Self-driven highly responsive p-n junction InSe heterostructure near-infrared light detector

Chandraman Patil; Chaobo Dong; Hao Wang; Behrouz Movahhed Nouri; Sergiy Krylyuk; Huairuo Zhang; Albert V. Davydov; Hamed Dalir; Volker J. Sorger

doi:10.1364/PRJ.441519

Two-dimensional (2D) semiconducting materials due to their bandgap have been studied as promising photodetector materials, by changing the layer numbers or forming van der Waals (vdW) heterostructures, owing to their high responsivity, fast response time, broadband photodetection, photodetectivity, and low dark current noise [1 –7]. The operation of these high-performance devices demands high bias voltage leading to large power consumption due to the Schottky barrier potential and poor photogenerated carrier collection. This limits technological applications for remote operation conditions under extreme environments, biomedical sensing, and portable devices [8 –11]. However, self-driven photodetectors are promising devices to solve energy consumption issues where the photon energy is higher than the bandgap of the material for a better signal-to-noise ratio. Recently, few self-driven 2D material-based photodetectors were demonstrated in Refs. [9,12 –14]. III–VI group 2D materials (InSe, GaSe, and GaTe) have been recently studied for light–matter interaction properties for optoelectronic applications [15 –17]. 2D InSe, with its direct bandgap ( $\sim 1.25 eV$ ) [18], has recently been investigated, showing higher ultrasensitive photodetection characteristics than other 2D semiconducting materials such as ${MoS}_{2}$ and ${WSe}_{2}$ [19 –22]. p- and n-doped materials used in heterostructure devices form atomically sharp p-n junctions [23 –25].

A thorough study has been performed on InSe-based photodetectors in the visible spectrum but is yet to be explored in detail for near-infrared (NIR) applications [26,27]. Optoelectronic devices for detection, modulation, and sensing applications in the mid-infrared spectrum are also limited by the lack of on-chip sources and the high cost of production [28 –30]. Optoelectronic devices based on InSe are usually designed in transistor configuration that requires high electrical bias gating and bias voltage [19,27,31]. Furthermore, the NIR wavelength at 980 nm is widely used for optical humidity sensing fields such as indoor air quality, industrial production process control, and agricultural instrumentation [32,33]. Also, with the technological advancement in building autonomous vehicles, LiDAR-based devices usually are designed at 940–980 nm wavelengths for short to medium range positioning and mapping [34,35]. Due to a direct bandgap (1.25 eV) [18], InSe is an attractive material for manufacturing optoelectronic devices in this range.

Here, we demonstrate a self-driven p-InSe/n-InSe heterostructure photodetector for NIR applications. The p- and n-doped 2D InSe flakes form the vertical heterostructure stack showing a $\sim 3$ times improvement in responsivity and $\sim 3.5$ times lower response time as compared to p- or n-type InSe photodetectors, thus demonstrating a novel fast and sensitive InSe heterojunction-based NIR photodetector suitable for low power optical sensors or detector devices.

Enhancing light–matter interaction by building heterostructures using 2D materials paves the way toward building high-performance and energy-efficient compact photodetectors [36]. The 2D material-based photodetectors are often limited by the high bias voltage required for photon-generated carrier collection due to high resistance and the Schottky barrier [37]. Here, we demonstrate a p-n junction-based photodetector using InSe for NIR detection or sensing applications as seen in Fig. 1(a). The 2D InSe flakes are mechanically exfoliated from bulk crystals grown by the vertical Bridgman method and transferred precisely on prefabricated Au/Ti metal contacts using the novel 2D material transfer system discussed in Ref. [38]. The optical microscope image of the device is shown in Fig. 1(b). The p-n InSe heterojunction is formed by the physical contact between Sn-doped InSe (n-type) and Zn-doped InSe (p-type) materials. The band diagram for the device structure can be seen in Fig. 1(c) where the built-in heterojunction potentially helps in collecting the photogenerated carriers in the absence of external electrical bias potential. The material quality was assured after transfer using Raman spectroscopy by monitoring the relative intensities of Raman active modes at $A_{1 g}^{1}$ , $A_{1 g}^{2}$ , and $E_{2 g}^{1}$ [18].

The photovoltaic properties of the p-InSe/n-InSe heterojunction-based, p-InSe-based, and n-InSe-based photodetector devices are studied at 980 nm under vertical illumination using a free-space optical setup. Figure 2(a) shows the current-voltage (I–V) characteristics for photocurrent measurement of the p-n-InSe junction, p-InSe, and n-InSe, where the illumination power is 111.6 μW. The p-n junction device exhibits an order of magnitude higher photocurrent as compared to the p- and n-InSe devices, thus indicating a higher photo-absorption at 980 nm for the vdW heterojunction characteristics. Figure 2(b) shows the power-dependent photocurrent response of the p-n junction device. It can be observed that the photocurrent saturates after reaching high optical incident power intensity. The photovoltaic effect is attributed to the built-in electric field in the heterojunction depletion region. Figure 2(c) shows the dark current mapped at different voltages for the device. This shows a very small change in the dark current with an increase in the bias voltage leading to low electrical energy loss. As a result of the built-in potential of the junction, the device can be operated under no external bias for collecting photogenerated carriers. Figure 2(d) shows the relationship between the photocurrent and the input optical power for p-n-InSe heterojunction, p-InSe, and n-InSe devices under zero-bias voltage at 980 nm light illumination. It can be seen that the built-in potential in the p-n junction device shows higher photocurrent generation as compared to the non-junction (p- and n-) type devices.

Further, the photodetector devices were tested for responsivity as a function of wavelength from 800 nm to 900 nm (supercontinuum source) and at 980 nm (diode laser) for NIR photodetection. Figure 3(a) shows the responsivity [ $R_{λ} = (I_{ph} - I_{dark}) / P_{in}$ ] of p-n heterojunction, p-, and n-InSe at the zero-bias voltage for the p-n device and at 2 V for p- and n-devices. The p-n junction device shows enhanced responsivity along the NIR wavelength as compared to n- and p-devices by 3.03 times. It can be found that the maximum responsivity of p-InSe/n-InSe heterojunction is about $0.5 {mA W}^{- 1}$ at 980 nm under zero bias. The photoluminescence (PL) spectra of the p-InSe, n-InSe, and p-n-InSe heterojunction from 800 nm to 1000 nm were recorded using a 532 nm laser for excitation. The exciton recombination peak at 980 nm (1.265 eV) in the PL spectrum shows the intensity of p-n-InSe heterojunction higher than that of p-InSe and n-InSe by $\sim 5.9$ times at 980 nm as seen in Fig. 3(b). It is known that the exciton PL is very sensitive to the presence of defects and surface contaminants. The enhanced PL intensity at the p-n junction, which is due to the increased photoexcitation volume, provides evidence of a clean interface between the p- and n-InSe flakes. The Raman spectroscopy mapping of the p-n junction stack also confirmed the clean heterojunction formed between the two flakes [Fig. 3(c)]. No additional peaks were observed due to contaminants or defects produced in the material during device fabrication [39].

The photoresponse of the p-n-InSe photodetector is expected to be higher than that of the p- and n-InSe due to the built-in potential inside the heterojunction. The on/off switching response for p-n-InSe (0 V), p- ( $- 2$ ), and n- (2 V) photodetector under 980 nm (111.6 μW) illumination was measured by modulating the optical source power supply. With an increase in the external electric field, the charge density in the InSe is high, leading to a decrease in mobility of photogenerated carriers because of the decrease in carrier drift velocity. The rise and fall time constants of the p-n junction device show a faster response as compared to the p- and n-only InSe photodetectors (Fig. 4). Due to the absence of junction for p- and n-InSe devices, the photocarrier collection is possible only after applying an electrical bias. Therefore, both the devices were tested at 2 V bias for the response time measurement; unlike the built-in electric field of the p-n junction, light-excited electron–hole pairs can be more effectively separated due to the photovoltaic effect versus relying on photoconductivity for the n- and p-only junction devices. The response time of the p-n-InSe device shows a rise time/fall time ( $τ_{r} / τ_{f}$ ) response of 8.3 ms/9.6 ms [Fig. 4(a)], which is reduced as compared to n- and p-InSe devices by 18 ms/40 ms [Fig. 4(b)] and 22 ms/29 ms [Fig. 4(c)], respectively. This is an interesting result since it shows the engineerability for designing fast response photodetectors by stacking the 2D materials and enhancing the optoelectronic properties of the material for various applications.

The response time of this device is limited by trap states generated in-between material to the substrate and 2D-2D interlayer boundaries. Such trap states help achieve signal detection effectively due to the accumulation of charges but adversely affect the speed response of the device. The trap states generated between the substrate and 2D material majorly contribute to the slow response. This can be improved further by building the device on a 2D substrate material like hBN for preserving the mobility of the material and achieving a faster response time. The low dark current of the device provides a stable noise floor owing to the presence of built-in potential. As seen in Fig. 2(c), the minute change in dark current with an increase in voltage shows that the device can be operated over a varying range of bias voltage without increasing the noise floor of the device. Such device engineering would provide better solutions to lower power sensing and wearable electronics applications.

A self-driven photodetector was realized for NIR applications at 980 nm, and an extremely low dark current in the range of a few picoamps was achieved. The photoresponse of the 2D InSe was enhanced by building a p-n vdW heterojunction using p- and n-doped InSe showing an increase of $\sim 3$ times in responsivity as compared to the control n-InSe and p-InSe photodetector devices. The p-n device also exhibits a faster response which is $\sim 3.5$ times lower than that of a p- or n-type InSe photodetector, thus demonstrating a novel fast and sensitive InSe heterojunction-based NIR photodetector suitable for low power optical sensors or detector devices. Such performance of this device shows a high potential for realizing an NIR photodetector for sensing and optical applications. Furthermore, by engineering the interlayer stacking to match closely with the lattice structure and improving the metal contact Schottky barrier, it is expected that the device performance in the NIR region will be further improved. Further investigations are required to determine the chemical and mechanical material stability of the InSe heterojunction structure under various environmental effects for applications like remote sensing and biological sensing.

Acknowledgment. A. V. D. and S. K. acknowledge support through the Material Genome Initiative funding allocated to the National Institute of Standards and Technology. H. Z. acknowledges support from the U.S. Department of Commerce, National Institute of Standards and Technology under the financial assistance. Disclaimer: Certain commercial equipment, instruments, or materials are identified in this paper to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.