Supporting information for:
Giant Enhancement of Photoluminescence Emission in WS
2
-
2D Perovskite Heterostructures
Arky Yang
1
*, Jean-Christophe Blancon
2
*, Wei Jiang
3
, Hao Zhang
2
, Joeson Wong
1
, Ellen Yan
1
,
Yi-Rung Lin
1
, Jared Crochet
2
, Mercouri G. Kanatzidis
4
, Deep Jariwala
1
**, Tony Low
3
, Aditya
D. Mohite
2
*** and Harry A. Atwater
1
***
1
California Institute of Technology, Pasadena, CA, 91125, USA.
2
Department of Chemical and Biomolecular
Engineering, Rice University, Houston, TX 77005, USA.
3
Department of Electrical and Computer Engineering,
University of Minnesota, Minneapolis, MN, 55455, USA.
4
Northwestern University, Evanston, IL, 60208, USA.*A.
Yang and J. Blancon contributed equally to this work. **Currently at Department of Electrical and Systems
Engineering, University of Pennsylvania, Philadelphia, PA 19104 USA.
***Corresponding authors: Aditya Mohite (
adm4@rice.edu), Harry Atwater (
haa@caltech.edu)
1. Spatial variation of PL emission
Photoluminescence at different regions of 2L WS
2
/ n=4 2DPVSK sample is shown in
Supplementary Fig. 2a, corresponding to spots labeled in Supplementary Fig. 2b. Bare 2DPVSK
shows varying emission intensity, likely due to differences in thickness and degree of degradation
caused by the transfer process. Bare WS
2
also shows spatial variation in PL intensity despite
Raman spectra that show the same number of layers. This might be attributed to variation in defect
types/ densities in different regions of the material, as well as inhomogeneous interaction with the
underlying substrate. Het-1 shows the highest emission among all heterostructure regions, in
agreement with the PL mapping shown in MS Fig. 3b. This can be caused by a combination of
variation on 2DPVSK quality, WS
2
defects, and interlayer interactions. The spatial inhomogeneity
indicates that interface between TMDC and 2DPVSK layer plays an important role in the PL
enhancement. More careful investigation and modeling of the interface is required to fully
understand the origin of the PL enhancement.
2. Effect of spacer layer on 2L WS
2
PL emission
To study the potential dielectric screening effect that 2DPVSK may impose on 2L WS
2
, control
samples 2L WS
2
/hBN/SiO
2
/Si and 2L WS
2
/butylammonium iodide(BAI)/SiO
2
/Si were fabricated.
WS
2
/hBN heterostructure was obtained by dry exfoliation and transfer, while WS
2
/BAI
heterostructure was obtained by first spin-coating BAI solution (1mg/ml in DMF) onto SiO
2
/Si
substrate, followed by dry transfer of WS
2
. Optical images of resulting heterostructures are shown
in Supplementary Fig 7 a and b. Photoluminescence spectra of corresponding region are shown in
Supplementary Fig. 7c. Comparing to bare 2L WS
2
, 2L WS
2
/hBN heterostructure does not show
enhanced PL intensity, but increased ratio between peak intensities of indirect and direct transition.
This observation indicates that introducing similar spacer effect cannot explain the PL
enhancement in 2L WS
2
/2DPVSK heterostructure. 2L WS
2
/BAI shows decreased PL emission as
well as emergence of lower energy peak. This indicates that bare BA cation does not induce the
PL enhancement. Comparing to BA in 2DPVSK, spin-coated BA is randomly oriented, which may
enhance surface roughness and reduce WS
2
emission. Raman spectra (Supplementary Fig. 7d) of
corresponding region do not show significant peak shift or change in intensity ratio between A
1
g
and E
2
g
peaks. Overall, simply shielding the effect from SiO
2
substrate on 2L WS
2
does not
enhance PL of WS
2
.
3. Optical characterizations on 1L WS
2
/ n=3 2DPVSK heterostructure
An optical image of a 1L WS
2
/ n=3 2DPVSK heterostructure is shown in Supplementary Figure
3a. In the Raman spectrum (Supplementary Figure 3b), a separation of 66 cm
-1
between E
1
g
and
A
1
g
vibration mode and higher intensity of E
1
g
mode for 1L WS
2
/n=3 2DPVSK heterostructure
indicate a monolayer thickness of WS
2
. Room temperature PL of WS2 and heterostructure is
shown in Supplementary Figure 3c. The heterostructure experiences laser curing effect similar to
2L WS
2
/n=4 2DPVSK heterostructure. Before curing, heterostructure exhibits 27 folds of
enhancement of PL emission over WS
2
. After curing, a broad peak at 1.71eV is observed in the
heterostructure area, which shows 81 folds enhancement in overall integrated PL intensity. In
comparison to the three peaks identified for 2L WS
2
/n=4 2DPVSK sample, monolayer (1L) WS
2
shows only two emission features, one broad peak at 1.84eV and another narrow peak at 1.98eV
(Supplementary Figure 3d). An overall 134-fold enhancement at an excitation power of
10μW
was
observed under low temperature, which is lower than what was observed for WS
2
/n=4 2DPVSK
sample. Reflection spectra (Supplementary Figure 3e) shows the presence of A exciton transition
and B exciton of WS
2
in both WS
2
and heteorstructure region. Absorption at 1.98eV in all three
regions corresponds to sharp peak at PL spectra.
3
PLE spectra (Supplementary Figure 3f) shows
overall 250 times and 70 times enhancement of intensity on heterostructure comparing to bare n=3
2DPVSK and 1L WS
2
. Specifically, peaks denoted as H
6
and H
8
are significantly enhanced in
heterostructure, concommitent with the observed enhancement of H
6
and H
8
in 2L WS
2
/n=4
2DPVSK heterostructure. Power dependence of heterostructure is shown in Supplementary Fig. 5.
Main exciton peak at 2.00-2.01eV is denoted as H
1
and the shoulder at 1.99-2.00eV as H
0
. Both
of the peaks show near-linear dependence on excitation power density (Supplementary Fig. 4b),
which confirms their excitonic nature. H
0
shoulder is likely defect-bound exciton confirmed with
disappearance at room temperature due to thermal excitation. In contrast to H
0
in 2L WS
2
/ n=4
2DPVSK heterostructure, neither of the peak and shoulder experiences observable peak shift with
increasing power. Power dependence of FWHM of these two peaks show slight decrease with
increasing power density, likely due to lower signal to noise ratio at lower excitation power.
Valence band position of n=3 2DPVSK is obtained from UPS measurements (Supplementary
Figure 6a), and band alignment of n=3 2DPVSK and 1L WS
2
shows that 1L WS
2
and n=3
2DPVSK form a type II band alignment (Supplementary Figure 6b). Similar to 2L/n=4
heterostructure, the exciton 2s of n=3 2DPVSK is also in resonance with A exciton 1s of 1L WS
2
,
indicating possible channel for resonant energy transfer. 1L WS
2
/n=3 2DPVSK heterostructure
show less emission enhancement of WS2 on heterostructure comparing to bare WS2
(Supplementary Figure 3c). Mapping on n=3 2DPVSK exciton peak does not show the absence of
2DPVSK PL on the heterostructure, due to an overlap in emission wavelength of the WS
2
exciton
and n=3 2DPVSK exciton (Supplementary Figure 6c). In contrast to the shorter lifetime of n=4
2DPVSK in heterostructure, almost identical lifetimes are observed in the n=3 heterostructure
(Supplementary Figure 6d), and are close to the system resolution of our experimental setup.
Methods
Sample Preparation.
2D Perovskite bulk crystals with the composition (BA)
2
(MA)
5
Pb
4
I
13
were
synthesized as previously reported.
4
WS
2
bulk crystals were purchased from HQ Graphene. A
schematic of the dry transfer process to fabricate TMD-2DPVSK heterostructure is shown in
Supplementary Fig. 1. 285nm SiO
2
/Si substrate (University Wafer) was cleaned with acetone and
isopropanol using ultrasonication (15 min each), and subsequently subjected to oxygen plasma
cleaning (5min, 100W, 300mTorr under O
2
flow). The bottom layer of the heterostructure
(2DPVSK) was exfoliated directly onto the cleaned SiO
2
/Si substrate using dicing tape (blue low
tack roll, Semiconductor Equipment Corp). Top WS
2
layers were transferred using a viscoelastic
dry transfer technique with a home-built setup in glovebox at room temperature without heating
steps.
5
WS
2
was exfoliated with Scotch tape onto PF-20-X4 Gel Film from Gel-Pak supported by
clear glass slides. Flakes are identified under optical microscope inside glovebox after each
exfoliation.
Optical Characterization.
Photoluminescence (PL), spatially-resolved PL, time-resolved
photoluminescence (TRPL), photoluminescence emission (PLE), and optical reflection were
performed with an in-lab-built confocal microscopy system.
3
The system focused a CW or 6-ps-
pulsed laser close to the diffraction limit, and detected output light with a CCD camera (EMCCD
1024B) integrated with a spectrograph (Spectra-Pro 2300i). PL spectra are obtained with laser
excitation at 550nm, with excitation intensity adjusted for different regions to obtain reasonable
signal-to-noise ratio while preventing degradation of 2DPVSK. Spatial resolution of low
temperature PL is achieved by rastering the laser beam onto sample surface with a fast steering
mirror, and wavelength selection was achieved by reducing slit width before detection beam
reaches Avalanche Photo-Diode (MPD-SPAD). TRPL was measured with a time correlated single
photon counter (PicoHarp 300) and MPD-SPAD. All measurements mentioned above were carried
out under vacuum (10
-5
-10
-6
torr) at liquid helium temperature (~7K). Raman and room
temperature PL are measured in ambient condition using Renishaw M1000 Micro Raman
Spectrometer System with 514nm laser excitation with 100x objective (spot size
~1μm).
Room
temperature PL map was extracted from hyperspectral map obtained by scanning over the sample
using mechanical motor stage integrated into Renishaw system. All PL peak positions and FWHM
were obtained from fitting of individual peaks using Gaussian function, and integration of the
peaks gives corresponding PL intensity. Due to low intensity of H
2
peaks comparing to H
0
and H
1
,
intensity of H
2
peaks were extracted from Gaussian peak fitting of spectra subtracting a
background of linear sections defined by valleys (Supplementary Fig. 8). Power dependent PL
emission intensities are normalized by dividing the original intensity by integration time for
comparison of intensity between heterostructure and WS
2
(Figure 3e). Raman peak positions are
obtained from fitting with Voigt functions, calibrated with Si peak at 520cm
-1
.
Ultraviolet photoelectron spectroscopy (UPS).
Thick flakes were exfoliated onto gold substrate to
expose clean surface right before UPS measurements. UPS measurements were performed using
a Kratos AXIS Ultra spectrometer with a He I ultraviolet source (21.2 eV) at a pressure of 1 × 10
-8
Torr in the analysis chamber. The photoelectron ejection vector was 90° with respect to the sample
surface plane, the electron-collection lens aperture was set to a 55
μm
spot size, and the analyzer
pass energy was 5 eV. The instrument energy scale and work function were calibrated using clean
Au, Ag, and Cu standards. Interceptions of linear fitting on high kinetic energy (KE) cutoff (Fermi
edge) and low KE cutoff of the spectra gives the width of the binding energy
(ΔE).
The difference
between
ΔE
and the He I source energy (21.22 eV) gives the binding energy of valence band
maximum electrons referenced to the vacuum level.
6
Transfer Matrix Calculation
. Transfer matrix calculations were carried out in MATLAB with code
designed according to methods described by Wong et al.
7,8
Refractive index(n) and absorbance(k)
of 2D-PVSK and WS
2
are obtained from Guo et al.
9
and Li et al.
10
, respectively. The n and k values
of SiO
2
and Si are obtained from literature reports.
11,12
Infinite thickness was assumed as a realistic
approximation for the Si substrate. Air-dielectric interface is defined as z=0. The thickness of the
2L WS
2
is determined to be 1.7nm from literature for typical bilayer WS
2
.
13
The thickness of the
n=4 2DPVSK is estimated to be 10nm from optical contrast.
Calculation method.
The structural and electronic properties of the 2D perovskites (2DPVSK),
transition metal dichalcogenides (TMD, WS
2
), and their heterostructures were studied using first-
principles methods based on density functional theory (DFT). We applied the generalized gradient
approximation exchange-correlation potentials plus the projector augmented wave method for the
electron-ion interaction for the calculations
14
, as implemented in Vienna
ab initio
simulation
package (VASP) code
15
. The heterostructures were built with
supercell of WS
2
on
5
×
3
2
×
2
supercell of 2DPVSK, after balancing the lattice mismatch and the size of the system given the
limitation of the DFT calculation. A vacuum region of more than 15 was introduced between
Å
slabs to avoid the interaction between slabs. DFT-D3 method was applied to properly treat the van
der Waals interaction between layers
16
. The dipole correctionswas applied to properly treat the
dipole moment due to the formation of heterostructure
17
. An electric field is applied along the
z
direction of the multilayer WS
2
thin films with the dipole corrections applied to study the E-field
effect. All self-consistent calculations were performed with a plane-wave cutoff of 400 eV. The
geometric optimizations were carried out without any constraint until the force on each atom is
less than 0.01 eV/ and the change of total energy per cell is smaller than 10
-4
eV. The Brillouin
Å
zone k-point sampling was set with a 3
3
1 -centered Monkhorst-Pack grids. The charge
×
×
Γ
transfer distribution was calculated by subtracting the charge of isolated WS
2
and 2DPVSK from
that of their heterostructure. To get the change of the plane-averaged potential (PAP) profile due
to the formation of the heterostructure, the PAP of isolated WS
2
and 2DPVSK were first aligned
with and then subtracted from that of their heterostructure.
Supplementary Figure 1.
Schematics of exfoliation and dry transfer process for heterostructure
fabrication. The whole process is carried out inside a glovebox under microscope objective with
remote-controlled mechanical manipulator.
Supplementary Figure 2.
(a) PL emission spectra at different regions of sample taken at 900
W/cm
2
(WS
2
) and 5W/cm
2
(heterostructure and 2DPVSK) at 550nm laser excitation. Vertical
scales of the three subplots are arbitrarily chosen and not the same. (b) Optical image of the sample
with spots labeled correspondingly. (c) Integrated PL intensity map over 1.6-1.7 eV, measured at
514nm 3kW/cm
2
laser excitation and integration time of 3s per point. (d) PL spectra extracted
from the map at heterostructure where 2DPVSK has degraded (point a) and not degraded (point
b).
Supplementary Figure 3.
Optical measurements on 2L WS
2
/n=3 2DPVSK heterostructure. (a)
Optical image of 2L WS
2
/n=3 2DPVSK heterostructure on SiO
2
/Si substrate. (b) Raman spectra
of the bare WS
2
and the heterostructure. (c) Room temperature photoluminescence spectra taken
with 514 nm laser excitation at 3kW/cm
2
. (d) PL emission of 1L WS
2
/n=3 2DPVSK
heterostructure at 7K with 561nm laser excitation at 1kW/cm
2
before interface curing.
Corresponding reflection (e) and photoluminescence excitation PLE (f).
Supplementary Figure 4.
Optical constant fitting on differential reflection spectra of 2L WS
2
/
n=4 2DPVSK sample. (a) Experimental differential reflection spectra and fitting line. (b)
Extinction coefficient and (c) refractive index obtained from fitting. (d) Experimental PLE spectra
measured at 600nW. The complex refractive index were obtained by Kramer’s-Kronig constrained
fitting of the differential reflection spectra. The spectral dependence of real part of refractive index
n, can be derived from imaginary part k using Kramer’s-Kronig relation, under the assumption that
the contribution beyond the spectral range can be represented by one additional constant. Using
this relation, we derived the best n and k couple fitting the experimental data.
Supplementary Figure 5.
Power dependence of PL emission of 1L WS
2
/n=3 2DPVSK at 7K. (a)
PL spectra of heterostructure taken at different incident laser power. Power dependence of (b)
integrated PL intensity and (c) peak position of the photoemission transitions of the heterostructure.
(d) Corresponding power dependent of the FWHM.
Supplementary Figure 6.
(a) UPS spectra for the 1L WS
2
/n=3 heterostructure at low KE region
with linear fitting at the edge (left) and high KE region with linear fitting at Fermi edge (right). (b)
(Left)Corresponding band positions of n=3 2DPVSK and 1L WS
2
, with grey boxes indicating
conduction and valance band. (Right) Energy diagram in the excitonic picture for the bare n=3
2DPVSK and the bare WS
2
taken from previous reports.
9,10
Black lines indicate the exciton
Rydberg series. (c) Map of photoluminescence at low temperature corresponding to
Supplementary Fig. 3d. (d) Time resolved PL of n=3 PVSK exciton peak and heterostructure at
low temperature corresponding to Supplementary Fig. 3d. Both samples show PL decay on par or
faster with the resolution of our TRPL system.
Supplementary Figure 7.
(a) Optical image of 2L WS
2
on hBN heterostructure. b) Optical image
of 2L WS
2
on spin-coated BAI heterostructure. (c) PL emission of three regions taken at 514nm
excitation under room temperature and ambient atmosphere. (d) Raman spectra of corresponding
regions under the same condition.
Supplementary Figure 8.
Example plot of Gaussian peak fitting for power dependence analysis.
The spectra were taken on heterostructure at excitation power density of 65W/cm
2
. (a) H
0
and H
1
are fitted from the raw data using two Gaussian peaks. (b) H
2
is fitted from raw data subtracted
with sections of linear background defined by valleys between peaks.
Supplementary Figure 9
. Heterostructures of WS
2
and 2DPVSK with different thickness. Side view of
heterostructure of (a) 2L WS
2
and n=4 2DPVSK, (b) 2L WS2 and n=2 2DPVSK, and (c) 1L WS2 and n=2
2DPVSK. (d) Top view of the heterostructure.
Supplementary Figure 10
. Plane-averaged potential (PAP) profiles of heterostrutures. PAP profiles for
(a) 1L WS
2
/n=2 2DPVSK, (b) 2L WS
2
/n=2 2DPVSK, and (c) 2L WS
2
/n=4 2DPVSK all show the formation of
dipole moment.
Supplementary Figure 11
. Charge transfer and dipole moment of the 2L WS
2
/n=2 2DPVSK. (a) and (b)
The structure and the charge transfer of the system due to the formation of heterostructure. Yellow and
blue color denote the electron and hole accumulation, indicating the charge transfer from 2DPVSK to 2L
WS
2
. (c) The plane average potential profile of heterostructure (Total, black), 2L WS2 (2T_shift, red), n=2
2DPVSK (2P_shift, blue), and the difference due to the formation of the heterostructure (Difference,
Cyan).
References
(1)
Fang, H. H.; Yang, J.; Tao, S.; Adjokatse, S.; Kamminga, M. E.; Ye, J.; Blake, G. R.; Even, J.;
Loi, M. A. Unravelling Light
‐
Induced Degradation of Layered Perovskite Crystals and Design of
Efficient Encapsulation for Improved Photostability.
Adv. Funct. Mater.
2018
,
28
(21).
(2)
Zhao, W.; Ghorannevis, Z.; Chu, L.; Toh, M.; Kloc, C.; Tan, P.-H.; Eda, G. Evolution of
Electronic Structure in Atomically Thin Sheets of WS2 AndWSe2.
ACS Nano
2013
,
7
(1), 791–
797.
(3)
Blancon, J.-C.; Tsai, H.; Nie, W.; Stoumpos, C. C.; Pedesseau, L.; Katan, C.; Kepenekian, M.;
Soe, C. M. M.; Appavoo, K.; Sfeir, M. Y.; et al. Extremely Efficient Internal Exciton Dissociation
through Edge States in Layered 2D Perovskites.
Science.
2017
, 355(6331), 1288-1292.
(4)
Stoumpos, C. C.; Cao, D. H.; Clark, D. J.; Young, J.; Rondinelli, J. M.; Jang, J. I.; Hupp, J. T.;
Kanatzidis, M. G. Ruddlesden-Popper Hybrid Lead Iodide Perovskite 2D Homologous
Semiconductors.
Chem. Mater.
2016
,
28
(8), 2852–2867.
(5)
Castellanos-Gomez, A.; Buscema, M.; Molenaar, R.; Singh, V.; Janssen, L.; van der Zant, H. S. J.;
Steele, G. A. Deterministic Transfer of Two-Dimensional Materials by All-Dry Viscoelastic
Stamping.
2D Mater.
2014
,
1
(1), 11002.
(6)
Chun, W. J.; Ishikawa, A.; Fujisawa, H.; Takata, T.; Kondo, J. N.; Hara, M.; Kawai, M.;
Matsumoto, Y.; Domen, K. Conduction and Valence Band Positions of Ta2O5, TaOn, and Ta3N5
by UPS and Electrochemical Methods.
J. Phys. Chem. B
2003
,
107
(8), 1798–1803.
(7)
Pettersson, L. A. A.; Roman, L. S.; Inganäs, O. Modeling Photocurrent Action Spectra of
Photovoltaic Devices Based on Organic Thin Films.
J. Appl. Phys.
1999
,
86
(1), 487–496.
(8)
Wong, J.; Jariwala, D.; Tagliabue, G.; Tat, K.; Davoyan, A. R.; Sherrott, M. C.; Atwater, H. A.
High Photovoltaic Quantum Efficiency in Ultrathin van Der Waal s Heterostructures.
ACS Nano
2017
,
11
(7), 7230–7240.
(9)
Guo, P.; Huang, W.; Stoumpos, C. C.; Mao, L.; Gong, J.; Zeng, L.; Diroll, B. T.; Xia, Y.; Ma, X.;
Gosztola, D. J.; et al. Hyperbolic Dispersion Arising from Anisotropic Excitons in Two-
Dimensional Perovskites.
Phys. Rev. Lett.
2018
,
121
(12), 127401.
(10) Li, Y.; Chernikov, A.; Zhang, X.; Rigosi, A.; Hill, H. M.; Van Der Zande, A. M.; Chenet, D. A.;
Shih, E. M.; Hone, J.; Heinz, T. F. Measurement of the Optical Dielectric Function of Monolayer
Transition-Metal Dichalcogenides: MoS
2
, MoSe
2
, WS
2
, and WSe
2
.
Phys. Rev. B - Condens. Matter
Mater. Phys.
2014
,
90
(20), 1–6.
(11) Marcos, L. V. R.; Larruquert, J. I.; Méndez, J. A.; Aznárez, J. A. Self-Consistent Optical
Constants of SiO2 and Ta2O5 Films.
Opt. Mater. Express
2016
,
6
(11), 3622–3637.
(12) Schinke, C.; Christian Peest, P.; Schmidt, J.; Brendel, R.; Bothe, K.; Vogt, M. R.; Kröger, I.;
Winter, S.; Schirmacher, A.; Lim, S.; et al. Uncertainty Analysis for the Coefficient of Band-to-
Band Absorption of Crystalline Silicon.
AIP Adv.
2015
,
5
(6), 67168.
(13) Chen, K.; Wan, X.; Wen, J.; Xie, W.; Kang, Z.; Zeng, X.; Chen, H.; Xu, J.-B. Electronic
Properties of MoS
2
–WS
2
Heterostructures Synthesized with Two-Step Lateral Epitaxial Strategy.
ACS Nano
2015
,
9
(10), 9868–9876.
(14)
G. Kresse and D. Joubert,
Phys. Rev. B
59
, 1758 (1999).
(15)
G. Kresse and J. Hafner,
Phys. Rev. B
47
, 558 (1993).
(16)
S. Grimme, J. Antony, S. Ehrlich, and S. Krieg,
J. Chem. Phys.
132
, 154104 (2010).
(17)
G. Makov and M.C.Payne,
Phys. Rev. B
51
, 4014 (1995).