Heterogeneous integration of two-dimensional materials and the conventional semiconductor has opened opportunities for next-generation semiconductor devices and their processing. Heterogeneous integration has been studied for economical manufacturing by substrate recycling and novel functionalities by a combination of incommensurate materials. However, utilizing the integration requires controlling locations of the integrated architectures. Here, we show area-selective deposition (ASD) of germanium on the graphene/MoS2 stack. Ge nucleation precisely occurred on the surfaces of the patterned graphene/MoS2 stack via dipole engineering. In this study, the growth temperature of ASD of Ge was significantly lower than that based on precursor desorption on SiO2. The first-principles calculations revealed that Ge deposited by ASD on the graphene/MoS2 stack was not affected by charge transfer. This work provides a viable way to utilize atomically thin materials for next-generation semiconductor devices, which can be applicable for “Beyond Moore” and “More Moore” approaches.

Two-dimensional (2D) materials are being considered essential building blocks for “Beyond Moore” and “More Moore” approaches to develop advanced semiconductor devices and systems.1,2 The main approach utilizing 2D materials in semiconductor devices is replacing the carrier transport channel with 2D materials in device architectures to deliver enhanced performances and realize novel functionalities of semiconductor devices, such as graphene-embedded photodetectors in infrared and visible wavelengths and 2D electrodes for on-chip optoelectronic devices.3–5 The recent progress in materials growth, represented by van der Waals epitaxy and remote epitaxy, has offered novel opportunities for 2D materials as substrates of conventional/three-dimensional (3D) materials, such as semiconductor thin films and membranes.6–9 Three-dimensional materials growth on 2D substrates has been mainly employed for economical manufacturing via substrate recycling or the formation of heterostructures for functionality.9–12 Specifically, 2D materials as substrates for 3D materials growth also provide another opportunity for selective nucleation via engineering of out-of-plane dipole moment.13 Therefore, 2D materials-based area-selective deposition (ASD) can be another application of 2D materials as substrates.

ASD of Si–Ge has been employed for various applications, such as the fabrication of carrier transport channels and electrodes14–21 and “chemical patterning” of metal layers.22 Therefore, ASD has become important because direct growth of Si–Ge on source/drain regions in advanced transistor technology nodes reduces parasitic series resistance, directly affecting the performances of fin field-effect-transistors (FETs) and 3D architecture FETs for future technology nodes.20,21,23,24 In addition to the FinFET applications, ASD of Ge has also been employed for optoelectronic device applications and heterogeneous integration.19,25,26

The main mechanisms of ASD of Si–Ge are kinetic control of adsorption and desorption of chloride-based precursors and contrast of crystallinity of Si–Ge deposited onto crystalline or amorphous substrates, resulting in a difference in the etching rates of single crystalline and polycrystalline Si–Ge. Another potential approach is providing a contrast of nucleation probability via dipole engineering without the change of conventional precursors of Si–Ge.13 The dipole engineering is realized by an induced out-of-plane dipole moment, which enhances the surface energy of a substrate, resulting in higher nucleation probability. A stack of 2D materials, such as graphene and hexagonal boron nitride (h-BN), can induce an out-of-dipole moment, which is strong enough for the nucleation of Ge. However, the formation of 2D/Si–Ge multi-dimensional structures also bring changes in the physical properties of Si–Ge via charge transfer and unintentional diffusion of atoms across the interfaces.27 Therefore, a combination of 2D materials and Si–Ge should be considered to avoid an unwanted change in carrier transport characteristics, which requires a novel device design approach. Graphene is a robust 2D material for diffusion barrier to prevent unintentional changes in the physical properties of Si–Ge grown on 2D materials.28,29 However, the inherently small out-of-plane dipole moment of graphene hinders Si–Ge from nucleation on graphene without introducing defects in the graphene.13 Although the graphene/h-BN stack induces Ge nucleation on the graphene, scalable preparation of the graphene/h-BN continuous thin film stack is severely limited by the growth temperature of the h-BN film (700–1500 °C) and inhomogeneity of the number of h-BN layers prepared by deposition.30 

Here, we report a solution for ASD of Ge on a stack of graphene and monolayer (1L) MoS2, which does not induce unintentional charge transfer and is available to prepare in a scalable manner. The first-principles calculations revealed that the graphene/1L-MoS2 stack induces an out-of-plane dipole moment, which is strong enough for the nucleation of Ge on graphene. We prepared a patterned graphene/1L-MoS2 stack on SiO2/Si substrates. Subsequently, Ge was selectively grown on the patterned graphene/1L-MoS2 regions by chemical vapor deposition (CVD). Additionally, the effect of the patterning process on Ge nucleation was studied by comparing Ar ion milling and oxygen plasma etching.

The out-of-plane dipole moment of a 2D material system is a crucial factor governing nucleation probability of Ge as reported.13 The density functional theory (DFT) calculations revealed that the out-of-plane dipole moment of 1L graphene is in the range of −4.32 × 10−5 and −1.1 × 10−4 debye (D), which is not enough to nucleate Ge on the graphene surface. According to the DFT calculations, the out-of-dipole moment of a stack composed of 1L graphene and 1L-MoS2 (−2.5 × 10−2 D) is about two orders larger than that of 1L graphene as well as that of 1L-MoS2 (−4 to −1.6 × 10−4 D). Figure 1 shows the charge distribution in the graphene/MoS2 stack. Electron accumulation in the vdW gap between graphene and MoS2 and electron depletion in graphene result in an interfacial dipole moment. The electron depletion in graphene is consistent with the reported results of photoemission spectroscopy of the graphene/MoS2.31 The two-orders magnitude increase in the out-of-plane dipole moment in the graphene/1L-MoS2 stack induces significant enhancement of the surface energy on graphene, implying a remarkably higher nucleation probability of Ge on the graphene surface.

FIG. 1.

Charge density difference in graphene/1L-MoS2 calculated by DFT. Side view of atomic configuration and charge density distribution in the graphene/1L-MoS2 stack.

FIG. 1.

Charge density difference in graphene/1L-MoS2 calculated by DFT. Side view of atomic configuration and charge density distribution in the graphene/1L-MoS2 stack.

Close modal
Another critical factor for the nucleation of Ge on graphene is the insignificant charge transfer between Ge and the graphene/MoS2 stack because the unintentional charge transfer alters the electrical characteristics of Ge grown on MoS2, as shown in a previous study.27  Figure 2 shows the charge distribution in the Ge/graphene/1L-MoS2 structure. In the Ge/graphene/1L-MoS2 structure, the out-of-plane dipole moment inducing nucleation of Ge is governed by the charge accumulation along the graphene/MoS2 interface. However, the out-of-plane dipole moment formed along the interface between Ge and graphene has a polarity opposite to that formed along the interface between graphene and MoS2. The total charge distribution in the Ge/graphene/1L-MoS2 heterostructure indicates that charge distribution in the Ge is not noticeably affected. The density of state (DOS) calculations also revealed insignificant charge transfer between the orbitals of the Ge, graphene, and MoS2 layers, as shown in Fig. S1. The DOS profiles of S, C, and Ge atoms near the interfaces between the adjacent layers were compared to those of S, C, and Ge atoms in pristine phases. The robust nature of DOS and chemical potential (0 eV in Fig. S1) show no net charge transfer in the Ge/graphene/MoS2 heterostructure, resulting in no effect on the physical properties of Ge in Ge/graphene/MoS2. Furthermore, the formation of the heterostructure is possible due to interfacial charge re-distribution. The formation energy was calculated by the following equation:
ΔE=EGe+graphene+MoS2EGeEGrapheneEMoS2.
FIG. 2.

Spatial distribution of the charge density difference in Ge/graphene/1L-MoS2.

FIG. 2.

Spatial distribution of the charge density difference in Ge/graphene/1L-MoS2.

Close modal

The calculated formation energy was −0.13 eV/C atom. The small formation energy indicates that the heterostructure was formed by physisorption.

ASD study of Ge was performed on the graphene/MoS2 stack along the theoretical prediction discussed above. Graphene and 1L-MoS2 continuous thin films were prepared by chemical vapor deposition (CVD) on Cu foils and metalorganic CVD on SiO2/Si substrates, respectively. The graphene film was separated from the Cu foil and, subsequently, transferred onto the 1L-MoS2/SiO2/Si substrate. Lithography and etching techniques were employed to define stripe patterns of the graphene/MoS2 stack with different widths, as shown in Fig. 4(a). Two sets of the patterned graphene/MoS2 stack were prepared by using different etching methods, such as Ar ion milling and oxygen plasma etching, to remove graphene layers, as previously reported.32–35 Raman and x-ray photoelectron spectroscopy studies (Figs. S2 and S3) revealed that there was no noticeable change in the surface characteristics of graphene and 1L-MoS2, induced by oxygen plasma etching and Ar ion milling with the conditions for the study. In the patterning procedure, the graphene was removed outside the defined stripe patterns to expose the 1L-MoS2 underneath the removed graphene regions. An optical microscopy image of the patterned graphene/1L-MoS2 stack after Ar ion milling is shown in Fig. 3(a). There is a clear contrast between the graphene/1L-MoS2 layers and exposed MoS2.

FIG. 3.

(a) Schematic of the patterning of graphene on 1L-MoS2 and the optical microscopy image of the patterned graphene/1L-MoS2 stack. (b) Detailed Raman maps of the graphene/1L-MoS2 stacks prepared by Ar ion milling. (c) Line profiles of the MoS2 E2g/A1g peak positions, the difference of the MoS2 E2g/A1g peak positions (A1g − E2g), and the Raman signal intensity of graphene 2D.

FIG. 3.

(a) Schematic of the patterning of graphene on 1L-MoS2 and the optical microscopy image of the patterned graphene/1L-MoS2 stack. (b) Detailed Raman maps of the graphene/1L-MoS2 stacks prepared by Ar ion milling. (c) Line profiles of the MoS2 E2g/A1g peak positions, the difference of the MoS2 E2g/A1g peak positions (A1g − E2g), and the Raman signal intensity of graphene 2D.

Close modal

Detailed Raman mapping [Fig. 3(b)] of the Ar ion milled graphene/MoS2 stack reveals that the graphene layer was patterned on the MoS2 layer as intended. The Raman shift originating from the graphene 2D peak was observed on the patterned graphene/MoS2 stripes only. On the other hand, the MoS2 Raman mapping demonstrates that even after Ar ion milling, 1L-MoS2 remains in the entire SiO2/Si substrate. Figure 3(c) shows the line profiles of MoS2 E2g/A1g peak positions and the maximum intensity of graphene 2D peaks. The green and yellow shaded boxes indicate the etched (MoS2 only) and graphene-covered areas (graphene/MoS2 stack) respectively. Compared to the MoS2 A1g peaks from the etched area (MoS2 only), the MoS2 A1g peaks measured on the graphene/MoS2 stack decreased by 1 cm−1. It can be explained by electron doping in the MoS2 layer when the MoS2 layer is covered by graphene.36 This result corresponds to the DFT calculations shown in Fig. 1 which expect that the out-of-plane dipole moment is caused by the charge density difference between the MoS2 and graphene layers.

Raman microscopy was performed for the Ar ion milled and oxygen plasma etched graphene/MoS2 stacks before Ge growth to investigate processing-dependent characteristics of the graphene/MoS2 stack as a substrate of Ge growth. Figure 4(a) shows the Raman mapping of the graphene/MoS2 stacks before Ge growth. For both Ar ion milled and oxygen plasma etched stacks, the characteristic Raman peaks of 1L-MoS2 were detected at 383.7 cm−1 (E2g1of MoS2) and 404.6 cm−1 (A1g of MoS2) without a noticeable shift, while the G band of graphene shows an apparent shift from 1592.3 cm−1 of the oxygen plasma etched stack to 1589.1 cm−1 of the ion milled stack. On the other hand, the D band at ∼1357.8 cm−1, originating from the defects of graphene, was clearly observed for the oxygen plasma etched graphene/MoS2 stack, whereas it was not noticeable for the Ar ion milled graphene/MoS2 stack. The D band signal intensities from two patterning methods were compared, as shown in Fig. 4(b). It shows more clearly that oxygen plasma etching makes the graphene layer defective. The difference of defects between the two patterning methods means that the graphene quality after patterning is process-dependent.

FIG. 4.

(a) Raman maps of the graphene/1L-MoS2 stacks prepared by Ar ion milling and oxygen plasma etching. (b) Line profiles of the Raman signal intensity of graphene D from the Ar ion milled and oxygen plasma etched graphene/1L-MoS2 stacks.

FIG. 4.

(a) Raman maps of the graphene/1L-MoS2 stacks prepared by Ar ion milling and oxygen plasma etching. (b) Line profiles of the Raman signal intensity of graphene D from the Ar ion milled and oxygen plasma etched graphene/1L-MoS2 stacks.

Close modal

Nucleation behaviors of Ge were investigated on the patterned graphene/MoS2 stack in which an out-of-plane dipole moment has spatial distribution. Chemical vapor deposition (CVD) of Ge was performed at 500 °C on the patterned graphene/MoS2/SiO2/Si substrates in a stainless steel reactor with germane as depicted in Fig. 5(a). Figure 5(b) shows an optical microscopy image of the patterned graphene/1L-MoS2 stack after Ar ion milling. There is a clear contrast depending on the presence of the graphene layer. After Ge growth on the patterned substrates for 30 min, scanning electron microscopy (SEM) images [Figs. 5(c) and 5(d)] show a clear contrast indicating that Ge was selectively grown on the graphene/1L-MoS2 layers. The Ge growth results from the out-of-plane dipole moment of the graphene/1L-MoS2 stack as we predicted. The ASD of Ge on the patterned graphene/MoS2 stack can be explained by the nucleation of Ge on the graphene surface with the enhanced out-of-plane dipole moment. Another interesting observation is that the ASD occurred on the graphene/1L-MoS2 stack, although the bilayer stack was surrounded with 1L-MoS2. The reported nucleation of Ge on 1L-MoS2 indicates that the dipole moment of 1L-MoS2 is enough to nucleate Ge.27 Nevertheless, nucleation of Ge preferentially occurred on the graphene/1L-MoS2 stack providing a higher out-of-plane dipole moment. The observation indicates that spatial distribution of the out-of-plane dipole moment on the same chemical composition can induce ASD. In our study, the growth temperature (500 °C) of Ge was significantly lower than that (800 °C) of the reported ASD of Ge on patterned SiO2/Si by CVD using the same precursor (germane).15 The previous study of ASD of Ge is based on the higher desorption rate of germane on SiO2 at 800 °C, while our study utilizes the controlled nucleation of Ge along the out-of-plane dipole moment.

FIG. 5.

(a) Schematic of the nucleation of Ge on the graphene/1L-MoS2 pattern. (b) Optical microscopy image of the patterned graphene/1L-MoS2. SEM images after Ge growth on the patterned graphene/1L-MoS2 stripes with different widths, prepared by Ar ion milling [(c), (e), and (g)] and oxygen plasma etching [(d), (f), and (h)].

FIG. 5.

(a) Schematic of the nucleation of Ge on the graphene/1L-MoS2 pattern. (b) Optical microscopy image of the patterned graphene/1L-MoS2. SEM images after Ge growth on the patterned graphene/1L-MoS2 stripes with different widths, prepared by Ar ion milling [(c), (e), and (g)] and oxygen plasma etching [(d), (f), and (h)].

Close modal

The ASD of Ge on the patterned graphene/MoS2 stack showed different behaviors along the patterning methods. The uniform color and contrast of the Ge layer grown on the graphene/MoS2 stack prepared by Ar ion milling, as shown in Fig. 5(c), indicates that the Ge thickness is uniform. Meanwhile, the Ge layer grown on the graphene/MoS2 stack prepared by oxygen plasma etching [Fig. 5(d)] is bright around the edges of the stripes and circular patterns and is darker in the regions away from the pattern edges. The spatial difference of contrast indicates that the Ge layer thickness varies around pattern edges and inner regions of the pattern. Magnified SEM images [Figs. 5(e)5(h)] show that Ge nucleation occurred uniformly for the graphene/MoS2 stack prepared by Ar ion milling. In contrast, preferred Ge nucleation along the edges of the patterns was observed for the graphene/MoS2 stack prepared by oxygen plasma etching. We assume that the thickness difference in ASD with oxygen plasma etching is due to the defective graphene layer based on the previous study about Ge nucleation on graphene.13 Oxygen plasma etching and ozonation of graphene induce defective sites through oxidation. The defective sites on graphene act as nucleation sites of Ge.

Another interesting feature of the Ge growth behavior is the size-dependent nucleation of Ge on the graphene/MoS2 stack. Figures 5(e) and 5(g) show that the coverage of Ge was smaller as the stripe width became narrower. The size-dependent nucleation and growth behavior can be explained by “size-dependent precursor adsorption.” A previous study revealed that the desorption coefficient of the incident precursor molecules is different at the edge and the center of a plane. In the case of enhanced desorption at the edges, the growth rate becomes lower as the size of the structure shrinks because of the higher probability of encountering edges before the precursor molecules are decomposed to supply adatoms for growth.37 The size-dependent adsorption/desorption behavior is prominent in the range of 200 nm and 3 µm, which matches that of the current study.

The out-of-plane dipole engineering achieved ASD of Ge on the graphene/MoS2 stack at a marginal temperature of 500 °C, significantly lower than the previous studies performed at 800 °C. This site-selective nucleation strategy for 3D material growth on a 2D material demonstrates that 2D/3D integration is applicable to semiconductor manufacturing. Moreover, edge-preferred nucleation of Ge on the graphene/MoS2 stack prepared by oxygen plasma etching shows that patterning process affects the nucleation of ASD. We anticipate that the ASD of Ge on the 2D material stack provides semiconductor manufacturing for electronic and photonic device applications. The progress in the synthesis of 2D materials and their heterostructures, providing clean interfaces and an improved crystalline quality of a 2D material, offers a scalable way of ASD of high-quality Ge, overcoming the disadvantages of the stack approach presented here.

Details of the experimental methods, additional density of states calculations of S, C, and Ge atoms (Fig. S1) in the pristine and the Ge/graphene/MoS2 structure, Raman and x-ray photoelectron spectroscopy characterization after patterning of graphene and MoS2 with 300 nm-thick PMMA layer are provided in the supplementary material.

This research was financially supported by the Laboratory Directed Research and Development program of Los Alamos National Laboratory (Grant Nos. 20230256ER and 20230014DR). This work was partly performed at CINT, a U.S. Department of Energy, office of Basic Energy Sciences User Facility at Los Alamos National Laboratory (Contract No. 89233218CNA000001), Sandia National Laboratories (Contract No. DE-NA-0003525), and Pacific Northwest National Laboratory (Contract No. DE-AC05-76RL0-1830). This research used resources of the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility located at Lawrence Berkeley National Laboratory, operated under Contract No. DE-AC02-05CH11231 using NERSC Award No. ERCAP0022610.

The authors have no conflicts to disclose.

J.Y. conceived and directed the main experimental idea and performed patterning and germanium growth; X.W., M.T.P., Y.L., and S.K. conducted Raman mapping and analyses; T.A. conducted theoretical simulations; Y.K. prepared graphene/MoS2 stacks; J.P. and K.K. prepared monolayer MoS2 thin films; W.S.Y. and Y.J.H. prepared graphene films. J.Y. is responsible for all research results. J.Y., T.A., X.W., and Y.L. co-wrote the manuscript. All authors discussed and commented on the manuscript.

Yeonjoo Lee: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (lead); Writing – review & editing (lead). Towfiq Ahmed: Data curation (equal); Formal analysis (equal); Visualization (equal); Writing – review & editing (equal). Xuejing Wang: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (supporting); Writing – review & editing (supporting). Michael T. Pettes: Data curation (equal); Formal analysis (equal). Yeonhoo Kim: Data curation (supporting); Formal analysis (supporting); Methodology (supporting); Visualization (supporting). Jeongwon Park: Investigation (supporting). Woo Seok Yang: Investigation (supporting). Kibum Kang: Investigation (supporting). Young Joon Hong: Investigation (supporting). Soyeong Kwon: Investigation (supporting). Jinkyoung Yoo: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (supporting); Methodology (supporting); Project administration (lead); Supervision (lead); Validation (lead); Writing – original draft (lead); Writing – review & editing (lead).

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

1.
M. C.
Lemme
,
D.
Akinwande
,
C.
Huyghebaert
, and
C.
Stampfer
,
Nat. Commun.
13
,
1392
(
2022
).
2.
D.
Akinwande
,
C.
Huyghebaert
,
C.-H.
Wang
,
M. I.
Serna
,
S.
Goossens
,
L.-J.
Li
,
H. S. P.
Wong
, and
F. H. L.
Koppens
,
Nature
573
,
507
(
2019
).
3.
Q.
Chu
,
L.
Li
,
C.
Zhu
,
Y.
Zang
,
S.
Lin
, and
Y.
Han
,
Mater. Lett.
211
,
133
(
2018
).
4.
H.
Cong
,
F.
Yang
,
C.
Xue
,
K.
Yu
,
L.
Zhou
,
N.
Wang
,
B.
Cheng
, and
Q.
Wang
,
Small
14
,
1704414
(
2018
).
5.
I. A.
Fischer
,
M.
Brehm
,
M.
De Seta
,
G.
Isella
,
D. J.
Paul
,
M.
Virgilio
, and
G.
Capellini
,
APL Photonics
7
,
050901
(
2022
).
6.
Y.
Kim
,
S. S.
Cruz
,
K.
Lee
,
B. O.
Alawode
,
C.
Choi
,
Y.
Song
,
J. M.
Johnson
,
C.
Heidelberger
,
W.
Kong
,
S.
Choi
,
K.
Qiao
,
I.
Almansouri
,
E. A.
Fitzgerald
,
J.
Kong
,
A. M.
Kolpak
,
J.
Hwang
, and
J.
Kim
,
Nature
544
,
340
(
2017
).
7.
J.
Jeong
,
K.-A.
Min
,
B. K.
Kang
,
D. H.
Shin
,
J.
Yoo
,
W. S.
Yang
,
S. W.
Lee
,
S.
Hong
, and
Y. J.
Hong
,
Appl. Phys. Lett.
113
,
233103
(
2018
).
8.
C.-H.
Ma
,
L.-S.
Lu
,
H.
Song
,
J.-W.
Chen
,
P.-C.
Wu
,
C.-L.
Wu
,
R.
Huang
,
W.-H.
Chang
, and
Y.-H.
Chu
,
APL Mater.
9
,
051115
(
2021
).
9.
Y. H.
Kim
,
J.
Watt
,
X.
Ma
,
T.
Ahmed
,
S.
Kim
,
K.
Kang
,
T. S.
Luk
,
Y. J.
Hong
, and
J.
Yoo
,
ACS Nano
16
,
2399
(
2022
).
10.
H. S.
Kum
,
H.
Lee
,
S.
Kim
,
S.
Lindemann
,
W.
Kong
,
K.
Qiao
,
P.
Chen
,
J.
Irwin
,
J. H.
Lee
,
S.
Xie
,
S.
Subramanian
,
J.
Shim
,
S.-H.
Bae
,
C.
Choi
,
L.
Ranno
,
S.
Seo
,
S.
Lee
,
J.
Bauer
,
H.
Li
,
K.
Lee
,
J. A.
Robinson
,
C. A.
Ross
,
D. G.
Schlom
,
M. S.
Rzchowski
,
C.-B.
Eom
, and
J.
Kim
,
Nature
578
,
75
(
2020
).
11.
J.
Jeong
,
Q.
Wang
,
J.
Cha
,
D. K.
Jin
,
D. H.
Shin
,
S.
Kwon
,
B. K.
Kang
,
J. H.
Jang
,
W. S.
Yang
,
Y. S.
Choi
,
J.
Yoo
,
J. K.
Kim
,
C.-H.
Lee
,
S. W.
Lee
,
A.
Zakhidov
,
S.
Hong
,
M. J.
Kim
, and
Y. J.
Hong
,
Sci. Adv.
6
,
eaaz5180
(
2020
).
12.
S.
Sundaram
,
X.
Li
,
Y.
Halfaya
,
T.
Ayari
,
G.
Patriarche
,
C.
Bishop
,
S.
Alam
,
S.
Gautier
,
P. L.
Voss
,
J. P.
Salvestrini
, and
A.
Ougazzaden
,
Adv. Mater. Interfaces
6
,
1900207
(
2019
).
13.
J.
Yoo
,
T.
Ahmed
,
R.
Chen
,
A.
Chen
,
Y. H.
Kim
,
K. C.
Kwon
,
C. W.
Park
,
H. S.
Kang
,
H. W.
Jang
,
Y. J.
Hong
,
W. S.
Yang
, and
C.-H.
Lee
,
Nanoscale
10
,
5689
(
2018
).
14.
B. D.
Joyce
and
J. A.
Baldrey
,
Nature
195
,
485
(
1962
).
16.
Q. S.
Wang
,
A.
Reisman
, and
D.
Temple
,
J. Electrochem. Soc.
141
,
593
(
1994
).
17.
R.
Soman
,
A.
Reisman
,
D.
Temple
, and
R.
Alberti
,
J. Electrochem. Soc.
147
,
1847
(
2000
).
18.
R.
Soman
,
A.
Reisman
,
D.
Temple
,
R.
Alberti
, and
C.
Pace
,
J. Electrochem. Soc.
147
,
1854
(
2000
).
19.
H. Y.
Yu
,
J. H.
Park
,
A. K.
Okyay
, and
K. C.
Saraswat
,
IEEE Electron Device Lett.
33
,
579
(
2012
).
20.
G.
Wang
,
C.
Qin
,
H.
Yin
,
J.
Luo
,
N.
Duan
,
P.
Yang
,
X.
Gao
,
T.
Yang
,
J.
Li
,
J.
Yan
,
H.
Zhu
,
W.
Wang
,
D.
Chen
,
T.
Ye
,
C.
Zhao
, and
H. H.
Radamson
,
Microelectron. Eng.
163
,
49
(
2016
).
21.
A.
Vohra
,
C.
Porret
,
D.
Kohen
,
S.
Folkersma
,
J.
Bogdanowicz
,
M.
Schaekers
,
J.
Tolle
,
A.
Hikavyy
,
E.
Capogreco
,
L.
Witters
,
R.
Langer
,
W.
Vandervorst
, and
R.
Loo
,
Jpn. J. Appl. Phys.
58
,
SBBA04
(
2019
).
22.
R.
Khan
,
B.
Shong
,
B. G.
Ko
,
J. K.
Lee
,
H.
Lee
,
J. Y.
Park
,
I.-K.
Oh
,
S. S.
Raya
,
H. M.
Hong
,
K.-B.
Chung
,
E. J.
Luber
,
Y.-S.
Kim
,
C.-H.
Lee
,
W.-H.
Kim
, and
H.-B.-R.
Lee
,
Chem. Mater.
30
,
7603
(
2018
).
23.
N.
Collaert
,
R.
Rooyackers
,
A.
Hikavyy
,
A.
Dixit
,
F.
Leys
,
P.
Verheyen
,
R.
Loo
,
M.
Jurczak
, and
S.
Biesemans
,
Thin Solid Films
517
,
101
(
2008
).
24.
C.
Qin
,
G.
Wang
,
M.
Kolahdouz
,
J.
Luo
,
H.
Yin
,
P.
Yang
,
J.
Li
,
H.
Zhu
,
Z.
Chao
,
T.
Ye
, and
H. H.
Radamson
,
Solid-State Electron.
124
,
10
(
2016
).
25.
T.
Mogami
,
S.
Akiyama
,
T.
Baba
,
S.
Okumura
,
J.
Fujikata
,
T.
Horikawa
, and
K.
Kinoshita
, in
2016 IEEE Silicon Nanoelectronics Workshop
(
IEEE
,
Honolulu, USA
,
2016
), p.
216
.
26.
Z.
Liu
,
F.
Yang
,
W. Z.
Wu
,
H.
Cong
,
J.
Zheng
,
C. B.
Li
,
C. L.
Xue
,
B. W.
Cheng
, and
Q. M.
Wang
,
J. Lightwave Technol.
35
,
5306
(
2017
).
27.
Y.-C.
Lin
,
I.
Bilgin
,
T.
Ahmed
,
R.
Chen
,
D.
Pete
,
S.
Kar
,
J.-X.
Zhu
,
G.
Gupta
,
A.
Mohite
, and
J.
Yoo
,
Nanoscale
8
,
18675
(
2016
).
28.
H.-Y.
Kim
,
C.
Lee
,
J.
Kim
,
F.
Ren
, and
S. J.
Pearton
,
J. Vac. Sci. Technol., B
30
,
030602
(
2012
).
29.
J.
Hong
,
S.
Lee
,
S.
Lee
,
H.
Han
,
C.
Mahata
,
H.-W.
Yeon
,
B.
Koo
,
S.-I.
Kim
,
T.
Nam
,
K.
Byun
,
B.-W.
Min
,
Y.-W.
Kim
,
H.
Kim
,
Y.-C.
Joo
, and
T.
Lee
,
Nanoscale
6
,
7503
(
2014
).
30.
M. H.
Khan
,
H. K.
Liu
,
X.
Sun
,
Y.
Yamauchi
,
Y.
Bando
,
D.
Golberg
, and
Z.
Huang
,
Mater. Today
20
,
611
(
2017
).
31.
H.
Coy Diaz
,
R.
Addou
, and
M.
Batzill
,
Nanoscale
6
,
1071
(
2014
).
32.
I.
Childres
,
L. A.
Jauregui
,
J.
Tian
, and
Y. P.
Chen
,
New J. Phys.
13
,
025008
(
2011
).
33.
P.
Jia
,
F.
Pan
, and
T.
Chen
,
IOP Conf. Ser.: Mater. Sci. Eng.
182
,
012030
(
2017
).
34.
H.
Zhu
,
X.
Qin
,
L.
Cheng
,
A.
Azcatl
,
J.
Kim
, and
R. M.
Wallace
,
ACS Appl. Mater. Interfaces
8
,
19119
(
2016
).
35.
H. M.
Gramling
,
C. M.
Towle
,
S. B.
Desai
,
H.
Sun
,
E. C.
Lewis
,
V. D.
Nguyen
,
J. W.
Ager
,
D.
Chrzan
,
E. M.
Yeatman
,
A.
Javey
, and
H.
Taylor
,
ACS Appl. Electron. Mater.
1
,
407
(
2019
).
36.
Y.
Zou
,
Q.-S.
Ma
,
Z.
Zhang
,
R.
Pu
,
W.
Zhang
,
P.
Suo
,
K.
Sun
,
J.
Chen
,
D.
Li
,
H.
Ma
,
X.
Lin
,
Y.
Leng
,
W.
Liu
,
J.
Du
, and
G.
Ma
,
J. Phys. Chem. Lett.
13
,
5123
(
2022
).
37.
J.
Yoo
,
S. A.
Dayeh
,
N. C.
Bartelt
,
W.
Tang
,
A. T.
Findikoglu
, and
S. T.
Picraux
,
Nano Lett.
14
,
6121
(
2014
).

Supplementary Material