Self‐Driven Fast‐Speed Photodetector Based on BP/ReS2 van der Waals Heterodiode

Broadband photodetectors based on 2D layered materials provide great potential applications in night vision, sensing, and communications. However, it remains a challenge for detectors to achieve both high photoresponsivity and fast response. Here, a high‐sensitive photodetector based on ReS2/BP van der Waals (vdW) heterodiode with fast speed, rising time (τr) of 770 ns, and decay time (τd) of 760 ns under a 637 nm laser is reported. The detection range is covered from visible to mid‐wave infrared (MWIR) 0.405–3.753 μm. In the visible range, a high photoresponsivity of 107.1 AW−1, competitive specific detectivity (D*) of 1.89 × 1010 cm Hz1/2 W−1, and a low noise equivalent power of 3.03 × 10−14 W Hz−1/2 are obtained. In the MWIR the D* of 3.26 × 108 cm Hz1/2 W−1 is demonstrated in the photovoltaic model. Notably, the photodiode realizes a high external quantum efficiency of 71.8%, and a high power conversion efficiency of 2.0%. This work provides a way to design broadband response and fast‐speed self‐driven photodetectors with great potential applications in weak light intensity.


Introduction
Photodetectors with broadband response, especially in the MWIR, have attracted extensive attention in the past few years due to their excellent performance and great potential applications in imaging, remote sensing, optical radar, and free-space communication. So far, higher performance photodetectors based on 2D semiconductors with narrow bandgaps such as DOI: 10.1002/adsr.202300029 black phosphorus (BP), [1,2] black arsenic phosphorus (b-AsP), [3] platinum diselenide (PtSe 2 ), [4] and palladium selenide (PdSe 2 ) [5] were demonstrated. However, the photodetectors based on the narrow bandgap 2D semiconductors were suffering from the large dark current and high background current noise. To obtain highly sensitive uncooled MWIR, the dark current should be depressed. Thanks to the dangling bonds-free 2D layered materials, a variety of vdW heterostructures could be fabricated by the try transfer technique [6] without being limited by lattice-mismatched, and an atomic sharp interface without contamination can be prepared in the laboratory. To date, BP is one of the most extensively studied materials due to its direct bandgap property, [7] high mobility both for hole and electron, [8][9][10] high light absorption, [11] and anisotropic crystal structure. [12,13] In bulk form, the bandgap of BP is approximated to ≈0.3 eV, [14] which can be used as MWIR photodetection. [2] The bandgap of the BP can be easily tuned by the strain, [15,16] electrical field, [17] and the thickness, [7] which can be used to broaden the detection spectral range. Using the vertical electrical field applied on a 5 nm-thick BP, the detection range can be extended from 3.7 to 7.7 μm. [17] BP is a p-type semiconductor, [18] and it can be staked  with n-type 2D materials to form atomic thin p-n junctions. Blackbody radiation MWIR photoresponses at room temperature were demonstrated based on BP phototransistor [1] and BP-MoS 2 , [19] and MoS 2 /BP/MoS 2 [20] heterostructures. Based on the BP/InSe heterostructure, the ballistic avalanche phenomena [21] in the MWIR were observed. Polarization-sensitive photodetectors were realized based on anisotropic BP, [22] b-AsP, [3] and BP/ReS 2 heterostructure. [23] In the 2D ReS 2 , the lowering of the lattice symmetry induced electronic anisotropic properties [24] and polarization sensitive photodetection [25] were demonstrated. Based on the ReS 2 phototransistor, ultrahigh photoresponsivity [26] with an excellent ability to detect weak light [27] was realized for the strong light absorption of the direct bandgap ReS 2 due to weak interlayer coupling. However, due to the long lifetime trap states and low electron mobility of ReS 2 , the speed of photoresponse time is very slow (>100 s), which can be enhanced by oxygen (O 2 ) plasma treatment. [26] After treatment, the photoresponse speed of the ReS 2 phototransistor enhances considerably from 16.7 to 0.67 s. However, the response speed is still very slow and needs further improve. The bandgap of ReS 2 is ≈1.45 eV [28] (corresponding to 855 nm) which is independent of the thickness. The operation spectral range of the ReS 2 photodetector has only covered the near-infrared. To realize broadband photoresponse, narrow bandgap materials should be induced. 2D vdW heterodiode can be fabricated using n-type ReS 2 [24,29] stacked with p-type BP. The built-in electrical field can depress the dark current and realize a highly sensitive pho-todetector. Here, we report an ultrafast photodetector vertically vdW heterodiode by staking p-type BP on an n-type ReS 2 . The BP/ReS 2 heterodiode was placed on a metal electrode to shorten the photocarriers' lateral transport distance and to achieve high photocarrier collection efficiency. The device exhibits ultrabroad band photoresponse from 0.405 to 3.753 μm, and ultrafast photoresponse speed with r = 770 ns and d = 760 ns under a 637 nm laser. The detection range is covered from visible to mid-wave infrared (MWIR). The competitive performance of the BP/ReS 2 heterodiode includes D* of 1.89 × 10 10 cm Hz 1/2 W −1 in the visible range and 3.26 × 10 8 cm Hz 1/2 W −1 in MWIR range operation in the photovoltaic model.

Results and Discussion
To investigate the optoelectric properties of the BP/ReS 2 vdW heterodiode, we fabricated the BP/ReS 2 through the dry transfer technique. The heterostructure was fully placed on a metal electrode to obtain good contact and shorten the photocarrier lateral transport distance. The schematic image of the heterodiode is presented in Figure 1a. The crystal structure of the ReS 2 (top view and side view) is shown in Figure 1b left panel. The previous report according to the DFT calculations result indicates that the hexagonal (H) phase of ReS 2 is unstable. [28] The stable phase is a distorted 1T structure, which is come from the octahedral (T) phase and goes through the Peierls distortion. There are two principal axes (highlighted by red arrows), the a and b axes. The b-axis The obtained ReS 2 thin flake was characterized by the Energy dispersive X-ray spectrometer (EDS) elements mapping images of Re and S, as shown in Figure S1a,b (Supporting Information) respectively. The EDS spectrum is shown in Figure S1c (Supporting Information). The atomic ratio of Re: S = 32.4:67.6, which is very close to 1:2. Figure S2a (Supporting Information) shows the AFM image of a typical ReS 2 /BP heterodiode device. The thickness of ReS 2 is ≈10.6 nm and BP is ≈7.5 nm, as shown in Figure  S2b,c (Supporting Information) respectively. The Raman spectra of ReS 2 and BP of the device are shown in Figure 1c. In the up panel of Figure 1c, the three Raman peaks were observed at 359, 438, and 464 cm −1 which were related to the A 1 g , B 2g , and A 2 g phonon vibration modes respectively. [12,30,31] Five major peaks of ReS 2 appeared at 137, 149, 160, 210, and 316 cm −1 , which is consistent with previous results. [24,28] Then we study the electrical transport properties of the BP/ReS 2 heterodiode. As shown in Figure 1d, the output curve of the heterodiode with a good rectification behavior was observed. The semilogarithmic plot of the output curve is shown in the inset of Figure 1d. The rectification ratio of 10 2 was obtained. The energy band alignment of BP and ReS 2 before contact is presented in Figure 2a. The affinity and bandgap of multi-layer BP and ReS 2 are 4.2, 0.3, [32] and 4.4, 1.3 eV [23,33] respectively. The work function of BP (≈4.5 eV) is higher than the ReS 2 (4.8 eV). After the ReS 2 contacted the BP, the electrons were transferred from BP to ReS 2 to build a new equilibrium state, and an interface barrier was formed. Figure 2b shows the energy band diagram of the new equilibrium state after contact, exhibiting type II band alignment, which is favorable for designing a photodetector. When the heterostructure device operated at a forward bias (positive bias at BP), the recombination of the majority carriers at the interface results in a larger current, as shown in Figure 2c. While the heterodiode device was operated in reverse bias (negative bias at BP), the minority carrier electrons in BP crossed the interface barrier to reach ReS 2 , as well as the minority carrier holes in ReS 2 to cross the potential barrier to BP and resulted in a smaller reverse current, as shown in Figure 2d. At the reverse bias, the BP/ReS 2 heterodiode device exhibits a very low dark current.
To explore the photoresponse of the BP/ReS 2 heterodiode in the visible range, we systematically investigated the photoresponse of the device under 405, 520, and 637 nm lasers. Figure 3a shows the I-V curves of the BP/ReS 2 heterodiode with and without 405 nm laser illumination. When the light illuminated the device, the current increased obviously. The photocur-rent (I P ) can be calculated by I P = I Light −I Dark . As light power was fixed at 1.46 μW, the photocurrent of 56.8 and 20.8 μA were obtained at 1 and −1 V bias respectively. The photocurrent at the forward bias is much higher than at the reverse bias. However, the light on/off ratio at 1 and −1 V of 4.8 and 42.6 was obtained, indicating the device operation at the reverse bias with a high signal-to-noise ratio. Figure S3a (Supporting Information) shows the semilogarithmic plot of I-V curves under various illumination powers of a 405 nm laser. The light power increased from 5.8 nW to 2.5 μW. To evaluate the performance of the device, we calculate the photoresponsivity (R) and EQE, which are defined as the ratio of the photocurrent I P to the incident light power and the ratio of the number of photoexcited carriers forming the photocurrent to the number of incident photons, respectively. The R = I P /P and EQE can be expressed as EQE = R h c/e , where h, c, e, and are the Plank's constant, speed of light, the elementary electronic charge, and wavelength of the incident light. Figure 3b shows the light power-dependent R and EQE of the device at a −1 V bias. The high photoresponse of 107.1 AW −1 and a corresponding EQE of 32778.9% was achieved at the incident light power of 21.5 nW. R and EQE decreased as the incident light power increased. When the light power was increased to 10.6 μW, R and EQE drop to 1.4 AW −1 and 416.0% respectively. The dependence of R on the incident light power could be attributed to the inevitable trap state that widely exists in 2D materials and the interface of the heterostructure. At the higher light power, most of the trap states were trapped carriers, and the number of available trap states was reduced. As the trap states were all filled, the probability of the photocarrier recombination increased, resulting in a light power-dependence photocarrier lifetime. The dependence of R on the incident power can be well described by the Hornbeck−Haynes model: [34] where is a coefficient, F is the photon absorption rate in units of s −1 , and F 0 is the photon absorption rate when trap saturation occurs. The purple dashed line in Figure 3b is fitted by the Hornbeck−Haynes model. At a higher light power, the experimental R is higher than that of the fit results, indicating that the traditional photoconductance effect plays a major role in photoresponse. To study the bias dependence R and EQE, we also measured the time-resolved photoresponse at −0.1 V bias by changing the incident power of the 405 nm laser, as shown in Figure 3c.   Figure S4a,b (Supporting Information) respectively. The extracted R and EQE as a function of incident light power are plotted in Figure S4b (Supporting Information) for −0.1 V and Figure S4d (Supporting Information) for 0 V bias. At −0.1 V bias, the R of 9.1 AW −1 was obtained under the illumination of 1.1 nW. For the photovoltaic response under 520 nm laser, the extracted R and EQE as a function of light power are shown in Figure S4d  The V oc shows the linear dependence on the ln(P) as expected from the conventional p-n junction theory. The V oc can be obtained by the formula qV oc = (E cp − E vn ) + k B Tln(np/N c 2 ), [30,35] where n and p are the ReS 2 electron and BP hole carrier density, E cp , E vn are the lowest unoccupied molecular orbital (LUMO) of ReS 2 and the highest occupied molecular orbital (HOMO) of BP respectively, N c is the effective density of state, and k B is the Boltzmann's constant. As the device is at the open-circuit condition, the equilibrium state was built at the balance between photogeneration (GR ∝ P) and recombination rates (RR). The RR can be obtained by the empirical equation [36] RR = n , where is the recombination prefactor, and is the recombination order, = 1 for monomolecular (Shockley Read Hall; SRH) recombination [37] and = 2 for bimolecular (Langevin) recombination. For the photon excitation (intrinsic excitation), n = p. By equating RR and GR, np ∝ P in 2/ . And we obtain the equation Adv. Sensor Res. 2023, 2, 2300029 That we extracted the = 0.95 by fitting the experimental data at a higher power range as shown in Figure 4b the purple dashed line, which is very close to 1 for SRH recombination to dominate the interlayer recombination. The fill factor (FF) and PCE are calculated as follows: FF = P max /I sc ·V oc , PCE = P max /P in , where P in is the incident light power. As shown in Figure 4c, the FF of 0.38 and PCE of 2.0% were obtained under an incident power of 10.4 μW. The electrical power P el , defined as P el = I ds ·V ds . As presented in Figure 4d, the P el increases with increasing incident power. For the highest illumination power, the P el reaches 0.57 μW. To estimate the performance of the device, measure the time-resolved photoresponse at the bias of 0.1, 0, and −1 V as shown in Figures S5a-c (Supporting Information), respectively. As shown in Figure S5a (Supporting Information), with five cycles of the switch on/off the light at 1000 Hz frequency, the device exhibited excellent repeatability. Figure S5b  The photoresponse speed is one of the most important performance indexes of a photodetector. The response speed of a device is usually evaluated by the rising time ( r ) and decay time ( d ), which are defined the time as the photocurrent increasing from 10% to 90% of the stable photocurrent when the laser is turned on and decreased from 90% to 10% of the stable photocurrent when the laser is turned off. The response time of the ReS 2 /BP heterogeneous diode under a 637 nm laser is shown in Figure 4f. At 0.1 V bias, the rising time r of 0.77 μs, and the decay time d of 0.76 μs were obtained. The speed of this device is much faster than other 2D material photodetectors. The PCE and photoresponse time of the 2D material-based photodetector is summarized in Tables S1 and S2 (Supporting Information).
Then, we studied the performance of the device in the shortwave infrared (SWIR) and MWIR region. In the SWIR range, we systematically investigated the photoresponse of the device under a 1650 nm laser. We measured the time-resolved photoresponse at 0.1 and 0 V biases by changing the incident power of the 1650 nm laser, as shown in Figure S6a,b (Supporting Information) respectively. The extracted R and EQE as a function of illumination power at 0.1 V bias are plotted in Figure S6c (Supporting Information). The R of 0.5 AW −1 corresponding to an EQE of 38.8% was obtained under 47.9 nW illumination. In Figure S6b (Supporting Information), as the incident light power was fixed at 243.2 μW, the photocurrent of 0.8 μA was obtained. Then, we measured the response time of the device under 1650 nm laser irradiation. As shown in Figure S6d (Supporting Information), the rise time of r = 48.7 μs and the decay time of d = 56.2 μs was demonstrated at 0 V bias. The photovoltaic response of the device in the MWIR was investigated. Figure S7a-c (Supporting Information) respectively shows the time-resolved photoresponse of the BP/ReS 2 heterodiode device under 3300, 3403, and 3753 nm MWIR lasers at 0 V bias with different incident light powers. Under a 3300 nm laser, the photocurrent of 20.9 nA was obtained at 21.84 μW. The device exhibited a good MWIR photovoltaic response. Figure 5a presents the wavelength dependence of R and EQE at a bias of 0 V. The ReS 2 /BP heterodiode device exhibits a good photovoltaic response from the visible range to mid-wave infrared (0.405-3.753 μm). From 405 to 637 nm of visible light, R increases from 0.17 to 0.37 AW −1 , and corresponding EQE increases from 51.3% to 71.8%. Then, with the increase of wavelength, R, and EQE present a decreasing trend. From 1650 to 3753 nm, R shows a sharp decrease from 0.33 AW −1 to 1.34 mAW −1 , and the corresponding EQE decreased from 24.8% to 0.044%. The sharp decrease of the R from SWIR to LWIR could be ascribed to that the photocarriers were blocked by the interface barrier, as shown in Figure S8 (Supporting Information). In the SWIR 1650 nm, the photon energy is ≈0.77 eV much higher than the bandgap of BP (0.3 eV). During the device under the SWIR, only BP absorption the light and the electrons can be excited to a higher energy level as shown in Figure S8a (Supporting Information). The photogenerated electrons with higher energy can get cross the interface barrier easily and form the current. While for the MWIR (3753 nm, corresponding to 0.33 eV), the electrons were excited to the conduction band minimum, as shown in Figure  S8b (Supporting Information). The photogenerated carriers were impeded by the interface barrier and most of them were recombined before from the photocurrent. To evaluate the sensitivity of the device, the specific detectivity D* and NEP were calculated. NEP is used to evaluate the photodetector's ability to distinguish the minimum light power from the noise and it can be calculated by NEP = i n /R, wherein is the noise current in the 1 Hz measurement electrical bandwidth. D* represents the detection limit of the photodetector, and it can be calculated by D* = (A·B) 1/2 /NEP, A is the effective area of the device, and B is the electrical bandwidth. To calculate NEP and D*, we measured the current noise power spectra S(f) at different biases 0, −0.1, and −1 V, as shown in Figure 5b. At low bias, the current noise power is much lower than at higher bias. At 0 V bias, the root-mean-square www.advancedsciencenews.com www.advsensorres.com noise current <i n 2> 1/2 = 1.12 × 10 −14 AHz −1/2 was calculated by The shot noise could be estimated by <i n 2> = 2eI d Δf = 1.09 × 10 −30 A 2 Hz −1 , which is two orders of magnitude lower than the measured result, indicating the shot noise is not the major contributor. Figure 5c shows the wavelength dependence of NEP and D* of the BP/ReS 2 heterodiode at V ds = 0 V. For the ReS 2 /BP heterostructure device, D* of 1.89 × 10 10 cm Hz 1/2 W −1 and NEP of 3.03 × 10 −14 W Hz −1/2 were obtained at 637 nm. From the SWIR 1650 nm to the MWIR 3753 nm, the D* shows a rapid decrease from 1.69 × 10 10 to 6.89 × 10 7 cm Hz 1/2 W −1 , and NEP increased from 3.39 × 10 −14 to 8.36 × 10 −12 W Hz −1/2 . Figure 5d summarizes the relationship between R and the response time of 2D material photodetectors. The devices located in region 4 marked by the pink ellipse, exhibit low response speed (≈10 ms) and R (≈1 mAW −1 ). The device distribution in region 3 marked by a blue ellipse, shows a low response speed (≈10 ms) but a high R (100 AW −1 ). The devices in region 2 marked by green ellipse present relatively fast response speed (10 μs) and moderate R (1 AW −1 ). Our device located in region 1 marked by a red ellipse exhibits a very fast response speed (0.7 μs) and high R (107.1 AW −1 ). Our response time is much faster than that of other 2D materials-based photodetectors.

Conclusion
In summary, we demonstrated a ReS 2 /BP vdW heterodiode photodetector with high photoresponse and fast response speed.

Experimental Section
Device Fabrication: The bottom electrodes were fabricated using the ultraviolet lithograph method. First, the photoresist (AZ5214) was spincoated on a N-type phosphor-doped silicon (0.5 mm, orient: <100>, resistivity 0.01-0.05 Ω cm) substrate covered by 300 nm SiO 2. Then the pattern mask was placed on the substrate and exposed under a 365 nm ultraviolet light using contact mode. After the standard development and fixer processes, a 5 nm/45 nm Ti/Au film was deposited by electron beam evaporation, and followed by a standard lift-off process, the bottom electrodes were obtained. The BP/ReS 2 heterostructure devices were fabricated using the dry transfer technique. Firstly, BP and ReS 2 multi-layer flakes were exfoliated to polydimethylsiloxane (PDMS) by standard mechanical exfoliation method. The prepared multi-layer BP and ReS 2 flakes on PDMS were transferred to a bottom electrode in a sequence of transfer ReS 2 first and then BP to build a vertically stacked BP/ReS 2 heterojunction. As the environmental instability of BP, [38] which affects the performance of the BP/ReS 2 heterodiode photodetector, the full process of the BP/ReS 2 heterodiode device fabrication was carried out in a high-purity nitrogen-filled glove box (moisture and oxygen content less than 0.1 ppm). The top electrodes were patterned by electron beam lithography (EBL). The top electrode metal film of 10 nm Cr and 80 nm Au was deposited using electron beam evaporation. To prevent ReS 2 /BP heterojunction device exposed to the air, a thin PMMA film was spin-coated before taking out of the glove box.
Materials Characterization: The thicknesses of the BP and ReS 2 flakes were examined by atomic force microscopy (HITACHI, AFM 5500 M). Raman spectroscopy was measured using a 532 nm laser as the excitation source to characterize black phosphorus (BP) and rhenium disulfide (ReS 2 ) sheets. EDS elements analysis was carried out using the Zeiss cross-beam 550L system.
Electrochemical Characterization: Electrical and optical response measurements were made using a dual-channel digital source meter (Keithley 2636B). A homemade wavelength-tuneable multichannel MWIR laser (2.5-4.2 μm) was used to study the MWIR performance of the photodetector. The diameter of the MWIR laser beam was ≈3 mm. The performance of the BP/ReS 2 heterodiode device in the visible and SWIR spectral range (from 405 to 1650 nm) was measured using semiconductor fiber lasers, which were focused on the device using a 20× objective lens. Noise current density spectra at various biases were measured in a metal-shielded box. Data were acquired using a noise measurement system (PDA NC300L, 100 kHz bandwidth).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.