Strong enhancement of photoresponsivity with shrinking the electrodes spacing in few layer GaSe photodetectors

A critical challenge for the integration of optoelectronics is that photodetectors have relatively poor sensitivities at the nanometer scale. Generally, a large electrodes spacing in photodetectors is required to absorb sufficient light to maintain high photoresponsivity and reduce the dark current. However, this will limit the optoelectronic integration density. Through spatially resolved photocurrent investigation, we find that the photocurrent in metal-semiconductor-metal (MSM) photodetectors based on layered GaSe is mainly generated from the region close to the metal-GaSe interface with higher electrical potential. The photoresponsivity monotonically increases with shrinking the spacing distance before the direct tunneling happens, which was significantly enhanced up to 5,000 AW−1 for the bottom Ti/Au contacted device. It is more than 1,700-fold improvement over the previously reported results. The response time of the Ti/Au contacted devices is about 10–20 ms and reduced down to 270 μs for the devices with single layer graphene as metallic electrodes. A theoretical model has been developed to well explain the photoresponsivity for these two types of device configurations. Our findings realize reducing the size and improving the performance of 2D semiconductor based MSM photodetectors simultaneously, which could pave the way for future high density integration of optoelectronics with high performances.


DEVICE MOBILITY MEASUREMENTS
The top contacted device was characterized first in the dark state by applying a constant drain-source voltage DS V = 10 V and sweeping the back-gate voltage BG V [ Figure S1]. The typical p type Field Effect transistor behavior was observed. The effective field-effect mobility of the device was estimated from the back gate sweep using the equation , where l = 1 μm is the channel length, W = 5 μm is the channel width, and i C = 1.3×10 -4 Fm -2 is the back gate capacitance ( i 0 r Cd   , r  = 3.9, d = 300 nm). For the device shown in Figure 4a in the manuscript, the field-effect mobility μ = 5×10 -3 cm 2 V -1 s -1 was obtained.
2 / 8 Figure S1. Gating characteristics of the nanosheet GaSe transistor presented in the main manuscript. Room-temperature transfer characteristic of the nanosheet GaSe phototransistor presented in the main manuscript. IDS -VBG sweep is performed at VDS =10 V in the dark state.

AFM RESULTS OF GASE NANOSHEET
The schematic crystal structure was shown in Figure S2c. The spacing distance between the neighbor layers is about 0.90 nm, which was confirmed by the AFM measurements based on the mechanically exfoliated monolayer GaSe [ Figure S2].
However, the monolayer GaSe devices show very large resistance which is beyond the measurement accuracy of our instruments. The bad electronic property may be because of easy oxidation of GaSe during the fabrication processing.
3 / 8 Figure S2. The thickness of the monolayer GaSe measured as 0.9 nm. a, AFM imaging of a few-layer GaSe flake on a silicon substrate with a 300 nm thick oxide layer; b, Plot along the white line in a determines the thinnest part; c, Three dimensional structure of GaSe, the thickness of the single layer is 0.9 nm.

PHOTOCURRENT AND PHOTORESPONSIVITY RELATIONSHIP WITH PHOTOINTENSITY AT DIFFERENT VOLTAGE.
With the bias voltage above 2 V for the bottom contacted photodetector with l =1 m, the photocurrent first sharply increases with increasing the light intensity and then the photocurrent shows very slow increase when the light intensity is large enough ( Figure S3a). Since the photoresponsivity is proportional to the ratio of photocurrent and the light intensity, thus it shows a reverse result ( Figure S3b

PHOTORESPONSIVITY DEPENDENCE OF DISTANCE AND CONTACTS
To demonstrate that concept related to the transport of the photo generated carriers in a metal-semiconductor-metal (MSM) photodetector, a numerical model was developed. The photogenerated electrons diffusive to the interface between the GaSe and the metal contact with higher electrical potential (using Xl  at forward bias for example), and then the electrons have the same possibility to pass through the interface and enter into the metal contact. For clarity and simplicity, the depletion region is equivalent to an infinitely thin Schottky barrier at 0 X  or Xl  depending on the bias direction (take forward namely Xl  at positive voltage side as example in this paper).
We firstly discuss the top contacted device. To simplify, considering the device as a one dimensional chain, the number of the photogenerated electrons per second at arbitrary position Xx  between the two electrodes is () Nx under global illumination ( Figure S4), where the distribution of () Nx will according to the Gaussian function. Then the number of the photogenerated electrons per second in the whole device is () N N x l  .

/ 8
Under forward bias, the photogenerated electrons will drift to Xl  side.
During this process, the electrons will be partially recombined. The remaining photogenerated electrons will be exponentially decrease with the travel distance, which can be written down as where D L   is the diffusion length. Under forward electrical bias, the number of the electrons reached to X = l (the high electrical potential) under global illumination can be written down as: where intensity L is the light intensity and ) have to diffusive to the right side and then enter into the metal contact below, which can be described similarly as the above formula. However, the photoexcited electrons in right contact side will have vertical rather than planar transport and then enter into the where l L is the width of left contact, ' c is the probability of vertical transport electrons enter into the metal contact, r L is the width of the right contact, d is the thickness of the GaSe layer and D L is the diffusion length. In the above equation, the first term in right hand side describes the photocurrent contribution from the left contact region and the device region between the two electrodes and the second term describes the photocurrent contribution from the right contact region. Only the device area between the two electrodes was counted for calculation of the photoresponsivity in the top contacted device. Thus the photoresponsivity for the bottom contacted device from Equation (3)