Red‐Sensitive Image Array Based on Defect‐Engineered PBDB‐T/WSe1.5‐WSe2/Graphene Heterostructure

2D transitional metal dichalcogenide (TMDC) materials have potential applications in image arrays for electronic eyes that mimick the human visual system. However, as a crucial requirement of the electronic eye, color discrimination is hard to achieve due to the broad absorption band of the TMDCs in the visible region. Here, a red‐sensitive photodetector based on the defect‐engineered WSe2‐Graphene heterostructure covered by PBDB‐T polymers is proposed. The WSe2‐Graphene heterostructure is partially modified by defects generated by the focused ion beam, forming a photovoltaic diode at the boundary of the engineered region. Then, PBDB‐T with optical absorption centered at the wavelength of 624 nm is coated onto the surface of the heterostructure, enhancing the photoresponse of the defect‐engineered WSe2‐Graphene to the red light (633 nm) ten times. Furthermore, a red‐sensitive image array with 6 × 10 pixels is constructed. This work paves a novel avenue for developing the 2D color‐resolvable photodetector array.


Introduction
As an active-pixel sensor, the photodetector is a critical element of the electronic eye mimicking the human visual system, [1][2][3] which has attracted continuous and intense attention due to its application potential in the bioinspired sensing system. Based on image sensor arrays, the object's shape and color can be sensed similarly to human eyes. [4][5][6][7][8] 2D heterostructures, consisting of diverse 2D materials with van der Waals contacts [9] (like MoS 2 -graphene), have been reported as human eye-inspired soft implantable optoelectric devices, stimulating the optic nerves for perceived images.
Besides, low energy consumption is another requirement for the electronic eye. As a highly integrated photoelectric device, the external power source increases the cost and volume of photodetectors but also causes the heating problem. One promising strategy for the low-consumption is the photovoltaic photodetector. 2D heterostructures have been widely studied as photovoltaic diodes due to their different semiconductor functions of 2D materials at the interface. The heterostructure for the 2D photovoltaic diode is commonly prepared by stacking different 2D materials together, the so-called vertical diode heterostructure. [35,36] For instance, the MoS 2 stacked with graphene and WSe 2 constructs a photovoltaic photodetector, [37] showing a broad photoresponse, high responsivity, and on/off ratio. However, preparing a vertical heterostructure requires manual mechanical transfer and alignment processes, which cannot automate. [38] Manual operation is appropriate for a one-pixel photodetector but impossible to prepare a photodetector array, as manual operation cannot guarantee the uniformity of each photodetector. Recently, selective defect engineering has been reported to be an automated approach to manipulating lateral photovoltaic diodes. [39] Defects are generated in the heterostructure via ion irradiation. Due to the contact potential difference at the boundary of the non-irradiated and irradiated region, a photovoltaic photodetector is formed with Figure 1. Photomicroscope images and schematic diagrams of device preparation. a) WSe 2 /G heterostructure on p-type Si substrate. b) WSe 1.5 -WSe 2 /G heterostructure fabricated by partial Ga + ion irradiation of WSe 2 /G. c) PBDB-T is coated on the surface of WSe 1.5 /G-WSe 2 /G, forming PBDB-T/WSe 1.5 -WSe 2 /G heterostructure. excellent performance of a broadband photoresponse and a high on/off ratio.
Here, we demonstrate the red-sensitive image array constructed from photovoltaic PBDB-T/WSe 1.5 -WSe 2 /G photodetectors composed of WSe 1.5 -WSe 2 /G heterostructures and PBDB-T polymers. The WSe 1.5 -WSe 2 /G is prepared via the focused ion beam (FIB) writing of one side of WSe 2 /G, leaving Sevacancies in the irradiated side (WSe 2-0.5 /G). The boundary of WSe 2-0.5 /G and WSe 2 /G forms a built-in field enabling the WSe 1.5 -WSe 2 /G heterostructure to have photovoltaic characteristics. Finally, PBDB-T polymers, with red-sensitive photoelectric properties, are coated onto the surface of WSe 2-0.5 -WSe 2 /G to construct the PBDB-T/WSe 1.5 -WSe 2 /G photodetector. The photodetector has a solid photoresponse at the wavelength of ≈630-640 nm (≈0.38-0.55 A W −1 ) in the photovoltaic mode achieving a filter-free color discrimination scheme. Utilizing the PBDB-T/WSe 1.5 -WSe 2 /G as one pixel, we fabricate an image array with 6 × 10 pixels to capture visible images, which realize selective photon detection and achieve clear red-color shape resolution for imaging. This work paves a novel avenue to developing a colordistinguishable electronic eye. Figure 1 shows the preparation processes of the PBDB-T/WSe 2-x -WSe 2 /G heterostructure photodetector, which has a sandwich structure with graphene, monolayer tungsten diselenide with heterogeneous distribution of defects (WSe 1.x -WSe 2 ), and graphene. As displayed in Figure 1a, first, graphene and the monolayer WSe 2 are successively covered onto 5/50 nm thick Cr/Au source (S) and drain (D) electrodes with graphene on the bottom, and electrodes have a distance of 10 μm. Then the left side of the monolayer WSe 2 is irradiated by the Ga + ion beam, which sputters Se atoms away from the top Se-atomic layer in the irradiated forming WSe 2-x on the left side, modifying the uniform WSe 2 /G into the equally divided WSe 2-x -WSe 2 /G. Figure 1b shows a clear boundary in the photograph of WSe 2-x -WSe 2 /G, distinguishing the irradiated and non-irradiated regions. Finally, PBDB-T is dissolved in CB with 8 mg mL −1 and stirred at 40°C overnight, and the solution of PBDB-T is spin-coated onto the WSe 2-x -WSe 2 /G at 3000 rpm for 50 s. After annealing under 100°C for 10 min, the spin-coated PBDB-T is solidified, forming a uniform PBDB-T film, which has a thickness of 60 nm. Figure 1c shows the image of the prepared PBDB-T/WSe 2-x -WSe 2 /G heterostructure.

Characteristics of WSe 2-x -WSe 2 /G and PBDB-T
We determine the atomic ratio of Se and W in the irradiated WSe 2-x via X-ray photoelectron spectroscopy (XPS) measurements. There are two distinct sets of peaks: the 4f 7/2 and the 4f 5/2 spin-orbit components for the W; and the 3d 5/2 and 3d 3/2 spinorbit components for the Se. Figure 2a shows the Se-3d peaks of tungsten disulfide with and without Ga + irradiation. The pristine WSe 2 has 3d 5/2 and 3d 3/2 (Se) located at 55.5 and 54.6 eV, and the irradiated one has a slight blue shift with 3d 5/2 and 3d 3/2 (Se) at 55.2 and 54.4 eV. Similar to Se-3d peaks, the W-4f 7/2 (W-4f 5/2 ) is slightly shifted from 32.4 (34.6 eV) to 32.1 eV (34.3 eV) as shown in Figure S1, Supporting Information. C-1s peaks have little variation after the ion irradiation indicating the undisturbed atomic layer of C (graphene). Integrating the area of the peaks, the element ratio of Se and W decreases from 2 to 1.5 after the ion irradiation (x = 0.5). It demonstrates that the Se atoms are sputtered away from the tungsten diselenide monolayer, leaving vacancies. Figure 2b shows the scanning electron microscope (SEM) images of the WSe 1.5 -WSe 2 /G, where a boundary appears between the irradiated (WSe 1.5 /G) and pristine (WSe 2 /G) areas indicating that the missing Se atoms modify the electronic characteristics of the heterostructure. To explore the variation of the heterostructure, we calculate the electronic structure of WSe 1.55 /G and WSe 2 /G by the first-principle calculation method, as shown in Figure 2c. WSe 2 /G (solid blue lines) has an electronic structure similar to the superposition of the graphene and WSe 2 monolayers, which has not only the characteristics of the WSe 2 band structure but also the Dirac points of graphene. While, in the WSe 1.55 /G (solid orange lines), Se defect states introduce more electronic states in the band structure, disturbing the original band structure of the tungsten diselenide. Besides, the Dirac point vanishes from K, indicating that the typical electronic characteristics of graphene have vanished after irradiation.
Moreover, the surface potential of the WSe 1.5 -WSe 2 /G is scanned by the Kelvin probe force microscope (KPFM), as shown in Figure 2d. The surface potential of WSe 1.5 /G is 0.2 V lower than the value of WSe 2 /G. The surface potential variation is directly related to the density of Se defects ( Figure S2, Supporting Information). It means the surface potential variation is induced by Se vacancies, which leave a lot of dangling bonds on the surface of WSe 2-x . Figure 2e displays the absorption spectrum of WSe 1.5 /G and WSe 2 /G. WSe 2 /G has a broad visible absorption ranging from 400 to 1100 nm, with absorption peaks at 421, 491, and 737 nm corresponding to the exciton resonances of the tungsten diselenide. Compared with WSe 2 /G, absorption peaks of WSe 1.5 /G at exciton resonances are slightly increased due to the external electronic states induced by Se defect sites. Meanwhile, Se defect sites also cause absorption peaks to broaden.
PBDB-T, as a conjugated polymer material, is capable of forming interchain aggregates in thin films with strong -stacking signals in the out-of-plane direction of the 2D grazing-incidence wide-angle X-ray diffraction patterns, indicating that the conjugated backbone tends to pack with face-on orientation referring to the substrate to facilitate the directional migration of carriers. [40][41][42] An absorption peak of PBDB-T is located at 624 nm (Figure 2f), which is complementary to the absorption bands of WSe 2 , enhancing the photon harvest of the layer. The photoluminescence (PL) spectra peak of PBDB-T is located at 700 nm (Figure 2g). At such a degree of crystallinity, the hole mobility (μ h ) of the PBDB-T films reaches 8 × 10 −4 cm 2 V −1 s −1 ( Figure S3, Supporting Information). Figure 3a shows the photocurrent mapping of the WSe 1.5 -WSe 2 /G by scanning photocurrent microscopy (SPCM) at zero bias with a 532 nm laser excitation (2 mW). We observe the unidirectional photocurrent between electrodes. The directional migration of the photoexcited carriers indicates the existence of the built-in electric field between electrodes. The built-in field is attributed to the electronic characteristic differences of WSe 1.5 /G and WSe 2 /G, which can be divided into the quasi-metal (WSe 1.5 /G) and graphene-like (WSe 2 /G) semiconductor materials depending on whether the Dirac point exists in the electronic structures displayed in Figure 2. Figure 3b shows the schematic diagram of WSe 1.5 -WSe 2 /G between electrodes. The boundary of the WSe 1.5 /G and WSe 2 /G is similar to the metal-graphene contact, where the surface potential difference between WSe 1.5 /G and WSe 2 /G drives the redistribution of charges resulting in a built-in electric field. Figure 3c gives the current-voltage bias (|I DS |-V DS ) curves of WSe 1.5 -WSe 2 /G without and with light illumination. In the dark, the device shows a typical rectifying behavior of a diode with a current rectification ratio of 10 2 at V ds = ±1 V. While WSe 1.5 -WSe 2 /G displays typical photovoltaic characteristics under the illumination of a 632.8 nm laser with the incident light power density (P laser ) of 0.31 W cm −2 , which has the maximum open circuit voltage (V OC ) of 0.12 V. The response time of WSe 1.5 -WSe 2 /G is determined to be 54 μs/60 μs for the rising/decay time ( Figure  S4, Supporting Information). Figure 3d shows the photoresponse of WSe 1.5 -WSe 2 /G at 0 bias with wavelengths of 450, 532, 633, and 780 nm, respectively. To quantitate its photodetection performance, we calculate the responsively (R) and specific detectivity (D*) at a range of laser wavelength between 450 and 780 nm in Figure 3e, following the equations below

Photoresponse of WSe 1.5 -WSe 2 /G Heterostructure
where I ph is the photocurrent, I d is the dark current, P laser is the laser power density, and S is the effective illuminated area. With the P laser of 0.31 W cm −2 , the value of R achieves 0.08 A W −1 (450 and 780 nm), 0.06 A W −1 (532 nm), and 0.13 A W −1 (633 nm), respectively. As one can see, the WSe 1.5 -WSe 2 /G has a broadband photoresponse in the visible range, and the photoresponse has a slight difference (± 0.03 A W −1 ) along with the wavelength, which can not achieve the color distinction depending on the specificity of the wavelength-dependent photoresponse. In addition, we carried out the current on/off ratio test at 0 V bias (633 nm, P laser = 3.1 W cm −2 ). As shown in Figure S14, Supporting Information, the on/off ratio of our photodetector reached 10 4 .

Photoresponse of PBDB-T/WSe 1.5 -WSe 2 /G Heterostructure
We detect photoelectronic performances of PBDB-T/WSe 1.5 -WSe 2 /G heterostructure following the same experimental processes of WSe 1.5 -WSe 2 /G. Figure 4a shows |I DS |-V DS curves under the illumination of 450, 532, 633, and 780 nm laser with a P laser of 0.31 W cm −2 . The PBDB-T/WSe 1.5 -WSe 2 /G heterostructure has a rectification of 2.8 × 10 2 in the dark and exhibits typical photovoltaic characteristics under light illumination, where the rectification is reduced to 40,75,9, and 70 at the wavelength of 450, 532, 633, and 780 nm, respectively. Figure. 4b shows the comparison of the photoswitching characteristics of WSe 1.5 -WSe 2 /G (dashed red line) and PBDB-T/WSe 1.5 -WSe 2 /G (solid red line) under the same illumination conditions. The PBDB-T/WSe 1.5 -WSe 2 /G not only has a wide photoresponse in the visible range but also dramatically increases the photocurrent under the 633 nm laser illumination due to the deposition of the PBDB-T organic film. Meanwhile, the photocurrent has no significant variation at 450, 532, and 780 nm.
The R and D * of PBDB-T/WSe 1.5 -WSe 2 /G heterostructure are shown in Figure 4c. The value of R reaches up to 0.55 A W −1 under the 633 nm laser illumination, ten times higher than lasers at 450, 532, and 780 nm. We further measure the photoresponse of the heterostructure in the range of ≈580-685 nm with the help of dichroic prisms and a white light source. Figure S5, Supporting Information, displays that the PBDB-T/WSe 1.5 -WSe 2 /G heterostructure has a highly sensitive photoresponse in the ≈630-640 nm band corresponding to the color red.
The enhancement of the photoresponse to the red light is owing to the out-of-plane charge transfer between PBDB-T and WSe 2 (WSe 1.5 ). As illustrated in Figure 4d, PBDB-T has a band gap of 1.8 eV corresponding to the strong excitation resonant absorption around the wavelength of 624 nm. Under the red light (≈630-640 nm) illumination, electrons are excited into the conduction band of the PBDB-T. As the conduction band of PBDB-T (−3.53 eV) is close to the one of WSe 2 (−3.55 eV, Figure S6, Supporting Information), the close contact with WSe 2 allows for the out-of-plane charge transfer between them. While the WSe 2-0.5 has a conduction band of −3.41 eV ( Figure S6, Supporting Information, slightly higher than the one PBDB-T extracting electrons from PBDB-T to WSe 2-0.5 . Subsequently, transferred carriers are driven by the built-in field in WSe 1.5 -WSe 2 /G, leading to the enhanced photocurrent. In contrast, there is less optical absorption of PBDB-T at wavelengths of 450, 532, and 780 nm, leading to no significant difference in photoresponse of heterostructures of WSe 1.5 -WSe 2 /G and PBDB-T/WSe 1.5 -WSe 2 /G. Consequently, the photoresponse behavior of WSe 1.5 -WSe 2 /G is dramatically enhanced at the red color (≈630-640 nm) through the simple deposition of PBDB-T organic film onto the WSe 1.5 -WSe 2 /G surface, and the PBDB-T/WSe 1.5 -WSe 2 /G heterostructure has an ultrahigh response to the red light (≈630-640 nm) achieving the distinction of red light.
Compared with reported 2D self-power photodetectors, our device delivers competitive and superior performance ( Figure S8, Supporting Information). In addition, as shown in Figure S9, Supporting Information, the WSe 1.5 -WSe 2 /G exhibits 59.5 A W −1 with 633 nm laser irradiation under a 1 V bias. To further improve the performance of our device, the WSe 2 /G was irradiated with a higher dose, and then the WSe 1.3 -WSe 2 /G heterostructure was obtained. We test the photoresponse and calculate the responsivity at different wavelengths, as shown in Figure S10, Supporting Information. Comparing the WSe 1.5 -WSe 2 /G, WSe 1.3 -WSe 2 /G heterostructure has higher responsivity (450 nm 0.14 A W −1 , 532 nm 0.11 A W −1 , 633 nm 0.2 A W −1 , and 780 nm 0.23 A W −1 ). A layer of PDBD-T film was also attached to the surface of WSe 1.3 -WSe 2 /G, and the performance test was carried out under the same laser irradiation condition, which also showed the red light resolution ability (450 nm 0.25 A W −1 , 532 nm 0.23 A W −1 , 633 nm 0.82 A W −1 , and 780 nm 0.23 A W −1 ). However, according to first-principles calculations ( Figure S11, Supporting Information), due to the existence of a large number of defect states in the WSe 1.3 -WSe 2 /G heterostructure. This is also proved by the experimental results shown in Figure S12, Supporting Information. The current levels obtained by the device under different tests are not stable.

PBDB-T/WSe 1.5 -WSe 2 /G Image Array
We construct a red-sensitive image array utilizing the PBDB-T/WSe 1.5 -WSe 2 /G heterostructure. Figure 5a shows an electrode array with 60 electrode couplers and dimensions of 120 × 140 μm 2 ( Figure S7, Supporting Information). Graphene and WSe 2 monolayers are transferred onto the electrode array and then partially irradiated by the focused Ga + ion beam. The Ga + ion beam irradiation process is automatically completed within 5 min by setting the program of the FIB machine. Figure 5b shows the local SEM image of the partially irradiated WSe 2 /G. Finally, the PBDB-T is spin-coated onto the WSe 1.5 -WSe 2 /G forming the PBDB-T/WSe 1.5 -WSe 2 /G photodetector array. Figure 5c shows the schematic diagram of the array under the illumination of lasers at different wavelengths (450, 532, and 633 nm) with a mask of numbers (0, 1, 2, and 3), respectively. Each number is observed as a single image from PBDB-T/WSe 1.5 -WSe 2 /G heterostructure arrays with the bias of 0 V as depicted in Figure 5d,e. As we can see, compared to the photocurrent of 450 ( Figure 5d) and 532 nm (Figure 5e) laser illumination for the numbers, only the pixels projected by 633 nm (Figure 5f) laser exhibited a sizeable photocurrent of around 0.4 μA. At the same time, the masked area displayed a feeble dark current, suggesting that all the number shapes can be observed in current contrast mapping with a clear resolution. These results suggest that the proposed array could be a promising device option for fabricating future broad image sensors. Based on this approach, photodetector arrays with 6 × 10 pixels become a great candidate for capturing visible images with different colors.

Conclusion
This work demonstrates a red-sensitive image array based on the heterostructure composited by 2D materials and polymers. FIB partially irradiates the WSe 2 /Graphene to construct a photovoltaic substrate (WSe 1.5 -WSe 2 /Graphene), and then the PBDB-T polymer, as a red-sensitive optoelectric material, is coated onto the WSe 1.5 -WSe 2 /Graphene substrate forming the PBDB-T/WSe 1.5 -WSe 2 /Graphene heterostructure. Photodetectors based on PBDB-T/WSe 1.5 -WSe 2 /Graphene heterostructures have a high sensitivity to red light enabling red-color discrimination. Utilizing the PBDB-T/WSe 1.5 -WSe 2 /Graphene heterostructure, we construct an image array with 6× 10 pixels and dimensions of 120 × 140 μm 2 , which realize the filter-free redcolor image. Our work proposes the hybrid structure of polymers and 2D materials as a novel avenue for developing the 2D colorresolvable image array.

Experimental Section
Sample Preparation: The monolayer WSe 2 monolayers were grown via chemical vapor deposition (CVD). The graphene and WSe 2 were then transferred to the Si substrate one after the other through the wetchemistry transfer process. A layer of PMMA was first spin-coated onto the substrate with WSe 2 or graphene and then baked at 100°C for 5 min. Next, it was immersed into a 2 mol L −1 KOH solution for 2 h to separate the PMMA/graphene (WSe 2 ) from the substrate, and it was swilled fully with DI water. It was then collected on the target substrate and baked at 80°C for 2 h. Finally, the film was immersed in acetone to remove the PMMA. The drain and source electrodes were fabricated via electron-beam lithography and evaporation (Au/SiO 2 /Si, Au: 50 nm, SiO 2 : 50 nm). The scattering cross-section was calculated to be ≈0.15-0.20 barn/sr via stopping and range of ions in matter (SRIM).
Ga Ion Beam Writing: The writing of focused Ga + ions was carried out using the FIB technique (model FEI START 400S). The acceleration voltage was 30 kV, while the beam current was 2.5, 9.3, and 21 nA.
Characterization: The fabricated devices were characterized using an optical microscope (Leica DM2700), an SEM (model FEI START 400S with accelerating voltage of 10 kV), an AFM, and a KPFM (Bruker, Dimension Icon). The electrical and optoelectronic properties of the devices were measured via a Keithley 4200A-SCS at room temperature and room pressure. WSe x /G heterostructures used for Raman spectra and XPS spectra measurements were produced by CVD and transferred on copper nets. The power of the laser used in Raman spectra measurements was 1 mW, and the wavelength was 532 nm.
DFT Calculations: First-principles calculations were performed using the density functional theory (DFT), adopting the plane-wave pseudopotential approach as implemented in the Vienna Ab initio Simulation Package (VASP). [43] The electron-electron interaction in the calculations was described via the Perdew-Burke-Ernzerhof (PBE) functional under the generalized gradient approximation (GGA). [44] The Van der Waals weak interaction between graphene and WSe 2 was corrected via the Grimme method (DFT-D3). For the electron wave function, a plane wave with a cutoff energy of 500 eV was used. A vacuum region of 20 Å was applied in the z-direction to avoid interactions between adjacent images. The convergence threshold parameters for the optimization and the subsequent calculations were set to be 10 −5 eV (energy) and 10 −2 eV Å −1 (force).