Light‐Driven WSe2‐ZnO Junction Field‐Effect Transistors for High‐Performance Photodetection

Abstract Assembling nanomaterials into hybrid structures provides a promising and flexible route to reach ultrahigh responsivity by introducing a trap‐assisted gain (G) mechanism. However, the high‐gain photodetectors benefitting from long carrier lifetime often possess slow response time (t) due to the inherent G–t tradeoff. Here, a light‐driven junction field‐effect transistor (LJFET), consisting of an n‐type ZnO belt as the channel material and a p‐type WSe2 nanosheet as a photoactive gate material, to break the G–t tradeoff through decoupling the gain from carrier lifetime is reported. The photoactive gate material WSe2 under illumination enables a conductive path for externally applied voltage, which modulates the depletion region within the ZnO channel efficiently. The gain and response time are separately determined by the field effect modulation and the switching speed of LJFET. As a result, a high responsivity of 4.83 × 103 A W−1 with a gain of ≈104 and a rapid response time of ≈10 µs are obtained simultaneously. The LJFET architecture offers a new approach to realize high‐gain and fast‐response photodetectors without the G–t tradeoff.

The I ds -V bg measurement of WSe 2 back-gated transistor is conducted by applying the sourcedrain voltage to the two side top-gate electrodes underneath the WSe 2 nanosheet (see Figure   1g). (b) A schematic of the WSe 2 -ZnO heterostructure device model, where R cw (R cz ) is the contact resistance between WSe 2 (ZnO) and the electrode; R w (R z ) represents the WSe 2 (ZnO) series resistance outside the overlapped region; R ow (R oz ) is the WSe 2 (ZnO) resistance in the overlapped region; and R i represents the junction resistance. (c-h) Back-gate tunable I-V characteristics of the WSe 2 -ZnO heterostructure device. The power intensity is 367.5 mW cm -2 (637 nm).
In order to study the rectifying characteristics of the WSe 2 -ZnO heterostructure with V bg modulation, we perform the I ds -V bg measurements of WSe 2 and ZnO back-gated transistors first. From Figure S4a, it can be seen that when V bg < -7.8 V, the carriers in ZnO (WSe 2 ) are depleted (enhanced); when -7.8 V < V bg < -5.8 V, there are some carriers in both ZnO and WSe 2 ; when V bg > -5.8 V, the carriers in ZnO (WSe 2 ) are enhanced (depleted). So, the applied V bg can cause one material to remain conductive and the other to remain non-conductive except for the V bg range between -7.8 V and -5.8 V. Figure S4b depicts a simplified device model for the WSe 2 -ZnO heterostructure. Figure S4c shows the I-V curves of the device at V bg = -20 V (the voltage was marked by the blue dashed line in Figure S4a). The currents are very low even under light illumination. The device does not show a good rectifying characteristic. This is because that the V bg of -20 V depleted the carriers in ZnO and enhanced the carriers in WSe 2 ( Figure S4d). ZnO functions as a large resistance for the heterostructure. The illumination can only change the resistance of WSe 2 but not that of ZnO, resulting in low currents under both dark and light conditions. Figure S4e shows the I-V curves of the device at V bg = -6.5 V. The device shows rectifying characteristics under dark and light conditions. This is because that both ZnO and WSe 2 remain conductive at V bg = -6.5 V (see Figure S4a and Figure S4f). The illumination can further decrease the resistance of WSe 2 . So, the device presents a rectifying state. Figure S4g shows the I-V curves of the device at V bg = 10 V. The device shows a rectifying characteristic only under light excitation. This is because the V bg of 10 V depleted the carriers in WSe 2 and enhanced the carriers in ZnO (see Figure S4a and Figure S4h). In the dark, WSe 2 functions as a large resistance for the heterostructure, resulting in a low dark current. Upon light illumination, the carriers are excited in WSe 2 (see Figure   S4h) and both ZnO and WSe 2 remain conductive. The rectifying state of the heterostructure is induced by light. This is our design basis of LJFET. Noise current measurement was conducted using a low-noise current preamplifier and a FFT spectrum analyzer. The device was kept in a metal box which provides a shielded and dark environment. The source-drain current is ~6.5 μA. It can be seen from Figure S12 that flicker noise (1/f noise) dominates at low frequencies and white noise dominates at high frequencies. According to the frequency dependence results, we calculate the specific detectivity (D * ) at f = 1 kHz. D * is defined as D * = (AΔf) 1/2 R /i n , where A is the effective area of the device, R is the responsivity, Δf is the electrical bandwidth, and i n is the noise current [3] .
Because our device possesses a broadband frequency response (Figure 4f Considering that the GaSe nanosheet does not conduct well even in the accumulation regime (see Figure S15b), the function of V bg is only to modulate the density of electrons in ZnO for the GaSe-ZnO LJFET. Figure S16a and 16b give the temporal response of device #1 measured at V bg of 10 V and 15 V, respectively. It can be seen that the variation of both the rise and fall time is within 7 μs (~4% change). Therefore, the effect of V bg on the response time (hundreds of microseconds) is negligible. Figure S16c and 16d give the temporal response of device #2 and device #3, respectively.
It should be noted that these two devices use the same ZnO channel but have different GaSe nanosheets (see the insets). To study the effect of quality of photosensitive material on response time, the same channel was used to exclude the possible effect of different ZnO channels. A large variation of ~100 μs (~30% change) is observed in both rise and fall time.
For the device fabrication, we used PDMS to take GaSe nanosheet #2 off and put GaSe nanosheet #3 onto the ZnO channel. Although we cannot acquire the GaSe nanosheets with the same morphology (area and thickness) using mechanical exfoliation, the results still can support the view that the quality of photosensitive material has a great impact on the response time.

Table S1
Response time of typical bare GaSe and WSe 2 photodetectors

Response time Reference
Mechanically exfoliated few-layer GaSe 20 ms [4] Mechanically exfoliated few-layer GaSe 270 μs [5] CVD-grown monolayer WSe 2 23 ms [6] Mechanically exfoliated 3L WSe 2 10 μs [7] CVD-grown 1L WSe 2 >1 ms [8] Additionally, the response time of all the GaSe-ZnO LJFETs (#1-#7) is basically at the level of hundreds of microseconds which is longer than that of WSe 2 -ZnO LJFET. Many researchers have studied the photoresponse of bare GaSe and WSe 2 detectors. It can be seen from Table S1 that different response times have been reported even for the same material.
The difference in photoelectric conversion efficiency and defect states could be the reason for these varying results. Therefore, in the LJFET, the type of photosensitive material also has a great impact on the response time.