High‐Sensitive and Fast Speed UV Photodetector Based on HfSe2/InSe Heterostructure

The type‐II band‐aligned van der Waals (vdW) heterostructures are favorable for photocarrier separation and are often used for designing high‐performance photodetectors. Inspired by this, a metal‐mirror electrode enhanced HfSe2‐InSe vdW heterostructure photodetector is designed and demonstrated excitement performance in UV light detection. It is demonstrated the moderate bandgap heterostructure can be configured as a high‐performance UV photodetector with excellent light on/off ratio of 106, high photoresponsivity of 47.3 AW−1, competitive high specific detectivity of 3.2 × 1012 cmHz1/2W−1 and very low noise equivalent power of 2.8 × 10−16 WHz−1/2. Notably, the photoresponse speed of the device is very fast, with a rise time of 4.1 µs and a decay time of 5.4 µs. The results indicate that 2D HfSe2‐InSe vdW heterostructure possesses great potential applications in UV photodetection.


Result and Discussion
Figure 1a shows a schematic image of the HfSe 2 and InSe heterojunction.The HfSe 2 -InSe heterostructure was placed on a mirror Au electrode.The junction area was well-defined by the electron beam lithography (EBL) pattern, and outside the heterojunction, the region covered metal electrodes.The bottom electrode is designed to shorten the lateral transport distance of the photocarriers.To prevent oxidation of the device when exposed to air, a thin layer of polymethyl methacrylate (PMMA) is spincoated.Figure 1b, c shows the crystal structure's top and side views of monolayer InSe and monolayer HfSe 2 respectively.The monolayer InSe possesses a trigonal prismatic H-phase, which is different from the monolayer HfSe 2 octahedral T-phase.Raman spectroscopy was used to characterize multilayer HfSe 2 and InSe nanosheets.The excitation light source is a 532 nm laser.
As shown in Figure 1d, three obvious Raman peaks were observed, which correspond to the vibration models of A 1g 1 , E 2g 1 , and A 1g 2 of InSe located at 113.9, 176.9, and 225.8 cm −1 , [45,46] respectively.For the HfSe 2 nanoflake, the Raman spectrum is presented in Figure 1e.Only an obvious A 1g model at 198.3 cm −1 was observed. [36,47,48] Then, we systematically investigate the photoresponse of HfSe 2 -InSe devices.The photoresponse of the heterojunction device in the UV wavelengths of 275 and 365 nm were carried out.Figure 2a shows the I-V curves of the device under dark and different powers from 0.05 to 5.71 nW of a 275 nm light-emitting diode (LED) illumination.The optical response under reverse bias is higher than that under forward bias.The linear plot of the I-V curves is presented in Figure S3a (Supporting Information).The photovoltaic effect with SBUV 275 nm light was observed.We extracted the open-circuit voltage (V OC ) and short-circuit current (I SC ) versus light power as shown in Figure S3b (Supporting Information).The V OC exhibits weak light power dependence and a bit of fluctuation from 0.36 to  2c.The light power is ranged from 2.36 to 9.48 nW.At the reverse bias, the photoresponse is much higher than at the forward bias, which is rarely observed in 2D material photodetectors.The time-resolved photoresponse of the HfSe 2 /InSe vdW heterostructure photodetector with different illumination power of 365 nm is presented in Figure S4b (Supporting Information).As the incident light power was fixed at 9.48 nW, the photocurrent up to 59.6 nA was obtained.Then we calculate the R and EQE of the HfSe 2 -InSe vdW heterostructure detector with the illumination power of 365 nm light, as shown in Figure 2d.The R fluctuated from 2.7 to 5.9 AW −1 as the incident light power ranges from 2.36 to 9.48 nW.When the indent optical power was fixed at 9.48 nW, the EQE is up to 2021.7%.Next, we measured the time-resolved photoresponse of the HfSe 2 /InSe vdW heterostructure photodetector at visible wavelengths of 405, 520, and 637 nm.We systematically investigated the photoresponse of the device under a 405 nm laser.As shown in Figure S5a (Supporting Information), the output curves of the HfSe 2 -InSe vdW heterostructure device under various light power and in the dark.The incident light power increased from 3.72 nW to 1.65 μW.The current under the laser is much higher than in the dark.A high light on/off ratio of 10 7 is demonstrated at 0 V bias.The time-resolved photoresponse at 2 V bias under various incident light power is plotted in Figure S5b (Supporting Information).When the laser is turned on, the current increased sharply and becomes stable.After three cycles of the switch on/off the laser, the photocurrent can be repeated steadily.The photocurrent was up to 10.6 μA as the incident light power was set at 6.51 μW.Then we extracted the dependence of photocurrent on the incident light powers, as shown in Figure S5c (Supporting Information).The power law relationship I P ∝ P  with  = 0.51 < 1 (sublinear behavior) is observed.This sublinear behavior was observed in 2D materials photodetectors with trap stated.The small value of the , the larger number of trap states participate in photocurrent contribution. [49]The dependence of R and EQE on the light power is shown in Figure S5d (Supporting Information).The highest R of 34.2 AW −1 and corresponding EQE of 10 458.7% were demonstrated at 2 V bias.The dependence of R on the incident light power can be fitted by the Hornbeck-Haynes model [50,51] where, C = e   2  l /(h c t ),  is the light absorption efficiency, F and F 0 in units of s −1 are the photon absorption rates at unsaturated and saturated trap centers,  l and  t is the lifetime of the photocarriers and the transport time respectively.F is proportionate to the incident light power P and the parameter n = 1 for simplicity. [51]R can be expressed as R = a b+P , where a and b are two fitting parameters.The fitted result as shown in Figure S5d (Supporting Information) the dashed purple line was obtained at a = 4.7 × 10 −6 and b = 1.405 × 10 −7 .When the light power is higher than 2 μW, the R becomes higher than the fit value.The traditional photoresponse without gain plays a major role in photoresponse.This can be explained as the light power was high enough, and the trap centers are saturated.The temporal photoresponse at 2 V bias at different incident light powers is shown in Figure S6 (Supporting Information).The photocurrent of 4.2 μA was demonstrated.Next, we investigate the performance of the device under a 637 nm laser.Figure 3a shows the I-V curves of the HfSe 2 -InSe vdW heterostructure device in the dark and under various illumination powers.As the light incident on the device, the current increase considerably.The light on/off ratio at 0 V bias is up to 10 7 , while the on/off ratio decreased to 10 6 as the bias decreased to -1 V.The time-resolved photoresponse at 2 V bias is shown in Figure S7a (Supporting Information).The incident light power increased from 0.10 to 22.59 μW and the photocurrent increased from 0.1 to 2.7 μA.The extracted R and EQE as a function of incident light power are shown in Figure S7b (Supporting Information).The R of 1.2 AW −1 and EQE of 236.53% were obtained at 0.1 μW light power.This HfSe 2 -InSe vdW heterostructure exhibits a good photovoltaic response.The I-V curves of the device at different incident light (637 nm laser) powers are plotted in Figure S7c (Supporting Information).Then we extract the I sc and  S1 and S2 (Supporting Information).
V oc with different light powers.As shown in Figure 3b, the V oc and I sc versus incident light power are presented.The V oc shows weak light power dependence and shows a bit of fluctuation from 0.50 to 0.55 V.The I sc shows the linear dependence on the light power without exhibiting the signature of saturation over the full measured power range.Output electrical power P el = I sc × V oc as a function of incident light power is presented in Figure S7d (Supporting Information).The maximum value of P el of 48.7 nW was obtained at 61.37 μW illumination.The fill factor (FF), which is one of the key parameters for the solar cell, can be calculated according to FF = P el,m /(I sc V oc ).The power conversion efficiency (PCE) is defined as PCE = P el,m /P I , where P el,m is the maximum output electrical power.The FF and PCE as a function of incident light power are presented in Figure 3c.The FF shows a bit of fluctuation from 0.21 to 0.30.The highest PCE of 0.27% is obtained which is comparable with the WSe 2 p-n junction (0.1%-0.6%) [52] and WSe 2 -MoS 2 (0.2%) [53] heterostructure.The photoresponse time of the device at V ds = 2 V under a 637 nm laser is plotted in Figure 3d.The photoresponse speed is very fast, with a rise time of  r = 4.1 μs, and a decay time of  d = 5.4 μs.The photoresponse times with 275 and 365 nm light are shown in Figure S8a and S8b (Supporting Information) respectively.The photoresponse speed in the UV spectral range is two orders of magnitude slower than that of in the visible range, which further confirms the trap stated induced a higher photogain in the UV spectral range.
To understand the photoresponse at the reverse bias is higher than that at the forward bias, we analyzed the band alignment of the HfSe 2 -InSe vdW heterostructure.The work function of InSe is ≈4.8 eV, [42] which is smaller than that of HfSe 2 5.5 eV. [43]As shown in Figure S9a (Supporting Information), the energy band profile of HfSe 2 and InSe before contact.Due to the larger work function of HfSe 2 , the Fermi level of HfSe 2 is lower than that of InSe.After the HfSe 2 is contacted with the InSe, the electrons flow from the InSe to HfSe 2 till the Fermi levels are at the same position, as shown in Figure S9b (Supporting Information).At the forward bias, the HfSe 2 is connected to the anode.The photogenerated electrons flow from the InSe to HfSe 2 with the help of bias voltage.While the hole flows from HfSe 2 to InSe will be impeded by the large valance band offset, as shown in Figure S9c (Supporting Information).While at the reverse bias, both the electron and hole can flow unimpeded as shown in Figure S9d (Supporting Information), which gives a large photoresponse.Figure 4a shows the wavelength-dependent R and EQE at −2 V bias.The R-value shows a rapid decline in both the ultraviolet and visible ranges The R decreased from 47.3 to 7.0 AW −1 as the wavelength increased from 275 to 365 nm.Then in the visible range, the R decreased from 46.7 to 1.6 AW −1, and the corresponding EQE decreased from 14 318.9% to 307.8% as the wavelength increased from 405 to 637 nm.The UV-vis rejection ratio was used to evaluate the spectral selection ability of a photodetector.The high UV-vis rejection ratio of R 275 /R 637 = 39.4.Finally, in Figure S10 (Supporting Information), we measured the current noise density spectrum under different deviations using a homemade noise spectrum analyzer (PDA NC300L, 100 kHz bandwidth).In the reverse bias, the noise current spectrum is much lower than that of a forward bias.NEP is defined as the signal incident light power incident on the detector when the detection signal-to-noise ratio is S/N = 1.It represents the lowest signal power that a detector can pick up from the background noise.Since the noise level is proportional to the sign of the measurement electrical bandwidth, NEP specifies the measurement results at a 1 Hz bandwidth condition.The NEP can be obtained by the formula NEP = i n /R, where the i n is the measured noise current.In the low-frequency range, the flicker noise (1/f) dominates the noise current contribution.We adopt the mean-square-root <i n 2 > 1/2 to represent the in at 1 Hz bandwidth.The <i n , where B is the measured electric bandwidth.At −2 V bias, <i n 2 > 1/2 = 1.35 × 10 −14 AHz −1/2 was obtained.The value of D* can be calculated by D* = (A d B) 1/2 /NEP, where A d is the area of the device.Figure 4b shows the NEP and D* as a function of wavelength.In the full spectral range, the NEP is lower than 8.5 × 10 −15 WHz −1/2 .In the SBUV 275 nm, the NEP of 2.8 × 10 −16 WHz −1/2 is obtained.The D* as high as 3.2 × 10 12 cmHz 1/2 W −1 was realized at 275 nm.In the full spectral range, the D* is higher than 1.0 × 10 11 cmHz 1/2 W −1 .The results indicate this HfSe 2 -InSe vdW heterostructure device has great potential application in SBUV detection.Figure 4c shows a comparison of photoresponse time and light on/off ratio with other SBUV photodetectors.Our HfSe 2 -InSe vdW heterostructure photodetector (red star) exhibits an ultra-high-light on/off ratio of 10 7 and fast speed compared with that based on 2D materials and wide bandgap material.The summary of the D* and response time of other 2D material SBUV detectors and wide bandgap semiconductor films is shown in Figure 4d.Our device exhibits both high sensitivity and high speed, which is highly desired in SBUV detection.

Conclusion
In conclusion, we design a metal mirror electrode-enhanced HfSe 2 -InSe vdW heterostructure photodetector.The device exhibits excellent performance in SBUV detection.The competitive performance includes ultra-high light on/off ratio of 10 7 , high R of 47.3 AW −1 , very low NEP of 2.8 × 10 −16 WHz −1/2 and a remarkable D* as high as 3.2 × 10 12 cmHz 1/2 W −1 .Notably, this vertical HfSe 2 -InSe vdW heteros'tructure photodetector with very fast speed with a decay time of 5.4 μs, and a rise time of 4.1 μs.These results prove that HfSe 2 -InSe holds great potential application in high-performance SBUV detection.
See the Supporting Information for the EDS elements mapping, additional data for the performance of photodetector at SBUV 275, 405, 520, and 637 nm lasers, energy band alignment of the HfSe 2 /InSe vdW heterostructure, and current noise density spectra and Tables S1 and S2 (Supporting Information).Summary of the performance of the SBUV photodetectors.

Figure 1 .
Figure 1.Schematic diagram and material characterization of HfSe 2 /InSe vdW heterogeneous photodetector.a) Schematic image of metal mirror electrode enhanced HfSe 2 /InSe vdW heterojunction photodetector.b) and c) Crystal structures of monolayer InSe and HfSe 2 respectively.d) and e) Raman spectra of few-layer InSe and HfSe 2 flakes respectively.
Figure S1a (Supporting Information) is the scanning electron microscope (SEM) image of a multilayer InSe flake.The elements were analyzed by energy dispersive x-ray spectroscopy (EDS) element mapping.The EDS mappings of In and Se are shown in Figure S1b,c (Supporting Information).The SEM image of the measured HfSe 2 flake is presented in FigureS1d(Supporting Information).Figure1e,f are the EDS element mappings with uniform distribution of Hf and Se.The elements ratio of In: Se = 4.0: 3.3 and Hf: Se = 1.3: 2.4 are very close to 1:1 and 1:2, as shown in FigureS1gand h (Supporting Information) respectively.The thickness of the device was measured by atomic force microscopy (AFM).The AFM image of a typical HfSe 2 /InSe vdW heterostructure is shown in FigureS2a(Supporting Information).The thicknesses of HfSe 2 and InSe are 18.4 and 9.6 nm, as shown in FigureS2band S2c (Supporting Information) respectively.

Figure 2 .
Figure 2. Photoresponse of the HfSe 2 /InSe vdW heterostructure device at UV spectral range.a) I-V curves of the device under different powers of illumination at 275 nm.b) R (left axis) and EQE (right axis) of a typical HfSe 2 /InSe vdW heterostructure device versus incident light power at a bias of −2 V. c) Output curves of the HfSe 2 /InSe vdW heterostructure device in the dark and under various incident light powers of a 365-nm illumination.d) R (left axis) and EQE (right axis) of a typical HfSe 2 /InSe vdW heterostructure device versus incident light power at a bias of −2 V.

Figure 3 .
Figure 3. Photoresponse of the HfSe 2 /InSe vdW heterostructure device at 637 nm wavelength.a) Output curves of the HfSe 2 /InSe vdW heterostructure device at 637 nm laser darkness and different incident light power.b) Extracted V oc and I sc as a function of incident light power.c) Illumination powerdependent fill factor (FF) and PCE.d) The photoresponse speed of the device with a rising time of  r = 4.1 μs and decay time of  d = 5.4 μs.

Figure 4 .
Figure 4. Sensitivity of the HfSe 2 /InSe vdW heterostructure device.a) R and EQE as a function of wavelength at −2 V bias.b) Wavelength dependence of the NEP and D* of a HfSe 2 /InSe vdW heterostructure device.c) Comparison of photoresponse speed and light on/off ratio among reported SBUV photodetectors.d) Summarized the response time and D* among reported 2D materials SBUV photodetectors.References to the selected work can be found in TablesS1 and S2(Supporting Information).