Research Article

www.advmat.de

Lattice Polarity Manipulation of Quasi-vdW Epitaxial GaN

Films on Graphene Through Interface Atomic Configuration
Fang Liu, Tao Wang, Zhihong Zhang, Tong Shen, Xin Rong, Bowen Sheng, Liuyun Yang,
Duo Li, Jiaqi Wei, Shanshan Sheng, Xingguang Li, Zhaoying Chen, Renchun Tao,
Ye Yuan, Xuelin Yang, Fujun Xu, Jingmin Zhang, Kaihui Liu,* Xin-Zheng Li,* Bo Shen,
and Xinqiang Wang*

1. Introduction
Quasi van der Waals epitaxy, a pioneering epitaxy of sp3-hybridized semicon-
ductor films on sp2-hybridized 2D materials, provides a way, in principle, to 2D materials represented by graphene are
achieve single-crystal epilayers with preferred atom configurations that are composed of ultrathin atomic layers, and
these atomic-scale monolayers (MLs) with
free of substrate. Unfortunately, this has not been experimentally confirmed
an in-plane hexagonal honeycomb lat-
in the case of the hexagonal semiconductor III-nitride epilayer until now. tice are combined through van der Waals
Here, it is reported that the epitaxy of gallium nitride (GaN) on graphene can (vdW) interactions.[1] Considering their
tune the atom arrangement (lattice polarity) through manipulation of the good growth compatibility with hexa­
interface atomic configuration, where GaN films with gallium and nitrogen gonal semiconductors such as wurtzite
III-nitrides,[2,3] 2D materials are regarded
polarity are achieved by forming CONGa(3) or COGaN(3) con-
as an ideal platform for the fabrication of
figurations, respectively, on artificial CO surface dangling bonds by atomic advanced optoelectronic devices beyond
oxygen pre-irradiation on trilayer graphene. Furthermore, an aluminum the substrate limitation.[4] For instance,
nitride buffer/interlayer leads to unique metal polarity due to the formation transferable III-nitride light-emitting
of an AlON thin layer in a growth environment containing trace amounts of diodes (LEDs)[5] and micro-LEDs[6] have
oxygen, which explains the open question of why those reported wurtzite been implemented on graphene, and
transferable high-electron-mobility tran-
III-nitride films on 2D materials always exhibit metal polarity. The reported
sistors have been prepared on hexagonal
atomic modulation through interface manipulation provides an effective boron nitride.[7] Notably, these devices can
model for hexagonal nitride semiconductor layers grown on graphene, which be transferred from the epitaxial substrate
definitely promotes the development of novel semiconductor devices. to any other desired substrate with high
heat dissipation capacity[8] or flexibility[9]

F. Liu, T. Shen, X. Rong, B. Sheng, L. Yang, D. Li, J. Wei, S. Sheng, X. Li, Y. Yuan
Z. Chen, R. Tao, X. Yang, F. Xu, K. Liu, X.-Z. Li, B. Shen, X. Wang Songshan Lake Materials Laboratory
State Key Laboratory for Mesoscopic Physics and Frontiers Science Dongguan, Guangdong 523808, China
Center for Nano-optoelectronics K. Liu, X.-Z. Li, B. Shen, X. Wang
School of Physics Collaborative Innovation Center of Quantum Matter
Peking University Peking University
Beijing 100871, China Beijing 100871, China
E-mail: [email protected]; [email protected]; [email protected] K. Liu, X.-Z. Li
T. Wang, J. Zhang Interdisciplinary Institute of Light-Element Quantum Materials
Electron Microscopy Laboratory Research Center for Light-Element Advanced Materials
School of Physics Peking University
Peking University Beijing 100871, China
Beijing 100871, China X.-Z. Li, X. Wang
Z. Zhang Peking University Yangtze Delta Institute of Optoelectronics
Beijing Advanced Innovation Center for Materials Genome Engineering Nantong, Jiangsu 226010, China
Beijing Key Laboratory for Magneto-Photoelectrical Composite
and Interface Science
Institute for Multidisciplinary Innovation
University of Science and Technology Beijing
Beijing 100083, China
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adma.202106814.

DOI: 10.1002/adma.202106814

Adv. Mater. 2022, 34, 2106814 2106814 (1 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

and so on[10] by breaking the weak coupling interface through semiconductor films grown on layered graphene through inter-
a simple mechanical separation method. These results demon- face manipulation, thereby promoting the industrial develop-
strate an effective method of combining commercial production ment of novel semiconductor devices.
equipment in the III-nitride industry, such as metal−organic
chemical vapor deposition (MOCVD), to intelligently manufac-
ture flexible and/or high-power devices with low cost and high 2. Results and Discussion
yield.[11] The breakthrough of epitaxial growth of III-nitrides
on 2D materials mainly resides on the fine modulation of the In this work, we use plasma-assisted molecular beam epitaxy
weak coupling interface formed by two different hybridized sys- (MBE) to explore the growth behavior of wurtzite GaN films on
tems,[12] which, in principle, can achieve single-crystal epilayers graphene and thus to build a proper GaN/graphene interface
with preferred atom configurations (i.e., lattice polarity).[13] platform. Instead of a graphene monolayer, three-monolayer-
However, it has not been experimentally confirmed in the case thick single-crystal graphene is used to shield the interaction
of III-nitride epitaxy on 2D materials until now. between the epitaxial GaN film and the Al2O3 substrate.[3,17]
It is known that the sp2-hybridized graphene surface lacks The schematic diagram of the epitaxial growth of GaN films
covalent dangling bonds that can serve as nucleation sites for on graphene is shown in Figure 1a, which includes four steps:
sp3-hybridized III-nitrides.[1,2] Perfect vdW epitaxy actually graphene transfer, pre-irradiation, GaN nucleation and GaN
makes the growth of relatively thick (for example, several-µm- growth. First, three graphene monolayers were sequentially
thick) III-nitride layers, such as GaN films and device struc- transferred to the surface of the Al2O3(0001) substrate. Raman
tures, very difficult since epitaxial films are easily peeled off scattering measurements (Figure S2a, Supporting Information)
during growth due to the weak interactions at the interface. confirmed the fine structural quality of the transferred trilayer
Therefore, epitaxy of wafer-scale III-nitrides is likely quasi- graphene, where the scattering peak G at 1581.9 cm−1 and peak
vdW epitaxy, requiring defects with covalent dangling bonds; 2D at 2693.1 cm−1 derived from graphene are obviously stronger
however, a lack of sufficient defects often leads to the growth than peak D at 1347.0 cm−1 arising from structural defects.[18]
of nanostructures rather than films.[14] To solve this problem, Then, atomic O pre-irradiation was applied to partially destroy
it is necessary to form some artificial dangling bonds on the the hexagonal carbon (C) lattice (Figure S2b, Supporting Infor-
surface of graphene to improve the nucleation capability of the mation), thereby forming unsaturated CO dangling bonds on
epitaxial film.[24,25] The quasi-vdW epitaxy of III-nitride films on the graphene surface (Figure S2c,d, Supporting Information).
2D materials seems to have become a paradigm by first using Subsequently, by providing Ga atoms and N atoms required for
a pretreatment method such as atomic nitrogen (N) pre-irradi- epitaxy, GaN nucleation started from the artificial CO dan-
ation to form artificial dangling bonds on the surface of gra- gling bonds on the pre-irradiated graphene surface. Finally,
phene and then adopting a high-nucleation-ability AlN buffer GaN nucleation islands continued to grow, coalescence, and
layer to improve the crystal quality of the film.[7,15,16] The epi- eventually form a continuous film with an ordered hexagonal
taxial layer and graphene are actually connected in a mixed lattice arrangement. Herein, 1 µm-thick single-crystal wurtzite
form of covalent bonds and vdW bonds. This paradigm, on GaN films were grown on the transferred single-crystal trilayer
the one hand, provides a unique opportunity to manipulate graphene on Al2O3(0001) substrates. It was found that there was
the interface and thus tune the growth and, on the other hand, a 2 nm-thick gap layer composed of in situ atomic O pre-irra-
leads to an open question of why those films have unique metal diated trilayer graphene, which completely separated the epi-
lattice polarity and thus lose freedom to fabricate N-polarity taxial GaN film from the Al2O3 substrate (Figure S3, Supporting
based opto-electronic devices. Information). Then, X-ray diffraction (XRD) measurements
Here, we report experimental evidence for the quasi-vdW were performed to evaluate the lattice arrangement of these epi-
epitaxial growth of high-quality single-crystal GaN films on taxial GaN films. In the XRD 2θ−ω scan, the diffraction peak
single-crystal graphene with controllable atomic configura- corresponding to the GaN (0002) plane at 34.4° was observed
tion. For this hexagonal semiconductor, there are two kinds of (Figure 1b). It indicates the ordered out-of-plane orientation of
atomic configurations along the c-direction due to the broken the GaN films. In the φ-scan for the (1015) plane of GaN, six
inversion symmetry, that is, Ga- and N-lattice polarity, where peaks with an interval of 60° confirm its in-plane orientation
each Ga atom exhibits 1 or 3 dangling bonds on the growing and hexagonal symmetry (Figure 1c). The combination of out-
surface, respectively (Figure S1, Supporting Information). For of-plane and in-plane orientations proves that these epitaxial
the first time, these lattice polarities are successfully controlled GaN layers all have a regular hexagonal lattice arrangement.[19]
by forming CONGa(3) configuration (for Ga-polarity GaN It is worth noting that the epitaxial relation between GaN and
film) or COGaN(3) configuration (for N-polarity GaN single-crystal graphene is (0001) [1010]GaN || (0001) [1120]graphene,
film) on artificial CO surface dangling bonds through atomic as shown in Figure S4a–c, Supporting Information. Moreover,
oxygen (O) irradiation on transferred trilayer graphene. Due the full widths at half maximum of the XRD ω-rocking curves
to the universal applicability of this lattice polarity modulation for the (0002) and (1012) planes of the 1 µm-thick GaN film
method, it is suitable for different epitaxy methods. Further- grown on the pre-irradiated trilayer graphene template are 0.29°
more, we confirmed that a few-nanometer-thick AlN buffer/ and 0.35°, respectively (Figure S4e, Supporting Information),
interlayer leads to unique metal polarity due to the formation which are comparable to those of the film grown directly on the
of an AlON ultrathin layer, which explains why the reported III- Al2O3 substrate. These results indicate that the transferred gra-
nitride films often exhibit metal polarity. This work can be seen phene layer provides the possibility of fabricating single-crystal
as an effective model of atomic modulation of hexagonal nitride III-nitride films with high crystal quality.

Adv. Mater. 2022, 34, 2106814 2106814 (2 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

Figure 1. Epitaxial growth of a hexagonal semiconductor GaN film on layered graphene. a) Schematic diagram of the growth process of wurtzite GaN
films on graphene (Gr), including transfer of Gr (I), pre-irradiation (II), GaN nucleation (III), and GaN growth (IV). b) XRD 2θ–ω scan of a 1 µm-thick
GaN film grown on a trilayer graphene/Al2O3(0001) substrate by MBE. c) XRD azimuthal off-axis φ-scan for the GaN (1015) reflection confirming its
sixfold symmetry.

Next, we focus on this GaN/graphene epitaxial interface to bonding paths of GaN on graphene are shown in Figure 2b.
explore the bonding mechanism of different hybridized mate- The COGa bond is the strongest (−3.017 eV), indicating
rials. The interface bonding of hexagonal semiconductor GaN that graphene tends to initiate GaN epitaxy by binding one Ga
and graphene mainly depends on two steps shown in Figure 1a atom on a CO dangling bond. The CON bond has the
of pre-irradiation and GaN nucleation, which actually deter- second strongest binding energy (−1.967 eV), which means that
mine the bonding sequence of Ga or N atoms on the CO although the CON bond is not as stable as the COGa
surface dangling bonds on graphene, thereby forming spe- bond, it can still be formed. In contrast, due to the positive
cific interface atomic configurations as initial GaN nucleation binding energy values, the CN bond (1.762 eV) and CGa
islands. Different from the artificial CN dangling bonds on bond (2.667 eV) are structurally unstable and cannot be formed
the graphene surface, which always tend to first bind to Ga in principle. Next, the nucleation behavior of GaN on the pre-
atoms to form a C−N−Ga−N(3) configuration and thus start the ferred COGa and CON bonds was discussed (Figure 2c).
epitaxy of N-polarity GaN,[20] the unsaturated O atoms in the The Ga atom in the COGa bond has three unsaturated
CO dangling bonds provide more GaN nucleation options. bonds, and it is most likely to trigger the nucleation growth
Density functional theory (DFT) calculations based on this sim- of N-polarity GaN on atomic O pre-irradiated graphene by
plified structure (Figure 2a) were performed to clarify the inter- forming the C−O−Ga−N(3) configuration. The difference is
face bonding path and explore the feasible lattice polarity mod- that the N atom in the CON bond tends to adsorb three Ga
ulation of the epitaxial GaN film on atomic O pre-irradiated atoms to form the C−O−N−Ga(3) configuration and thus to ini-
graphene. The essence of this simplified structure is the chem- tiate the epitaxy of Ga-polarity GaN on atomic O pre-irradiated
ical adsorption model of Ga or N atoms on the CO dangling graphene.
bonds on the surface of monolayer graphene. More details are Following this principle, we developed a lattice polarity
provided in the Tables S1 and S2, Supporting Information, and modulation model for epitaxial GaN films on graphene. The
Experimental Section. Four possible interface atomic configura- core technology is in situ atomic O irradiation of graphene in
tions were considered: 1) one Ga atom combines with the O the MBE chamber at 500 °C. The optimization of the process
atom of the CO bond (COGa), 2) one N atom combines parameters was briefly discussed elsewhere (Figure S2b, Sup-
with the O atom of the CO bond (CON), 3) one Ga atom porting Information). By prioritizing Ga atoms rather than N
replaces the O atom of the CO bond (CGa), and 4) one N atoms during high-temperature (HT) GaN growth, the nuclea-
atom replaces the O atom of the CO bond (CN). The cal- tion and film growth of N-polarity GaN film on pre-irradiated
culated binding energy values of the four possible interfacial graphene were realized. In detail, the shutter in front of the

Adv. Mater. 2022, 34, 2106814 2106814 (3 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

Figure 2. Theoretical calculation of the hexagonal semiconductor GaN/graphene interface bonding configuration. a) Top-view and side-view diagrams
of the optimized structure of a single O atom chemisorbed on a graphene monolayer. b) Calculated binding energy values of various cases of one Ga
or N atom absorption on a monolayer graphene surface through an unsaturated CO bond. c) Schematic diagram of the lattice polarity modulation
mechanism of wurtzite GaN films grown on graphene through the preferred interface configuration, where the C−O−N configuration tends to adsorb
three Ga atoms to from C−O−N−Ga(3) configuration, and the C−O−Ga configuration tends to adsorb three N atoms to from C−O−Ga−N(3) configu-
ration. Herein, the C−O−N−Ga(3) configuration always corresponds to Ga-polarity GaN, and the C−O−Ga−N(3) configuration always corresponds to
N-polarity GaN.

Ga source was opened in advance to obtain a 1 nm-thick Ga exhibit N-polarity due to the dissolution of initial C−O−N/C−
adsorption layer to protect the epitaxial interface from the O−N−Ga(3) configuration and the formation of other interface
destruction of N atoms provided by the N-plasma source. bonding such as C−O−Ga−N(3) between GaN and graphene.
Then, the N-plasma source was stricken, and the epitaxy of To probe the interface bonding structure and the origin of the
GaN started at 780 °C. Its polarity was confirmed by the typical lattice polarity of these GaN films on the graphene template,
(3 × 3) reconstructed reflection high-energy electron diffraction high-resolution scanning transmission electron microscopy
(RHEED) patterns (Figure S4b, Supporting Information) and (STEM) measurements were performed on the N-polarity GaN
the serious deterioration of the surface morphology (Figure 3a) film (Figure 3c) and the Ga-polarity GaN film (Figure 3h).
after etching with sodium hydroxide (NaOH) solution.[21] In For the N-polarity one, a space gap originated from 3-ML-
contrast, by opening the shutter in front of the N-plasma thick graphene between the GaN film and the Al2O3 substrate
source in advance and adopting a 5 nm-thick low-temperature (Figure 3b). The integrated differential phase contrast (iDPC)
(LT) GaN buffer layer at 500 °C, the epitaxy of Ga-polarity GaN STEM images of GaN near the interface region are shown in
film at 780 °C was also achieved on pre-irradiated trilayer gra- Figure 3d, where the array of Ga and N atoms along the [0001]
phene/Al2O3 template (Figure S4f, Supporting Information). growth direction indicates the N lattice polarity. Moreover, as
The Ga-polarity was confirmed by the (1 × 1) unconstructed shown in Figure 3e, this N-polarity lattice arrangement is likely
RHEED patterns (Figure S4c, Supporting Information) and to start from the GaN/graphene interface with a discontinuous
the stable surface morphology after etching with NaOH solu- monolayer of O atoms (i.e., the CO dangling bonds on trilayer
tion (Figure 3f). It should be pointed out that the quasi-vdW graphene surface).[23] This is consistent with the calculation
epitaxy of Ga-polarity films requires a more complicated pro- results of N-polarity GaN on graphene through the expected
cess, due to the fact that the C−O−N configuration is not stable C−O−Ga−N(3) configuration (Figure 2c). The following STEM
enough (Figure 2b). Therefore, a LT-GaN buffer layer is nec- measurements of the Ga-polarity film on trilayer graphene
essary to protect the C−O−N configuration and subsequent further confirm the interfacial atomic model based on the cal-
C−O−N−Ga(3) configuration against being broken at high culation results (Figure 3g,i). Notably, the Ga lattice polarity of
temperature and thus protect the upper Ga-polarity GaN lat- the film results from the GaN/graphene interface, where GaN
tice arrangement.[22] Otherwise, the GaN film may eventually is most likely to bond with the uppermost graphene through a

Adv. Mater. 2022, 34, 2106814 2106814 (4 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

Figure 3. Exploration of the interface atomic structure of epitaxial GaN films on graphene. a) AFM images of a 1 µm-thick epitaxial GaN film grown
on a trilayer graphene/Al2O3(0001) template via the expected C−O−Ga−N(3) configuration before and after NaOH solution etching. The surface dete-
rioration phenomenon during etching is due to the high surface chemical activity of the N-polarity film. b,c) iDPC-STEM and HAADF-STEM images of
the interface atomic structure of this N-polarity film. The adhesion between the graphene and the Al2O3 substrate is not perfect, causing the graphene
surface to fluctuate, which leads to the tilt of the GaN lattice in some areas. d,e) iDPC-STEM images of the lattice arrangement of this N-polarity GaN
film. f) AFM images of a 1 µm-thick Ga-polarity GaN film on a trilayer graphene/Al2O3 template through the expected C−O−N−Ga(3) configuration
before and after NaOH solution etching. g,h) iDPC-STEM and HAADF-STEM images of the interface atomic structure of this Ga-polarity film. The
slightly blurred interface may come from structural damage caused by the high-energy electron beam during STEM measurement. i,j) iDPC images of
the lattice arrangement of this Ga-polarity film. Herein, these images (b–e,g–j) viewing both along the [1120] orientation of GaN. The [000 1] growth
direction of N-polarity GaN is marked as the −c direction, and the [0001] growth direction of Ga-polarity GaN is marked as the +c direction.

discontinuous O atomic layer (Figure 3j), that is, the expected expected interface atomic configuration and prevent it from
C−O−N−Ga(3) configuration. Therefore, theoretical prediction being destroyed at higher growth temperatures. In addition,
(Figure 2) coincides with the experimental results (Figure 3 and we can achieve the separation of quasi-vdW epitaxial GaN film
Figure S4, Supporting Information), which indicates that the from the Al2O3 substrate with the help of thermal release tape
hexagonal lattice of the epitaxial GaN film is determined by the (Figure S4f,g, Supporting Information). Because of this lat-
three-monolayer-thick single-crystal graphene layer rather than tice polarity modulation model of epitaxial GaN on graphene
the substrate and that the lattice polarity is modulated by the through interfacial atomic configuration rather than the lat-
interfacial atomic configuration between GaN and graphene. tice arrangement of the substrate, we were able to achieve the
Thus far, we have achieved the quasi-vdW epitaxy of wurtzite desired GaN-based epitaxial structures for multifunctional
GaN films and established a lattice polarity modulation model devices on various substrates.
of epitaxial GaN films on graphene through interfacial atomic Considering that our MBE system is used for the epitaxy
configuration. of nitride and oxide semiconductors, the oxygen content in
Notably, this model does not depend on the epitaxy tech- the chamber may be slightly higher than that of the conven-
nique. Herein, we also realized the epitaxy of both Ga- and tional MBE system. From this viewpoint, the growth ambient
N-polarity GaN films on atomic O-pre-irradiated trilayer single- is somehow close to that of MOCVD. Finally, we would like
crystal graphene through MOCVD (Figure S5, Supporting to answer the open question of why the GaN layer grown by
Information). In this case, we used ammonia (NH3) as the N MOCVD on graphene and some other 2D materials with the
source and a relatively thick LT-GaN buffer layer to protect the help of an AlN buffer layer always exhibits Ga lattice polarity.

Adv. Mater. 2022, 34, 2106814 2106814 (5 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

Figure 4. Lattice polarity reversal effect of the AlN interlayer on the upper III-nitride layer. a) HAADF-STEM image of the epitaxial GaN/AlN/GaN sand-
wich structure on an atomic O pre-irradiated trilayer graphene/Al2O3(0001) template through the expected C−O−Ga−N(3) configuration. b–d) iDPC-
STEM images of the bottom GaN (I) and upper GaN (II) layers, where the bottom layer has N-polarity and the upper layer has Ga-polarity. e) An
iDPC-STEM image of the AlN/GaN interface area, showing the lattice polarity inversion phenomenon existing in the AlN insertion layer. f) A HAADF-
STEM image and g) an EDX line scan analysis performed on the AlN/GaN interface showing the enrichment of O atoms in the polarity inversion region
of the AlN interlayer. h) A HAADF-STEM image and i) a schematic diagram of the AlN layer above the GaN interface, including the planar inversion
domain boundary. Since it is impossible to accurately confirm the ratio of O and N atoms at each position, the N lattice positions in the ideal AlN are
filled with two-color small balls to represent AlON. The yellow parts of these Al atom balls indicate possible vacancies. Herein, these images (a–f,h)
viewing both along the [1120] orientation of GaN.

To eliminate the disturbance of the imperfect graphene surface in the white area and GaN in the gray area) to ≈0.08 (AlN in
and slightly disordered AlN nucleation conditions, we designed the gray area), as depicted in Figure 4g. However, the atomic N
an experiment using a GaN/AlN/GaN sandwich structure fraction and atomic O fraction exhibited opposite trends with
grown on an atomic O-pre-irradiated trilayer graphene/Al2O3 the spatial position. Even at three specific points of AlN in the
template by plasma-assisted MBE (Figure 4a). Here, before gray area, the atomic O fraction obviously increased to ≈0.20,
AlN epitaxy, an 100 nm-thick GaN layer was grown on the and the atomic N fraction decreased to ≈0.14 (marked by the
transferred trilayer graphene/Al2O3 template by the expected black arrow). The enriched O atoms will tend to partially replace
C−O−Ga−N(3) configuration to obtain an atomic-scale clear the N atoms in the hexagonal AlN lattice to form a thin AlON
AlN/GaN interface. According to the lattice polarity modulation layer. The 5-monolayer-thick O-rich AlN layer with an obvious
model of epitaxial GaN discussed above, this structure should contrast difference with the surroundings (Figure 4h) further
has N lattice polarity, similar to the results shown in Figure 3d. proved this viewpoint, where the upper AlN has Al-polarity, and
However, this method does not seem to be suitable for this GaN/ the bottom AlN has N-polarity. Considering the structure of
AlN/GaN sandwich structure because the bottom GaN has this AlON thin layer,[22,24] we can propose a reasonable lattice
N-polarity (Figure 4d), while the upper GaN has Ga polarity polarity reversal model in the AlN interlayer (Figure 4i). Since
(Figure 4b,c). In other words, the AlN interlayer reverses the lat- the precise ratio of O and N components is difficult to directly
tice polarity of GaN from N lattice polarity to Ga lattice polarity. determine, we used two-color balls to represent nonmetallic
Further iDPC-STEM measurements (Figure 4e) show that the atoms. For N-polarity AlN, the higher surface chemical activity
lattice polarity of the AlN interlayer is indeed reversed from facilitates the incorporation of higher fractions of O atoms,[25]
N- to Al-polarity near the spatial position about 3 nm above the which in turn forms a few-nm-thick AlON layer and thus
AlN/GaN interface. To illustrate this unusual lattice polarity reverses the lattice polarity of the upper AlN to the relatively
reversal mechanism, high-angle annular dark-field (HAADF) chemically inert metal polarity. The metal polarity in the AlN
STEM and energy-dispersive X-ray spectroscopy (EDX) meas- layer is mainly related to the incorporation of O atoms from
urements were performed on this AlN/GaN interface. As shown the epitaxial environment, epitaxial substrates, and source
in Figure 4f, there is obvious enrichment of the O composition materials, which is somehow free of growth equipment. It has
near the lattice polarity inversion region in AlN compared with been reported that the MOCVD-grown AlN layer on pretreated
GaN or upper AlN, as confirmed by the contrast difference. An graphene initially has nitrogen lattice polarity through the CN
EDX line scan performed along the marked direction shows dangling bond to form C−N−Al−N(3) configuration, and then
that the average atomic O fraction increased from ≈0.03 (AlN reverses to the metal lattice polarity by forming a thin AlON

Adv. Mater. 2022, 34, 2106814 2106814 (6 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

layer.[15,19] In other words, in MOCVD growth, a small amount periodic images. The lattice parameters of the graphene unit cell were
of oxygen will lead to the formation of AlON layer which locks a = b = 2.468 Å, α = β = 90°, and γ = 120°. More detailed information is
metal lattice polarity of the upper epilayer. Therefore, the lat- provided in Section S6, Supporting Information.
MBE Growth of Wurtzite GaN Films: 1) N-polarity GaN films grown on
tice polarity modulation of epitaxial GaN on graphene reported atomic O pre-irradiated trilayer graphene. In the MBE growth chamber,
in this manuscript provides freedom for fabricating multifunc- about 3-ML Ga atoms were first supplied by a thermal evaporation
tional devices based on III-nitrides with both kinds of lattice source before HT-GaN epitaxy at 780 °C. Then, the N-plasma source was
polarities. stricken, and bottom GaN film was deposited at 780 °C. The deposition
process is described as follows: turn on the shutters of Ga source for
6 s, the corresponding beam equivalent pressure was 5.5 × 10−7 mbar;
then turn on the shutter of N-plasma source for 30 s to provide
3. Conclusion sufficient N atoms to combine with Ga atoms to form bottom GaN (the
N2 flow rate was 0.25 sccm, and the plasma power was 220 W); repeat
We have demonstrated the lattice polarity manipulation of the above steps ten times (about 10-monolayer-thick GaN) to protect
quasi-vdW epitaxial GaN on graphene by interface atomic con- the expected C−O−Ga−N(3) configuration. Finally, 1 µm-thick N-polarity
figuration engineering. By using atomic O pre-irradiation and GaN films were deposited at 780 °C with a growth rate of 3 nm min−1.
atomic Ga and N supply sequences to form the C−O−N−Ga(3) During GaN epitaxy, the N2 flow rate was 0.8 sccm, the plasma power
was 350 W, and the beam equivalent pressure of the Ga atomic beam
and C−O−Ga−N(3) configurations, Ga- and N-lattice polarity
was 5.5 × 10−7 mbar. 2) Ga-polarity GaN films grown on atomic O pre-
GaN films were achieved on transferred graphene, respectively. irradiated trilayer graphene. According to theoretical guidance, 5 s of
This polarity-control rule is not affected by growth methods gentle atomic nitrogen treatment was first performed at 500 °C to form
such as MBE and MOCVD and is free of either crystalline or an expected CON configuration. Then, a LT-GaN buffer/protection
noncrystalline substrates. It makes epitaxy of wurtzite III- layer (5 nm-thick) was finely deposited at 500 °C, in which the shutter
nitrides with preferred lattice polarity possible on graphene in front of the Ga source was opened 3 s earlier than that of the
according to objective demands and definitely improves the N-plasma sources. During the abovementioned steps, the N2 flow rate
was 0.25 sccm, the radio frequency plasma power was 220 W, and the
capability to fabricate more kinds of devices with multiple func- beam equivalent pressure of the Ga atomic beam was 5.0 × 10−8 mbar.
tions, which is believed to play an important role in next-gener- The small N atomic beam was mainly used to protect these fragile
ation semiconductor devices. interface states on the graphene surface. Considering the influence of
growth temperature on the thermal evaporation rate of Ga atoms on the
graphene surface, the Ga beam flux here for LT-GaN epitaxy was also
smaller. Subsequently, 1 µm-thick Ga-polarity GaN films were deposited
4. Experimental Section at 780 °C. During GaN deposition, the N2 flow rate was 0.8 sccm, the
Graphene Synthesis and Transfer: Monolayer single-crystal graphene plasma power was 350 W, and the beam equivalent pressure of the Ga
was grown on a single-crystal Cu(111) substrate by chemical vapor atomic beam was 5.5 × 10−7 mbar. 3) Epitaxy of AlN interlayer. The AlN
deposition. The growth of graphene was conducted at 1030 °C. The interlayer was deposited at 780 °C, and its growth rate was 3 nm min−1.
detailed procedure is described in the previous report.[26] Then, trilayer During AlN deposition, the N2 flow rate was 0.8 sccm, the plasma power
graphene templates were achieved by the poly(methyl methacrylate) was 350 W, and the beam equivalent pressure of the Al atomic beam was
(PMMA)-mediated transfer method. Monolayer graphene/Cu was 2.5 × 10−7 mbar.
first spin-coated with PMMA (950 K, 4.0 wt% in ethyl lactate). The Characterization: The structural quality of transferred graphene was
PMMA/graphene film was pasted onto another graphene/Cu sample characterized by Raman scattering spectroscopy using a 532 nm laser
after etching the Cu substrate (etchant: 4.0 wt% (NH4)2S2O8 aqueous (CNI MSL-200 mW) as the excitation source and a HORIBA-iHR550
solution). By repeating the above process again and then removing spectrometer. XPS (Kratos, Axis Supra) was performed to quantitatively
the Cu substrate, free-standing PMMA/trilayer graphene was obtained. estimate the chemical composition of graphene before and after atomic
Finally, the dried PMMA/trilayer graphene was transferred onto the oxygen activation. The surface morphology of epitaxial GaN film on the
Al2O3(0001) substrate, and the PMMA was removed by hot acetone graphene template and the morphology of the GaN film after etching by
(70 °C). NaOH solution (10%, AR) at 60 °C for 20 min were measured by AFM in
Graphene Pre-Irradiation: The transferred graphene templates were tapping mode (Bruker Dimension ICON-PT). High-resolution spherical
kept at 500 °C in a plasma-assisted MBE chamber with an O-plasma aberration-corrected STEM characterization was performed using a FEI
source. Then, 3 min of atomic O pre-irradiation was performed with an Titan Cubed Themis G2 300.
O2 flow rate of 0.8 standard cubic centimeters per minute (sccm) and
a radio frequency plasma power of 250 W to form unsaturated CO
dangling bonds on the trilayer graphene surface as nucleation sites for
wurtzite GaN epitaxy (Figure S2b, Supporting Information). The CO Supporting Information
dangling bonds formed on the surface of pre-irradiated graphene can Supporting Information is available from the Wiley Online Library or
be first combined with Ga or N atoms, which makes possible the fine from the author.
modulation of the interfacial atomic configuration between the GaN film
and the graphene.
Theoretical Simulation: DFT calculations were implemented by
the Vienna ab initio simulation package code. Projector augmented Acknowledgements
wave pseudopotentials were used for electron–ion interactions.
The generalized gradient approximation was used for the exchange F.L., T.W., Z.H.Z., and T.S. contributed equally to this work. This work
correlation functional as proposed by Perdew–Burke–Ernzerhof. The was partly supported by the National Key R&D Program of China (No.
energy cutoff for plane wave expansion was 600 eV. In the calculation, a 2021YFA0716400), the Beijing Outstanding Young Scientist Program (No.
4 × 4 × 1 Γ centered k-point mesh was used, and spin polarization was BJJWZYJH0120191000103), the National Natural Science Foundation
considered. This monolayer graphene unit cell model contained a 20 Å of China (Nos. 61734001, 61521004, 52025023, 11774003, 11934003,
vacuum layer along the c-axis direction to avoid any interactions between 61904002, and 62104010), the Beijing Natural Science Foundation (Nos.

Adv. Mater. 2022, 34, 2106814 2106814 (7 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

Z200004 and JQ19004), the Guangdong Key Research and Development J. H. Ryu, Y. S. Katharria, C. J. Choi, C. H. Hong, Nat. Commun.
Program (Nos. 2019B010132001, 2020B010189001, 2019B010931001, 2013, 4, 1452.
and 2018B030327001), the Pearl River Talent Recruitment Program [6] J. Jeong, Q. X. Wang, J. Cha, D. K. Jin, D. H. Shin, S. Kwon,
of Guangdong Province (No. 2019ZT08C321), the China National B. K. Kang, J. H. Jang, W. S. Yang, Y. S. Choi, J. Yoo, J. K. Kim,
Postdoctoral Program for innovative Talents (No. BX2021007) and C. H. Lee, S. Lee, A. A. Zakhidov, S. Hong, M. J. Kim, Y. J. Hong, Sci.
the China Postdoctoral Science Foundation (Nos. 2020T130023 and Adv. 2020, 6, eaaz5180.
2020M680234). X.W. thanks Prof. Weikun Ge (Tsinghua University) for [7] a) N. R. Glavin, K. D. Chabak, E. R. Heller, E. A. Moore,
helpful discussions. The computational resources were provided by the T. A. Prusnick, B. Maruyama, D. E. Walker, D. L. Dorsey, Q. Paduano,
High-Performance Computing Platform of Peking University, China.
M. Snure, Adv. Mater. 2017, 29, 1701838; b) M. J. Motala,
E. W. Blanton, A. Hilton, E. Heller, C. Muratore, K. Burzynski,
J. L. Brown, K. Chabak, M. Durstock, M. Snure, N. R. Glavin, ACS
Conflict of Interest Appl. Mater. Interfaces 2020, 12, 21837.
[8] T. Ayari, S. Sundaram, X. Li, Y. El Gmili, P. L. Voss, J. P. Salvestrini,
The authors declare no conflict of interest. A. Ougazzaden, Appl. Phys. Lett. 2016, 108, 171106.
[9] G. Yang, Y. Jung, C. V. Cuervo, F. Ren, S. J. Pearton, J. Kim, Opt.
Express 2014, 22, A812.
[10] a) Y. Kobayashi, K. Kumakura, T. Akasaka, T. Makimoto, Nature
Data Availability Statement 2012, 484, 223; b) H. Kum, D. Lee, W. Kong, H. Kim, Y. Park, Y. Kim,
The data that support the findings of this study are available from the Y. Baek, S. H. Bae, K. Lee, J. Kim, Nat. Electron. 2019, 2, 439.
corresponding author upon reasonable request. [11] a) Y. Yu, T. Wang, X. F. Chen, L. D. Zhang, Y. Wang, Y. F. Niu,
J. Q. Yu, H. T. Ma, X. M. Li, F. Liu, G. Q. Deng, Z. F. Shi,
B. L. Zhang, X. Q. Wang, Y. T. Zhang, Light: Sci. Appl. 2021, 10,
117; b) Y. Qi, Y. Y. Wang, Z. Q. Pang, Z. P. Dou, T. B. Wei, P. Gao,
Keywords S. S. Zhang, X. Z. Xu, Z. H. Chang, B. Deng, S. L. Chen, Z. L. Chen,
epitaxial growth, interface atomic configuration, lattice polarity, layered H. N. Ci, R. Y. Wang, F. Z. Zhao, J. C. Yan, X. Y. Yi, K. H. Liu,
graphene, wurtzite gallium nitride H. L. Peng, Z. Q. Liu, L. M. Tong, J. Zhang, Y. J. Wei, J. M. Li,
Z. F. Liu, J. Am. Chem. Soc. 2018, 140, 11935; c) X. Li, S. Sundaram,
Received: August 29, 2021 Y. El Gmili, T. Ayari, R. Puybaret, G. Patriarche, P. L. Voss,
Revised: October 30, 2021 J. P. Salvestrini, A. Ougazzaden, Cryst. Growth Des. 2016, 16,
Published online: December 9, 2021 3409.
[12] a) D. D. Liang, T. B. Wei, J. X. Wang, J. M. Li, Nano Energy 2020, 69,
104463; b) J. D. Yu, L. Wang, Z. B. Hao, Y. Luo, C. Z. Sun, J. Wang,
Y. J. Han, B. Xiong, H. T. Li, Adv. Mater. 2020, 32, 1903047.
[1] a) A. K. Geim, K. S. Novoselov, Nat. Mater. 2007, 6, 183; b) X. Z. Xu, [13] a) M. Hetzl, M. Kraut, T. Hoffmann, M. Stutzmann, Nano Lett. 2017,
Z. H. Zhang, L. Qiu, J. N. Zhuang, L. Zhang, H. Wang, C. N. Liao, 17, 3582; b) P. Wang, Y. Yuan, C. Zhao, X. Q. Wang, X. T. Zheng,
H. D. Song, R. X. Qiao, P. Gao, Z. H. Hu, L. Liao, Z. M. Liao, X. Rong, T. Wang, B. W. Sheng, Q. X. Wang, Y. Q. Zhang, L. F. Bian,
D. P. Yu, E. G. Wang, F. Ding, H. L. Peng, K. H. Liu, Nat. Nano- X. L. Yang, F. J. Xu, Z. X. Qin, X. Z. Li, X. X. Zhang, B. Shen, Nano
technol. 2016, 11, 930; c) Y. Kubota, K. Watanabe, O. Tsuda, Lett. 2016, 16, 1328.
T. Taniguchi, Science 2007, 317, 932; d) H. Yu, M. Z. Liao, W. J. Zhao, [14] a) H. Yoo, K. Chung, Y. S. Choi, C. S. Kang, K. H. Oh, M. Kim,
G. D. Liu, X. J. Zhou, Z. Wei, X. Z. Xu, K. H. Liu, Z. H. Hu, K. Deng, G. C. Yi, Adv. Mater. 2012, 24, 515; b) I. M. Hoiaas, A. Liudi Mulyo,
S. Y. Zhou, J. A. Shi, L. Gu, C. Shen, T. T. Zhang, L. J. Du, L. Xie, P. E. Vullum, D. C. Kim, L. Ahtapodov, B. O. Fimland, K. Kishino,
J. Q. Zhu, W. Chen, R. Yang, D. X. Shi, G. Y. Zhang, ACS Nano 2017, H. Weman, Nano Lett. 2019, 19, 1649; c) V. Kumaresan, L. Largeau,
11, 12001; e) A. Koma, Thin Solid Films 1992, 216, 72. A. Madouri, F. Glas, H. Z. Zhang, F. Oehler, A. Cavanna, A. Babichev,
[2] a) K. Chung, C. H. Lee, G. C. Yi, Science 2010, 330, 655; b) J. W. Ben, L. Travers, N. Gogneau, M. Tchernycheva, J. C. Harmand, Nano
X. K. Liu, C. Wang, Y. P. Zhang, Z. M. Shi, Y. P. Jia, S. L. Zhang, Lett. 2016, 16, 4895; d) S. Fernandez-Garrido, M. Ramsteiner,
H. Zhang, W. J. Yu, D. B. Li, X. J. Sun, Adv. Mater. 2021, 33, 2006761; G. H. Gao, L. A. Galves, B. Sharma, P. Corfdir, G. Calabrese,
c) J. H. Park, X. Yang, J. Y. Lee, M. D. Park, S. Y. Bae, M. Pristovsek, Z. D. Schiaber, C. Pfuller, A. Trampert, J. M. J. Lopes, O. Brandt,
H. Amano, D. S. Lee, Chem. Sci. 2021, 12, 7713. L. Geelhaar, Nano Lett. 2017, 17, 5213; e) J. Jeong, D. K. Jin, J. Cha,
[3] W. Kong, H. S. Li, K. Qiao, Y. Kim, K. Lee, Y. F. Nie, D. Lee, B. K. Kang, Q. X. Wang, J. Choi, S. W. Lee, V. Y. Mikhailovskii,
T. Osadchy, R. J. Molnar, D. K. Gaskill, R. L. Myers-Ward, V. Neplokh, N. Amador-Mendez, M. Tchernycheva, W. S. Yang,
K. M. Daniels, Y. W. Zhang, S. Sundram, Y. Yu, S. H. Bae, S. Rajan, J. Yoo, M. J. Kim, S. Hong, Y. J. Hong, ACS Appl. Nano Mater.
Y. Shao-Horn, K. Cho, A. Ougazzaden, J. C. Grossman, J. Kim, Nat. 2020, 3, 8920; f) A. KovACS, M. Duchamp, R. E. Dunin-Borkowski,
Mater. 2018, 17, 999. R. Yakimova, P. L. Neumann, H. Behmenburg, B. Foltynski,
[4] a) J. Kim, C. Bayram, H. Park, C. W. Cheng, C. Dimitrakopoulos, C. Giesen, M. Heuken, B. Pecz, Adv. Mater. Interfaces 2015, 2,
J. A. Ott, K. B. Reuter, S. W. Bedell, D. K. Sadana, Nat. Commun. 1400230.
2014, 5, 4836; b) Y. Kim, S. S. Cruz, K. Lee, B. O. Alawode, C. Choi, [15] Z. L. Chen, Z. Q. Liu, T. B. Wei, S. Y. Yang, Z. P. Dou, Y. Y. Wang,
Y. Song, J. M. Johnson, C. Heidelberger, W. Kong, S. Choi, K. Qiao, H. N. Ci, H. L. Chang, Y. Qi, J. C. Yan, J. X. Wang, Y. F. Zhang,
I. Almansouri, E. A. Fitzgerald, J. Kong, A. M. Kolpak, J. Hwang, P. Gao, J. M. Li, Z. F. Liu, Adv. Mater. 2019, 31, 1807345.
J. Kim, Nature 2017, 544, 340; c) P. K. Mohseni, A. Behnam, [16] a) J. Ning, C. C. Yan, Y. Q. Jia, B. Y. Wang, Y. Zeng, J. C. Zhang,
J. D. Wood, X. Zhao, K. J. Yu, N. C. Wang, A. Rockett, J. A. Rogers, D. Wang, Y. Hao, ACS Appl. Nano Mater. 2020, 3, 5061; b) Y. Chen,
J. W. Lyding, E. Pop, X. L. Li, Adv. Mater. 2014, 26, 3755; d) H. Oh, H. Zang, K. Jiang, J. W. Ben, S. L. Zhang, Z. M. Shi, Y. P. Jia, W. Lu,
Y. J. Hong, K. S. Kim, S. Yoon, H. Baek, S. H. Kang, Y. K. Kwon, X. J. Sun, D. B. Li, Appl. Phys. Lett. 2020, 117, 051601; c) P. Wang,
M. Kim, G. C. Yi, NPG Asia Mater 2014, 6, e145. A. Pandey, J. Gim, W. J. Shin, E. T. Reid, D. A. Laleyan, Y. Sun,
[5] N. Han, T. V. Cuong, M. Han, B. D. Ryu, S. Chandramohan, D. H. Zhang, Z. Liu, Z. H. Zhong, R. Hovden, Z. T. Mi, Appl. Phys.
J. B. Park, J. H. Kang, Y. J. Park, K. B. Ko, H. Y. Kim, H. K. Kim, Lett. 2020, 116, 171905.

Adv. Mater. 2022, 34, 2106814 2106814 (8 of 9) © 2021 Wiley-VCH GmbH

 15214095, 2022, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202106814 by Beijing University Of, Wiley Online Library on [23/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de

[17] a) S. Zhang, B. Y. Liu, F. Ren, Y. Yin, Y. Y. Wang, Z. L. Chen, [21] G. Naresh-Kumar, J. Bruckbauer, A. Winkelmann, X. Yu,
B. Jiang, B. Z. Liu, Z. T. Liu, J. Y. Sun, M. Liang, J. C. Yan, B. Hourahine, P. R. Edwards, T. Wang, C. Trager-Cowan,
T. B. Wei, X. Y. Yi, J. X. Wang, J. M. Li, P. Gao, Z. F. Liu, Z. Q. Liu, R. W. Martin, Nano Lett. 2019, 19, 3863.
Small 2021, 17, 210098; b) F. Ren, B. Y. Liu, Z. L. Chen, Y. Yin, [22] S. Mohn, N. Stolyarchuk, T. Markurt, R. Kirste, M. P. Hoffmann,
J. Y. Sun, S. Zhang, B. Jiang, B. Z. Liu, Z. T. Liu, J. W. Wang, R. Collazo, A. Courville, R. Di Felice, Z. Sitar, P. Vennegues,
M. Liang, G. D. Yuan, J. C. Yan, T. B. Wei, X. Y. Yi, J. X. Wang, M. Albrecht, Phys. Rev. Appl. 2016, 5, 054004.
Y. Zhang, J. M. Li, P. Gao, Z. F. Liu, Z. Q. Liu, Sci. Adv. 2021, 7, [23] Z. P. Dou, Z. L. Chen, N. Li, S. Y. Yang, Z. W. Yu, Y. W. Sun, Y. H. Li,
eabf5011. B. Y. Liu, Q. Luo, T. B. Ma, L. Liao, Z. F. Liu, P. Gao, Nat. Commun.
[18] J. H. Lee, E. K. Lee, W. J. Joo, Y. Jang, B. S. Kim, J. Y. Lim, S. H. Choi, 2019, 10, 5013.
S. J. Ahn, J. R. Ahn, M. H. Park, C. W. Yang, B. L. Choi, S. W. Hwang, [24] H. Yang, J. G. Lozano, T. J. Pennycook, L. Jones, P. B. Hirsch,
D. Whang, Science 2014, 344, 286. P. D. Nellist, Nat. Commun. 2015, 6, 7266.
[19] Y. X. Feng, X. L. Yang, Z. H. Zhang, D. Kong, J. Zhang, K. H. Liu, [25] a) N. Stolyarchuk, T. Markurt, A. Courville, K. March, J. Zuniga-
X. Z. Li, J. F. Shen, F. Liu, T. Wang, P. F. Ji, F. J. Xu, N. Tang, T. J. Yu, Perez, P. Vennegues, M. Albrecht, Sci. Rep. 2018, 8, 14111;
X. Q. Wang, D. P. Yu, W. K. Ge, B. Shen, Adv. Funct. Mater. 2019, 29, b) D. Wang, K. Uesugi, S. Y. Xiao, K. Norimatsu, H. Miyake, Appl.
1905056. Phys. Express 2020, 13, 095501.
[20] F. Liu, Z. H. Zhang, X. Rong, Y. Yu, T. Wang, B. W. Sheng, [26] X. Z. Xu, Z. H. Zhang, J. C. Dong, D. Yi, J. J. Niu, M. H. Wu, L. Lin,
J. Q. Wei, S. Y. Zhou, X. L. Yang, F. J. Xu, Z. X. Qin, Y. T. Zhang, R. K. Yin, M. Q. Li, J. Y. Zhou, S. X. Wang, J. L. Sun, X. J. Duan,
K. H. Liu, B. Shen, X. Q. Wang, Adv. Funct. Mater. 2020, 30, P. Gao, Y. Jiang, X. S. Wu, H. L. Peng, R. S. Ruoff, Z. F. Liu, D. P. Yu,
2001283. E. G. Wang, F. Ding, K. H. Liu, Sci. Bull. 2017, 62, 1074.

Adv. Mater. 2022, 34, 2106814 2106814 (9 of 9) © 2021 Wiley-VCH GmbH