1 Introduction
Since its discovery in 2004, graphene (Gr) has triggered extensive research attention due to its unique two-dimensional (2D) structure and excellent physical properties, including high carrier mobility, excellent mechanical strength, outstanding thermal conductivity, and broadband optical absorption [
1]. More and more researchers have focused on other 2D materials with similar properties, such as hexagonal boron nitride (h-BN), black phosphorus (BP), metal halide perovskites (MHPs), transition-metal oxides (TMOs), and transition-metal dichalcogenides (TMDs) [
2−
13]. In particular, TMDs materials exhibit diverse applications in photodetectors, photodiodes, transistors, sensors, photocatalysts, resonators, memristors, logic circuits, and smart devices due to their high carrier mobility, tunable bandgap, strong light absorption, and light-material interactions [
14−
35].
Photodetectors are optoelectronic devices that can convert an optical signal into an electrical signal and are employed in several fields, including ray measurement and detection [
36,
37], imaging [
38−
40], and optical communication [
41,
42]. With the in-depth study of the photoelectric properties of 2D materials, TMDs have shown great potential in the application of photodetectors. Chen
et al. [
43] fabricated a photodetector based on a MoS
2/MoSe
2 heterojunction with high performance, demonstrating responsivity (
R) of 1.3 A/W, detectivity (
D*) of 2.6 × 10
11 Jones, and external quantum efficiency (EQE) of 263.1%. Lopez-Sanchez
et al. [
44] used mechanical stripping to obtain a monolayer of MoS
2 and constructed photodetectors. The devices exhibited a broad spectral response in the 400−680 nm range and
R of 880 A/W. However, the long carrier lifetime limited the response time (
τ) of the devices to just 4 seconds. Although these photodetectors have shown excellent performance, they often require an additional external power source to facilitate the directional movement of the carriers, which results in increased energy consumption. In the face of the growing energy crisis, self-powered photodetectors (SPPDs) offer a viable solution. They not only reduce energy consumption but also decrease the weight and size of the devices, thereby achieving applications in many challenging environments.
SPPDs, characterized by lightweight, small size, and low energy consumption, can independently realize light detection without relying on an external power source. SPPDs can be classified according to the self-powered principle into three main categories: heterojunction [
45−
48], Schottky junction [
49,
50], and photoelectrochemical [
51,
52]. Heterojunction photodetectors have been widely studied, exhibiting characteristics such as excellent photoresponse, fast
τ, low dark current, and a large optical switching ratio (
Ion/
Ioff). The self-powered principle of heterojunction photodetectors is the photovoltaic effect (PVE). When light strikes a semiconductor material, the material absorbs the photon’s energy and drives electrons from the valence band to the conduction band, forming an electron−hole pair. The built-in electric field (BEF) drives holes to flow from the n-type region to the p-type region, while electrons flow from the p-type region to the n-type region. When the circuit is connected, the separated charge carriers form a photocurrent, thus realizing the direct conversion of light energy into electrical energy. Consequently, heterojunctions can exhibit photodetection capabilities even at zero bias.
SPPDs based on TMDs can be classified into vertical heterojunction photodetectors and lateral heterojunction photodetectors. Due to the absence of dangling bonds on the surface of TMDs, vertical heterojunction photodetectors can be prepared by the arbitrary stacking of two different materials without considering lattice mismatch. This provides more possibilities for the design of heterojunction photodetector. Lateral heterojunction photodetectors are engineered to provide photoresponse capabilities across the entire channel, facilitating the efficient collection of photogenerated carriers. In addition, lateral homojunction photodetectors have been investigated through the modulation of doping levels, which eliminates the need for complex material matching and interfacial state considerations.
It has been determined that the Ion/Ioff ratio of SPPDs is typically limited, the optical response range is constrained, and τ and R require further enhancement. In addition to exploring other novel photovoltaic materials, strategies such as designing more rational band-matching structures, micro-modulation using interface engineering, and optimization of device structures using structural engineering can be employed to optimize the BEF. Moreover, the carrier type and concentration can be modulated by surface charge transfer techniques, thickness modulation, and metal contact doping. These strategies improve the τ and R of SPPDs, thereby enhancing their device performance and application range.
A summary of recent research on SPPDs offers valuable theoretical and experimental guidance for the design and construction of high-performance photodetectors, as illustrated in Fig.1.
In this review, we briefly introduce the optical and electronic properties of TMDs. Next, heterojunction photodetectors based on TMDs are classified into vertical and lateral structures according to their structural types. For the vertical heterojunctions photodetectors, we review the research on applying energy band alignment and interface modulation to improve device performance in recent years. We also discuss the research on lateral heterojunctions to address the problems of uncontrollable growth and synthesis, as well as relatively limited carrier modulation. Finally, we summarize the current challenges and issues faced by SPPDs based on TMDs and provide a forward-looking perspective. Our review provides theoretical references for further improving the performance of SPPDs.
2 Structure and properties of TMDs
2D TMDs have become a hotspot at the forefront of scientific research. These materials adopt an MX
2 configuration, where M represents transition metals such as molybdenum (Mo) and tungsten (W), and X represents chalcogen elements such as sulfur (S) or selenium (Se). Single-layer TMDs are composed of three atomic layers, with metal atoms sandwiched between two chalcogen layers, as shown in Fig.2(a). This structure enables them to exhibit various electronic properties. For example, MoS
2 and WS
2 typically exhibit semiconducting properties, whereas NbS
2 and TaS
2 exhibit metallic properties [
53].
TMDs have demonstrated many advantages in the field of photodetectors. In terms of optical properties, most TMDs share a similar characteristic regarding their energy band structure, where the bandgap is influenced by the number of layers. Fig.2(b) and (c) show the bandgap values obtained through first-principles calculations. The bandgap of MoS
2 is found to be 1.9 eV for monolayers, 1.6 eV for bilayers, and about 1.2 eV for bulk materials. The bandgap of WS
2 is found to be 2.1 eV for monolayers and about 1.3 eV for bulk materials [
54,
63]. Moreover, due to their tunable bandgap, TMDs can achieve broad-spectrum absorption ranging from the visible to the near-infrared region, thereby enabling broad-spectrum detection.
In terms of electrical properties, the electron and hole mobilities of TMDs are usually high. For example, MoTe
2 has a mobility of 2500 cm
2·V
−1·s
−1, and PdSe
2 has a mobility of 294 cm
2·V
−1·s
−1 [
55]. Although Gr has a higher carrier mobility of 200 000 cm
2·V
−1·s
−1 [
56], its zero bandgap limits its application in photodetection. In a monolayer structure, TMDs exhibit a direct bandgap, which gives single-layer TMDs an advantage in terms of light emission efficiency. Fig.2(d) summarizes the carrier mobility of 2D monolayer TMDs from theoretical studies.
In terms of material preparation, TMDs have a graphene-like atomic structure and properties, and the thickness of the molecular layers of TMDs is about 6 to 7 Å. This atomic-level stacking results in weak interlayer van der Waals forces, enabling facile exfoliation into bilayers or monolayers, thereby providing a high degree of manipulability. Meanwhile, due to the absence of dangling bonds on the surface of TMDs, layered TMDs can form heterojunctions with other 2D materials via van der Waals forces without considering lattice matching. Therefore, TMDs hold significant value in optoelectronic devices, including photodetectors [
57], photodiodes [
58], photosensors [
59], and integrated circuits [
60−
62].
3 SPPDs based on TMDs with vertical heterojunctions
The 2D TMDs can be exfoliated from bulk materials and subsequently stacked to form a variety of novel vertical heterojunctions. A larger contact area is beneficial for carrier separation and transport in vertical heterojunction photodetectors, resulting in excellent optoelectronic performance of the device [
65,
66]. In previous reports, vertical heterojunctions can be prepared by either the direct growth method or the transfer method. Lee
et al. [
67] employed a direct growth method to fabricate highly ordered WSe
2/MoS
2 heterojunctions, achieving an
R of 5.39 A/W and a
τ of 16 μs, which is attributed to the clean and densely packed heterojunction interface. In addition, Cheng
et al. [
68] constructed a WSe
2/MoS
2 photodiode by vertically stacking n-type MoS
2 with p-type WSe
2 via mechanical exfoliation and transfer. Under 512 nm laser irradiation, the PVE is evident. The device exhibits an open-circuit voltage (
Voc) of 0.27 V and a short-circuit current (
Isc) of 0.22 μA. Although the WSe
2/MoS
2 heterojunction has not been directly applied to the design of SPPDs, its strong BEF indicates potential for SPPDs. In recent years, energy band alignment engineering, interface engineering, and structural engineering have been widely applied to improve the performance of SPPDs. In the following, we summarize the research and progress in vertical heterojunction SPPDs in recent years.
3.1 Energy band alignment engineering
Energy band alignment engineering typically entails the combinatorial control of the energy band structure of the material. This approach has been demonstrated to effectively promote the separation and transport of photogenerated electron−hole pairs [
69]. Energy band alignments are typically classified into three categories: straddling (type-I), staggered (type-II), and broken gaps (type-III), as shown in Fig.3.
In type-I heterojunctions, the carriers are confined within the smaller bandgap material, which facilitates the recombination of photogenerated carriers [
70,
71]. This phenomenon arises due to the higher conduction band minimum and lower valence band maximum of the smaller bandgap material relative to those of the larger bandgap material. Gui
et al. [
72] fabricated a PtSe
2/MoS
2 heterojunction by directly growing PtSe
2 thin films on MoS
2 nanosheets using a two-step method [as shown in Fig.4(a)−(c)]. The photodetector exhibits type-I energy band alignment [Fig.4(d)], with its spectral response expanding from 405 nm to 1550 nm. Under a
Vds of 1 V, the device achieves a high
R of 5.42 A/W,
D* of 2.52 × 10
10 Jones, and τ
rise/τ
fall of 92 μs/112 μs. In addition, the device demonstrates significant self-powered capability and high photodetection performance. At 0 V bias, the device achieves a
R of 1.35 A/W and a
D* of 2.95 × 10
9 Jones.
He
et al. [
73] fabricated a p-Te/n-WSe
2 heterojunction photodetector with type-I energy band alignment, as shown in Fig.4(e)−(g). This design promotes effective charge separation and strong interlayer coupling at the interfaces, leading to excellent optoelectronic performance. Under a −2 V bias and 405 nm illumination, the photodetector exhibits a current rectification ratio exceeding 10
4, a dark current of 10
−11 A, and an
Ion/
Ioff ratio of up to 5657. Significantly, the p-Te/n-WSe
2 photodetector demonstrates strong self-powered characteristics with an
Ion/
Ioff ratio exceeding 5 × 10
4, a
R of 196 mA/W, a
D* of 7.5 × 10
11 Jones, and a
τ of 18 μs at 0 V bias, as shown in Fig.4(h).
Type-II heterojunctions are particularly advantageous for the design and optimization of SPPDs. The staggered energy band alignment between materials facilitates the effective separation of photogenerated electrons and holes, thereby enhancing photoelectric conversion efficiency [
74]. He
et al. [
75] fabricated an InSe/MoTe
2 type-II heterojunction SPPD [as shown in Fig.5(a)] with asymmetric contacts by stacking exfoliated multilayer MoTe
2 and InSe onto Au electrodes using a dry transfer method. Benefiting from the reduced barrier at the Au/InSe interface and the BEF formed at the InSe/MoTe
2 and MoTe
2/Au interfaces in the same direction, the PVE and the performance of self-powered photodetection have been significantly enhanced. The InSe/MoTe
2 heterojunction exhibits outstanding photovoltaic performance with a large
Voc of 390.19 mV and a
Isc of 20.32 nA. Furthermore, the device exhibits an
Ion/
Ioff ratio up to 10
4, τ
rise/τ
fall of 99 μs/117 μs, a
R of 433.88 mA/W, and a
D* of 1.65×10
12 Jones [Fig.5(c,d)] at 0 V bias.
Zhang
et al. [
76] prepared a GeSe/WS
2/MoS
2 heterojunction SPPD, as shown in Fig.5(e). The energy band diagram of the device is shown in Fig.5(f), where the three materials form a stepped type-II energy band alignment. Benefiting from this ingenious heterojunction energy band design, the GeSe/WS
2/MoS
2 heterojunction SPPD exhibits excellent performance at 0 V bias, including an
Ion/
Ioff ratio of 3.2 × 10
3, a dark current of 1.1 × 10
−13 A, a
R of 14 mA/W, an EQE of 4.1%, a
D* of 7.3 × 10
8 Jones, and
τrise/
τfall of 2.4 ms/5.2 ms [as shown in Fig.5(g, h)]. Furthermore, the device demonstrates stable operation under continuous high-frequency switching cycles.
Unlike type-I and type-II energy band alignments, type-III energy band alignment enables significant band shifts and band-to-band tunneling (BTBT) at the heterojunction interface. This facilitates interfacial charge transfer and further enhances response speed [
77]. Pan
et al. [
78] proposed ReSe
2/SnSe
2 SPPDs with type-III energy band alignment, as shown in Fig.5(i). On this basis, it was found that the multilayer graphene (MLG)/ReSe
2/SnSe
2 van der Waals heterojunction, after simple combination with MLG, exhibited excellent performance, as shown in Fig.5(g)−(l). The introduction of the ReSe
2 unilateral depletion region results in a strong BEF, which significantly reduces the dark current. More importantly, due to the synergistic effect of unilateral depletion and photoinduced tunneling current, the MLG/ReSe
2/SnSe
2 device demonstrated excellent self-powered detection properties. Under 0 V bias and 635 nm illumination,
τrise/
τfall is 752 μs/928 μs,
R is 144 mA/W, and
D* is 2.4 × 10
10 Jones. Fig.5(m) shows the type-III energy band alignment of the device at different bias voltages. Under reverse bias (
Vds < 0 V), the depletion of ReSe
2 and the electron accumulation in SnSe
2 result in minimal charge flow through the heterojunction. When a small forward bias (0 <
Vds < 0.1 V) is applied, a triangular potential barrier is formed in the heterojunction due to the conduction band offset between ReSe
2 and SnSe
2, which hinders the movement of electrons from SnSe
2 to ReSe
2. As a result, diffusion transport behavior dominates the forward current. Under a larger forward bias voltage (
Vds > 0.1 V), the triangular barrier region narrows and sharpens, leading to the continuous generation of field-enhanced tunneling transport behavior and a dramatic increase in forward current.
In summary, energy band alignment engineering plays a pivotal role in the heterojunction. It significantly improves the optoelectronic performance of devices, including
R,
D*,
τ, and spectral response range. However, each type of band alignment faces different advantages and challenges. For instance, type-I energy band alignment, due to the slow injection of charge carriers over the barrier, struggles to achieve both low dark current and fast response speed at longer wavelengths. Type-II energy band alignment, as the most widely used band structure, can achieve extremely low dark current and fast response speed through the BEF it generates. However, it faces issues such as limited
R. Type-III energy band alignment can achieve large band offsets and BTBT at the heterojunction. A few 2D material photodetectors with type-III energy band alignment, such as Bi
2O
2Se/MoTe
2 [
79] and SnSe
2/MoTe
2 [
80], have been constructed and used in various applications. An external electric field is required to enhance the BTBT effect and achieve ultrafast response speeds. However, achieving BTBT using type-III energy band alignment at 0 V bias is challenging. For BTBT to occur, the valence band maximum of one material in the heterojunction must overlap with the conduction band minimum of the other material, which requires careful material selection and precise band alignment control. We look forward to further research and application of band alignment engineering in SPPDs.
3.2 Interfacial engineering
During the formation of vertical heterojunctions, the interfacial properties of 2D material heterojunctions are more sensitive to atomic-scale interfaces. The carrier concentration difference between the two material components determines the interlayer bonding in van der Waals heterojunctions [
81]. If the difference is too large, Shockley−Read−Hall (SRH) recombination caused by interfacial defects/traps becomes the dominant mechanism [
77,
82]. Therefore, optimizing interface defects at the microscopic interface is a key consideration for improving the performance of SPPDs. Feng
et al. [
83] constructed a WSe
2/Gr/MoTe
2 heterojunction SPPD [Fig.6(a)]. Fig.6(b) is a schematic diagram of the energy bands of the heterojunction under illumination. It was found that the self-powered function can be significantly enhanced with the introduction of Gr. This improvement is attributed to better contact between the interfaces, which promotes the transport of photogenerated carriers by reducing interfacial charge traps [
84,
85]. Fig.6(c) shows that the WSe
2/Gr and Gr/MoTe
2 interfaces separately form a Schottky junction, with the two BEF oriented in the same direction, implying that the electric fields cannot inhibit each other. Besides, the two Schottky junctions based on Gr are advantageous for photogenerated carrier separation and transport, thus exhibiting excellent photovoltaic properties. The device exhibited a
R of 40.84 mA/W and a
D* of 1.21 × 10
11 Jones under 550 nm illumination. The device demonstrated a significant PVE both in the dark and under different irradiation wavelengths, as shown in Fig.6(c). Fig.6(d) shows
τrise/
τfall of 468 ms/428 ms at 0 V bias under 520 nm irradiation.
In another study, Shin
et al. [
86] developed a WSe
2/WO
x/MoS
2 heterojunction photodetector based on oxygen plasma doping [Fig.6(e)]. The formation of a WO
x layer on WSe
2 not only realizes p-type doping of WSe
2 but also serves as an interfacial oxide layer, which contributes to the formation of a stronger built-in potential at the heterojunction interface. The WSe
2/WO
x/MoS
2 heterojunction photodetector is observed to suppress the dark current to 5.5 × 10
−13 A. Additionally, the detector exhibited high photoresponse and a broad detection spectrum spanning the visible to near-infrared region. The device exhibited a photocurrent of 1.79 × 10
−6 A, a
R of 11.75 mA/W, and a
D* of 2.78 × 10
10 Jones under 520 nm irradiation (0 V bias). Fig.6(f) shows the scanning photocurrent mapping under 520 nm irradiation. In addition, under 852 nm irradiation, the device exhibited a
Isc of 3.95 × 10
−8 A and an
Voc of 0.3 V. The
D* and
R of the undoped and oxygen plasma-doped devices were compared, as illustrated in Fig.6 and 6h, where the highest
R and
D* are 0.19 mA/W and 4.59 × 10
8 Jones, respectively.
In the process of modulating 2D materials for SPPDs, although these doping methods can precisely control the doping level to adjust the electronic properties of semiconductors, they may impact the stability of the material and require more delicate processing conditions. Pan
et al. [
87] introduced a technique to incrementally adjust the precursor reaction concentration during the growth process, enabling the fabrication of a p-Si/MoS
2 heterojunction, as depicted in Fig.6(i). By controlling the temperature of the sulfur source, high-quality MoS
2 nanosheets can be formed. The reason is that an appropriate sulfur source temperature can effectively increase the probability of covalent bonding between Mo and S, thereby suppressing the formation of sulfur vacancies (V
S) and other defects, as shown in Fig.6(g). By suppressing the formation of V
S and other defects, the quality of the heterojunction interface is improved [
88]. Finally, the photovoltage and response speed of the device are enhanced several-fold, as shown in Fig.6(k). The SPPD built with optimized MoS
2 nanosheets exhibited a
R of 330.14 A/W, τ
rise/τ
fall times of 40 μs/133 μs, and a
D* of 1.0 × 10
10 Jones. Interface engineering plays a crucial role in the optimization of SPPDs. By improving interface quality, device detection performance is enhanced, which is attributed to the facilitation of photogenerated carrier transport and the reduction of carrier recombination.
3.3 Structural engineering
Several studies have demonstrated that external fields, including gate voltage, can effectively regulate the BEF of photodetectors [
89]. However, these strategies inevitably require an external power source, which increases the weight and complexity of devices. This contradicts the portability of SPPDs. Therefore, it is feasible to simulate the effect of an external electric field without increasing the complexity of the device through clever structural engineering. Shang
et al. [
90] constructed an InSe/WSe
2/SnS
2 van der Waals heterojunction photodetector, as shown in Fig.7(a). The simulation of an additional negative gate voltage in the WSe
2 layer was achieved by forming a phototropic vertical electric field at the SnS
2/WSe
2 interface, which enhanced the BEF of the InSe/WSe
2 heterojunction. This resulted in a significant improvement in the separation efficiency of the photogenerated carriers and a reduction in carrier drift time. To further observe the effect of SnS
2 on the device, Fig.7(b) shows the photocurrent mapping under 400 nm illumination with 0 V bias. Although photocurrent existed in InSe/WSe
2, the photocurrent mainly existed in the InSe/WSe
2/SnS
2 region, proving that SnS
2 plays a major role in the photoresponse. As shown in Fig.7(c), the device exhibited
τrise/
τfall times of 110 μs/120 μs, which is about an order of magnitude faster than InSe/WSe
2, demonstrating exceptional optoelectronic performance. Thus, the photodetector exhibited a
R of 550 mA/W and a
D* of 1.9 × 10
13 Jones, as shown in Fig.7(d). By carefully designing the device structure, we can significantly improve optoelectronic performance. For example, incorporating a matching intermediate layer can serve as an absorption layer for low-energy photons, enhancing the dissociation of photogenerated carriers. Additionally, tailoring the heterojunction architecture based on its intrinsic properties further optimizes device performance.
Zhang
et al. [
91] constructed a vertical heterojunction of MoS
2/PdSe
2/WSe
2, sandwiching a 2D semimetal between two semiconductor layers to form a p-type semiconductor/semimetal/n-type semiconductor (PSN) structure, as shown in Fig.7(e). In self-powered mode, the device demonstrates broad spectral response from the visible to the infrared band (405 to 1550 nm), a
R of 0.56 A/W, a
D* of 5.63 × 10
11 Jones, and
τrise/
τfall of 190 μs/74 μs at room temperature, as shown in Fig.7(f)−(h). The photodetector has an extremely low dark current (1.8 pA), resulting in high sensitivity. This performance is attributed to the architecture, where the top and bottom layers generate a BEF, while the middle layer acts as an absorption layer for low-energy photons and promotes the dissociation of photogenerated carriers.
Wu
et al. [
92] demonstrated a broadband SPPD that utilizes a vertical transport channel (Gr/BP/MoS
2/Gr) within a 2D heterostructure, as shown in Fig.7(i). The device is composed of a BP/MoS
2 heterojunction sandwiched between two layers of Gr. Thanks to the protective effect of the Gr layers, the device exhibits excellent long-term stability, maintaining its performance even after one month. More importantly, the device shows a rapid and stable response to an incident light power range from 132 nW to 14.97 μW at 0 V bias [Fig.7(j)]. Fig.7(k) and (l) compare the room-temperature performance of photodetectors with vertical and lateral transport channels. Through systematic research on the impact of channel length on device performance, the significant advantages of vertical channels over traditional lateral channels are confirmed. The device has the advantage of ultra-wide spectral detection, with the EQE exceeding 15% across the range of 325−3600 nm, peaking at a
R of 0.66 A/W at 3600 nm. In terms of
D*, the device surpasses 3.88 × 10
6 Jones across the entire tested wavelength range (325−3800 nm), reaching 2.38 × 10
7 Jones at 3600 nm. This study not only introduces a novel structural design but also validates the advantages of the vertical channel configuration through a systematic experimental investigation into the impact of channel length on device performance.
Another aspect is that heterojunctions based on 2D materials face the challenge of poor light absorption due to their nanoscale thickness. The localized surface plasmon resonance (LSPR) effect is a key strategy to enhance the light absorption of 2D materials. By integrating metal nanoparticles (such as gold, silver, and copper) with 2D materials, LSPR can be induced at specific frequencies, significantly improving the photoresponse. Wang
et al. [
93] utilized LSPR technology to deposit gold nanoparticles on the surface of the heterojunction, constructing a high-performance Au@MoS
2/p-Si SPPD, as illustrated in Fig.7(m). Through the LSPR effect, not only was light absorption enhanced, but carrier separation efficiency was also significantly improved, thereby achieving simultaneous improvements in higher
R and faster
τ, as shown in Fig.7(n) and (o). The
R and
D* of the device were 1498 mA/W and 1.96 × 10
12 Jones, respectively, which are 24 times higher than those of the MoS
2/p-Si photodetector. Under 0 V bias, the
Isc after LSPR optimization increased by 138 times, reaching 1.8 mA, and the
Voc increased from 0.23 V to 0.38 V, demonstrating the enhancement of the BEF.
The application of structural engineering enhances the performance of SPPDs and is in line with the fundamental objective of SPPDs. This offers a novel approach to the development of SPPDs. By ingeniously designing multi-heterojunction structures, it is possible to improve photodetection performance. Additionally, structural engineering can help reduce the dark current, thereby enhancing the sensitivity and dynamic range of the detectors. In the future, it is worth considering the combination of clever device structure design with other advanced technologies, such as LSPR and quantum dots, to achieve even higher optoelectronic performance.
4 SPPDs based on TMDs with lateral heterojunctions
While mechanical exfoliation and transfer techniques have been successful in preparing vertical heterojunctions, the inevitable introduction of impurities during the stacking process leads to the formation of interface states, thereby affecting carrier transport and hindering further advancements. Conversely, lateral heterojunctions synthesized via epitaxial growth form sharp interfaces and offer easier modulation of electronic properties [
94−
96]. The lateral structure features a narrow space charge region with a strong BEF, leading to superior optoelectronic properties such as enhanced photogenerated carrier separation efficiency, reduced dark current, and improved rectification ratios. These attributes make lateral heterojunctions particularly suitable for integration into electronic circuits, attracting significant interest. However, two major challenges persist: (i) Due to issues such as defects at the material interfaces and the complexity of the fabrication processes, the quality of fabricated lateral heterojunctions remains suboptimal. (ii) Modulating the type and concentration of charge carriers in regions of the material is crucial for the fabrication of heterojunction devices. However, the means of carrier modulation currently available are relatively limited. The following section discusses strategies to improve the optoelectronic performance of lateral heterojunction devices, with a focus on synthesis methods and regulatory approaches.
4.1 Growth control strategies
Lateral heterojunctions can be synthesized via epitaxial growth using one-step or two-step chemical vapor deposition (CVD) or physical vapor deposition (PVD) methods. The covalent bonding of TMDs ensures a tighter contact interface in lateral heterojunctions, resulting in rapid τrise/τfall and high R. Therefore, controlling the growth conditions of lateral heterojunctions to achieve ideal dimensions and morphology is important for the development of high-performance SPPDs. In this section, the one-step and two-step CVD methods are emphasized.
4.1.1 One-step CVD
The one-step CVD method, featuring a single heating cycle, is simple, cost-effective, and widely used to synthesize lateral heterojunctions. Li
et al. [
97] constructed monolayer MoS
2/WS
2 lateral heterojunctions with sharp interfaces and well-defined lateral dimensions, as shown in Fig.8. These devices exhibit excellent rectification characteristics, with an
Ion/
Ioff ratio of 10
5, an EQE of 1874%, a
τ of 120 ms, and a
D* of 5.5 × 10
11 Jones. Fig.8(c) and (d) show the
I−
V characteristics of the two heterojunction devices under dark and illumination conditions. Fig.8(e) and (f) present the time-dependent photoresponse characteristics of the lateral heterojunction samples with a 1.51 mW/cm
2 power density incident laser. The
τrise and
τfall of the sharp interface lateral heterojunctions could be estimated as 120 ms and 290 ms, respectively, while those of the graded interface heterojunctions were 550 ms and 230 ms, respectively. The difference in
τrise and
τfall is due to the quality of the interface. The alloy region of the graded interface heterojunction is wide and has many defects. The carriers are subject to impurity scattering during transport, which reduces their mobility and slows down the response time. On the contrary, the fast response speed of the sharp interface MoS
2/WS
2 lateral heterojunction is attributed to its excellent interface quality. These results underscore that enhancing the interfacial quality can significantly improve the optoelectronic performance of lateral heterojunctions.
Najafidehaghani
et al. [
98] presented a straightforward, one-step CVD method for fabricating large-area MoSe
2/WSe
2 lateral heterojunctions, as shown in Fig.9(a). This approach involves differential thermal modulation of the partial pressure of the metal oxide precursor during the growth phase, followed by selective evaporation of the metal oxide precursor. Fig.9(b) shows that obvious
Voc and
Isc were observed. The device exhibited a dark current of 135 pA and a
R of 110 mA/W at 1 V bias. At 0 V bias, the SPPDs exhibited
τrise/
τfall of 0.16 s/0.15 s and an
Ion/
Ioff ratio of approximately 10
4 [Fig.9(c)]. Jia
et al. [
99] successfully synthesized WSe
2/MoSe
2 heterojunctions by controlling the number of precursors and the growth conditions using a one-step CVD method [Fig.9(d)]. Optical microscopy revealed that the heterojunction interface possesses a sharp and straight edge. Under 543 nm laser irradiation, the
I−
V curve shifted downward and crossed into the fourth quadrant, indicating the presence of the PVE in comparison to the dark condition [Fig.9(e)]. Additionally, the lateral heterojunction demonstrated a
τ of 6 ms at 0 V bias, as shown in Fig.9(f). Zheng
et al. [
100] synthesized a series of WS
2/WS
(2₋x)Se
x/WSe
2 lateral heterojunctions [Fig.9(g)] with atomically sharp interfaces and tunable band alignment using a one-step vapor deposition method. The energy-band arrangement and conduction type of these multi-heterostructures can be effectively tuned by controlling the chemical composition of the alloy in the intermediate region. However, this alloying interface is challenging to avoid in CVD growth, impacting the quality of the heterojunction interfaces.
4.1.2 Two-step CVD
The two-step CVD is an efficacious methodology for obtaining high-quality heterojunction interfaces. Lateral heterojunctions obtained via a two-step CVD typically exhibit atomically sharp interfaces. This two-step CVD process, with its spatial and dimensional controllability, shows significant promise for integrated fabrication applications. Moreover, this method’s growth process proficiently curbs the cross-contamination of disparate elements. Specifically, coexisting gas-phase substances may undergo unpredictable growth, frequently yielding thermodynamically preferred TMDs ternary alloy formation [
101]. Different TMDs combinations can be actualized by managing the sequence of TMDs growth. Ye
et al. [
102] reported the synthesis of lateral bilayer WS
2/MoS
2 heterojunctions using a two-step CVD method [Fig.10(a)]. Fig.10(b) illustrates the two-step CVD process employed for the growth of the WS
2/MoS
2 heterojunctions. In the first step, bilayer WS
2 crystals were grown on a Si/SiO
2 substrate using WO
3 and S as precursors via CVD. In the second step, using the bilayer WS
2 crystals obtained in the first step as a template, a mixture of MoS
2 and NaCl was employed as the precursor to epitaxially grow bilayer MoS
2 on the periphery of WS
2 via CVD, thereby forming WS
2/MoS
2 heterojunctions. This two-step CVD method successfully achieved a sharp interface between the WS
2 and MoS
2 domains. Fig.10(c) and (d) demonstrate the presence of a BEF of 100.78 mV within the heterojunction. At 5 V bias, the device exhibited remarkable performance, including a
R of 6.72 × 10
3 A/W, an EQE of 3.09 × 10
13 Jones, and a
τ of less than 50 ms. Sheng
et al. [
103] investigated the fabrication of WS
2/MoSe
2 lateral heterojunctions using a two-step CVD method, as shown in Fig.10(e). The procedure of the two-step CVD method is as follows: First, a monolayer of MoSe
2 was grown on a 4-inch SiO
2/Si substrate using atmospheric pressure chemical vapor deposition (APCVD). Subsequently, the as-grown MoSe
2 was treated with ultraviolet ozone (UVO). Fig.10(f) shows the optical image of the monolayer MoSe
2 after growth, with the inset displaying the MoSe
2 after 2 minutes of UVO treatment. The UVO treatment introduces Mo−O transition states at the edges of MoSe
2, preventing the rapid substitution of Se atoms by S atoms during the subsequent growth process. Finally, a monolayer of WS
2 was grown on the MoSe
2 monolayer obtained in the first step using low-pressure chemical vapor deposition (LPCVD). Fig.10(g) illustrates the optical image of the WS
2/MoSe
2 lateral heterojunction after growth. Under illumination at 340 nm and at 3 V bias, the device exhibited a maximum photocurrent of 99.01 nA, a dark current of 1.45 nA, and an
Ion/
Ioff ratio of 68. Fig.10(i) shows that a BEF of 0.06 V exists within the WS
2/MoSe
2 heterojunction, which facilitates the effective separation of photogenerated electron−hole pairs, implying that the device has self-powered capabilities.
In summary, the CVD method offers irreplaceable advantages in the controllable synthesis of heterojunctions. It allows for the direct growth of 2D materials on substrates, enabling the formation of controllable heterojunctions through the adjustment of growth parameters such as temperature, precursor ratios, and carrier gas flow rates [
104−
106]. However, several challenges remain, including conflicting growth conditions, interface contamination, and difficulties in large-scale production, which currently limit its commercial viability. Recent studies have employed machine learning to assist in process design by simulating the growth process and predicting optimal growth parameters. Meanwhile, in-situ characterization techniques have been utilized to monitor the growth process in real-time, thereby optimizing interface quality.
4.2 Carrier modulation strategies
In the current research landscape, various methods exist to regulate lateral heterojunctions without necessitating the complex stacking processes inherent to vertical structures. Techniques such as electric field modulation [
107,
108], surface charge transfer doping [
109], thickness modulation [
110], and width modulation [
111] have been employed to enhance photodetector performance. These approaches can be leveraged to develop photodetectors with enhanced performance by modulating the doping level of the material and changing the energy band of the material, thereby altering the potential barrier of the BEF. The subsequent section delineates the more prevalent techniques utilized in developing SPPDs.
4.2.1 Surface charge transfer doping
The carrier concentration and type in lateral heterojunction photodetectors are typically modulated via localized chemical doping, plasma treatment, and surface charge-transfer doping (SCTD) [
112]. However, localized chemical doping and plasma treatment can induce surface damage and degrade electrical properties. Conversely, owing to its simplicity, effective doping, and precise control over the surface-doped layer, SCTD has garnered widespread usage in research. Zhan
et al. [
113] employed 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F4-TCNQ) as a p-type dopant to dope WSe
2, with undoped WSe
2 serving as an intrinsic layer and ReS
2 as an n-type semiconductor to construct p−i−n junctions [as shown in Fig.11(a)]. Fig.11(b) depicts the time-dependent photoresponse at 0 V bias under different wavelengths with a light intensity of 100 nW before and after doping. The photocurrents increased from 7.28 nA to 25.2 nA at 635 nm and from 8.03 nA to 27.8 nA at 785 nm. This enhancement in photocurrent is attributed to the strengthened BEF, which accelerates the separation of photogenerated electron-hole pairs. These heterojunctions exhibit a favorable PVE via surface charge transfer. Under 0 V bias and 785 nm laser irradiation, the device demonstrated a
R of 0.29 A/W, a
D* of 8.02 × 10
12 Jones, and an EQE of 46.29%. Fig.11(c)−(f) show the relationship between the photocurrent,
R,
D*, and EQE of the photodetector and the laser power intensity at 0 V bias. Jian
et al. [
114] utilized deionized water as a surface-charge dopant to enable the rapid and straightforward transformation of p-type WS
2, employing h-BN as a localized mask for water treatment. They successfully constructed a lateral homojunction based on multilayered WS
2, as depicted in Fig.11(g). The device demonstrated a
R of 57 mA/W,
τrise/
τfall of 17.9 ms/15.9 ms, a
D* of 5.8 × 10
11 Jones, and a large linear dynamic range of 60 dB, as shown in Fig.11(h) and (i). The WS
2 lateral homojunction exhibited self-powered characteristics with a
Isc of 7.55 nA and an
Voc of 0.15 V at 0 V bias.
4.2.2 Thickness Modulation
The quantum confinement effect observed in 2D materials results in a thickness-dependent energy band structure. Consequently, the thickness of TMDs can be modulated during the preparation of homojunctions to enhance the band alignment of the resulting devices. Tan
et al. [
115] constructed a WSe
2/WSe
2 lateral homojunction photodiode with a unilateral depletion region through thickness modulation [Fig.12(a)]. This photodiode is capable of transitioning from n-type to p-type via simple electrically controlled tuning, resulting in a significant improvement in the rectification ratio from 1 to 12 000. Fig.12(b) and (c) show the photovoltaic response of the WSe
2/WSe
2 lateral homojunction photodiode at 0 V bias, which exhibits a significant PVE, with an
Voc of 0.49 V, a
Isc of 0.125 nA, and an EQE of 2.6%. The presence of a single-side depletion region results in a
D* of 4.4 × 10
10 Jones and a
τ of 0.18 ms at 0 V bias. The modulation of the homojunction by thickness enables the creation of a more suitable bandgap and higher absorbance, resulting in a more intense PVE. Different band structures are formed when illumination is applied at different thicknesses of the homogeneous junction, as shown in Fig.12(d)−(f). When the Fermi level at the thin end is higher than that at the thick end, electrons will migrate from the thin segment to the thick segment, resulting in electron accumulation on the thin side at equilibrium, as depicted in Fig.12(d). When the thin edge is illuminated, the light-induced holes move freely toward the thick edge due to the large built-in potential in the unilateral depletion region, while the photoinduced electrons move to the thin region, producing a negative
Isc [Fig.12(e)]. When the thick edge is illuminated, the photoinduced electrons are blocked by a relatively high barrier, as shown in Fig.12(f).
4.2.3 Metal-contact doping
In a lateral heterojunction, metal contact doping is more straightforward in some respects compared to other doping treatments, as it involves the selection and deposition of metallic materials. Early studies have postulated that n-type or p-type conductivity can be achieved by contacting TMDs with metals whose Fermi energy levels are proximate to the conduction or valence band edges of semiconductors, as metal contacts facilitate carrier injection. Li
et al. [
116] constructed a metal-contact-based TiN−WSe
2−Ni (TWN) homojunction self-powered SPPD with a short-channel TWN lateral homojunction. EQE was approximately 150%, with a
τ of 10 μs, markedly superior to that of a vertical homojunction constructed by metal-contact doping. The current density was close to or exceeded 0.5 A/W. Notably, when comparing different channel widths in the study, the fill factor (FF) of the TWN channel with a width of approximately 3.8 micrometers is slightly lower than that of the other two materials with shorter channels and similar FF values (above 0.4). Regarding other parameters, such as power conversion efficiency (PCE),
R, and EQE, the channel width of approximately 1.2 μm performs significantly better than that of approximately 3.8 μm, as shown in Fig.13(b)−(d). This discrepancy can be attributed to the limitation of the doping length due to the metal-contact effect, which results in the formation of a p−i−n homojunction in TWN (3.8 μm) devices rather than homojunctions with shorter channels. The application of metal contact doping to shorten the channel length has the potential to significantly enhance device performance.
In summary, carrier modulation strategies have already demonstrated unique advantages in SPPDs. First, SCTD can achieve quantitative control of doping concentration, with minimal limitations on material selection. Moreover, this technology has the potential to be combined with CVD techniques to enable mass production, holding a promising future. However, it still faces challenges regarding stability and reproducibility after doping.
Second, the structure and fabrication process of SPPDs based on metal-induced doping are simple. However, the materials must be ambipolar, such as WSe2 and MoS2, which allow the Fermi level to be tuned towards the conduction or valence band through metal-induced doping.
Lastly, thickness modulation leverages the characteristic that the band structure of TMDs changes with the number of layers, avoiding issues associated with doping processes. However, modulating the band structure of 2D materials through thickness modulation is limited by the number of layers that can be modulated, resulting in a relatively small BEF.
5 Summary and prospect
In this review, we have summarized the recent advancements of SPPDs based on TMDs. Based on vertical and lateral heterojunctions, we focus on discussing various interface modulation strategies to enhance the performance of SPPDs. Tab.1 shows a comparison of SPPDs using various performance enhancement strategies. SPPDs based on the PVE exhibit impressive performance, such as a broad detection range, large Ion/Ioff ratios, high D*, and ultra-fast τ. Although significant progress has been made in exploring SPPDs based on TMDs, several challenges remain in material optimization, performance improvement, and functional applications.
(i) Current research on SPPDs based on 2D TMDs is still largely at the laboratory stage. The primary challenge lies in the large-scale, uniform synthesis of TMDs materials. While significant progress has been made, the large-area growth of TMDs is still in development, with only a few TMDs (e.g., MoS2) being produced as single-crystal, wafer-scale films. Recently, the 2D Czochralski method has been proposed as a promising technique to rapidly grow centimeter-scale single-crystal MoS2 domains without grain boundaries. Controlled, large-area growth of TMDs holds great potential for the commercial development of SPPDs.
(ii) A large number of interface modulation strategies have been well developed to enhance the efficiency of photogenerated carrier separation and to passivate heterojunction interfaces, thereby achieving high-performance TMDs-based SPPDs. However, the nanoscale thickness of TMDs results in low light absorption, which remains a critical factor limiting the performance of SPPDs and hinders their development into high-performance photodetectors. To enhance the absorption efficiency of SPPDs, further advancements in absorption enhancement technologies are required, such as the incorporation of LSPR structures and Fabry–Pérot cavities in SPPDs.
(iii) With the trend towards miniaturization and low power consumption in optoelectronic devices, multifunctionality within a single device has become an important development trend. The latest reported optoelectronic devices have integrated photodetectors and optoelectronic memristors, enabling both photodetection and biomimetic artificial synapse functions. These devices show great potential in fields such as image recognition and artificial neural networks. However, current related research primarily employs conventional oxide materials. These emerging TMDs-based SPPDs also have great potential for use in multifunctional optoelectronic devices in the future.
To conclude, while significant strides have been made in enhancing the performance of SPPDs based on TMDs, addressing the remaining challenges will be crucial for realizing their full potential in advanced optoelectronic applications. Continued research and innovation in these areas will pave the way for the next generation of high-performance optoelectronic devices.