Photocurrent Generation and Collection in a WSe₂-Based Composite Detector

Abstract

Two-dimensional (2D) van der Waals materials have been actively investigated for broadband, high-sensitivity, low-power-consumption photodetection owing to their highly customizable band structures and fast interfacial charge transfers. Studying photocurrent generation mechanisms provides insights into charge carrier dynamics in WSe₂-based detectors, linking spatial factors (e.g., photocurrent generation/collection) with interfacial band alignment. Here, we employ scanning photocurrent microscopy to spatially resolve the processes of photocurrent generation and collection in WSe₂-based composite structures. Photocurrent polarity and magnitude at interface reflects interfacial band alignment and potential gradients at metal–WSe₂ and WSe₂–In₂Se₃ junctions. Strong electric fields at metal–WSe₂ interfaces drive more efficient electron–hole separation and yield higher photocurrents, compared with WSe₂–In₂Se₃ interfaces. The photodetector exhibits broadband detection capabilities from visible to infrared light, achieving a high responsivity of 17.7 A/W and an excellent detectivity of 3.7 × 10¹² Jones, as well as fast response times of <113 µs. Furthermore, object imaging with a resolution better than 0.5 mm was successfully demonstrated, highlighting the potential of this photoresponse for practical imaging applications. This work reveals that photocurrent is distributed with a clear dependence on device configuration, offering a new avenue for optimizing 2D material-based photoelectric devices.

1. Introduction

Transition metal dichalcogenides (TMDCs), exemplified by monolayer tungsten diselenide (WSe₂), have emerged as promising optoelectronic materials due to their layer-dependent bandgaps, high carrier mobilities, and strong light–matter interactions [1]. Beyond their potential in ultrathin electronics, WSe₂-based heterostructures exhibit exceptional photoresponse characteristics, making them ideal candidates for next-generation photodetection systems [2]. Unlike zero-bandgap graphene, WSe₂ possesses a direct bandgap (~1.6 eV) in monolayer form, enabling efficient photon absorption across visible to near-infrared spectra [3,4]. The direct bandgap also facilitates fast exciton generation and dissociation, which is essential for high-performance photodetectors. Moreover, WSe₂ is an intrinsic p-type semiconductor, which is highly valuable for enabling hole transport, realizing complementary device architectures, and enhancing charge separation efficiency in electronic and optoelectronic systems [5]. Furthermore, vertical stacking with other 2D materials (e.g., graphene) creates tailored interfacial band alignments, which are crucial for efficient photocurrent generation and extraction [6,7]. Such heterostructures allow for precise control over charge transfer dynamics, enhancing both responsivity and detectivity in optoelectronic devices [8].

Recent advances in WSe₂-based photodetectors [9] have established three dominant architectures: photoconductors, phototransistors, and photodiodes, each leveraging distinct operational mechanisms to optimize performance. Electrically, WSe₂ exhibits intrinsic p-type conductivity yet allows dynamic carrier-type modulation through doping, electrostatic gating, or contact engineering [10], enabling flexible integration with both Schottky and Ohmic contacts. Photoconductive configurations employ these interfacial designs to regulate carrier transport through Schottky barrier modulation [11], achieving high responsivities, microsecond-scale response times, and low dark currents. Breakthroughs in van der Waals heterojunctions (WSe₂/MoS₂, WSe₂/In₂Se₃, WSe₂/graphene) exploit type-II band alignment or vertical p-n junctions to generate built-in potentials [12,13,14]. These heterostructures leverage interfacial charge transfer efficiencies and built-in electric fields to enhance carrier separation, with WSe₂/In₂Se₃ systems demonstrating broadband detection and long interlayer exciton lifetimes [11,15]. Graphene–WSe₂ hybrids further enable fast photocarrier extraction through ballistic transport channels. Such engineered interfaces position WSe₂ as a versatile platform for next-generation optoelectronics, including field-effect transistors [16], photovoltaic devices achieving external quantum efficiency. Nevertheless, critical gaps persist in our understanding of photocurrent dynamics within WSe₂-based detectors. The tripartite interplay between interfacial band engineering, multidimensional carrier routing, and spatially heterogeneous potential landscapes remains largely uncharted, demanding systematic investigation through advanced spatial-resolution techniques.

The spatially resolved scanning photocurrent (SPCM) technique, previously applied to 1D nanostructures (carbon nanotubes/nanowires) [17,18,19] with constrained current paths, contrasts with WSe₂’s 2D architecture, enabling multidirectional photocurrent pathways. In this study, we constructed WSe₂-based detectors via dry-transfer methods, and found that laser-localized photocurrent generation and electrode-specific collection revealed potential-dependent transport dynamics of photon-generated carriers. Compared with pure WSe₂ and the WSe₂–In₂Se₃ heterojunction, the multiple-pair metal-WSe₂ heterojunction exhibited an enhancement in photoresponse of nearly two orders of magnitude, highlighting the effectiveness of the asymmetric multiple-metal–WSe₂ configuration for high-performance visible-to-infrared photodetection, particularly under low-light conditions and in high-resolution imaging applications.

2. Device Fabrication and Characterization

To preserve the integrity of WSe₂ and optimize the interface quality, all devices were fabricated on Si/SiO₂ substrates (300 nm oxide) with laser-direct-written Au electrodes (electron-beam evaporated, 50 nm thick). Three types of device architectures were prepared: (i) A photodetector structure with a single pair of electrodes, where dry-transferred WSe₂ flakes were aligned onto Au electrodes to form metal–semiconductor Schottky junctions. (ii) A photodetector structure with multiple interdigitated electrodes, where WSe₂ flakes were dry-transferred onto pre-patterned arrays to form multiple metal–semiconductor junctions. (iii) A vertical van der Waals heterostructure, formed by sequentially stacking mechanically exfoliated WSe₂ and In₂Se₃ flakes between Au electrodes using a PDMS-assisted dry-transfer method. This technique ensures clean interfaces and well-controlled charge transport across the heterojunction. This methodology avoids photolithographic damage and guarantees clean van der Waals contacts, which are essential for preserving the intrinsic properties of the layered materials and achieving high-performance and reproducible photoresponse.

Optical and spectroscopic measurements were performed using an Olympus optical microscope and a 532 nm micro-Raman system. All electrical and optoelectronic measurements were conducted in ambient atmosphere at room temperature. A dual-channel digital source meter (Keithley 6482; METATEST CORPORATION, Nanjing, China) was used to simultaneously apply bias and measure current. Photocurrent signals were collected using a lock-in amplifier (MStarter 200; METATEST CORPORATION, Nanjing, China), and spatial photocurrent mapping was carried out by scanning the sample in both lateral directions using a precision-controlled probe stage. Devices were securely mounted on a custom sample holder to ensure measurement stability.

3. Results and Discussion

To investigate the photoresponse performance of WSe₂-based devices, we fabricated a WSe₂ detector with a pair of electrodes (Device 1). Figure 1a,b present the schematic diagram and corresponding optical microscope image of the WSe₂ device with a metal–semiconductor–metal (MSM) structure. The morphology of the WSe₂ flakes is clearly observed, the electrodes are uniformly covered by the material, and the overall device structure appears intact.

Figure 1c displays the output current–voltage characteristic curves under dark conditions. The rectifying behavior of the Au/WSe₂/Au device shows voltage-dependent nonlinearity in the dark state, indicating the presence of an inherent electric field. Figure 1d shows the I–V characteristics of the device under different light power densities at 638 nm, as well as in the dark. The photocurrent amplitude also increases with rising light intensity (see Figure S1 in the Supporting Information for more power-dependent and bias-dependent photoresponse data at 638 nm). To further characterize the advantages of our device in terms of response speed, the rise and fall times, typically defined as the duration required for the photocurrent to transition from 10% to 90% of its maximum value, are shown in Figure 1e. The rise time is ~370 μs, and the fall time is ~405 μs. This rapid response speed is attributed to fast separation and transportation of photogenerated carriers facilitated by the strong built-in electric field at the metal–WSe₂ Schottky junction. Figure 1f presents the time-resolved photoresponse of the device under laser illumination ranging from visible to near-infrared light of 520–940 nm at zero external bias, further confirming that the device maintains a strong photoresponse capability in the near-infrared range (see Figures S2 and S3 in the Supporting Information for power-dependent photoresponse at 520 nm and 940 nm, respectively).

Figure 1g presents the photocurrent generated at positions P1 and P2 under focused optical illumination, highlighting the spatial dependence of the photoresponse. The photocurrent polarity observed at zero bias voltage, 4.6 nA at position P1 and −7.5 nA at position P2, indicates the presence of a built-in electric field at the interface. Analysis of the zero-bias photocurrent direction indicates that the internal electric field points from WSe₂ flakes toward the Au electrode. The schematic diagrams of the energy band structure are illustrated in Figure 1h. A Schottky barrier is formed due to the difference in work function between Au and WSe₂ [20], leading to band bending and the formation of a built-in electric field. The inherent symmetry of the dual Schottky junction architecture leads to photocurrent cancellation under global illumination; however, localized excitation breaks this symmetry, enabling a net photoresponse that provides a robust basis for probing the influence of composite structures on photodetector performance.

To further confirm the influence of metal–WSe₂ Schottky junctions on the photoresponse, a WSe₂-based photodetector (Device 2) with multiple electrode pairs was fabricated. Figure 2a illustrates the schematic structure of the device, while Figure 2b shows an optical microscope image of a representative device fabricated based on this design. The interdigitated electrode gaps are uniformly distributed, and the WSe₂ flake extends across multiple electrode fingers, forming a large effective photosensitive area to facilitate a more comprehensive analysis of the photoresponse mechanisms. In the lower panel of Figure 2a, adjacent WSe₂ electrodes may be seen to generate opposing built-in fields, enabling directional separation of photogenerated carriers for enhanced photocurrent. The scanning photocurrent microscopy characterization under zero bias shown in Figure 2c reveals spatially localized photocurrent at WSe₂/Au interfaces under 638 nm illumination. The mapped photoresponse exhibits five polarity inversion regions between neighboring electrode pairs, governed by counter-oriented built-in potentials across multiple WSe₂/Au Schottky junctions. This spatial photoresponse pattern directly correlates with interfacial band bending dynamics, demonstrating the critical role of electrode architecture in steering carrier transport pathways.

The I–V curves of the device under dark conditions and 638 nm laser illumination are shown in Figure 2d. Across the entire voltage of −1 V to 1 V, the photocurrent under 638 nm illumination is significantly higher than the dark current, indicating a strong photoresponse. Time-resolved zero-bias photoresponse measurements were performed under various incident-light power densities at 638 nm, as shown in Figure 2e (see Figure S4 in the Supporting Information for the time-resolved photoresponse at 0.5 V under varying optical power). As the light intensity increases, the photocurrent response amplitude gradually increases, demonstrating good power-dependent behavior. Figure 2f depicts the temporal response at zero bias voltage, with rise and fall times of approximately 365 and 340 μs, respectively, indicating rapid carrier dynamics under drift conditions (see Figure S5 in the Supporting Information for temporal photoresponse under 520 nm and 940 nm illumination).

Gate voltage modulation is an effective strategy for optimizing the performance of photodetectors [1]. Figure 2g shows the transfer characteristics of Device 2 at different bias voltages in dark condition. A significant suppression of the dark current occurs as the gate voltage sweeps from −20 V to 20 V, at both positive and negative bias voltage. Specifically, the dark current of the device decreases by 97.2% (from 142 pA to 4 pA) at 0.5 V and by 96% (from −102 pA to −5 pA) at –0.5 V. The transfer characteristics of the device under illumination are presented in Figure 2h. The photocurrent displays different modulation behaviors at positive and negative bias voltages. At a positive bias voltage of 0.5 V, the photocurrent increases by five-fold, from 0.17 μA to 0.9 μA, while at a negative bias voltage of −0.5 V, the photocurrent decreases by approximately 95%, from −0.56 μA to −0.08 μA. Moreover, the photocurrent saturates rapidly under positive gate voltage. In Figure 2i, as the gate voltage increases, the photocurrent increases while the dark current decreases, leading to a 509-fold improvement in the on–off ratio, which rises from 369 at −20 V to 188,000 at 20 V. Thus, by leveraging interdigitated electrode architectures and gate modulation, we systematically probed the interfacial role of Au/WSe₂ junctions, and established photocurrent performance optimization strategies through composite structural engineering.

Raman spectroscopy was performed to verify the formation of the heterostructure. Figure 3c presents the Raman spectra of WSe₂ (purple line), In₂Se₃ (yellow line), and the WSe₂/In₂Se₃ heterostructure region (green line). The results show characteristic E_2g and A_1g peaks of WSe₂ near 250 cm⁻¹ [21], while In₂Se₃ exhibits distinct peaks in the range of 80–200 cm⁻¹. The heterojunction spectrum retains both material signatures with preserved crystallinity, demonstrating non-destructive van der Waals integration. Figure 3d reveals the formation of two distinct built-in electric fields arising from the differences in bandgap and electron affinity among WSe₂, Au, and In₂Se₃. The WSe₂/In₂Se₃ interface exhibits type-II band alignment, with the conduction and valence bands of In₂Se₃ positioned below those of WSe₂. This built-in field directed from WSe₂ to In₂Se₃ spatially separates photogenerated electrons into In₂Se₃ and holes into WSe₂. Concurrently, the Au/WSe₂ contact with a Schottky junction from Au to WSe₂ enhances hole extraction to the electrode via work-function mismatch. These two fields synergistically suppress recombination and improve charge collection [22]. Their coexistence creates an asymmetric potential gradient across the device, enabling bias-free photocarrier separation/transport and supporting self-powered operation.

Figure 3e displays the spatial photocurrent distribution under 638 nm illumination at zero bias voltage. The observed asymmetry in photocurrent distribution reflects the asymmetric built-in electric fields of the device. A strong photoresponse region (−30–44 nA) appears at the metal–WSe₂ contact interface, while the WSe₂/In₂Se₃ region diminishes. Figure 3f shows the current–voltage curves of the device under 638 nm laser illumination at various optical power densities, as well as in the dark (see Figure S6 in the Supporting Information for the current–voltage characteristics under 940 nm illumination at different power levels). Compared with the WSe₂-based devices, the heterostructure device exhibits a significantly lower dark current across the entire bias range, primarily due to the strong internal built-in electric field, which effectively suppresses leakage and reduces power consumption.

Moreover, time-resolved zero bias photoresponse in Figure 3g demonstrates strong optical power dependence, with photocurrent increasing steadily as optical power rises to 638 nm. External bias modulation (−0.5 V to +0.5 V) reveals distinct carrier transport regimes in Figure 3h: negative bias enhances photocurrent, while positive bias suppresses carrier extraction through potential barrier elevation (see Figure S7 in the Supporting Information for power-dependent photoresponse at different bias voltages). This suggests that the direction of the external electric field significantly modulates the separation and transport of photogenerated carriers. Additionally, Figure 3i presents the response time of the device under 638 nm illumination with a bias voltage of 0 V (see Figure S8 in the Supporting Information for response time at −0.5 V). The fast photoresponse, with a time of about 110 μs, indicates excellent dynamic performance. These findings further highlight the device’s excellent zero-bias photoresponse performance, attributable to the synergistic contributions from intrinsic material properties, electrode-induced Schottky barriers, heterointerface effects, and additional built-in electric fields.

To comprehensively evaluate the photoresponse performance of the three devices, the value of I_ph, the I_ph/I_dark ratio, and the response time of each device under 638 nm illumination at zero bias voltage was calculated. As shown in Figure 4a, the photocurrents of Device 1, Device 2, and Device 3 under 638 nm illumination at zero bias voltage were measured to be 6 nA, 5 nA, and 7 nA, respectively. Figure 4b presents the light-to-dark current ratios of the devices, which reached 640, 11388, and 16,661 in the cases of Device 1, Device 2, and Device 3, respectively. Furthermore, the response times of the three devices, as illustrated in Figure 4c, were 375 μs, 365 μs, and 113 μs. The rapid response of Device 3 is attributed to its optimized carrier transport pathways.

To further assess the performance of the three devices, we quantitatively calculated the responsivity and specific detectivity of each device. Responsivity (R) is an important indicator for evaluating the performance of photodetectors. It is defined as: R = I_ph/(P_in·S) [23], where I_ph, P_in, and S represent the photocurrent, the illumination power, and the effective illumination area, respectively. Specific detectivity (D*) is another critical parameter for assessing the performance of photodetectors. This is defined as: D* = (S^1/2·R)/(2q·I_dark) ^1/2 [24], where q is the elementary charge and I_dark is the dark current. The variations in responsivity (R) and specific detectivity (D*) for the three devices under varying bias voltages are presented in Figure 4.

As shown in Figure 4d, Device 1 reaches its highest responsivity of 0.14 A/W and detectivity of 3.7 × 10¹² Jones at a bias of −1 V. In the case of Device 2 (Figure 4e), D* decreases from 3.66 × 10¹² Jones at −1 V to 3.03 × 10¹¹ Jones at 1 V, while R increases from 0.058 A/W to 17.7 A/W over the same voltage range. In contrast, Device 3 in Figure 4f displays a decrease in responsivity from −1 V to 0.4 V, followed by an increase to 0.03 A/W at 1 V. Meanwhile, D* exhibits triphasic behavior: it initially rises from −1 V to 0 V (peaking at 2.16 × 10¹² Jones), then declines until 0.4 V, and subsequently increases again towards 1 V. Overall, the detectivity across the devices spans over two orders of magnitude, from 1.32 × 10⁹ Jones to 2.16 × 10¹² Jones. Specifically, the detectivity’s peak-valley behavior versus bias voltage originates from heterojunction internal and external electric field modulation. At zero bias, detectivity peaks as the device relies solely on built-in fields from heterojunction and Schottky contacts. This minimizes dark current (suppressing noise) while maintaining sufficient field strength for efficient photocarrier separation without external bias, maximizing signal-to-noise ratio. Conversely, a detectivity valley emerges at bias voltage of ~0.4 V, where the applied field partially cancels the built-in field, degrading carrier separation efficiency and lowering photocurrent. These results highlight the crucial role of self-driven operation, enhanced carrier separation induced by Schottky barriers, and efficient charge transport across the heterointerface. The synergistic effects of these factors collectively contribute to the observed high photocurrent, rapid photoresponse, and low-power photoelectric detection capabilities. A comparison of the performance of the device fabricated in this study with recently reported photodetectors based on WSe₂ and other two-dimensional TMD materials is provided in Table S1 of the Supporting Information [25,26,27,28,29,30,31,32,33,34,35].

To validate the stability and imaging performance of our device, we carried out long-term stability testing over 1000 s. The results shown in Figure 5a reveal photocurrent decay under 638 nm illumination at zero bias, with consistent response amplitudes throughout multiple switching cycles. To further demonstrate the imaging capability of our photodetector, we constructed an imaging system for a demonstration experiment, as illustrated in Figure 5b. The imaging system consists of a laser source, lens, focusing optics, target object, stepper motor, lock-in amplifier, preamplifier, and a computer. The laser beam is emitted from the source, focused by the optics, and then directed onto the object. Under the precise control of the stepper motor, the object moves point-by-point in both horizontal (x) and vertical (y) directions to ensure complete scanning coverage of the entire area. The transmitted light passing through the object is collected by our photodetector, which converts the optical signal into an electrical signal. The generated photocurrent is processed through a preamplifier and a lock-in amplifier, then collected and recorded by the computer along with positional information. Finally, based on the spatial distribution of the photocurrent, the contour and features of the target object are reconstructed.

The leaf and butterfly both feature complex boundaries and intricate details, such as fine edge contours and internal cut patterns, which presented a greater challenge for our detector during point-by-point scanning and imaging. The real picture and the imaging of the leaf are shown in Figure 5c, where the size of the leaf pattern is 30 mm × 25 mm. The real picture and the imaging of the butterfly are shown in Figure 5d, with a pattern size of 35 mm × 30 mm. Even under zero-bias conditions, the photodetector successfully achieved high-quality imaging of these complex-shaped patterns, producing images with high contrast and clearly defined contours of <0.5 mm resolution. These results confirm the stable and excellent detection performance of our device and demonstrate the application potential of the photodetector, laying a solid foundation for the development of low-power, high-resolution image sensors.

4. Conclusions

This study presents a systematic design and investigation of a WSe₂-based photodetectors featuring three structures: a paired-electrode WSe₂ device, an interdigitated-electrode WSe₂ device, and a WSe₂/In₂Se₃ heterojunction. Scanning photocurrent imaging of WSe₂ enables deeper understanding of its photoresponse characteristics across two-terminal and multiterminal device architectures. We anticipate that the incorporation of spectrally resolved techniques will provide a powerful multi-resolution platform for understanding the properties of other two-dimensional materials in a much more detailed fashion. Additionally, comparing the mapping images of WSe₂-based devices helps elucidate the respective roles of metal contacts and heterojunctions. These findings lay a solid foundation for optimizing the low-power, high-sensitivity performance of WSe₂-based electronic and nanomechanical devices.