Atomically thin layered tungsten diselenide (WSe2) has attracted tremendous research attention for its potential applications in next-generation electronics. This article reports the synthesis method of high-quality monolayer to trilayer WSe2 by molten-salt-assisted chemical vapor deposition. With the optimization of different types of molten salts and depths of corundum boat, large trilayer WSe2 films can be grown with domain size up to 80 µm for the first time. A systematic study of the electrical properties of the n-type field-effect transistor has been carried out based on WSe2 with the above three different layer thicknesses. The trilayer WSe2 devices exhibit higher drive current, mobility, on/off ratio, and lower contact resistance than both bilayer and monolayer counterparts. Moreover, short channel transistors using the trilayer WSe2 with a channel length of 230 nm have been fabricated, exhibiting an excellent on/off ratio up to 108 and a high current density of 187 µA/μm. This facile synthesis of high-quality large-area multilayer WSe2 provides a pathway for future high-performance two-dimensional electronic devices.

Transition metal dichalcogenides (TMDCs) have received increasing attention for electronic and optoelectronic applications due to their excellent transport properties and ultrathin body nature.1–4 WSe2, one of the TMDC materials, exhibits ambipolar transport behavior and is capable of operating in both n-type and p-type transistor modes with different growth conditions, as well as interface and contact engineering.5–8 However, the field-effect transistors made from WSe2 so far are largely based on mechanical exfoliation, especially the n-type ones.9–12 Large-area synthesis using chemical vapor deposition (CVD) has been proved to be a successful method in previous work for graphene and MoS2 and is considered to be one of the most facile approaches for low-cost, high-quality material growth.13,14 Furthermore, multilayer WSe2 beyond monolayer can provide better electrical contact with metal and less interface scattering while maintaining the ultrathin body nature, which has yet to be explored, especially on the electrical performance.15–18 Previous work showed bilayer CVD WSe2 growth with p-type behavior with a single-crystal edge length less than 40 μm.20–22 Recently, molten salts have received increasing attention for their role in the CVD growth because they can increase mass flux by reducing the melting points of the metal precursors, thus increasing the reaction rate, which can yield a much larger single crystal size of the bilayer of about 120 μm.19 As for the electrical performance, the previous best n-type WSe2 transistors are built on mechanical exfoliation flakes with a carrier mobility of 44 cm2/V s, an on/off ratio of 106, and a current density of 210 µA/μm.9 The current density of WSe2 transistors based on CVD growth is only one-tenth of that of the best exfoliated ones.16–18,27

In this article, we report systematic material synthesis for growing large-area monolayer to trilayer WSe2 using molten salt-assisted low-pressure chemical vapor deposition (LPCVD). With the optimized salt type and adjusting metal precursor boat depth, the large size nuclei of the metal precursor lead to high-quality multilayer growth of single-crystalline WSe2. The resulting WSe2 grown on sapphire was transferred onto a silicon substrate with a 20 nm high-κ HfO2 dielectric on top for the back-gate transistor fabrication. The n-type field-effect transistors built on the trilayer channel show much higher performance compared with the monolayer and bilayer counterparts, including contact resistance, mobility, and output current. The trilayer short channel (230 nm) device exhibits a record high on/off ratio of 108 and an output current of 187 µA/μm among all CVD grown results. This low-cost, high-quality growth process shows the potential for larger domain crystal growth for future high-performance 2D electronic devices.

WSe2 domains from monolayer to trilayer were grown on c-plane sapphire substrates by the molten-salt-assisted low-pressure chemical vapor deposition (LPCVD) process. The growth process for WSe2 by the LPCVD method was carried out in a 1-in. diameter quartz tube furnace as depicted in Fig. 1(a). First, c-plane (0001) sapphire substrates were cleaned with acetone, isopropanol alcohol, and deionized water. Then, 60 mg WO2 powder and 3 mg KCl were added in a clean corundum boat with a depth of 6 mm. A piece of sapphire substrate was placed above the WO2 solid source. 300 mg Se powder was added in a ceramic boat, which was able to provide a relatively low mass flux of the Se precursor compared with the corundum boat, which is beneficial to form single crystalline films rather than polycrystalline films.19 Se powder and WO2 powder were located in a quartz tube, keeping a distance of around 19 cm from each other. Before heating the furnace, the base pressure was pumped down below 4 Pa to remove excess oxygen gas and moisture.23 Then, a mixture of gas with 8 SCCM H2 and 105 SCCM Ar was introduced into the furnace while maintaining pressure at around 1700 Pa. As depicted in Fig. 1(b), the two heating zones were heated to 300 and 890 °C, respectively, with a ramping rate of 30 °C/min. After reaching the target temperature, it was maintained at the growth temperature for 15 min. Finally, the furnace was cooled down naturally to room temperature.

FIG. 1.

(a) LPCVD setup for the synthesis of multilayer WSe2 on sapphire. (b) The temperature profile for precursors Se and WO2. (c) Optical microscopic image of WSe2 from monolayer to trilayer. The scale bar is 10 µm.

FIG. 1.

(a) LPCVD setup for the synthesis of multilayer WSe2 on sapphire. (b) The temperature profile for precursors Se and WO2. (c) Optical microscopic image of WSe2 from monolayer to trilayer. The scale bar is 10 µm.

Close modal

As depicted in Fig. 1(c), WSe2 single crystals from monolayer to trilayer were grown on c-plane sapphire substrates. The molten salt KCl can reduce the melting temperature ofWO2, thus yielding large size nuclei on sapphire which provide enough sources for the layer-to-layer growth ofWSe2. WO2 was used here for the metal precursor instead of WO323–28 because WO3 sublimates readily above 850 °C and can easily produce a high mass flux of the metal precursor leading to high-density nuclei, which typically result in polycrystalline films rather than high-quality multilayer single crystalline films.

It should be noted here that in order to grow multilayer WSe2, we adopted KCl as the salt assistance, in addition to a few other commonly used salts, such as sodium. In previous work, only metal and chalcogen precursors were used, which typically resulted in the growth of monolayer or bilayer of large domain size.16–18,31 Large-size multilayer WSe2 is difficult to obtain partly due to the high melting points of W and WO3. Recently, molten-salt-assisted methods were employed to facilitate the growth of monolayer and bilayer WSe2.17 In our growth process, we studied the growth effect of three kinds of molten salts, as shown in Figs. 2(a)2(d). First, clean and large monolayer triangle shape single crystals were grown without any salt assistance, with an average size of about 100 μm. Then, three different salts NaI, KI, and KCl were added to the WO2 precursor.27 The typical optical micrographs of CVD WSe2 of the NaI-assisted growth are shown in Fig. 2(b). Compared to Fig. 2(a), some white points can be observed on the surface of the substrate and on top of the monolayer WSe2 crystals, which can be identified as WOxSeyIz. Mixing WO2 and NaI will produce a molten solution, with a much higher vapor pressure than WO2, which promotes the reaction rate and results in the formation of large nuclei. The large nuclei are essential for layer-by-layer growth to form multilayers in the end. Despite the emergence of the nuclei, NaI is still insufficient for any sizable layer-by-layer growth. Next, KI was introduced to replace NaI, and irregular bilayer regions can be grown on the triangular monolayer WSe2 crystals, with a size typically below 20 μm as shown in Fig. 2(c). Finally, KCl was introduced and the growth resulted in not only large size bilayer crystals but also trilayer WSe2 regions with an average size of about 50 μm.

FIG. 2.

(a) Optical micrographs of CVD grown WSe2 without molten salts. (b)–(d) Optical micrographs of CVD grown WSe2 with different molten salts; the corresponding molten salts are NaI, KI, and KCl, respectively. The scale bar is 50 µm.

FIG. 2.

(a) Optical micrographs of CVD grown WSe2 without molten salts. (b)–(d) Optical micrographs of CVD grown WSe2 with different molten salts; the corresponding molten salts are NaI, KI, and KCl, respectively. The scale bar is 50 µm.

Close modal

In order to optimize the crystal size of the multilayer WSe2, we used the corundum boats with different depths to change the distances between the substrate and the WO2/KCl precursors, as shown in the schematic diagram in Fig. 3(a). When the furnace is heated, the WO2 precursor mixture with KCl leads to the formation of volatile compound WO2Cl2. Under H2/Ar atmosphere carrying the selenium precursor, WO2Cl2 forms WOxSeyClz, which forms the nucleation centers for WSe2 growth. The density and size of the nuclei will determine the geometries of the resulting crystal films. The density of nuclei is usually decided by the mass flow rate and will mainly affect the resulting domain size of single crystals. On the other hand, multilayer growth requires large size nuclei because each layer grows from the same nuclei, provided that multilayers are concentric. The layer-by-layer growth will be limited from source supply if nuclei are small.19 Here, the corundum boats used to carry WO2/KCl precursors inside also supports the sapphire substrate on top, and the depth of the boat is the distance between the substrate and the WO2/KCl precursors. In order to achieve the optimal depth for growing large-size films, we used corundum boats with six different depths. When the boat depth is around 5 mm, both the size and density of few-layer domains decrease, and when it is much deeper denser triangles can be observed as shown in Fig. 3(b).

FIG. 3.

(a) Schematic diagram of layer size and thickness optimization of WSe2 by controlling the nucleation conditions. (b) Optical micrographs of CVD grown WSe2 with the boat depth of 5 mm. The scale bar is 50 µm. (c) The correlation of the average size of the triangle crystal domains with various boat depths.

FIG. 3.

(a) Schematic diagram of layer size and thickness optimization of WSe2 by controlling the nucleation conditions. (b) Optical micrographs of CVD grown WSe2 with the boat depth of 5 mm. The scale bar is 50 µm. (c) The correlation of the average size of the triangle crystal domains with various boat depths.

Close modal

When the boat depth is above 8 mm, only monolayer WSe2 under 20 μm can be obtained. When the depth decreases from 8 to 6 mm, bilayer and trilayer regions begin to grow and the largest domain can be achieved at the depth of 6 mm as shown in Fig. 3(c). The nuclei growth and the film growth along the nuclei can be viewed as a competing process during the whole growth period. When the boat is deep enough, the nuclei size remains small as the in-plane growth dominates, which results in large area monolayer crystals. When the depth is reduced gradually, the nuclei size will grow larger, and out-of-plane growth, i.e., multilayer vertical growth, and in-plane lateral growth happen concurrently. Under this circumstance, a larger monolayer on the bottom will be grown, while out-of-plane growth will start from the nuclei followed by the in-plane growth of the second layer, which repeats similarly for the third layer. When the depth is reduced too much, the nuclei will become too thick and large, preventing the lateral growth of large-area few-layer WSe2.

As shown in Fig. 4(a), the morphology of monolayer, bilayer, and trilayer CVD WSe2 has been determined using an atomic force microscope (AFM) according to the optical image as shown in Fig. 4(b), and the step heights are found to be about 0.8 nm for each layer, consistent with previous reports.29,30 Raman spectroscopy characterization was used to distinguish WSe2 with different layers after transferring the film onto 100 nm SiO2 on a Si substrate. Raman spectra collected from the monolayer and bilayer WSe2 show peaks at 250 and 251 cm−1, respectively, as shown in Fig. 4(c), corresponding to the E12g mode, in agreement with previous reports.31,32 As shown in Fig. 4(d), the intensity of Raman mapping spectra at peaks is consistent with the spectrum position, and the color contrast distribution indicates a good uniformity of the film.

FIG. 4.

(a) AFM image (left) of the multilayer WSe2 transferred from sapphire to the 100 nm SiO2/Si substrate and the corresponding height profiles (right) of WSe2 from monolayer to trilayer. The scale bar is 1 µm. (b) Optical micrograph of trilayer WSe2. (c) Corresponding Raman spectra of WSe2 from monolayer to trilayer. (d) Corresponding Raman mapping spectra of trilayer WSe2 in (b). The scale bar is 10 µm.

FIG. 4.

(a) AFM image (left) of the multilayer WSe2 transferred from sapphire to the 100 nm SiO2/Si substrate and the corresponding height profiles (right) of WSe2 from monolayer to trilayer. The scale bar is 1 µm. (b) Optical micrograph of trilayer WSe2. (c) Corresponding Raman spectra of WSe2 from monolayer to trilayer. (d) Corresponding Raman mapping spectra of trilayer WSe2 in (b). The scale bar is 10 µm.

Close modal

In order to study the layer-dependent electrical properties of WSe2, back-gated transistors were fabricated with the device structure shown in Fig. 5(a). After growth on sapphire, WSe2 films were transferred onto 20 nm high-κ HfO2 dielectrics, which were deposited by atomic layer deposition (ALD) on a highly degenerated p-type Si substrate. As for the transfer process, a 400-nm thick PMMA layer was spin-coated on the as-grown WSe2 on sapphire, and then, a layer of heat release tapes (HRTs) was applied to the PMMA layer followed by soaking in deionized water for 2 h. Then, the HRT peeled off the PMMA layer and WSe2 from the sapphire substrate and was transferred onto 20 nm HfO2 on silicon. Finally, the HRT was removed by heating, and PMMA was dissolved with acetone. A post-annealing process step was performed for WSe2 to remove the PMMA residual before device fabrication. As for the device fabrication, the WSe2 channel was patterned by electron beam lithography (EBL). Then, S/D ohmic contacts were formed with 20 nm Ni/60 nm Au metal deposition by E-beam evaporation. The channel lengths range from 0.23 to 3 µm.

FIG. 5.

(a) Schematic view of the WSe2 transistors. (b) Layer-dependent transfer characteristics at 1 V with 1 µm channel length back-gated WSe2 transistors. (c) Layer-dependent contact resistance extracted using the transfer length method with different channel lengths. (d) On-current and electron mobility comparison of CVD WSe2 transistors with different layers.

FIG. 5.

(a) Schematic view of the WSe2 transistors. (b) Layer-dependent transfer characteristics at 1 V with 1 µm channel length back-gated WSe2 transistors. (c) Layer-dependent contact resistance extracted using the transfer length method with different channel lengths. (d) On-current and electron mobility comparison of CVD WSe2 transistors with different layers.

Close modal

Figure 5(b) shows the transfer characteristics of the layer-dependent WSe2 transistors at Vds = 1 V with Lch = 1 μm. The off-current is almost constant for the devices with different layers. The trilayer WSe2 transistor shows the highest performance among the three types of devices, exhibiting one order of magnitude higher on/off ratio (2 × 108), one order of magnitude higher on-current (Ion), and five times higher mobility than the monolayer device. As shown in Fig. 3(c), the contact resistance (Rc) can be extracted from the transfer characteristics for three different channel lengths, resulting in 19, 11, and 3.2 kΩ μm for the monolayer, bilayer, and trilayer devices, respectively. The contact resistance is the smallest for the trilayer devices due to the higher density of states of thicker semiconductors and smaller bandgap as well, which is beneficial for high-performance devices. As shown in Fig. 5(d), both the mobility and on/off ratio are the highest for the trilayer devices.

Figures 6(a) and 6(b) show the transfer and output characteristic curves of a short channel (230 nm) back-gated trilayer WSe2 transistor. It shows typical n-type behavior with a field-effect mobility of 32 cm2/V s, an on/off ratio of 2.6×108, and a maximum drain current of 187 µA/μm. Decent current saturation can also be observed from the output characteristics with clear Ohmic contact using Ni.35 As shown in Fig. 6(c), the on-current increases from 35 to 187 µA/μm as the channel length decreases from 3 to 0.23 µm, which is the largest on-current among CVD WSe2n-type transistors. In addition, as shown in Fig. 6(d), our trilayer WSe2 transistors exhibit the largest on/off ratio (Imax/Imin in the transfer characteristics at Vds = 1 V) among CVD WSe2n-type transistors.16,17,27,33,34

FIG. 6.

(a) The transfer and (b) output characteristics at drain-to-source voltages of 0.05 and 1 V with 0.23 µm channel length back-gated trilayer WSe2 transistors. (c) On-current and (d) on/off ratio benchmark of the trilayer WSe2 with different channel lengths.16,17,27,33,34

FIG. 6.

(a) The transfer and (b) output characteristics at drain-to-source voltages of 0.05 and 1 V with 0.23 µm channel length back-gated trilayer WSe2 transistors. (c) On-current and (d) on/off ratio benchmark of the trilayer WSe2 with different channel lengths.16,17,27,33,34

Close modal

In conclusion, this paper reports the synthesis of large single-crystal monolayer to trilayer WSe2 on sapphire and high-performance transistor characteristics. The largest trilayer domain size achieved is up to 80 µm by the molten salt-assisted method with an optimized mass flow of precursors and the size growth of nuclei on the substrate. As the layer number of WSe2 increases, the mean mobility increases from 3.4, 8.4, to 22 cm2/V s for the monolayer, bilayer, and trilayer, respectively. Field-effect transistors fabricated on the trilayer WSe2 show a much higher on/off ratio and on-current compared with previous work. This work demonstrates the potential of CVD trilayer WSe2 for future high-performance short channel electronic devices.

See the supplementary material for the statistical distribution of the crystal sizes and electrical performance of the transistors and other related electrical characteristics.

This work was supported by the National Key Research and Development Program of China (Grant No. 2020AAA0109005), the National Natural Science Foundation of China (Grant Nos. 62090034 and 61874162), and the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB30000000).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
B.
Radisavljevic
,
A.
Radenovic
,
J.
Brivio
,
V.
Giacometti
, and
A.
Kis
,
Nat. Nanotechnol.
6
,
147
(
2011
).
2.
M.
Huang
,
S.
Li
,
Z.
Zhang
,
X.
Xiong
,
X.
Li
, and
Y.
Wu
,
Nat. Nanotechnol.
12
,
1148
(
2017
).
3.
X.
Li
,
L.
Yang
,
M.
Si
,
S.
Li
,
M.
Huang
,
P.
Ye
, and
Y.
Wu
,
Adv. Mater.
27
,
1547
(
2015
).
4.
X.
Li
,
X.
Xiong
,
T.
Li
,
S.
Li
,
Z.
Zhang
, and
Y.
Wu
,
ACS Appl. Mater. Interfaces
9
,
044602
(
2017
).
5.
L.
Yu
,
A.
Zubair
,
E. J. G.
Santos
,
X.
Zhang
,
Y.
Lin
,
Y.
Zhang
, and
T.
Palacios
,
Nano Lett.
15
,
4928
(
2015
).
6.
G. V.
Resta
,
Y.
Balaji
,
D.
Lin
,
I. P.
Radu
,
F.
Catthoor
,
P.-E.
Gaillardon
, and
G.
De Micheli
,
ACS Nano
12
,
7039
(
2018
).
7.
Ansh
,
J.
Kumar
,
G.
Sheoran
,
R.
Mishra
,
S.
Raghavan
, and
M.
Shrivastava
,
IEEE Trans. Electron Devices
67
,
383
(
2019
).
8.
W.-M.
Kang
,
I.-T.
Cho
,
J.
Roh
,
C.
Lee
, and
J.-H.
Lee
,
Semicond. Sci. Technol.
31
,
105001
(
2016
).
9.
W.
Liu
,
J.
Kang
,
D.
Sarkar
,
Y.
Khatami
,
D.
Jena
, and
K.
Banerjee
,
ACS Nano
13
,
1983
(
2013
).
10.
C.
Pang
,
T.
Hung
,
A.
Khosravi
,
R.
Addou
,
R. M.
Wallace
, and
Z.
Chen
,
IEEE Electron Device Lett.
41
,
1122
(
2020
).
11.
M.
Si
,
C.
Jiang
,
W.
Chung
,
Y.
Du
,
M. A.
Alam
, and
P. D.
Ye
,
Nano Lett.
18
,
3682
(
2018
).
12.
K.
Cho
,
J.
Pak
,
S.
Chung
, and
T.
Lee
,
ACS Nano
13
,
9713
(
2019
).
13.
M.
Tian
,
X.
Li
,
T.
Li
,
Q.
Gao
,
X.
Xiong
,
Q.
Hu
,
M.
Wang
,
X.
Wang
, and
Y.
Wu
,
ACS Appl. Mater. Interfaces
10
,
20219
(
2018
).
14.
Q.
Gao
,
Z.
Zhang
,
X.
Xu
,
J.
Song
,
X.
Li
, and
Y.
Wu
,
Nat. Commun.
9
,
4778
(
2018
).
15.
H. G.
Ji
,
P.
Solís‐Fernández
,
D.
Yoshimura
,
M.
Maruyama
,
T.
Endo
,
Y.
Miyata
,
S.
Okada
, and
H.
Ago
,
Adv. Mater.
31
,
1903613
(
2019
).
16.
Y.-C.
Lin
,
B.
Jariwala
,
B. M.
Bersch
,
K.
Xu
,
Y.
Nie
,
B.
Wang
,
S. M.
Eichfeld
,
X.
Zhang
,
T. H.
Choudhury
,
Y.
Pan
,
R.
Addou
,
C. M.
Smyth
,
J.
Li
,
K.
Zhang
,
M. A.
Haque
,
S.
Fölsch
,
R. M.
Feenstra
,
R. M.
Wallace
,
K.
Cho
,
S. K.
Fullerton-Shirey
,
J. M.
Redwing
, and
J. A.
Robinson
,
ACS Nano
12
,
965
(
2018
).
17.
M.
Asghari Heidarlou
,
P.
Paletti
,
B.
Jariwala
,
J. A.
Robinson
,
S. K.
Fullerton-Shirey
, and
A. C.
Seabaugh
,
IEEE Trans. Electron Devices
67
,
1839
(
2020
).
18.
L.
Fang
,
X.
Yuan
,
K.
Liu
,
L.
Li
,
P.
Zhou
,
W.
Ma
,
H.
Huang
,
J.
He
, and
S.
Tao
,
Nanoscale
12
,
3715
(
2020
).
19.
J.
Zhou
,
J.
Lin
,
X.
Huang
,
Y.
Zhou
,
Y.
Chen
,
J.
Xia
,
H.
Wang
,
Y.
Xie
,
H.
Yu
,
J.
Lei
,
D.
Wu
,
F.
Liu
,
Q.
Fu
,
Q.
Zeng
,
C.-H.
Hsu
,
C.
Yang
,
L.
Lu
,
T.
Yu
,
Z.
Shen
,
H.
Lin
,
B. I.
Yakobson
,
Q.
Liu
,
K.
Suenaga
,
G.
Liu
, and
Z.
Liu
,
Nature
556
,
355
(
2018
).
20.
L.
Xu
,
J.
Zhou
,
Z.
He
,
J.
Hu
, and
M.
Liu
,
Mater. Lett.
284
,
128994
(
2021
).
21.
A.
Han
,
A.
Aljarb
,
S.
Liu
,
P.
Li
,
C.
Ma
,
F.
Xue
,
S.
Lopatin
,
C.-W.
Yang
,
J.-K.
Huang
,
Y.
Wan
,
X.
Zhang
,
Q.
Xiong
,
K.-W.
Huang
,
V.
Tung
,
T. D.
Anthopoulos
, and
L.-J.
Li
,
Nanoscale Horiz.
4
,
1434
(
2019
).
22.
Z.
Chu
,
A.
Han
,
C.
Lei
,
S.
Lopatin
,
P.
Li
,
D.
Wannlund
,
D.
Wu
,
K.
Herrera
,
X.
Zhang
,
A. H.
MacDonald
,
X.
Li
,
L.-J.
Li
, and
K.
Lai
,
Nano Lett.
18
,
7200
(
2018
).
23.
Z.
Zhang
,
X.
Xu
,
J.
Song
,
Q.
Gao
,
S.
Li
,
Q.
Hu
,
X.
Li
, and
Y.
Wu
,
Appl. Phys. Lett.
113
,
202103
(
2018
).
24.
M. H.
Chiu
,
H. L.
Tang
,
C. C.
Tseng
,
Y.
Han
,
A.
Aljarb
,
J. K.
Huang
,
Y.
Wan
,
J. H.
Fu
,
X.
Zhang
,
W. H.
Chang
,
D. A.
Muller
,
T.
Takenobu
,
V.
Tung
, and
L. J.
Li
,
Adv. Mater.
31
,
1900861
(
2019
).
25.
26.
J.-K.
Huang
,
J.
Pu
,
C.-L.
Hsu
,
M.-H.
Chiu
,
Z.-Y.
Juang
,
Y.-H.
Chang
,
W.-H.
Chang
,
Y.
Iwasa
,
T.
Takenobu
, and
L.-J.
Li
,
ACS Nano
8
,
923
(
2014
).
27.
S.
Li
,
S.
Wang
,
D.-M.
Tang
,
W.
Zhao
,
H.
Xu
,
L.
Chu
,
Y.
Bando
,
D.
Golberg
, and
G.
Eda
,
Appl. Mater. Today
1
,
60
(
2015
).
28.
L.
Chen
,
B.
Liu
,
M.
Ge
,
Y.
Ma
,
A. N.
Abbas
, and
C.
Zhou
,
ACS Nano
9
,
8368
(
2015
).
29.
J.
Xia
,
X.
Huang
,
L.-Z.
Liu
,
M.
Wang
,
L.
Wang
,
B.
Huang
,
D.-D.
Zhu
,
J.-J.
Li
,
C.-Z.
Gu
, and
X.-M.
Meng
,
Nanoscale
6
,
8949
(
2014
).
30.
J.
Xiao
,
J.
Liu
,
K.
Sun
,
Y.
Zhao
,
Z.
Shao
,
X.
Liu
,
Y.
Yuan
,
Y.
Li
,
H.
Xie
,
F.
Song
,
Y.
Gao
, and
H.
Huang
,
J. Phys. Chem. Lett.
10
,
4203
(
2019
).
31.
S. V.
Mandyam
,
M.-Q.
Zhao
,
P. M.
Das
,
Q.
Zhang
,
C. C.
Price
,
Z.
Gao
,
V. B.
Shenoy
,
M.
Drndić
, and
A. T. C.
Johnson
,
ACS Nano
13
,
10490
(
2019
).
32.
H.
Zeng
,
G.-B.
Liu
,
J.
Dai
,
Y.
Yan
,
B.
Zhu
,
R.
He
,
L.
Xie
,
S.
Xu
,
X.
Chen
,
W.
Yao
, and
X.
Cui
,
Sci. Rep.
3
,
1608
(
2013
).
33.
A.
Kozhakhmetov
,
B.
Schuler
,
A. M. Z.
Tan
,
K. A.
Cochrane
,
J. R.
Nasr
,
H.
El-Sherif
,
A.
Bansal
,
A.
Vera
,
V.
Bojan
,
J. M.
Redwing
,
N.
Bassim
,
S.
Das
,
R. G.
Hennig
,
A.
Weber-Bargioni
, and
J. A.
Robinson
,
Adv. Mater.
32
,
2005159
(
2020
).
34.
X.
Zhang
,
F.
Zhang
,
Y.
Wang
,
D. S.
Schulman
,
T.
Zhang
,
A.
Bansal
,
N.
Alem
,
S.
Das
,
V. H.
Crespi
,
M.
Terrones
, and
J. M.
Redwing
,
ACS Nano
13
,
3341
(
2019
).
35.
S.
Das
and
J.
Appenzeller
,
Appl. Phys. Lett.
103
,
103501
(
2013
).

Supplementary Material