was picked up and placed onto
the target substrate. The PS coating
was removed with toluene
after baking the sample at 80
°C for 60 minutes.
PL
and TRPL
Characterizations
: The
PL
spectra were taken with a Renishaw InVia
Raman
spectrometer system using a 514.3 nm laser (2.41 eV) as the excitation source. A 50
objective lens
with a numerical aperture of 0.75 and a 2400 lines/mm and 1800 lines/mm grating were chosen
during the measurement to achieve better signal
-
to
-
noise
ratio.
The time
-
resolved PL measurements