Mechanical Self-Assembly Technology for 2D Materials

Abstract

Self-assembled mechanical instabilities can offer a new technology roadmap for micro/nanopatterns of two-dimensional (2D) materials, which depends on the deterministic regulation of mechanical instability-induced self-assemblies. However, due to atomic thinness and ultra-low bending stiffness, different types of non-designable and non-deterministic multimode coupling mechanical instabilities, such as multimode-coupled crumpling, chaotic thermal-fluctuation-induced rippling, and unpredictable wrinkling, are extremely easy to be triggered in 2D materials. The above mode-coupled instabilities make it exceedingly difficult to controllably self-assemble 2D nanocrystals into designed morphologies. In this chapters, we will introduce a novel micro/nanopatterning technology of 2D materials based on mechanical self-assemblies. Firstly, a post-curing transfer strategy is proposed to fabricate multiscale conformal wrinkle micro/nanostructures of 2D materials. Secondly, we report a deterministic self-assembly for programmable micro/nanopatterning technology of atomically thin 2D materials via constructing novel 2D materials/IML/substrate trilayer systems. Finally, based on the micro/nanopatterning technology of 2D materials, we proposed a new fabrication method for the flexible micro/nano-electronics of deterministically self-assembled 2D materials including three-dimensional (3D) tactile and gesture sensors. We fundamentally overcome the key problem of self-assembly manipulation from randomness to determinism mode by decoupling mono-mode mechanical instability, providing new opportunities for programmable micro/nanopatterns of 2D materials. Moreover, mechanical instability-driven micro/nanopatterning technology enables simpler fabrication methods of self-assembled electronics based on 2D materials.

1. Introduction

Mechanical self-assemblies in two-dimensional (2D) materials have aroused great scientific interest due to their significant promotion of basic understanding of special physical and chemical phenomena, spanning selective chemical reactivity, abnormal thickness fluctuations, fusion-fission behaviors, and anisotropic friction [1, 2, 3, 4, 5, 6, 7], as well as exciting applications such as strain sensors, photodetectors, anticounterfeiting, and energy storage [8, 9, 10, 11, 12, 13]. Compared to three-dimensional (3D) bulk materials, the atomically thin and inherently planar 2D materials are more susceptible to being self-assembled into complex 3D architectures via rolling [14, 15], folding [16, 17], kirigami/origami [18, 19, 20], and rippling [21, 22, 23].

As an interesting self-assembly behavior, mechanical instability is expected to develop a micro/nanopatterning technology driven by deterministic mechanical self-assembly. However, 2D materials are very sensitive to external stimuli external stimuli due to the ultrathin thickness and exceptionally low-bending rigidity nature, such as thermal vibration [24, 25], high-temperature growth-induced lattice mismatch [26, 27], built-in tension [28, 29], van der Waals interactions [30], and compressive stress [31, 32, 33]. Therefore, it is very challenging to controllably assemble 2D materials into designed 3D morphologies, which often results in unavoidable restacking and aggregation. 2D materials are randomly corrugated in 3D via intrinsic rippling induced by thermal vibrations [34]. The abnormal mechanical, electronic, and thermal performances of ideal 2D materials can be greatly weakened, which tremendously increase the difficulty of manufacturing and transferring 2D nanomaterials and reduces the quality of materials [26]. In order to overcome the above shortcomings, extensive research has been conducted on the formation mechanism and suppression methods of spontaneous mechanical instabilities and non-deterministic and uncontrolled self-assemblies, resulting in ultra-flat or even strictly 2D materials [24, 26, 27, 35, 36, 37, 38]. However, it is still very challenging but interesting to implement deterministic mechanical self-assemblies and allow better control over the self-assembly-driven micro/nanopatterns of 2D materials.

Mechanical instabilities in 2D materials can be utilized to introduce non-invasively tunable out-of-plane 3D topological deformations [39, 40, 41], further adjust charge transfer and carrier scattering [42, 43], and even create flat bands in 2D materials [44]. The key point to realizing deterministic self-assembly technology is to make mechanical instabilities in 2D materials controllable. Nevertheless, complex coupled multimodal mechanical instabilities, such as crumpling, rippling, wrinkling, folding, and buckling-induced delamination, can easily be triggered at the same time [27]. They result in the corresponding self-assembled morphologies suffer poor designability. The ripples caused by thermal fluctuations [24, 25] and the inherent random wrinkles generated during the growth and transfer process [26, 27] are random behaviors, making it difficult to predict and control the instability modes, scales, and directions of self-assembly morphologies [26]. Uniaxial strain can trigger quasi one-dimensional (1D) ordered self-assembled structures, but they are still uncontrollable multidirectional structures [31, 32, 33]. In addition, by coupling folding mode instabilities with wrinkle mode instabilities at an uncertain rate, crumpling is easily formed in 2D materials, which can lead to multidirectional and multiscale 3D self-assembly morphologies [31, 33]. The chaotic self-assembled structures induced by multimode coupling mechanical instabilities with uncontrollable scales and directions cannot precisely adjust the deformation of 3D structures in any way, which largely restrict the potential applications of self-assembled 2D materials. Consequently, decoupling multimode coupling mechanical instabilities and obtaining deterministic self-assembly with controlled structural configurations in 2D materials are crucial, but highly challenging.

2. Deterministic mode mechanical self-assembly

A complementary metal oxide semiconductor-compatible programmable tuning strategy has been developed to decouple uncontrollable multimode-coupled mechanical instabilities into controllable mode-decoupled mechanical instability, and to achieve deterministic mode self-assembly of 2D materials (Figure 1A). As shown in Figure 1A, II, the key point that realizes mode decoupling of mechanical instability in 2D nanosheets is to introduce designable intermediate multifunctional layers (IMLs) into bilayer systems. IMLs, including hydrogen silsesquioxane (HSQ) and poly(methyl methacrylate) (PMMA), play a variety of roles in generating and controlling deterministic self-assemblies driven by mono-mode mechanical instability, including deterministic mode, direction, and scale self-assemblies. First, IMLs work as carrier materials to avoid unwanted mode mechanical instability that triggers deterministic mode self-assembly in 2D materials during transfer. Second, IMLs work as a structural layer to enhance the effective bending stiffness of 2D materials/IML structures for self-assembly with deterministic dimensions. Third, IMLs also work as a functional layer and protective layer to manufacture programmable deterministic directional self-assembly patterns of 2D materials.

Figure 1.

Deterministic mode mechanical self-assembly in 2D materials driven by mode-decoupled mechanical instability. (A) Diagrammatic sketch of mechanical instabilities in tri-layered systems: (I) uncontrollable multimode-coupled mechanical instabilities in bilayer systems; (II) mode-decoupled mechanical instability triggered in 2D materials/IMLs/substrate three-layer systems with programmable IMLs. Microscopic morphologies of self-assembled MoS₂ induced by mono-mode instability: (B) microscopic image of mono-mode self-assembled morphologies with deterministic wavelength; (C, D) 3D and cross-sectional topology curves of laser scanning confocal microscopy (LSCM) images of self-assembled MoS₂.

Decoupled mono-mode instability and corresponding self-assembled patterns with deterministic wavelength of MoS₂ are observed in the designed three-layer systems in Figure 1B and C. Further, Figure 1D indicates that the deterministic self-assembly morphologies driven by mono-mode instability are mono-scale wrinkles with amplitude of $1.544\mathrm{\mu m}$ and wavelength of $16.392\mathrm{\mu m}.$ In stark contrast to the uncertain multiscale and multimode coupling instability, completely different decoupled single-mode instability and the resulting deterministic modes, namely single-scale self-assembly morphology, were observed in 2D materials. The decoupled mono-mode instabilities and the resulting deterministic mode, mono-scale self-assembly morphologies of 2D materials observed in Figure 1B and C are completely different from the non-deterministic multiscale and multimode coupling instabilities [21, 22, 39, 40, 41]. This is crucial for the success of self-assembly-based micro/nano-patterning technology in 2D nanocrystals. The more deterministic mode self-assembly morphologies of other 2D materials, including graphene, boron nitride (BN), tungsten disulfide (WS₂), and tungsten diselenide (WSe₂), are also uniform with highly deterministic mono-scale (Figures 2–4).

Figure 2.

Mono-mode instability-driven deterministic mode self-assembled MoS₂ in MoS₂/IML/beGr-PDMS trilayer systems: (A) microimages of large area, high-quality deterministic mode self-assembled structures of MoS₂ with 4% PMMA layer spin-coated at the speed of 6000 rpm; (B) the enlarged view of (A); (C) the 3D laser scanning confocal microscope (LSCM) images of the self-assembled structures of MoS₂; (D) 2D topological curves of the mono-mode instability-induced deterministic scale self-assembled structures at three different positions, which show the $\lambda =\mathit{8.470}\pm \mathit{0.130}\mathit{\mu m},$ $2\mathit{A}=\mathit{0.790}\pm \mathit{0.058}\mathit{\mu m}$ for 4% PMMA under 6000 rpm.

Figure 3.

Mono-mode instability-driven deterministic mode self-assembled WSe₂ in WSe₂/IML/beGr-PDMS trilayer systems: (A) microimages of transferred single-layer WSe₂ prepared by the low-temperature post-curing transfer method without the heat treatment triggering mono-mode instability; (B, C) the 2D topological curves of the deterministic mode self-assembled structures of WSe₂ under different thickness 4% PMMA layer controlled by different spin coating speeds 6000 rpm and 3000 rpm, respectively, where $\lambda =\mathit{7.770}\pm \mathit{0.400}\mathit{\mu m},\mathit{2}\mathit{A}=\mathit{0.760}\pm \mathit{0.040}\mathit{\mu m}$ for 4% PMMA at 6000 rpm, $\lambda =\mathit{15.500}\pm \mathit{0.550}\mathit{\mu m},\mathit{2}\mathit{A}=\mathit{1.200}\pm \mathit{0.060}\mathit{\mu m}$ for 4% PMMA at 3000 rpm; (D, E) 3D LCSM images of the self-assembled WSe₂ under different thickness 4% PMMA layer controlled by different spin coating speeds 6000 rpm and 3000 rpm, respectively.

Figure 4.

Mono-mode instability-driven deterministic mode self-assembled WS₂ and BN in 2D materials/ IML/beGr-PDMS trilayer systems: (A, D) microimages of transferred single-layer WS₂ and BN prepared by the low-temperature post-curing transfer method without the heat treatment triggering mono-mode instability; (B, E) 3D LCSM images of the deterministic mode self-assembled structures of WS₂ and BN under 4% PMMA layer controlled by spin-coating speed 3000 rpm; (C, F) 2D topological curves of the mono-mode instability-driven self-assembled structures of WS₂ and BN under 4% PMMA layer at spin-coating speeds 3000 rpm, respectively.

The results indicate that the mode-decoupled mechanical instability of 2D materials is achieved by constraining unwanted multimode instability (folding, collapsed folding, and buckling induced delamination in Figure 1A, I) to mono-mode instability (wrinkling in Figure 1A, II). As shown in Figure 5, the wavelength and amplitude of the self-assembly morphologies induced by mono-mode instability in three-layer systems can be given by

Figure 5.

Mechanical modeling for conformally self-assembled 2D materials in 2D materials/IML/beGr-PDMS trilayer system: (A) the trilayer systems with stiff surface film including 2D materials and IML, and photothermal-responsive soft brGr-PDMS substrate, where the thickness and Young’s modulus of 2D nanosheets are h_2D and E_2D, the thickness and Young’s modulus of IML are h_IML and E_IML, the thickness and Young’s modulus of compliant substrates are h_s and E_s, respectively; (B) the modified core-shell model of surface elasticity of the effective surface film in the trilayer system, where the effective thickness $h_f=h_{2D}+h_{\mathit{IML}}$ and the effective Young’s modulus of surface film E_f; (C) the self-assembled state of the trilayer system, where $\lambda$ and A are the wavelength and amplitude of self-assembled structures, ${\epsilon }_0$ is the initial strain; (D) schematic diagram of the maximum and minimum strains occurring at the valleys and crests of the structures.

$$\lambda =2\pi h_f{\left({\overline{E}}_f/3{\overline{E}}_s\right)}^{1/3}$$E1

$$A=h_f\sqrt{{\epsilon }_0/{\epsilon }_c-1}$$E2

where E_f and $h_f=h_{2D}+h_{\mathit{IML}}$ are the effective flexural modulus and thickness of 2D materials/IMLs surface films, respectively; ${\epsilon }_c$ is the critical strain that triggers the wrinkling mechanical instability. The feature parameters of orientation, amplitude, wavelength, and dynamic behaviors of the deterministic mechanical self-assembly patterns can be well adjusted, which is a key factor driving the progress of micro/nanopatterning technology based on mechanical self-assembly.

A new low-temperature post-curing transfer strategy is developed to achieve desirable and controllable mode decoupling mechanical instability in atomic thin 2D materials. The transfer strategy has three key points. It is extremely difficult to decouple and control the multimode-coupled mechanical instability of single-layer or even few layer 2D materials due to the extremely low bending stiffness [27, 39, 45]. Firstly, the well-designed IMLs we introduced to build 2D material/IML/substrate three-layer systems instead of two-layer 2D material/substrate systems to tackle the problem of extremely low flexural stiffness of 2D materials (Figures 6, 7, and 8A–E).

Figure 6.

Schematic diagrams of the low-temperature post-curing transfer process of graphene onto the PDMS: (A, B) a IML is spin-coated onto CVD-grown graphene/Ge surface, note that the IML layer works as a carrier material to prevent the unnecessary surface instabilities or fractures of graphene during transfer; (C) beGr-PDMS fabricated by mixing the uncured PDMS with bubble-exfoliated graphene (beGr) sheets; (D) the IML/graphene-Ge laminate is placed onto the bottom of container C₂ and the uncured beGr-PDMS is slowly poured onto the surface of the IML/graphene/substrate laminate, which is cured at low temperature (30°C) for 48 h to make beGr-PDMS fully cured; (E) the residual cured beGr-PDMS on the surface of Ge is peeled off; (F, G) the Ge layer is etched with HF:H₂O₂:H₂O solution for 4 h and the samples are rinsed with deionized water and dry on the heating plate to fully flatten graphene/IML film; (H) the mono-mode instability of graphene is triggered by the heating treatment (110°C) of the graphene/IML/beGr-PDMS system.

Figure 7.

Schematic diagrams of the low-temperature post-curing transfer process of 2D transition metal dichalcogenides (TMDs) materials, such as MoS₂, WS_2, and WSe₂, onto the PDMS: (A, B) the carrier material IML is spin-coated onto CVD-grown TMD_S/sapphire surface; (C) beGr-PDMS fabricated by mixing the uncured PDMS with bubble-exfoliated graphene(beGr) sheets; (D) the IML-TMDs/sapphire composite layer is placed onto the bottom of container C₂ and the uncured beGr-PDMS is slowly poured onto the surface of IML-TMDs/sapphire laminate, which is cured at low temperature (30°C) for 48 h to make beGr-PDMS fully cured; (E) the residual cured beGr-PDMS on the surface of sapphire is peeled off; (F, G) the sapphire layer is slightly etched with KOH solution for 4 h and the samples are rinsed with deionized water and dry on the heating plate to fully flatten TMDs/IML film; (H) the mono-mode instability of 2D TMDs materials is triggered by the heating treatment (110°C) of the 2D TMDs/IML/beGr-PDMS trilayer system.

Figure 8.

Raman characterizations of graphene membranes transferred onto the PMMA-PDMS substrates: (A) D, G, 2D Raman, and other Raman spectra of graphene and PMMA in the samples; (B) 2D Raman results of I_2D/I_G for the samples after transfer.

However, it also presents a new challenge that the interface strength among 2D materials, IMLs, and substrate is too low. The interfacial liquid introduced during the pre-curing transfer significantly lowers the interfacial adhesion energy between the pre-curing substrate and the 2D material, which cannot transfer stress (Figure 9). Note that the low-temperature post-curing transfer process of CVD-grown h-BN on Cu foil is similar to that of the CVD-grown graphene, the only difference is the grown substrate etching, in which the etchant is replaced by FeCl₃ solution. Unlike the conventional wet transfer methods of 2D materials [45], the carrier material in the graphene/IML laminate is not removed after they are transferred onto substrates, which is used as a structure layer to increase the bending stiffness of surface film.

Figure 9.

Schematic diagrams of different interface conditions in film-substrate systems after 2D material transfer with different transfer methods: (A) the cross-sectional diagram of 2D materials/IML-substrate trilayer systems prepared by the pre-curing transfer method in [45], where interfacial liquid is trapped at the interface between IML and PDMS; (B) the cross-sectional diagram of 2D materials/IML-substrate trilayer systems prepared by the high temperature (70°C) post-curing transfer method, where no interfacial liquid is trapped at the interface between IML and PDMS, but high-temperature post-curing of PDMS can cause an gradient interface between IML and PDMS and multiscale self-assembled morphologies; (C) the cross-sectional diagram of 2D materials/IML-substrate trilayer systems prepared by the low-temperature (30°C) post-curing transfer method, where no interfacial liquid and gradient interface is existing at the interface between IML and PDMS, and mono-mode instability-induced deterministic scale self-assembled morphologies can be triggered.

Therefore, producing sufficient interfacial adhesion energy after transfer is the second key point, which can make the interface enable to subject to critical strain to generate the wanted mode mechanical instabilities. To improve interfacial strength between the 2D material and the substrate, we propose a post-curing transfer method for 2D materials, where the uncured beGr-PDMS is directly poured and then cured on the upper surface of IMLs (Figure 10B and C). However, the accelerated diffusion of uncured PDMS polymers by high temperatures can introduce an inevitable gradient interface between beGr-PDMS and IMLs during the post-curing process, which can lead to uncontrollable multiscale self-assemblies of 2D material in the three-layer system (Figure 11).

Figure 10.

Post-curing transfer strategy of 2D materials at low temperature: (A) different types of 2D materials grown by chemical vapor deposition (CVD); (B) IMLs spin-coating; (C) beGr-PDMS low-temperature curing; (D) substrate removal; (E) decoupled mono-mode mechanical instability and deterministic mode self-assembly triggered in trilayer systems with surface films (2D-materials and IMLs) and beGr-PDMS substrates. The mechanisms of interfacial strength enhancement caused by the post-curing transfer: (F, G) micro-mechanical interlocking in polymer networks; (H) liquid-free interfaces.

Figure 11.

(A) The optical micro-image of the cross sections of the samples prepared by the low-temperature posting-curing method, note that an obvious interface is observed in the sample and the red-dashed box is the PMMA IML; (B) 3D LSCM image of the cross sections of the 2D materials/PMMA-PDMS layered samples, where the thickness of 10% PMMA spin-coated by 1000 rmp is $2.582\mathrm{\mu m}.$ (C, D) 3D LSCM image of the top view and height profile in the 2D materials/PMMA-PDMS layered samples, where an obvious step with $\mathit{2.542}\mathit{\mu m}$ is observed in the sample by peeling off 2D materials/PMMA surface films. Note that the height $\left(h_f=\mathit{2.582}\mathit{\mu m}\right)$ of cross-sectional characterization is almost equal to the height $\left(h_f=\mathit{2.542}\mathit{\mu m}\right)$ of top-view morphology characterization, which indicates that the obvious interface observed in cross-sectional characterization in (A, B) is the interface observed in (C, D).

Figure 12.

Raman characterizations of single-layer MoS₂ (A, B) on SiO₂/Si substrates before transfer, and (C, D) on PMMA-PDMS substrates after transfer.

Therefore, the third point is avoiding the formation of diffusion-induced gradient interfaces between IMLs and beGr-PDMS substrate. To reduce polymer diffusion caused by Brownian motion, we cured the beGr-PDMS at low temperature (Figure 10C). To confirm this, we conducted cross-sectional characterization, which indicates that a sharp interface between IMLs and beGr-PDMS substrate was observed instead of a gradient interface in the samples prepared by the low-temperature post-curing transfer method (Figure 9). In additions, Raman characterizations indicate that high-quality 2D materials with low defects are well transferred to target substrates (Figures 8–14). The results indicate that the low-temperature post-curing process can not only generate sufficient interface adhesion strength to withstand the critical interfacial stresses triggering mode-coupled mechanical instabilities, but also effectively avoid gradient interfaces induced by high temperatures-accelerated PDMS polymer diffusion.

Figure 13.

Raman characterizations of single-layer WSe₂ (A, B) on sapphire substrates before transfer, and (C, D) on PMMA-PDMS substrates after transfer.

Figure 14.

Raman characterizations of single-layer WS₂ (A, B) on sapphire substrates before transfer, and (C, D) on PMMA-PDMS substrates after transfer. Note that Raman results in Figures S6–S8 indicate transferred single-layer 2D materials are highly uniform and of high quality.

We revealed the potential mechanism of the post-curing transfer-induced interfacial strength enhancement effect after solidification (Figure 10F–H). The enhanced interfacial strength between 2D materials and IMLs is caused by the thermal evaporation-driven topological entanglement of IMLs polymer chains during the high-temperature baking process (Figure 10F). Similarly, the interface strength enhancement between beGr-PDMS and IMLs is benefited by the topological entanglement between the IML and beGr-PDMS polymer network during the post-curing process (Figure 10G). The results indicate that the micro-mechanical interlocking induced by topological entanglement is the micro-mechanism of the interface strength enhancement [46]. Furthermore, a sharp decrease in interface strength during substrate etching is effectively prevented due to the liquid-free interfaces between beGr-PDMS, IML, and 2D materials (Figure 10D and H).

Moreover, more types of 2D materials grown by CVD, including insulator (h-BN), semi-conductors (2H-WS₂) and semi-metals (graphene, 1 T-WSe₂, 1 T-MoS₂), were transferred to prove the versatility of the transfer strategy (Figure 10A). As shown in Figures 15–17, the experimental data indicate that our transfer strategy has good universality and can be extended to other diverse 2D materials.

Figure 15.

Deterministic mode self-assembled MoS₂ in MoS₂/IML/beGr-PDMS trilayer systems: (A) microimages of large area, high-quality deterministic mode self-assembled structures of MoS₂ with 4% PMMA layer spin-coated at the speed of 6000 rpm; (B) the enlarged view of (A); (C) the 3D laser scanning confocal microscope (LSCM) images of the self-assembled structures of MoS₂; (D) 2D topological curves of the mono-mode instability-induced deterministic scale self-assembled structures at three different positions, which show the $\lambda =\mathit{8.470}\pm \mathit{0.130}\mathit{\mu m},2\mathit{A}=\mathit{0.790}\pm \mathit{0.058}\mathit{\mu m}$ for 4% PMMA under 6000 rpm.

Figure 16.

Deterministic mode self-assembled WSe₂ in WSe₂/IML/beGr-PDMS trilayer systems: (A) microimages of transferred single-layer WSe₂ prepared by the low-temperature post-curing transfer method without the heat treatment triggering mono-mode instability; (B, C) the 2D topological curves of the deterministic mode self-assembled structures of WSe₂ under different thickness 4% PMMA layer controlled by different spin coating speeds 6000 rpm and 3000 rpm, respectively, where $\lambda =\mathit{7.770}\pm \mathit{0.400}\mathit{\mu m},\mathit{2}\mathit{A}=\mathit{0.760}\pm \mathit{0.040}\mathit{\mu m}$ for 4% PMMA at 6000 rpm, $\lambda =\mathit{15.500}\pm \mathit{0.550}\mathit{\mu m},2\mathit{A}=\mathit{1.200}\pm \mathit{0.060}\mathit{\mu m}$ for 4% PMMA at 3000 rpm; (D, E) 3D LCSM images of the self-assembled WSe₂ under different thickness 4% PMMA layer controlled by different spin coating speeds 6000 rpm and 3000 rpm, respectively.

Figure 17.

Deterministic mode self-assembled WS₂ and BN in 2D materials/IML/beGr-PDMS trilayer systems: (A, D) microimages of transferred single-layer WS₂ and BN prepared by the low-temperature post-curing transfer method without the heat treatment triggering mono-mode instability; (B, E) 3D LCSM images of the deterministic mode self-assembled structures of WS₂ and BN under 4% PMMA layer controlled by spin-coating speed 3000 rpm; (C, F) 2D topological curves of the mono-mode instability-driven self-assembled structures of WS₂ and BN under 4% PMMA layer at spin-coating speeds 3000 rpm, respectively.

3. Cross-scale mechanical self-assembly

In order to achieve cross-scale self-assembly of 2D materials, IML was introduced as a structural layer to control the scale of deterministic mode self-assembly morphological features.

The IMLs are introduced to work as structural layers to control the characteristic scale of deterministic mode self-assembled patterns and achieve cross-scale self-assembly of 2D materials. The amplitude and wavelength of self-assembled micro/nanostructures in 2D materials can be controlled by the thickness of IMLs, judging from Eqs. (1)–(3). Furthermore, the thickness of IMLs can be programmatically regulated by spin-coating speeds${\omega }_{\mathit{\text{spin}}}$and IMLs concentrations c_IML. As shown in Figure 18A, when the PMMA IMLs changes from 0.5% at 6000rmp to 10% at 1000rmp, the amplitude of self-assembled wrinkles increases from $8.5\pm 4.0\mathrm{nm}$ to $5.203\pm 0.605\mathrm{\mu m}.$ The wavelength has increased from $456\pm 46\mathrm{nm}$ to $159.960\pm 8.500\mathrm{\mu m}.$ The experimental results indicate that wavelength and amplitude can be effectively tuned by changing c_IML and ${\omega }_{\mathit{\text{spin}}}$, and the tuning range is up to three orders of magnitude (Tables 1 and 2; Figures 19 and 20). Judging from the mechanical optoelectronic coupling physical model of self-assembly in 2D materials/IMLs/substrate three-layer systems, cross-scale self-assembly is crucial for the micro/nano patterning technology based on self-assembly of 2D materials.

Figure 18.

Mode-coupled instability-induced cross-scale and directed self-assemblies of 2D materials in three-layer systems. (A) the amplitude and wavelength of micro/nanowrinkles in self-assembled 2D materials under different thickness IMLs, and the insets show the LSCM images of different scale self-assembled patterns in 2D materials. (B–D) 2D directed self-assemblies of 2D materials: (B) directed self-assembly of graphene with “gr”-shape; (C) EBL-defined mirrored “MoS₂” pattern before transfer; (D) 2D directed self-assembly of MoS₂ with “MoS₂” shape, in which the partial enlarged views are highly ordered MoS₂ with the structure crest orientations perpendicular to EBL-defined boundaries.

PMMA	0.5%, 6kr	1%, 6kr	2%, 6kr	4%, 6kr	6%, 6kr	10%, 6kr
$\mathit{\lambda }\left(\mathbf{\mu m}\right)$	$0.456\pm 0.046$	$0.760\pm 0.048$	$3.530\pm 0.380$	$8.125\pm 0.775$	$16.08\pm 0.775$	$59.97\pm 2.125$
$\mathit{2}\mathit{A}\left(\mathbf{\mu m}\right)$	$0.017\pm 0.008$	$0.051\pm 0.007$	$0.215\pm 0.021$	$0.765\pm 0.045$	$1.867\pm 0.085$	$2.163\pm 0.132$

Table 1.

The tunable wavelength and amplitude of mono-mode instability-induced conformal self-assembled structures of 2D materials at spin-coating 6000 rpm with different IML concentrations.

PMMA	0.5%, 3kr	1%, 3kr	2%, 3kr	4%, 3kr	6%, 3kr	10%, 3kr	10%, 1kr
$\mathit{\lambda }\left(\mathbf{\mu m}\right)$	$0.566\pm 0.067$	$1.261\pm 0.266$	$5.700\pm 0.740$	$14.300\pm 1.850$	$25.220\pm 3.786$	$81.340\pm 5.560$	$159.960\pm 8.500$
$\mathit{2}\mathit{A}\left(\mathbf{\mu m}\right)$	$0.026\pm 0.005$	$0.085\pm 0.011$	$0.402\pm 0.042$	$1.020\pm 0.085$	$1.985\pm 0.136$	$2.866\pm 0.204$	$10.405\pm 1.210$

Table 2.

The tunable wavelength and amplitude of mono-mode instability-induced conformal self-assembled structures of 2D materials at spin-coating 3000 rpm and 1000 rmp with different IML concentrations.

Figure 19.

Morphological characterizations of self-assembled nanostructures in the samples with IML thickness under 0.5% PMMA at spin-coating 3000 rpm: (A) LCSM images, (B) 2D topological curves of the mono-mode instability-induced self-assembled nano-structures. (C) Thickness characterization of IML under 0.5% PMMA at spin-coating 3000 rpm.

Figure 20.

LCSM morphological characterization of self-assembled structures in the samples with different thickness IMLs: (A) under 2% PMMA A2 at spin-coating 6000 rpm and 3000 rpm; (B) under 4% PMMA A4 at spin-coating 6000 rpm and 3000 rpm, (C) under 6% PMMA A6 at spin-coating 6000 and 3000 rpm; (D) under 10% PMMA A10 at spin-coating 6000 and 3000 rpm.

The wavelength and amplitude of mono-mode instability-induced conformal self-assembled 2D materials can be tuned by selecting different IML concentrations c_IML and spin-coating speeds ${\omega }_{\mathit{\text{spin}}}$. The thickness of spin-coated IML layer can be given by

$$h_{\mathit{IML}}=k_{\mathit{IML}}{c}_{\mathit{IML}}^n{\omega }_{\mathit{\text{spin}}}^{-\gamma }$$E3

where k_IML, n, and $\gamma$ are the coefficients depended on the IML type, viscosity value, and solvent vaporizability rate. As shown in Eq. (3), the thickness of IML h_IML can be tuned by the concentrations c_IML and spin-coating speeds ${\omega }_{\mathit{\text{spin}}}$ of IML.

4. Directed mechanical self-assembly

As shown in Eqs. (1) and (2), the deformation orientation, anisotropic structural stiffness, and functionalization of 2D materials in three-layer systems of 2D materials/IMLs/substrate can be adjusted through directed self-assembly. Therefore, it is important for the applications of deterministic self-assembly to transform non-directed into directed self-assembly with designed directions, especially for self-assembly-driven directed micro/nanopatterning technology and 2D materials-based flexible devices. Via CMOS-compatible direct writing methods including electron beam lithography (EBL), the high-resolution designed mechanical boundaries in 2D material/IMLs/beGr-PDMS three-layer system are defined to realize directed self-assembly of 2D materials. It is worth noting that IMLs play a functional and protective role in directed self-assembly (Figure 21).

Figure 21.

IMLs-fueled programmable directed mechanical self-assembly of 2D materials: (A, B) photochemical reaction of IMLs (electron beam photoresist: PMMA and HSQ); (C–H) EBL-defined high-resolution mechanical boundaries, where (C, D) photochemical reaction of spin-coated IMLs is induced by the electron beam radiation, (E) the exposed IMLs are dissolved by the developer solution MIBK (Methyl Isobutyl Ketone): 2-proponol (IPA), (F) the 2D materials are removed via reactive ion etching (RIE) and the IMLs work as the protective layer to protect the 2D materials covered by IMLs, (G, H) the post-curing of PDMS and the etching of the substrates.

The above mechanical boundaries can be used to direct 2D materials self-assemble into ordered structures. Figures 18B–D and 22–24 show directed self-assembly morphologies of “MoS2” and “SJTU”-alphabetic-shaped MoS₂, and “Gr”-alphabetic-shaped graphene, in which IMLs, as electron beam resists, are used to programmatically fabricate ordered patterns. In the 2D oriented self-assembled graphene and MoS₂ nanosheets, the orientations of structural peaks are perpendicular to the EBL-defined boundaries (Figure 18B and D). Note that the EBL-defined “Gr” letter pattern was well transferred onto the target substrates by the low-temperature post-curing method without quality degradation. Moreover, the stress concentration-induced local self-assembly phenomenon can be harnessed to local strain engineering with large strains on demand by designing different curvature corners to control stress concentration and stress relaxation. We developed a revised shear-lag model reveals the mechanical mechanism of directed self-assembly structures induced by finite width soft constrained mechanical boundaries, indicating that mechanical boundaries can lead to asymmetric stress relaxation and asymmetric in-plane stress distribution.

Figure 22.

“Gr” letter-shaped 2D directed self-assembled morphologies of graphene in layered systems with the IML under the condition of (4% PMMA, 3000 rmp): (A) mirrored “Gr” letters short for graphene before transfer, where the EBL-defined pattern is obtained by selectively dissolving the exposure area by the developer solution MIBK (Methyl Isobutyl Ketone): 2-proponol (IPA) for 45 s; (B) after transfer; (C) directed self-assembling structures of graphene with PMMA layer after heat treatment, where the red elliptic dashed box areas show the stress concentration-induced local mechanical instability with local increasing amplitudes around the inter corners (included $\mathit{\text{angle}}<\mathit{180}°$), and the blue rectangle dashed box areas show the stress relaxation-induced local instability-free patterns around the outer corners (included $\mathit{\text{angle}}>\mathit{180}°$).

Figure 23.

“Gr” letter-shaped 2D directed self-assembled morphologies of graphene in layered systems with the IML under the condition of (4% PMMA, 6000 rmp): (A) mirrored “Gr” letters before transfer; (B) after transfer; (C) optical micro-images of directed self-assembling structures of graphene after heat treatment, similar with Figure S21, the stress concentration-induced local directed self-assembling structures around the inter corners and stress relaxation-induced local instability-free patterns around the outer corners are also observed.

Figure 24.

“Gr” letter-shaped 2D directed self-assembled morphologies of MoS₂ in layered systems: (A) EBL-defined mirrored “SJTU” letters short for Shanghai Jiao Tong University before transfer and (B) LCSM images of 2D directed self-assembling structures after transfer and heat treatment, where the partial enlargement maps show the detailed directed self-assembling morphologies of “S” and “T” letters.

5. Dynamic mechanical self-assembly

In addition to the quasi-static above-mentioned cross-scale and directed self-assemblies, the dynamic self-assembly of 2D materials can be acquired through effective dynamic adjustment of mode-coupled instability. Via an interesting experimental design, a non-contact dynamic adjustment strategy is proposed to achieve in situ, real-time dynamic operation and synchronous self-assembly observation (Figure 25A). Firstly, graphene/conjugated polymer composite (beGr-PDMS) is constructed by mixing beGr [46] into PDMS. Rapid temperature changes and the resulting photothermal forces can be generated by near-infrared (NIR) irradiation of the beGr-PDMS substrate. Interestingly, we use multilayer 2D materials from the substrate to dynamically tune the self-assembly of single-layer 2D materials in the surface films. In addition, a reflector was designed to reflect and allow near-infrared light to enter the sample from the back of the substrate and to improve the incidence rate (Figure 25A).

Figure 25.

Dynamic self-assembly of 2D materials in three-layer systems of 2D materials/IMLs/beGr-PDMS: (A) a platform for in situ dynamic operation and real-time synchronous self-assembly morphology characterization; (B) when NIR is turned off at 90s, the time-varying wavelength and amplitude of the self-assembled structures; (C) NIR-controlled dynamic self-assembly.

During the dynamic operation of different NIR ON/OFF switches, the wavelength remains almost unchanged, indicating that the self-assembly morphology of 2D materials remains unchanged in the self-assembly mode and high mode-scale characteristics (Figure 25B). As indicated in Figure 25B, the amplitude of the self-assembled patterns can be dynamically adjusted by NIR light. When $\mathrm{t}=0\mathrm{s},2A=1.549\pm 0.059\mathrm{\mu m};$ when $t=90\mathrm{s},2A=0\mathrm{\mu m}.$ It is indicated that the self-assembled structures have been completely erased (Figure 25C). After NIR is turned off at $t=470\mathrm{s},$ wrinkles gradually recover and $2A=1.417\pm 0.070\mathrm{\mu m}.$ We developed a thermodynamic model to study the dynamic manipulation mechanism and reveal the laws of dynamic self-assembly evolution of 2D materials in three-layered systems as follows

$$\mathrm{\Delta }T\left(t\right)=\left\{\begin{array}{cc}{\kappa }_1{t}^{{\alpha }_1}+T_r& t\le t_{\mathit{off}}\\ T_r+\left(T_m-T_r\right)e^{-{\alpha }_2\left(t-t_{\mathit{off}}\right)}& t>t_{\mathit{off}}\end{array}\right.$$E4

$${\sigma }_0\left(t\right)=\frac{E_f\left({\alpha }_s-{\alpha }_f\right)\mathrm{\Delta }T\left(t\right)}{\left(1-{\nu }_f\right)}$$E5

$$A\left(t\right)=h_f\sqrt{{\sigma }_0\left(t\right)/{\sigma }_c-1}$$E6

where ${\kappa }_1,{\alpha }_1,{\alpha }_2$ are the constants related to photothermal energy conversion efficiency; $T_r,T_m={\kappa }_1{t_{\mathit{off}}}^{{\alpha }_1}+T_r$ are the room and highest temperatures; A(t) is the time-varying amplitude of self-assembled structures; ${\sigma }_0\left(t\right)$ and $\mathrm{\Delta }T\left(t\right)$ are the photothermal force and temperature difference controlling by NIR light; ${\alpha }_s,{\alpha }_f$ are the thermal expansion coefficients of substrate and stiff thin film, respectively.

As shown in Eqs. (4)–(6) and Figures 26 and 27, the time-varying amplitude of self-assembled structures can be tuned by the time-varying photothermal force and temperature difference. The ratio of beGr, and power density and incidence rate of NIR can be used to modulated the photothermal response speed of beGr-PDMS.

Figure 26.

Photothermal effect of multilayer graphene/conjugated polymer composites beGr-PDMS. (A) Schematic illustration of in situ surface temperature measuring platform of samples by using an infrared thermography camera during NIR ON/OFF switch. (B) the variation curve of surface temperature with respect to irradiation time when NIR is off at 90 s, which can be depicted by piecewise function from Eq. (4) that well agrees with the experimental date. (C) the corresponding temperature distribution nephograms when the sample is illuminated by NIR light (808 nm, 2.0 W/cm²) for 90 s. note that the beGr-PDMS heats up rapidly upon NIR irradiation when $t\le t_{\mathit{off}}$, which indicates that the high surface-area-to-mass ratio of beGr can result in high photothermal energy conversion of the beGr-PDMS.

Figure 27.

Reversible dynamic evolution of self-assembled graphene when NIR is off at 90 s, where the IML (PMMA layer) is prepared by spin-coating 4% PMMA at 3000 rmp onto graphene surface and the power density of NIR is 2.0 W/cm². The amplitude of self-assembled structures decreases after the NIR ON, and the self-assembled structures are thoroughly erased $\left(A=0\right)$ when $t=90\mathrm{s}$ (from I to V). The amplitude of self-assembled structures increases after the NIR OFF and the self-assembled structures gradually reappear (from V to I).

The photothermal performance of NIR-responsive beGr-PDMS can be well described by Eq. (4) As shown in Figure 25, beGr-PDMS heats rapidly under near-infrared radiation, indicating a high conversion rate of photothermal energy. After the NIR is closed $\left(t>t_{\mathit{off}}\right)$, the sample cools rapidly and tends to room temperature, which meets Newton’s law of cooling (Figure 25B and C). The theoretical dynamic evolution of self-assembled wrinkles estimated by Eqs. (5) and (6) is very consistent with the experimental data (Figure 25B). As indicated in Eq. (6), the dynamic adjustment of strain and optoelectronic properties of 2D materials can be achieved by non-contact dynamic self-assembly, which opens up a way for the development of programmable micro/nano-patterning technology and dynamic tunable devices of 2D materials.

6. Self-assembled flexible devices

The deterministic self-assembled structures can bring some new unique properties into 2D materials, such as anisotropic piezoresistive, flexoelectric, piezoelectric, and mechanical optoelectronic coupling effects of 2D directed self-assembled 2D materials (Figure 18B–D). Moreover, the dynamic self-assembly can be used to realize the in situ dynamically tunable optoelectronic characteristics (Figure 25). The above new characteristics of controllable deterministic self-assembly can offer unprecedented feasibility opportunities for the self-assembled electronic devices of 2D materials.

In this work, as shown in Figure 28, a novel concept is proposed to manufacture deterministic self-assembled flexible electronic devices of 2D materials driven by deterministic mono-mode instability. A single self-assembled structure can be considered as an independently adjustable self-assembled electronic component, such as a transistor, a capacitor, a resistor, or a converter (Figure 28A). The independently adjustable electronic component can be assembled into complex circuits and interact with the external environment via self-assembled patterns of 2D materials. It is worth noting that self-assembled electronic components can be post-programmed through a determined mode circuit connection to form an adjustable self-assembled circuit network, such as resistors that can be linked in parallel, in series, or series parallel (Figure 29). In stark contrast, the 2D disordered multiscale self-assembly structures (Figure 29A) can lead to a non-deterministic connection network of random resistors (Figure 29D). The key to the self-assembled electronics of 2D materials driven by mono-mode instability is the deterministic self-assembly and circuit connections between electronic components.

Figure 28.

Deterministically self-assembled flexible electronic devices: (A) schematics of self-assembled electronics; (B) flexible and wearable self-assembled electronics for gesture detection; (C) 3D anisotropic tactile sensors for normal and shear force discrimination.

Figure 29.

The new concept of deterministically self-assembled micro/nano-electronics fueled by mode-decoupled mechanical instabilities in 2D materials. (A) Multimode coupling mechanical instabilities and 2D non-directed multiscale self-assembling structures; (B) multimode coupling mechanical instabilities and multiscale 1D directed self-assembling structures; (C) decoupled mono-mode mechanical instability and single-scale 1D directed self-assembling structures. The self-assembled micro/nano-electronics: (D) non-deterministic self-assembled networks of resistors, where resistors with random resistance values $R_{\mathit{si}}\left(i=\mathit{1},\mathit{2},\mathit{3},\dots N\right)$ are connected in a random manner, not deterministic series or parallel manner because 2D non-directed multiscale self-assembling structures are randomly aligned; (E) deterministic series self-assembled networks of resistors, where resistors with random resistance values are connected in deterministic series because 1D directed multiscale self-assembling structures are aligned in parallel; (F) deterministic series-parallel self-assembled networks of resistors, where resistors with deterministic resistance values R are deterministically connected in series and then in parallel because three stripes of 1D directed single-scale self-assembling structures are aligned in parallel.

As shown in Figure 28B, an ultra-fast response wearable flexible sensor for gesture detection was manufactured based on the deterministic self-assembly of 2D materials in 2D material/IML/substrate three-layer systems. When the hand is in the initial horizontal position, the sensor is connected to the arm near the wrist (Figure 28B). The amplitude and wavelength of initial self-assembled wrinkles are $\lambda$ and A, and the initial resistance of the corresponding self-assembled resistor is R, respectively. The sensor is stretched when the wrist is rotated downward by 30 degrees, and the amplitude of self-assembled wrinkles decreases by $△A,$ the resistance of the self-assembled resistor also increases by $△R_{\mathit{\text{down}}}$ and the voltage of the sensor increases by 40 mV (Figure 28B, I–IV). Furthermore, the sensor is compressed when the wrist rotates upward 30 degrees, resulting in a decrease in the amplitude of the self-assembled structures $△A,$ in the voltage and resistance of the sensor 40 mV and $N△R_{\mathit{up}},$ where N is the number of the structures, respectively. The sensor has an extremely fast response speed and also is sensitive to wrist movements, where the response and recovery times $T_{\mathit{rec}}=4\mathrm{ms},T_{\mathit{rec}}=3\mathrm{ms},$ respectively (Figure 30).

Figure 30.

The test platform of deterministically self-assembled electronics: (A) a signal generator is used to generate a constant current to drive electronic devices; (B) self-assembled electronics including a deterministically self-assembled 2D-materials/IML/substrate trilayer system, patterned Au electrodes, and a PDMS protective layer; (C) data acquisition device; (D) data analysis; and (E) fabricated sensors.

Moreover, by utilizing directed self-assembly of 2D materials in 2D material/IMLs/substrate three-layer systems, we fabricated a new functional oriented flexible electronic device (Figure 28C) [47]. A 3D anisotropic tactile sensor with ultra-fast response is presented by employing the 1D directed self-assembly of 2D materials to effectively identify normal and shear forces. The 1D ordered self-assembled structure of 2D materials can work as sensing element, which is the key to the tactile sensing function of the anisotropic 3D tactile sensor. Each 1D ordered self-assembly structure can be equivalent to a resistor with the same resistance value, and the resistor is determined to be connected in parallel before being touched by fingers (Figure 28C, II). As shown in Figure 28C, III–V, the tactile sensor has three touch modes: normal force touch, x-direction shear force touch, and y-direction shear force contact. As indicated in Figure 28C, VI, the self-assembled structures are locally squeezed into a flat state, and the resistance $\left(△R_n\right)$ and voltage signal changes are observed due to the local structural deformation and piezoresistive effect of 2D materials. Due to the ultra-fast response characteristics of the sensor, it can accurately record the finger pressing time, where the response and recovery times $T_{\mathit{rec}}=4\mathrm{ms}$ and $T_{\mathit{rec}}=3\mathrm{ms},$ respectively (Figure 31).

Figure 31.

Electric characterization of the ultra-fast responsive 3D tactile sensors in response to: (A) under finger press with increasing dwell times (from 1 s to 10s); (B) the enlarged view of V-T curves of the sensors under one press-release cycle; (C, D) V-T curves exhibiting ultra-fast responsive and recovery times, where the responsive time $T_{\mathit{res}}=\mathit{4}\mathit{ms}$ and the recovery time $T_{\mathit{rec}}=\mathit{3}\mathit{ms}.$

Furthermore, the sensor can also sense the shear force caused by finger sliding friction. Owing to 1D ordered self-assembly of 2D materials, the tactile sensors show anisotropic sensing characteristics. The self-assembled structure with a symmetrical sinusoidal shape undergoes significant deformation when the fingers slide along x direction, forming an asymmetric structure (Figure 28C, IV), with significant changes in resistance $\left(△R_{\mathit{sx}}\right)$ and sensing signal (Figure 28C, VII). However, the self-assembled structure undergoes very small deformation when the fingers slide along y direction (Figure 28C, V), without significant resistance changes $\left(△R_{\mathit{sy}}«△R_{\mathit{sx}}\right)$ and sensing signals (Figure 28C, VII). The reason for that is the effective shear stiffness in x direction is much less than the stiffness in y direction of the 1D directed self-assembled structures. It is indicated that the sensor can effectively separate different direction shear forces. Moreover, the sensor can also distinguish between normal and shear forces. As shown in Figures 31B and 32B, the V-T voltage signal curves of the sensors under one press-release and finger sliding cycle remain unchanged when the finger remains pressed. However, when the finger slides on the sensor, the voltage signal repeatedly jumps up and down. The mechanical mechanism is that the self-assembled structures can repeatedly rebound after the shear force is eliminated (Figure 33). Therefore, the 3D anisotropic tactile sensors can effectively recognize the pressure and different direction shear forces from the fingers, owing to the directed self-assembled wrinkle-induced anisotropic piezoresistive effect of 2D materials. The deterministic self-assembly technology driven by mono-mode instability can provide a new and simpler manufacturing strategy for flexible electronic micro/nanodevices with inspiring functions.

Figure 32.

Electric characterization of the ultra-fast responsive tactile sensors in response to: (A) under finger sliding motion with the shear force in x direction; (B) the enlarged view of V-T curves of the sensors under one finger sliding cycle; (C, D) V-T curves exhibiting ultra-fast responsive and recovery times, where $T_{\mathit{res}}=\mathit{4}\mathit{ms}$ and $T_{\mathit{rec}}=\mathit{3}\mathit{ms}.$

Figure 33.

The normal and shear force discrimination mechanical mechanism of anisotropic 3D tactile sensors: (A) the deformation mode and normal stress-sensitive sensing signals of the sensors under finger pressure; (B) the deformation mode and normal stress-sensitive sensing signals of the sensors when the fingers slide along x direction (the direction perpendicular to structural crest), where the sinusoidal self-assembled structure will rebound from left to right after the shear force is removed; (C) the deformation mode and normal stress-insensitive sensing signals of the sensors when the fingers slide along x direction (the direction parallel to structural crest).

7. Conclusions

We have fundamentally overcome the key problem of mechanical instability mode manipulation caused by the extremely low bending stiffness of atomically thin 2D crystals to realize the highly designable decoupled mono-mode mechanical instability and deterministic mode self-assembly of 2D materials. The key to decoupling multimode-coupled mechanical instabilities in ultra-thin 2D materials is to construct 2D material/IMLs-substrate three-layer systems by introducing a universal, programmable IMLs and micro-mechanical interlocking bonding.

The decoupled mono-mode instability of atomic thin 2D materials opens up a new opportunity for the determination of third-order cross-scale and directed self-assemblies in 2D nanocrystals. Interestingly, a non-contact in situ dynamic manipulation strategy was proposed for the post-tunable dynamic self-assembly of 2D materials. In addition, the above-mentioned programmable deterministic self-assembly endows 2D materials with new characteristics, such as anisotropic mechanical optoelectronic coupling and piezoresistive effects. A new concept is presented to manufacture deterministically self-assembled micro/nano-electronic devices based on 2D materials, such as ultra-fast response 3D anisotropic tactile sensors and wearable gesture sensors by using the above deterministic self-assembly inducing properties. The deterministic self-assembly driven by mono-mode instability can greatly promote CMOS-compatible micro/nanopatterning technology for 2D materials and deformable self-assembled electronic devices.

Acknowledgments

We strongly appreciate discussions with Yu-Long Chen, Shuai-Chen, Shu-Zheng Yan, Tian-Jiao Ma, Zhong-Ying Xue. We also thank Guang-Yu Zhang’s group for providing 2D materials, Qi Sun for the EBL process, Gang Wang for providing beGr, and Tian-Tian Li for LCSM measurements. The authors would like to acknowledge the supports by the National Science Foundation (12121002, 12032015, 12172216), Science and Technology Innovation Action Plan of Shanghai (21190760100), the Program of Shanghai Academic/Technology Research Leader (19XD1421600), the State Key Laboratory of Mechanical System and Vibration (Grant No. MSVZD202105), and Double First-Class Construction project of Shanghai Jiaotong University (WH220402002).