of 20
Supp
lementary
Information
All
-
optical control of high
-
purity trions in nanoscale waveguide
Hyeongwoo Lee,
1
Yeonjeong Koo,
1
Shailabh Kumar,
2
,3
Yunjo Jeong,
4
Dong Gwon Heo,
5
Soo Ho
Choi,
6
Huitae Joo,
1
Mingu Kang,
1
Radwanul Hasan Siddique,
2
,3
Ki Kang Kim,
6,7
Hong Seok Lee,
5
Sangmin An,
5
Hyuck Choo
,
2,
8
and Kyoung
-
Duck Park
1
1
Department of Physics, Pohang University of Science and
Technology (POSTECH), Pohang 37673, Republic of Korea
2
Department of Medical Engineering,
California Institute of Technology (Caltech), CA 91125, USA
3
Meta Vision Lab, Samsung Advanced Institute of Technology (SAIT), CA 91101, USA
4
Institute of Advanced
Composite Materials, Korea Institute of Science and Technology,
Jeonbuk 55324, Republic of Korea
5
Department of Physics,
Research Institute of Physics and Chemistry
,
Jeonbuk National University, Jeonju 54896, Republic of Korea
6
Center for Integrated Nan
ostructure Physics,
Institute for Basic Science (IBS), Suwon 16419, Republic of Korea
7
Department of Energy Science, Sungkyunkwan
University (SKKU), Suwon 16419, Republic of Korea
8
Advanced Sensor
Lab, Device Research Center,
Samsung Advanced Institut
e of Technology (SAIT), Suwon 16678, Republic of Korea
Supplementary Fig.
1
.
Raman scattering characteristics of the suspended MoS
2
ML.
(a) R
aman
spectra
of MoS
2
ML transferred on Au substrate (black) and suspended on
trionic
waveguide (red). (b)
Lorentz fitted Raman spectra of the MoS
2
ML transferred on Au substrate (top) and suspended on
trionic
waveguide (bottom).
R
ed, blue, and green filled peak
s
indicate
E
1
2g
,
A
1g
, and
2
LA
(
M
)
Raman mode
s
.
P
eak position image (c) and linewidth image (d) for
E
1
2g
mode
Raman spectra
.
S
cale bar is 400 nm.
In order to
induce
the
in
homogeneous strain on the
ultrathin
trionic
waveguide device, the
formation of the well
-
suspended MoS
2
ML on the
device is important.
We confirm the well
-
suspended MoS
2
ML by comparing the Raman spectra measured at the Au substrate and the
trionic
waveguide, as shown in
Supplementary Fig.
1a
-
b [1]. When the MoS
2
ML
is suspended,
both
E
1
2g
and
A
1g
mode are redshifted with noticeable spectral broadening.
Supplementary
Fig.
1c
-
d show the redshifted peak position with the spectral broadening at the nanogap region,
which are in good agreement with the previous studies [2, 3].
Supplementary Fig.
2.
PL spectra of MoS
2
ML under ambient condition without (black) and with H
2
O
molecules (blue) on crystal surface. Zoomed in plot indicate Raman spectra of H
2
O molecule.
Supplementary Fig.
3. Fabrication process of
lateral MIM waveguide.
The coupling of light and SPP propagation rely on a lateral MIM geometry where the top and
sidewalls of the nanogap are gold
-
coated, whereas bottom of the channel is SiO
2
. For
fabrication of the channel, we start with a SOI wafer, a
nd deposit 150 nm of Au on it. Then we
perform a FIB milling step to etch into the SiO
2
. As seen in the
Supplementary Fig.
3
, after
milling the resultant nanogaps have a thin layer of gold on top, but much of the sidewall is
SiO
2
. This milling process
results in a slight taper to the sidewalls, which is crucial for
subsequent gold deposition step. The top layer of gold is then removed using a gold etchant.
A fresh layer of Au (50 nm) is deposited using e
-
beam, and coats the top, sidewalls, and
bottom of
the nanogap. A second round of milling is then performed to remove the gold from
bottom of the nanogap channel. As the figure illustrates, the first milling process is an oxide
etch step, whereas the second milling is an Au etch step to remove gold from b
ottom of the
channels.
The parameters for fabricating lateral MIM waveguide, such as gap size and height, are
optimized in the way of maximizing the efficiency of SPP coupling to the nanogap while also
minimizing losses during SPP propagation in the nanoga
p. Specifically, the nanogap lateral
MIM geometry (Au
-
SiO
2
-
Au) has been designed to provide a high contrast between the
effective refractive index inside the channel and the refractive index of the substrate (nSiO
2
=
1.46) providing a higher coupling effic
iency. Furthermore, coupling of incoming laser light with
lateral MIM nanogap to activate SPP mode proceeds through scattering from the edges and
sidewalls of the gap. This method where light is coupled to Au
-
SiO
2
-
Au MIM nanogap through
edge/tail
-
end illum
ination has been discussed in detail in previous works [
4, 5, 6
].
Supplementary Fig.
4. 2D image of Fig. 2 in the main text.
SPP images with excitation polarization
across (a) and along (b) waveguide. (c
-
e) X
0
PL images of with different excitation polarizations. (f
-
h)
PL images of X
-
/X
0
ratio with different excitation polarizations. White dashed line is guideline for
waveguide structure based on
Supplementary Fig.
5.
Supplementary Fig.
5. Rayleigh scattering image of the trionic waveguide.
Guideline for scan area
in Fig. 2 is based on this Rayleigh scattering image.
Supplementary Fig.
6.
XPS spectra of as
-
grown (a) and transferred (b) MoS
2
monolayer.
To exclude
the defect
-
related electron generation, we perform X
-
ray photoelectron
spectroscopy (XPS) before and after transferring MoS
2
monolayer onto the target substrate.
As a result, we obtain Mo:S ratio of 1:1.976 for as
-
grown MoS
2
monolayer (
Supplementary
Fig.
6a) while 1:1.998 for transferred MoS
2
monolayer (
Supplementary Fig.
6b). After the
transfer of MoS2 onto the substrate, the Mo:S ratio is highly close to 1:2, indicating negligible
generation of defects during the transfer. Note tha
t slightly dominant detection of Mo can be
attributed to the Molybdenum precursor.
Supplementary Fig.
7. Charge density distribution in MoS
2
ML transferred onto Au substrate.
(a)
A
1g
center versus
E
2g
center measured at various spots. Probability histogram of
E
2g
center (b) and
A
1g
center (c).
To investigate the charge distribution of MoS
2
ML, we obtain Raman spectra at various spots
and extract the peak position of
E
2g
and
A
1g
peak, as shown in
Supplementary Fig.
7
a.
Supplementary Fig.
7
b
-
c shows the probability distribution of
E
2g
and
A
1g
peak, respectively.
The linewidth of the fitted probability distribution exhibits ~0.2 cm
-
1
, demonstrating the
homogeneity of the charge distribution [
7, 8
].
Supplementary Fig.
8. Power dependence of exciton
-
to
-
trion conversion.
(a) Excitation
-
power
-
dependent PL spectra at Au substrate. (b) Extracted X
0
intensity (blue) and X
-
intensity (red). (c)
Extracted X
-
intensity at low excitation powers. (d)
Excitation
-
power
-
dependent PL spectra at SPP mode.
(e) Extracted X
0
intensity (blue) and X
-
intensity (red). (f) Extracted X
-
intensity at low excitation powers.
To confirm the role of SPP mode on the exciton
-
to
-
trion
conversion, while excluding the effect
of defect
-
related provision of electrons, we measure excitation
-
power
-
dependent PL spectra
at Au substrate and SPP mode. At Au substrate, the linewidth of the PL spectra increases as
increasing the excitation power,
attributed to the emerging trion peak, as shown in
Supplementary Fig.
8a. In contrast to gradually increased X
0
intensity as a function of
excitation power, the X
-
intensity shows negligible changes at the low excitation power (<10
μW), which is possibly
due to the inactivation of defect
-
induced electrons at low excitation
power (
Supplementary Fig.
8b
-
c) [
9
]. On the other hand, the dominant X
-
peak are
continuously observed with increasing excitation power at the SPP mode, as shown in
Supplementary Fig.
8d. Correspondingly, the X
-
intensity linearly increases as increasing
excitation power even at the low excitation power, as shown in
Supplementary Fig.
8e
-
f. This
behavior well indicates the plasmon
-
induced hot electron injection and consequently
inc
reased X
-
intensity with high exciton
-
to
-
trion conversion efficiency [
10
].
Supplementary Fig.
9.
Second derivative curve (top) obtained from PL spectrum in Fig. 3d, with
excitation polarization of 45º (bottom). Two minima in second derivative are
assigned to neutral exciton
peak and trion peak.
Supplementary Fig.
1
0
. Polarization degree of excitons and trions.
Polar plot for intensity of X
0
(a)
and X
-
(b) as function of detection angle at Au substrate. Polar plot for intensity of X
0
(c) and X
-
(d) as
function of detection angle at SPP mode.
To investigate the strain
-
induced optical characteristics of X
0
and X
-
emission, we obtain PL
intens
ities of X
0
and X
-
with changing detection angle. At Au substrate, both X
0
and X
-
emissions exhibits negligible polarization degree (
Supp
lementary Fig.
1
0
a
-
b), in good
agreement with previous study [
11
]. By contrast, X
0
emission shows the noticeable polarization
degree of ~0.52 (
Supplementary Fig.
1
0
c) in contrast to still negligible polarization degree of
X
-
emission (
Supplementary Fig.
1
0
d), attributed to the SPP
-
induced excitation of additional
X
0
[
12, 13, 14
]. No
te that SPP mode is excited with the vertical excitation polarization, which
corresponds to waveguide axis.
Supplementary Fig.
1
1
.
Change in trion
density
(∆
X
-
) as function of change in neutral exciton density
(∆
X
-
), with activation of SPP mode.
To investigate the possible energy transfer between the excitons and the SPP, we obtain the
PL spectra with and without the SPP mode. We then extract X
0
a
nd X
-
intensities by fitting PL
spectra with Lorentz function.
Supplementary Fig.
1
1
shows the change of trion density
(∆
X
-
)
as a function of the change of neutral exciton density
(∆
X
0
). Without the influence of the SPP,
decrease in X
0
density should cor
respond to increase in X
-
density with dominating highly
efficient exciton
-
to
-
trion conversion process at strain gradient geometry. However,
Supplementary Fig.
11
exhibits the extra increase in the trion
density. This is probably due to
the SPP
-
induced excitation of excitons [
1
3
, 1
4
] and these additionally generated excitons are
converted to trions as enough number of electrons are confined at the center of nanogap.
Supplementary Fig.
12. Optical microscope image of lateral MIM waveguide device.
MoS
2
ML is
transferred on the waveguides #2
-
4 excepting the waveguide #1.
We fabricate four identical lateral MIM waveguide structure (sample #1
-
4). The MoS
2
ML is
transferred on the three lateral MIM waveguides (sample #2
-
4) while sample #1 remains bare.
As shown in Fig. 4c
-
d, SPP signal is spectrally overlapped with MoS2 PL. Therefore, to find
the optical phase mask with the SPP signal without the influence of MoS
2
PL, we used the
waveguide #1. Because all lateral MIM waveguides are identical, they have identical SPP
mode and share an optimal phase mask. Therefore, we optimize the phase mask at the device
#1 and then move onto the device #2 to perform the main exper
iments.
Supplementary Fig.
13
.
Optimal phase mask
of
Fig.
4 in main text.
Supplementary Fig.
1
4
.
On/off switching of exciton
-
to
-
trion conversion.
T
ime
-
series PL
spectra
during on/off switching of exciton
-
to
-
trion
conversion, before
normalization of
Fig.
4e
.
Supplementary Fig.
1
5
.
Reproducibility of modulating exciton
-
to
-
trion conversion location.
(a)
T
ime
-
series PL response during on/off switching of exciton
-
to
-
trion
conversion. (b) Normalization of (a).
(c) Representative PL spectra with
(red)
and without
(black)
optimal phase mask.
The controllability and reproducibility
are
confirmed by conducting a wavefront shaping at the
different
weak
SPP
region
.
Supplementary Fig.
15
a shows the
dynamic switching between
X
0
dominant emission and
X
-
dominant emission with the phase mask optimized for the current
location. The normalized
spectral
image in
Supplementary Fig.
15
b and the representative
PL
spectra in
Supplementary Fig.
15
c show the distinct transition between
X
0
domi
n
ant and
X
-
dominant emission
s
.
Supplementary Fig.
1
6
.
W
ork function of MoS
2
ML on nanogap of
trionic
waveguide.
T
opography
image (a) and corresponding work function
image (b)
of
nanogap
without MoS
2
ML. T
opography image
(c) and
corresponding work function image (d)
of suspended MoS
2
ML on nanogap.
Suppleme
ntary Fig.
1
7
.
Estimation of the strain profile.
(a)
I
mage of fitted line shape function
based
on experimentally measured topography profile of MoS
2
ML on nanogap of
ultrathin
trionic
waveguide.
(b) Corresponding strain image estimated by continuum theory for a thin and elastic plate
[
1
5
].
Supplementary Fig.
18.
Estimation of spatial distribution of X
0
and X
-
.
(a) Spatial distribution of the
photoexcited excitons before considering mass action model with (red) and without (black) strain
gradient. Contribution of X
0
(green) and X
-
(red) in total photoexcited excito
n density (black) for
α = 0.04
(b) and
α = 1.2
(c).
As mentioned in the main text, we subtract the photoexcited exciton density obtained without
the strain gradient (black, in
Supplementary Fig.
18a) from the exciton density with the strain
gradient (red, in
Supplementary Fig.
18a) to exclude the effect of the optical excitation. At this
stage, we exclude the contribution of X
-
to clearly investigate the role of strain gradient by the
nanogap. T
hen, we adopt the mass action model to estimate the density ratio of X
0
and X
-
depending on the background electron density
α,
as shown in
Supplementary Fig.
18b and
S18c.