Fabrication and applications of van der Waals heterostructures

Van der Waals heterostructures (vdWHs) are showing considerable potential in both fundamental exploration and practical applications. Built upon the synthetic successes of (two-dimensional) 2D materials, several synthetic strategies of vdWHs have been developed, allowing the convenient fabrication of diverse vdWHs with decent controllability, quality, and scalability. This review first summarizes the current state of the art in synthetic strategies of vdWHs, including physical combination, deposition, solvothermal synthesis, and synchronous evolution. Then three major applications and their representative vdWH devices have been reviewed, including electronics (tunneling field effect transistors and 2D contact), optoelectronics (photodetector), and energy conversion (electrocatalysts and metal ion batteries), to unveil the potentials of vdWHs in practical applications and provide the general design principles of functional vdWHs for different applications. Besides, moiré superlattices based on vdWHs are discussed to showcase the importance of vdWHs as a platform for novel condensed matter physics. Finally, the crucial challenges towards ideal vdWHs with high performance are discussed, and the outlook for future development is presented. By the systematical integration of synthetic strategies and applications, we hope this review can further light up the rational designs of vdWHs for emerging applications.


Introduction
Since the discovery of graphene (Gr) [1], 2D layered materials (2DLMs) [1] with varying compositions and electronic 1 These authors contributed equally to this work. * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
To study the novel properties, and more importantly, to realize practical applications of vdWHs, tremendous efforts have been put into preparation/synthesizing vdWHs by various methods, including the physical combination [51-58], deposition [59][60][61][62][63], solvothermal methods [64][65][66][67][68][69] and synchronous evolution (one-step synthesis) [70][71][72] (as illustrated in figure 1). Among them, the most intuitional and versatile technique is the physical combination which became popular after the initial breakthrough in the electronic device based on Gr/hBN vdWH via such mechanical technique, which has relatively high mobilities and carrier inhomogeneities assembled [13,73]. However, despite tremendous effort, the scalability of mechanical assembly is questionable, due to the small sample size and poor output [13,74,75]. Naturally, more efforts are devoted to the scalable synthetic strategies of vdWHs built upon the existing success in 2DLM syntheses, such as chemical vapor deposition (CVD) [76,77], physical vapor deposition (PVD) [78], and solvothermal methods. For instance, both in-plane and vertical Gr/hBN structures have been successfully synthesized by two-step CVD deposition benefitting from the similar lattice structures of hBN and Gr [79,80]. Most recently, GaSe/MoSe 2 vdWHs with large lattice mismatch have been fabricated, further expanding the application potential of deposition methods [81]. Compared to the CVD deposition, the self-assembly of vdWHs in the liquid phase is a high-throughput and low-cost alternative.  and consequent assembly in a colloidal mixture. Unlike CVD methods, solvothermal methods have relatively weak control over the morphology and uniformity, and it is very suitable for energy application due to more exposed active edges and defects resulting from the amorphous shape and porous structures [11]. The methodologies toward high-quality and scalable vdWHs have continued to evolve in the past decade.
2D vdWHs have been widely reviewed in recent years due to the comprehensive exploration and rapid development of their synthetic strategies and related applications. Almost all of them are focused on a specific 2D material system (such as Mxenes [82], TMDs [83][84][85]), or fabrication method (such as mechanical stacking [86,87], CVD [88,89]) or application (such as electronics [90][91][92], optoelectronics [93][94][95], energy-related application [84,96]), due to the extensive research on 2D vdWHs. However, it is not easy to construct the relationship between synthetic strategies and the application of 2D vdWHs for the novice, and following choose a proper synthetic strategy and design the heterostructures for a specific application. This review first introduces the major synthetic strategies: physical combination, gas-phase deposition, solvothermal synthesis and synchronous evolution. The fundamental principles, development tendencies, as well as strengths and weaknesses of each strategy are elaborated. Next, we will further illustrate the integration design based on the vdWHs in electronics, photodetectors and energy-related applications, and discuss the research on moiré superlattice based on resonant emissions as the representative of novel physical phenomena. Then discuss the potential opportunities and challenges arising in the vdWHs synthesis and applications. Combined with the comprehensive discussion of synthesis and applications, we hope this review can enlighten more rational designs of vdWHs for various emerging applications.

Synthetic strategies of vdWHs
Up to now, the synthetic strategies of 2D vdWHs can be separated into four main categories: physical combination, gasphase deposition, wet chemical synthesis and synchronous evolution. Different synthetic strategies favor different key properties of the vdWHs, such as layer number, cleanliness, defect density, and interface uniformity, which determine their application performance. Hence, it is essential to understand the mechanical principle and characteristics of diverse synthetic strategies. In this section, the primary mechanism and processes of diverse methods are illustrated and discussed.

Physical combination
Physical combination, which mechanically stacks 2DLMS layer-by-layer, is the easiest yet most intuitive way to fabricate vdWHs. It produces layered heterostructures with incommensurate layers at will, regardless of the lattice matching. The first step is to obtain ultrathin nanosheets as required through diverse methods such as CVD [76,77], PVD [78], mechanical exfoliation [13,97], solution exfoliation [98], and so on. The next step is to transfer layers and integrate them into vdWHs as designed. To improve the interface cleanness and produce efficiency, research in mass on the transfer and fabrication techniques has been conducted, including wet transfer, dry transfer, and self-assembly in solution, which will be introduced in the following.

Wet transfer.
Wet transfer and integration is the most widely used method to stack vdWHs with a gentle-energy procedure [101]. A polymer supporting and/or sacrificial layer is essential in the wet transfer method. The polymer types significantly affect the cleanness of interfaces and the efficiency of fabrication procedures, and the most widely adopted one is PMMA (polymethyl methacrylate). Figure 2(a) illustrates the wet transfer method based on PMMA to represent a typical wet transfer process [57,58]. Firstly, the 2D nanosheet deposited on a substrate, either mechanically exfoliated or synthesized, is spin-coated with PMMA, followed by baking at an elevated temperature to remove the solvent. Then, the PMMA-coated substrate bearing the 2D nanosheet was immersed in water. The hydrophobic PMMA layer bearing the 2D nanosheet will be gradually separated from the hydrophilic substrate (typically SiO 2 /Si) when the water slowly enters the interface between PMMA and the substrate. The PMMA/2D nanosheet stack will float on water with the 2D materials in contact with water on the bottom. It can be fished out by supporting material and later stacked onto other 2D material on the substrate at the desired position, followed by baking at a mild temperature (100 • C) to remove the residual water and gas molecules trapped between the 2D layers. PMMA is eventually dissolved to produce the target vdWHs on the substrate (including flexible ones), as shown in figure 2(b) [57,102]. Depending on the used 2D materials and target application, variants in the general wet-transfer method are introduced. For example, hydrophobic polymer (cellulose acetate butyrate) can substitute for PMMA [58], and the water intercalation can be replaced by HF or KOH etching [51,56,103]. Due to such methods being carried out in liquid, the fabricated heterostructures inevitably introduced trapped water blisters or wrinkles [104]. Since most 2D materials are sensitive to water and some are chemically unstable, dry transfer and integration methods have been developed.

Dry transfer.
The essential difference between dry transfer and wet transfer methods is that in dry transfer methods, water or solvents do not directly contact the surface of the 2D nanosheet that forms the van der Waals interface. The dry transfer allows the fabrication of vdWHs based on chemically sensitive 2D materials and, more importantly, a clean van der Waals interface by avoiding the contaminants and wrinkles caused by the involvement of liquids. It is imperative to achieve high-performance devices and probe the intrinsic physical property of vdWHs [13,92,105]. Dry transfer can be loosely divided into two categories: one is assisted with PMMA as the supporting layer and PVA (polyvinyl alcohol, water-soluble) as the sacrifice layer, while the others are realized by thermal release tape (TAT). Figure 2(c) illustrates the more widely used dry transfer methods assisted with the supporting layer/water-soluble layer. Figure 2(d) shows the optical images of Gr and hBN before and after transfer [13]. Specifically, the 2D nanosheet is firstly exfoliated onto the SiO 2 /Si substrate pre-covered with a supporting (PMMA) layer on top of a water-soluble layer (PVA), as shown in the left of figure 2(c). The nanosheets/PMMA stack is then detached from Si/SiO 2 substrate by dissolving the soluble interlayer in the water, while the top nanosheet never touches the water. After flipping over the stack, the 2D nanosheet is facing down and stacked with the target 2D nanosheet on the substrate to form the vdWHs, similar to wet transfer methods. This method avoids the contact of water on the top surface of the nanosheet, therefore avoiding the contamination of the interface between the two nanosheets forming the vdWHs. However, this method is only suitable for the nanosheets obtained via mechanical exfoliation, and the execution is relatively complex. Alternatively, a temperature-sensitive supporting layer instead of PMMA is used in thermal release methods, as illustrated in figure 2(e) [99]. The temperaturesensitive supporting layer can be chosen from various polymers such as Elvacite [106], MMA (methyl methacrylate) [107], TAT [99], and VEPLS (viscoelastic polymer support layer) [55]. In Elvacite and MMA methods, the supporting layers will melt during the thermal release process, and then the melted supporting materials are removed by acetone, while in TAT and VEPLS methods, the nanosheets will be left on substrates directly and separated from supporting layers during the heating process. Furthermore, to future improve the cleanness of the vdWH interfaces, researchers can choose to stack  [13]. (d) Optic images of Gr and BN before transfer, and that of the Gr/BN vdWH after all-dry transfer assisted with PMMA/PVA [13]. Reprinted by permission from Springer Nature Customer Service Centre GmbH: Springer Nature, Nature Nanotechnology [13], Copyright (2020). (e) Procedure of thermal release method [99]. Reprinted with permission from [99]. Copyright (2020) American Chemical Society. (f) Procedure of self-assembly in solution [100]. (g) Cross-sectional HR-TEM images and structural models of LDH/TMD vdWH fabricated by self-assembly method [100]. Reprinted with permission from [100]. Copyright (2010) American Chemical Society. the structures in a vacuum environment or glove box [108,109]. Though mechanical-assembly methods are deterministic and controllable, the yields are meager, and the operations are technologically complex.
2.1.3. Self-assembly. Self-assembly of 2D nanosheets in the solution can achieve alternatively stacked multilayer structures with high yields. Firstly, two types of solution-processable 2D nanosheets are respectively prepared by liquid exfoliation methods, such as ultrasound-assisted exfoliation, electrochemical exfoliation, chemical intercalation exfoliation, and chemical oxidative exfoliation. Then based on stable mixtures of the prepared nanosheets, the vdWHs can be self-assembled via ultrasonication in solution or ball milling in a solid state [52,53,110,111]. However, such assembly is a random restacking, and the qualities of assembled vdWHs are uncontrollable. In order to achieve ordered self-assembly, the two 2D components are chemically or physically modified with opposite charges. Then layer-by-layer self-assembly occurs based on electrostatic attraction (Coulomb interactions) to form vdWHs with good interfacial contact. For example, figure 2(f) shows the layer-by-layer self-assembly of positively charged LDH with negatively charged MoS 2 [100]. The cross-sectional HR-TEM images confirmed the interstratification of alternating LDH and TMD monolayers (figure 2(g)). Poly-(diallyldimethylammonium chloride) is widely explored to make nanosheets positively charged, such as Gr [112], MoS 2 [113] and reduced Gr oxide (rGO) [114]. Combined with negatively charged 2D nanosheets, such as Li-intercalated TMDs, HF-treated MXenes, and MnO 2 nanosheets, MoS 2 /rGO, MXene/rGO, and MnO 2 /rGO with alternately restacked vdWHs have been successfully synthesized [52, 115, 116].

Gas-phase deposition
Direct gas-phase deposition is an efficient strategy to synthesize a variety of 2DLMs with high qualities and yields, including CVD [89], metal-organic chemical vapour  [61]. Reprinted with permission from [61]. Copyright (2016) American Chemical Society. (c) The synthesis procedure of the MoS 2 /Gr vdWH by chemical transformation [121]. (d), (e) Cross-sectional and in-plane HR-TEM images of the MoS 2 /Gr vdWH. Respectively [121]. Reprinted from [121], Copyright (2019), with permission from Elsevier. (f) Procedures of seeded growth method [75]. (g) Photograph of VSe 2 /WSe 2 vdWH arrays on SiO 2 /Si substrate [75]. Reprinted by permission from Springer Nature Customer Service Centre GmbH: Nature. deposition (MOCVD) [117], PVD [78], atomic layer deposition [118], molecular beam epitaxy [119], electron beam evaporation [120]. Fine-tuning these deposition processes and playing with the sequences can directly synthesize vdWHs. Compared with physical combination, deposition methods have a reasonably high yield while introducing fewer interfacial contaminants between nanosheets in the vdWHs. The most common and intuitive strategy is direct layer-bylayer deposition, which generally requires good lattice matching between different 2D materials. Most recently, alternative deposition methods beyond compatible lattices are also demonstrated [81]. The following will introduce the three deposition strategies for vdWHs: direct deposition, chemical transformation, and seeded growth.

Direct deposition.
The CVD growth of vdWHs has been reviewed in many review articles [89,122]. The qualities of vdWHs, such as morphology, layer space, layer numbers, and lattice orientations, can be controlled by the growth parameters, including temperature, substrate, and precursor [77]. Additionally, the deposited materials are susceptible to substrate types, with the most common substrates being metals (such as Au [123], Cu [124], Ag [125], Ge [126] and Ir [127]) and SiO 2 /Si. The substrates usually become defective after only one deposition process, which impedes the further deposition of new materials on them [89,122,128]. Hence, many CVD-based methods must first utilize a transfer method to prepare the 2D nanosheets on the substrate before the deposition process [88,89,122]. Furthermore, since CVD deposition has strict requirements of lattice compatibility, the vdWHs can be synthesized by direct CVD methods are limited. Fortunately, it has been recently demonstrated that many 2DLMs can deposit on Gr and hBN, expanding the application potentials of CVD direct methods [88,89]. For example, such a two-step CVD method was used to fabricate large-scale MoS 2 /Gr vdWH, as shown in figures 3(a) and (b) [61]. Gr is synthesized on Cu foil and then transferred to SiO 2 /Si substrate for the further deposition of the MoS 2 layer. On the other hand, the direct deposition of vdWHs without transfer processes is also explored. For example, taking molten Au as substrate, a single crystalline hBN was grown, and then successive in situ Gr growth was conducted [59]. Still, the transfer process is also inevitably involved after vdWHs synthesis because it is impossible to fabricate devices on a molten substrate. Most recently, through controlling the deposition conditions, some vdWHs with large lattice misfits have been successfully constructed, such as GaSe/MoSe 2 [81] and GaSe/MoS 2 [62], which provides prospects for more vdWHs synthesis by direct deposition methods.

Chemical transformation.
For TMDs, an essential group of 2D materials, chemical transformation is an exciting and reliable method to synthesize the vdWHs. As shown in figures 3(c)-(e), molybdenum is first deposited on Gr to fabricate Mo/Gr heterostructure, then sulfurized in H 2 S/Ar plasma atmosphere into MoS 2 /Gr vdWH [121]. This deposition method is also suitable for the vdWHs composed of TMDs, such as WS 2 /MoS 2 [129]. Furthermore, two synthetic routes can be chosen if the non-metallic element of each van der Waals layer is the same (sulphur or selenium): (1) the bottom layer is deposited and then sulfurized or selenized, and then deposit and transform the upper layer [130]; (2) the precursors of vdWHs are deposited in order, and then sulfurize or selenize in one batch the whole structure [60]. It is worth noting that clear moiré patterns are observed for the MoS 2 /WS 2 vdWH fabricated by the second synthetic route, providing a scalable synthetic strategy of vdWHs with moiré patterns beyond mechanically twisting the van der Waals layers for the exploration of novel condensed matter physics. In addition to the metal precursors, metal oxides are also used in the chemical transformation methods [130]. Inspiringly, this method has been extended to group 10 (PtSe 2 /PtS 2 [101]) and group 14 (SnS 2 /SnS [131]), providing more opportunities for fabricating other novel vdWHs.

Seeded growth methods.
When the second layer is directly deposited on the other layer to form vdWHs, the nucleation positions and growth are random if the bottom materials are nearly perfect (with minimal defects and contaminants). Hence seeded growth methods are developed to control the process of vdWHs fabrication. These strategies control nucleation positions by artificially producing defects with high energy and chemical disorder. As illustrated in figures 3(f) and (g) [75], after the deposition of the first thin film of 2D materials, a periodic array of defects is introduced by the focused laser irradiation combined with a raster scan. Then, selective growth of the second layer of 2D materials at the desired position can be realized by seeded growth. A series of high-quality vdWHs arrays can be precisely fabricated via this strategy, including VSe 2 /WSe 2 , NiTe 2 /WSe 2 , CoTe 2 /WSe 2 , NbTe 2 /WSe 2 , VS 2 /WSe 2 , VSe 2 /MoS 2 and VSe 2 /WS 2 . Furthermore, the ideal interfaces with clear moiré patterns are observed, providing a versatile platform for exploring exotic physics.
In short, all three deposition strategies (direct deposition, chemical transformation, and seeded growth) can effectively produce vdWHs. The direct deposition method is the most common choice but is limited to the lattice compatibility between layers and the stability of the bottom layer of materials. Diverse substrates and deposition conditions have been explored to fabricate more potential vdWHs with large lattice mismatches. The chemical transfer methods are generally controllable in the quality of the vdWHs, which are, however, limited in applicable materials. On the other hand, the qualities of vdWHs synthesized by the seeded growth method are determined by two factors: the crystal quality of the bottom layers (the higher, the better to avoid extra nucleation sites) and the growth condition of the second layer.

Solvothermal synthesis
In addition to the exfoliation and deposition methods, solvothermal synthesis is well known for the cost-effective preparation of 2DLMs in high quantity [84]. Specifically, molecule precursors are dissolved and sealed in a solvothermal reactor under high-pressure environments to form products with different morphologies via controlling temperature and precursor concentration. Based on past success in solvothermal synthesis, cost-effective wet chemical synthesis of vdWHs has been developed. Similar to the previously discussed deposition methods, nanosheets of one 2D material are always needed as templates to both provide nucleation position for another 2D material and control the nucleation and growth of the second layer to form the desired vdWHs [132]. Besides, synthetic methods based on intercalation and chemical transformation are also explored. According to the different reaction processes, the solvothermal synthesis of vdWH is categorized as (1) template-assisted methods and (2) space-confined synthesis (intercalation and transformation).

2.3.1.
Template-assisted methods. TMD-based 2D vdWHs have been widely synthesized via template-assisted wetchemical methods. Many chemically stable and well-dispersed 2D materials have been used as the 2D templates, including rGO [64,69], BP [67], Mxenes [65] and graphydiyne (GD) [133]. Figure 4(a) illustrates the standard procedures of template-assisted methods, which composes the preparation, dispersion and mixture of template nanosheets (GD) and precursors (WCl 6 ) in the DMF solution [133]. The temperature and precursor concentration are important parameters that determine the intense degree of interaction between templates and precursors, ensuring nucleation position on templates rather than forming separate products. As mentioned above, the vdWHs fabricated by solvothermal methods are usually applied to energy-related applications due to the amorphous structures and abundant defects. However, solvothermal methods are relatively uncontrollable compared with deposition methods, leading to self-restack into thicker layers with decreased active sites and inferior interfacial interaction, hindering the catalytic performance (figure 4(b)). In order to solve this problem, a modified method with polyvinylpyrrolidone (PVP) surfactant and thioacetic acid (TAA) has been applied to fabricate uniform MoS 2 /rGO heterostructure and assisted with TAA and PVP, small-sized, and non-aggregated MoS 2 /rGO vdWHs are obtained [136]. Furthermore, space-confined synthesis has been explored to fabricate well-defined vdWHs as desired.
for the embedded precursors, enabling chemical transformation to occur under proper conditions. Excitingly, inspired by the space-confined methods in solutions and combing with the property of red phosphorus (RP), which can convert into layered BP under high pressure, BP/rGO heterostructure with high energy-storage performance has been synthesis at room temperature [135]. In this research, the RP precursor is sublimated and intercalated into the interlayer space of rGO to assemble RP/rGO, and then BP/rGO is obtained under the high-pressure condition (figures 4(d)-(f)). As shown in the high-resolution TEM image (figure 4(g)), the BP crystal lattice spacing can be observed forming coherent crystallites with dimensions of about 100 nm.
In short, solvothermal methods (solvothermal methods) are productive in constructing vdWHs, which include two different preparation routes, template-assisted and spaceconfined. The difference between them is the nucleation and growth position of the subsequent 2D layer. Template-assisted methods are widely used to fabricate TMDs-based vdWHs with proper temperature and precursor concentration, but the vdWH uniformity of the system is usually uncontrollable. Space-confined methods provide a more effective and desired route to synthesize vdWHs. By space-confined syntheses, several well-defined vdWHs have been successfully synthesized and show promising potential in energy-related applications.

Synchronous evolution (one-step synthesis)
Regardless of the methods mentioned above, it is inevitable to introduce contaminants or defects during transfer or fabrication processes. Recently, the synchronous evolution (onestep synthesis) has been explored to realize multiple vdWHs, consisting of a selective multi-junction sequence. The resulting vdWHs possess high crystallinity and hyper-clean interfaces, resulting in high device performance [144]. In order to achieve synchronous evolution, the synthesis route must be well-designed and the synthetic condition precisely controlled to ensure that different 2DLMs can nucleate and grow at proper positions in order.
A co-segregation method has been proposed to synthesize Gr/hBN by controlling the diffusion of atoms from a specifically designed sandwiched substrate, as illustrated in figure 5(a) [72]. During the vacuum annealing processes, dissolved C atoms first segregate from the C-doped Ni Top layer to the surface, nucleating and forming Gr on the top surface. Subsequently, B and N atoms diffuse to the top surface, nucleating and forming hBN between Gr and the top surface of the substrate. Figure 5(b) shows uniform Gr/hBN vdWHs in wafer-scale are synthesized without inherent size limitation. Furthermore, a similar strategy was employed to synthesize Gr/Mo 2 C vdWHs with the assistance of Cu foil to govern the Mo atom diffusion and tune the chemical reaction [71]. Beyond the co-segregation method, a one-step CVD method has also been developed. By preciously controlling the precursor position, heating temperature and flow rate, the heterostructures can be synthesized directly (figure 5(c)) [70]. For example, at a fixed temperature, when the S powder was 15 cm away from the MoO 3 precursors, MoS 2 nanosheets are synthesized only ( figure 5(d)), while only MoO 3 nanosheets can be observed when the distance increases over 19 cm (figure 5(e)), and MoO 3 /MoS 2 vdWHs can only be synthesized in between (figure 5(f)) [70]. The position-dependent phenomenon in CVD synthesis arises from the time and mass dependence of the sublimation of the solid precursors and spatially nonuniform growth dynamics [72].  [70]. Reproduced from [70]. © IOP Publishing Ltd All rights reserved. Overall, compared with other synthetic methods of vdWH, synchronous evolution (one-step synthesis) can avoid contaminants entirely. However, precisely controlling the diffusion and migration of atoms is quite tricky, and the successes are limited. It is worth noting that solvothermal methods can also achieve one-step heterostructure synthesis [145][146][147]. However, due to the relatively uncontrollable nucleation and combination, the synthesized heterostructures usually exhibit low crystallinity with amorphous morphology, which could be favored in electrochemical applications. Finally, the general limitations and advantages of existing synthetic strategies of vdWHs are summarized in table 1.

Applications based on vdWHs
The successful synthesis of various vdWHs opens unprecedented opportunities for electronic, optoelectronic, and energy-related devices with novel designs and exceptional performance. This section will summarize these three typical application fields with novel designs of vdWHs based on the diverse 2D materials and flexible integration strategies of vdWHs.

Vertical field effect transistors.
As conventional Si transistors approach the projected limit of about 5 nm channel lengths, 2DLMs began to be applied to channel materials to shorten the channel lengths [148]. However, the lateral FETs based on a single 2D semiconducting material usually suffer from considerable contact resistance resulting from high Schottky barriers between metal electrodes and semiconducting materials [149]. Inspiringly, vertical FETs (VFETs) based on vdWHs settled the contact problem through the novel design of devices. Furthermore, creating VFETs promises the ultimate channel lengths, which only depend on the thickness of the 2D semiconductors [150,151]. Moreover, VFETs based on vdWHs have great potential to reduce the supply voltage, solving the high-power consumption of conventional devices.
FETs based on vdWHs have attracted much attention as an alternative to conventional transistors. The first reported TFET is based on Gr/hBN/Gr vdWHs with an ON/OFF ratio of less than 10 due to the large effective tunnelling barrier (figures 6(a)-(e)) [22]. The tunable-Fermi level range is much smaller than the energy difference between the Dirac points of Gr and the upper edge of the valence band of hBN, leading to an overall tuning range by tunneling and a small ON current. To increase the ON current, TMDs, with much smaller bandgaps than hBN, have been introduced to replace hBN as barrier materials. However, due to the pinned effect near the top and bottom junctions, the ON current of FET based on Gr/MoS 2 is still small for practical applications (over 0.1 µA µm −2 ), though the ON-OFF ratio is much higher than Gr/hBN/Gr. By replacing top Gr with metal, the ON current density increases to 5 × 10 −5 A µm −2 , though the ON/OFF ratio drops to 10 3 at room temperature (still acceptable for practical application) [25]. The increase of ON current is attributed to the thermionic emission, which can also explain the temperature dependence of the performance of TFET based on Gr/WS 2 /Gr vdWH [24]. The TFET based on Gr/WS 2 /Gr vdWH shows excellent performance with an ON current of 2 × 10 −6 A µm −2 and an ON/OFF ratio of over 10 6 at room temperature (figure 6(g)), attributed to the proper bandgap alignment (figures 6(f) and (h)-(j). When the Fermi level in Gr is tuned to align with the conduction band of WS 2 , the barrier height dramatically decreases, leading to a thermionic current. In addition, with the rational design of band alignments of the vdWH of TFETs, TFETs with ultralow subthreshold swing (SS) reached lower than 60 mV dec −1 based on different materials have been fabricated [152,153]. However, the barrier height is not enough to block the thermal current at the OFF state, resulting in excellent temperature dependence on the performance of this transistor.
As mentioned above, many methods have been tried to simultaneously increase the ON current density and ON/OFF ratio to improve Gr-based VFETs [18,25]. To further simultaneously increase the ON current density and ON/OFF ratio of VFET, the Gr-based resonant tunneling FET has been fabricated and studied due to the enhancement of tunneling probability via interlayer scatterings [44]. For Gr/hBN/Gr resonant tunneling FET, the resonant scattering can arise from pot associated with the short-range disorder and the moiré pattern at the interface of Gr and hBN. Compared with straightforward FET based on Gr, the performance of resonant tunneling FET is improved, but the ON current density remains unsatisfactory [25]. To obtain interlayer scattering between the top and bottom Gr, a large momentum transfer is needed, which can be achieved by twisting Gr layers [43]. As a result, the bias voltage for the resonant peak is only determined by the twisted angle between the top and bottom Gr layers. As a result, the ON current density can increase to over hundreds of nA µm −2 .
VFETs based on stacking n-type and p-type 2D semiconductors with type II band alignment are also promising, such as Mos 2 /WSe 2 [154], BP/SnSe 2 [155], GaTe/MoS 2 [156], WSe 2 /SnS 2 [157]. In addition to the high ON/OFF ratio (10 3 -10 7 ), the p-n junction FETs show an excellent tunable rectification ratio, the mechanism of which remains controversial [154][155][156][157]. Furthermore, a low SS of 37 mV dec −1 is achieved by SnSe 2 /WSe 2 vertical p-n junction FET [158], exceeding the thermionic limit at room temperature in conventional FETs. Moreover, the energy band can be tuned from type II to type III when changing the gate voltage from negative to positive, leading to the device working as FET instead of a diode.
The vast exploration has demonstrated VFETs based on vdWHs as a promising platform for electronic applications. However, understanding electronic physics remains arbitral, and novel or hybrid theories are required to advance the field.

Van der Waals contact.
Besides novel vertical transistor structures, 2D vdWHs benefit 2D FETs by establishing pinning-free contact with semiconductors [159,160], which remains challenging. Gr has been considered a promising contact material for 2D semiconductors due to the gate-tunable work function without band-matching, as proved in the WSe 2 transistors with Gr contacts [161]. Afterwards, Gr was used as back electrodes to achieve ohmic contact with MoS 2 even at cryogenic temperatures (figures 7(a)-(c)) [17], thus exhibiting significantly enhanced on-current and apparent two-terminal mobility performance compared with the previous MoS 2 . Furthermore, the performance can be further improved via hBN passivation and encapsulation (figures 7(d)-(f)) [162], all of which are based on van der Waals integrations. In addition to Gr contacts, other 2D materials have also been demonstrated to suppress contact resistance and boosting performance. For example, WSe 2 FET with metallic NiSe contacts deposited by the CVD method shows low resistance and record-high current density [163]. Alternatively, the introduction of lateral multijunction between semi-metallic and semiconducting 2D materials has shown promising performance with significantly reduced contact resistance [164][165][166][167]. For example, the most successful strategy is in-planar heterojunction between semiconducting and semi-metallic MoS 2 , showing record low contact resistance [166]. The atomical cleanness of the vdW interface is essential to push the performance limit of 2D transistors based on 2D semiconductors.

Optoelectronics
With the wide variety of 2D materials with tunable electronic and optoelectronic properties, a wide range of 2D vdWHs and devices may be designed for creating novel photodetectors, light-harvesting devices, and LEDs [52,168]. Among them, infrared photodetectors based on vdWHs have attracted considerable attention due to their high sensitivity and broad detection range [10,12,169]. To make a response to infrared light, it is required that the bandgap of materials should be narrow, which can be achieved by choosing 2DLMs with proper energy band alignments. Furthermore, vertical vdWHs have ultrashort charge transfer channels determined by the thickness of 2DLMs, enabling the photodetectors with ultrafast response speed and high response sensitivity. In addition, the  [32]. (c) Semilogarithmic plot of I ds -V ds characteristic curves with (red) and without (dark) the light on [32]. (d) The extracted wavelength dependent photoresponsivity (red) and noise equivalent power (blue) of the PdSe 2 /MoS 2 photodetector [32] [31]. Reprinted with permission from [32]. Copyright (2019) American Chemical Society. efficient electron-hole pairs separation can be achieved by the solid build-in electrical field inside vdWHs with type II band alignments, contributing to high response speed.
The photodetectors based on p-type TMDs/Gr/n-type TMDs have been demonstrated as a possible choice for infrared detection with an ultrafast response and high detectivity [170]. However, due to the weak optical absorption of the Gr, the external quantum efficiency referring to the responsivity and noise level of the photodetector is not ideal. Alternatively, BP is projected to be a better absorption layer with a narrow bandgap (from 0.3 eV for bulk to 2 eV for monolayer) [171]. In addition, the bandgap of BP cane is effectively tuned by alloying with As (B-AsP), as shown in figure 8(a) [31]. A mid-infrared photodetector based on the structure of p-type B-AsP/n-type MoS 2 is fabricated, showing good responsivity, external quantum efficiency and fast response speed [30]. Furthermore, the photodetectors with high external quantum efficiencies (over 30%) are fabricated with the BP/MoS 2 and BP/InSe heterostructure, further demonstrating the potential of BP-based vdWHs in photodetectors [33,172].
TMD/TMD vdWHs-based photodetectors had largely been limited to visible wavelength ranges due to the large bandgaps.
Fortunately, abide the integration of new 2D materials, such as PdSe 2 and PtSe 2 (with tunable bandgap from semimetal for bulk materials to about 1.2 eV for monolayer), novel TMD/TMD-based vdWHs have attracted wide attention for infrared photodetection. For example, the photodetector based on MoS 2 /PdSe 2 vdWH shows a wide achievable range from 450 nm (visible) to 10.6 µm (long-wavelength infrared), as shown in figures 8(b) and (c) [32]. Interestingly, the highest photoresponsivity occurs at 4 µm, possibly due to the formation of the interlayer exciton. Furthermore, the MoS 2 /PdSe 2 photodetector also shows excellent stability, which shows no sign of degradation after one year in the air atmosphere.

Energy conversion and storage
Various 2D materials have been intensively explored and optimized as promising alternatives in energy-related fields [84,173]. However, it is hard to realize the stability and electrochemical activity simultaneously. For example, Gr has high stability and carrier mobility but lacks sufficient active sites for electrocatalysis. On the other hand, TMDs and TMOs are highly electrochemically active with abundant redox sites, but they are chemically unstable with poor conductivity. 2D vdWHs provide possibilities to integrate the best of both worlds and overcome the intrinsic shortcomings of materials [174]. In this section, the application of vdWHs in electrocatalytic water splitting, metal ion batteries, and photovoltaic devices are summarized as representatives of their application in energy conversion and storage.

Electrocatalysis for water splitting reactions.
Hydrogen is a promising future energy source, which can be obtained by electrochemically converting from water. To achieve scalable water electrolysis, proper catalysts are required for effective and efficient energy conversion by hydrogen evolution reaction (HER) and oxygen evolution reaction (OER). Especially to replace current commercialized Pt-based catalysts, many 2DLMs have been explored as candidates for HER catalysts with low cost and high efficiency, such as Mxenes [175,176], TMDs [177], and TMOs [178]. Among them, TMDs, especially MoS 2 , remain the most promising non-noblemetal electrocatalyst. However, despite significant research efforts, their performance remains far from practical. Additional chemical/morphology engineering is required to elevate their performance, including chemical doping and conversion [179], defect engineering [180], field modulation [181], and many more. Alternatively, vdWHs have been demonstrated as promising alternatives with improved activity and stability.
The vdWHs synthesized by solvothermal methods, such as MoS 2 on other 2D material templates, have been explored to improve the overall conductivity of 2D catalysts, showing improved catalytic performance compared to pristine materials [39,40,67,182]. Furthermore, the van der Waals interaction between different 2D materials can induce charge redistribution at the interface of vdWHs, which can modulate the electronic structures and modify the hydrogen adsorption energy, leading to enhanced electrocatalytic performance. We took MoS 2 /BP vdWH as an example to explain the influence of electronic-structure modulation on electrocatalysis, which is fabricated by conventional self-assembly in ethanol [67]. In this vdWH, the P 2P, S 2p, and Mo 3d peaks shift 0.1 eV to higher energy, 0.8 eV to lower energy and 0.7 eV to lower energy, respectively. The results indicate electron transfer from BP to MoS 2 , which attribute to the enhanced reaction kinetics of MoS 2 in HER. As a result, the vdWH shows good electrochemical performance with an overpotential of 85 mV at 10 mA cm −1 and a Tafel slope of 68 mV dec −1 . On the other hand, OER is generally more complicated than HER due to the involvement of multi-electron and multi-intermediate. NiFe/MoS 2 vdWH synthesized via space-confined method exhibits extraordinary OER performance, with a small overpotential of 260 mV at 10 mA cm −2 and a low Tafel slope of 48 mV dec −1 , which is much better than the performance of pure NiFe or MoS 2 [183].
On the other hand, the relationship between interface effect and catalytic performance in vdWHs with amorphous shapes is difficult to attribute [185]. Hence, vdWHs with large sizes and controllable morphology are employed to decouple the detailed mechanism of vdWH-based electrocatalysis. The large-scale vdWH composed of exfoliated BP (EBP) and N-doped Gr (NG) has been synthesized by self-assembly, showing impressive bifunctional activities towards both HER and OER [184]. The lower Fermi level of EBP than that of NG allows EBP to extract electrons from NG (figures 9(a) and (b)). The redistribution of the electronic structure at the interface of the vdWH not only optimizes H adsorption and desorption over EBP but also introduces abundant positive charged sites onto NG to favor the formation of the OER intermediates, simultaneously enhancing HER and OER energetics (figures 9(c)-(f)). It is worth noting that both BP and Gr are considered electrochemically inert 2D materials. However, the EBP/NG shows outstanding electrocatalytic performance, highlighting the importance of the rational design of vdWHs based on interface effects.

Metal ion batteries.
Layered materials are seen as promising candidates for metal ion batteries due to their unique 2D channels between layers, which provide sufficient ion storage sites. However, current 2D materials still fail to satisfy all the parameters required for ion metal batteries, such as high specific capacity, lower diffusion and good cycle stability. Li-ion batteries (LIBs) still dominate in the current consumer society due to better performance compared with other counterparts such as Na ion batteries (SIBs) and K ion batteries (PIBs). TMDs, especially MoS 2 and WS 2 , have been widely investigated as anode materials in LIBs. TMDs possess higher specific capacity due to abundant redox sites but insufficient cycling stability compared with traditional graphite LIBs. The fast structure dissociation of MoS 2 during the lithiation process results from the insufficient conductivity to meet the charge balance on the electrode with fast electron exchange, though it can be alleviated to a certain degree with phase engineering by introducing metallic metastable phase during Li interaction [186,187]. To this end, MoS 2 /Gr vdWHs [188][189][190] have been developed via templated-assisted solvothermal methods as the anode electrode in LIBs with reversible capacity up to 1800 mA h g −1 , which is much higher than the theoretical evaluated value of Li x MoS 2 (460 mA h g −1 ) [191]. The much-improved performance in the LIBs based on vdWHs structures is attributed to the synergetic effects of the 2D confinement effect, improved conductivity and enhanced structural stability. Based on the design criteria, many other vdWHs, such as MXenes/TMDs [192], TMOs/Gr [52], and BP/Gr [193] have also been reported for high-performance LIBs.
Though LIBs have excellent performance in energy storage, the relative scarcity of Li element necessitates the further development of non-Li batteries such as SIBs and PIBs. To this end, TMDs/Gr vdWHs have been utilized in both SIBs and PIBs [194]. Unlike LIBs, Na and K tend to be absorbed on the TMDs surface of vdWHs, rather than intercalate into the interlayer [195]. Hence, the enhanced performance of vdWHs in SIBs and PIBs mainly results from the increased electron transfer efficiency of TMDs and following higher pseudocapacitance contribution [196]. To this end, the capacities of vdWHs for SIBs and PIBs largely depend on the layer number of TMDs, which highly influence the electron transfer properties. On the other hand, the quantity of Gr of vdWHs also influences the performance. That is why MoS 2 /Gr synthesized by ball milling with relatively poor contact between vdWH layers in the vdWHs exhibits high-rate capability and great cycling stability [197]. Additionally, BP/Gr [141,198], TMOs/Gr [190,199] and other vdWHs [19,22,200] have been attempted for SIBs and PIBs.
vdWHs based on 2D materials also emerged as competitive candidates for solar cells [201]. The most conventional option for realizing solar cells is the formation of p-n junctions where photogenerated charge carrier separation occurs. A natural p-n junction can be fabricated by stacking two 2D semiconductor flakes with opposite conduction types together, such as n-type MoS 2 /p-type WSe 2 [202], p-type BP/n-type MoS 2 [203], p-type WSe 2 /n-type MoSe 2 [204], p-type α-MoTe 2 /ntype MoS 2 [205], p-type SnS 2 /n-type MoS 2 [206], n-type MoS 2 /p-type Bi 2 Te 3 [207], p-type GaSe/n-type MoSe 2 [208]. One major limitation in the vdWH strategy is the scarcity of intrinsic p-type 2D semiconductors [209]. In this regard, many modulation methods, such as electrostatic gating, contact engineering and chemical doping, have been explored to create p-type conducting and achieve desired photovoltaic devices [210][211][212]. In addition to the 2D TMDs, 2D halide perovskites have gained unprecedented interest in vdWHs for solar cell applications [213,214]. Among them, Dion-Jacobson and alternating-cation-interlayer 2D halide perovskites exhibit the most promising photovoltaic performance with superior chemical stability and intriguing anisotropic properties [215].
Overall, with the continuous successes in synthetic methods, vdWHs have attracted increasing interest in energy devices, showing unique properties and high performance.

Fundamental study of emerging physics
In addition to functional devices, many attractive physical phenomena emerge from unique vdWHs, particularly the moiré superlattice with a twisted angle of the component layers. A typical example is the 'magic angle (1.1 • ) twisted bilayer Gr', which has a flat band formed by the hybridization of Dirac cones (figures 10(a) and (b)), leading to a strong correlation of electrons and Mott-like insulator behavior at half-filling of this flat band (figure 10(c)) [216,217]. Furthermore, superconducting can be achieved via electrostatic doping of the Gr away from the correlated insulting states with a critical temperature of up to 1.7 kelvin (figure 10(d)) [218]. Another notable example is resonances caused by moiré potentials wells related to moiré patterns. Such resonant phenomenon was observed in WSe 2 /MoSe 2 /BN vdWH with interlayer resonance peaks in photoluminescence spectra, proving the formation of interlayer moiré excitons (figure 10(e)) [219]. In addition, the gate tunability of 2D vdWHs is promising to construct resonant tunnelling FETs, as discussed above, and study the charge doping on moiré excitons (figure 10(f)) [220]. It is worth noting that the dry transfer and assembly is most used in the moiré patten-related research, which requires an exceptionally clean van der Waals interface and high controllability.
In addition, the significance of the twist angle between different layers of vdWHs has been widely recognized. This new degree of freedom breaks through the symmetry restrictions imposed by the thermodynamics in natural crystals and makes vdWHs attractive for investigating novel physical phenomena.

Conclusions and outlook
In this review, we firstly systematically summarize the main synthetic strategies of vdWHs separated into four categories: physical combination methods, gas-phase deposition methods, solvothermal methods, and synchronous evolution. We then reviewed the applications of 2D vdWHs synthesized via introduced strategies, including electronics, optoelectronics and energy-related applications. Furthermore, the importance of vdWHs for exploring novel many-body physics was illustrated by the example of the moiré pattern.
Remarkable progress has been made in the fabrication and applications of vdWHs in the last decade. However, the full potential of vdWHs is still out of our scope, and their further development is still challenges. Firstly, a general synthesis method enabling scalable and controllable fabrication of vdWHs by design is still beyond reach. High yield and high quality are like two sides of the same coin that cannot be satisfied simultaneously. Mechanical assembly is the most intuitional and maneuverable strategy with high interfacial quality. Hence, it has become the standard protocol for exploring novel properties of the vdWHs, such as the 2D moiré patterns and various (opto) electronic devices [38]. Nevertheless, contaminants and defects are inevitably introduced due to the complex operation, even in dry transfer methods operated in a vacuum environment. Furthermore, the labor-intense and lowyield fabrication remains the most critical intrinsic limitation of these mechanical 'peel and stack' methods. On the other hand, gas-phase deposition is a promising synthesis strategy with much anticipation. However, most of them involve transfer procedures due to the sensitivity of nucleation and growth of 2DLMs to the substrates, inevitably leading to interfacial contaminants. In addition, the synthesized vdWHs usually have poor crystalline quality with multiple grain boundaries and massive grain boundaries, which is caused by the highdensity nucleation and relatively random growth during the deposition of the second 2D material layer. Hence, the current efforts focus on the control of the nucleation and growth of 2D materials. Among them, chemical transformation methods have been proposed to synthesize high-quality TMDs-based vdWHs via sulfurization or selenization. In addition, wetchemistry-based solvothermal methods shine in the high-yield fabrication of vdWHs. It is, however, limited to amorphous structures and binary systems. Moreover, the defective structures inherited from wet-chemical synthesis may be suitable for catalytic applications with abundant activity sites. Finally, synchronous evolution seems to be the ideal method to synthesize high-quality and desired vdWHs at large scales. However, the choice of suitable materials is still extremely limited, and the synthetic conditions to satisfy are very strict, requiring the precious control of the diffusion of precursor atoms or molecules and dedicated reaction procedures and conditions.
To further the field, a fundamental understanding of nucleation and growth of materials is required, and more synthesis strategies should be explored and applied. For example, reverse flow chemical vapor deposition [221,222] MOCVD [117] and vertical chemical vapor deposition [104] have demonstrated considerable success in synthesizing various vdWHs, including a series of TMDs with variable structural configurations and tunable electronic and optical properties [104,117,221,222].
vdWHs based on varying 2D materials have been widely explored as a promising platform to implement applications in the fields of electronics, optoelectronics, and electrochemistry. Each field is facing diverse key challenges. For electronic applications, the lack of comprehensive and cohesive fundamentals of device physics at low dimensions is problematic. For example, Richardson's law is still the primary theoretical basis for bulk materials to explain the thermionic emission transporting over the Schottky barrier between semiconducting 2DLMs and metals or Gr. However, the thickness of the nanosheets is often similar or even smaller compared with de Broglie wavelength. Moreover, thermionic carrier emission across high interfacial barriers involves energetic electrons at positions far away from the Dirac point. Therefore, the conventional theories and approximation models are insufficient to investigate the electronic physics of vdWHs, though some revisions have been attempted [223]. For photodetectors, getting a low dark current is still challenging, though the high barriers at the interfaces of vdWHs can suppress the dark current to a certain degree. Currently, the high sensitivity and fast response cannot be satisfied simultaneously in vdWHs based on infrared photodetectors, severely limiting their practical applications. In electrocatalysis, on the one hand, the desirable interfacial coupling effects are highly dependent on the quality of vdWHs; on the other hand, the amorphous structures can enhance the catalytic performance due to more edge sites. vdWHs with optimized catalytic performance require balancing from both. Furthermore, the stability of the vdWH electrocatalysts remains the biggest challenge for practical applications. More efforts should be made to understand the electrochemical fundamentals of the emerging 2D materials, and their vdWHs are in demand.
Novel chemical and physical phenomena caused by the interaction among different layers make vdWHs particularly attractive. The superlattices resulting from the twisting angle have many exciting and potential properties that have only just begun to be explored. Although many vdWH-based superlattice systems have been synthesized and studied, most are found by coincidence, besides the limited superlattices fabricated by deterministically mechanical placement. However, the crystallographic orientation of the neutrally grown material is usually fixed, leading to a fixed twist angle between different layers. Until now, the most feasible and controllable synthetic methods of such superlattices are based on dry transfer and integration, in which the twist angle can only be an inferred posterior by the transport or scanning probe measurements. It is, therefore, highly desirable to develop a scalable production of vdWH superlattice with twist angles, likely gas-deposition methods.
Despite challenges everywhere, vdWHs are already a force to be reckoned with, reaching performance beyond conventional materials and unveiling novel physical phenomena previously inaccessible to conventional methodologies. More work is anticipated in this exciting new field.

Acknowledgments
Q H thanks the support from the Grants (9229079, 9610482, 7005468) from City University of Hong Kong and Early Career Scheme Project 21302821 from Research Grants Council.

Conflict of interest
The authors declare no conflict of interest.