Highly Sensitive Photodetectors Based on Monolayer MoS2 Field-Effect Transistors.

Molybdenum disulfide (MoS2) is a promising candidate for the development of high-performance photodetectors, due to its excellent electric and optoelectronic properties. However, most of the reported MoS2 phototransistors have adopted a back-gate field-effect transistor (FET) structure, requiring applied gate bias voltages as high as 70 V, which made it impossible to modulate each detecting device in the fabricated array. In this paper, buried-gate FETs based on CVD-grown monolayer MoS2 were fabricated and their electric and photoelectric properties were also systematically investigated. A photoresponsivity of around 6.86 A/W was obtained at 395 nm, under the conditions of zero gate bias voltage and a light power intensity of 2.57 mW/cm2. By application of a buried-gate voltage of 8 V, the photoresponsivity increased by nearly 10 times. Furthermore, the response speed of the buried-gate MoS2 FET phototransistors is measured to be around 350 ms. These results pave the way for MoS2 photodetectors in practical applications.

Introduction

Since the discovery of graphene, two-dimensional (2D) materials have been considered to be promising candidates for applications in next-generation optoelectronic devices due to their high carrier mobility,[1] broad coverage of their adjustable band gaps,[2] strong interaction between light and matter,[3] and high flexibility.[4,5] Among these, photodetectors based on graphene and MoS2 have been intensely studied. Due to their extremely high carrier mobility, graphene photodetectors always show a high photoresponse speed.[6−8] However, graphene’s zero-band-gap structure limits the photocurrent generation within the interface region between graphene and the metal electrode.[9,10] Thus, the photocurrents and photoresponsivities of graphene photodetectors are both very small. In contrast, monolayer MoS2 not only has a direct band gap (1.8 eV)[11,12] but also exhibits a higher light absorption rate (11%). Furthermore, MoS2 is less vulnerable to environmental changes, whose robustness to harsh vacuum and solution processes enables the tuning of its optoelectronic properties.[13] Therefore, MoS2 has recently attracted more and more research attention in high-performance photodetectors.[7,14]

Since the first monolayer MoS2 photodetector was demonstrated with a photoresponsivity of 7.5 mA/W and a response speed of 50 ms,[15] various MoS2 photodetectors have been studied and reported. Kis et al. demonstrated a photodetector based on an exfoliated monolayer MoS2 membrane, which realized an extremely high photoresponsivity of 880 A/W (at 561 nm) and a response speed (rise time) of 4 s by improvement of the mobility, contact quality, and positioning technique.[16] Through insertion of a very thin TiO2 layer into the interface between the exfoliated MoS2 membrane and metal electrodes to optimize the interfacial contact, Roqan et al. demonstrated a MoS2 photodetector with an increase in photoresponsivity from 3 to 9 A/W (at 450 nm) and a reduction of the response speed from 5 s to less than 1 s.[13] Hu et al. improved the photoresponsivity of a monolayer MoS2 photodetector from 16.9 to 377 A/W (at 360 nm) and reduced the response speed from 31 to 7.5 s by combining CVD-grown MoS2 with carbon quantum dots.[17] Recently, Zhang et al. increased the photoresponsivity of few-layer MoS2 photodetectors from 13.46 to 26 A/W (at 660 nm) and obtained a photoresponse speed of about several seconds, by integrating gold nanoellipse arrays on a few-layer MoS2 nanosheet.[18] It is worth noting that all of the above MoS2 photodetectors employed a back-gate field-effect transistor (FET) structure, which required very high gate voltages of up to 70 V.[16,19−21] The high gate voltages limit their practical applications in low-voltage and energy-saving areas. The back-gate FET structure also makes it impossible to independently modulate each device in an array, which is not conducive to applications in sensing and imaging arrays.

In this paper, monolayer MoS2 FETs with a buried-gate structure were fabricated and demonstrated and their electric and photoelectric properties were systematically investigated. Under the condition of zero gate bias voltage, a photoresponsivity as high as 6.86 A/W was obtained. By application of a voltage as low as 8 V to the buried-gate electrode, the photoresponsivity can be enhanced 10 times. Furthermore, the photoresponse speed was about 350 ms. These results indicate that the buried-gate MoS2 FET photodetectors have superb combination properties.

Experimental Section

The monolayer MoS2 membranes were grown by the chemical vapor deposition (CVD) method, purchased from ACS Material, LLC. First, a 200 nm thick silicon nitride (SiNx) passivation layer was deposited on a highly doped p-type silicon substrate. Then, a chromium (Cr)/gold (Au) electrode was sputtered on the SiNx layer, acting as the buried-gate electrode, where the thicknesses of Cr and Au layers were 10 and 30 nm, respectively. After that, a silicon dioxide (SiO2) dielectric layer with a thickness of 30 nm was deposited on top of the buried-gate electrode. Next, the monolayer MoS2 membrane was transferred onto the SiO2 dielectric layer and patterned. Subsequently, Cr/Au (10 nm/50 nm) source and drain electrodes were evaporated on the MoS2 layer, and finally, MoS2 FETs with a buried-gate structure were obtained. Detailed information about the fabrication processes can be found in our previous work.[14]

The fabricated buried-gate MoS2 FETs were measured using an environmental scanning electron microscope (FEI Quanta 200 ESEM FEG) operating at a beam voltage of 15 kV in high-vacuum mode. Raman spectroscopy (LabRam HR-800, Horiba Jobin Yvon) was carried out in the conductive channel of the FET to confirm the existence of MoS2 and determine its layer number. The electrical properties of the devices were characterized by using a probe station (Summit 12000, Cascade Microtechnology) and a semiconductor parameter analyzer (B1500A, Keysight). By integration of a light-emitting diode (LED) light-curing system (CEL-LEDS35) into the electric testing system, the photoelectrical properties of the buried-gate MoS2 FETs were systematically investigated. Unless stated otherwise, all of the electric and photoelectric experiments were performed at room temperature and under ambient conditions.

Results and Discussion

Figure A shows the schematic structure of the buried-gate MoS2 FET. An SEM image of the three parallel devices fabricated on the same SiN passivation layer is shown in Figure C. The dimensions of the MoS2 conductive channel are 30 μm × 400 μm, while those of the contact pad are 100 μm × 100 μm. An enlarged SEM image of one of the buried-gate MoS2 FETs shown in Figure D demonstrates that the device was well constructed. It can be obviously seen that the gate electrode is located in the middle of the MoS2 conductive channel. The widths of the source, drain, and gate electrodes are all 10 μm. In comparison with a top-gate FET structure, this structure enables the MoS2 conductive channel to absorb more incident light. The existence of the MoS2 membrane and its layer number were verified and determined by Raman measurements, as shown in Figure B. Two peaks are seen in the figure, representing E2g1 and A1g peaks. The E2g1 peak is related to the in-plane vibration of two S atoms and one Mo atom inside the MoS2,[22,23] which is observed at around 384.8 cm–1. However, the A1g peak results from the out-of-plane vibration of only S atoms in inverse directions,[24,25] located at about 403.8 cm–1. The energy difference between the E2g1 and A1g Raman modes is 19 cm–1. These results agreed well with observations of the previously reported monolayer MoS2 devices[17,24−26] and verified the existence of a monolayer MoS2 membrane.

Figure 1

Structure characterization and Raman spectroscopy of the buried-gate MoS2FETs. (A) Schematic of the buried-gate MoS2 FET. (B) Raman spectrum for the conductive channel of the MoS2 FET. (C) SEM image of three parallel fabricated buried-gate MoS2 FETs. (D) Enlarged SEM image of the MoS2 conductive channel.

Figure A shows the transfer characteristics of the buried-gate MoS2 FET, where the source-drain current (Ids) turns on at a gate voltage (Vgs) of 2.5 V. Hence, a typical n-type enhancement mode behavior is clearly seen. Ids increases significantly with an increase in Vgs above the threshold voltage, exhibiting an on/off ratio of over 105. Figure B shows the output characteristics of the buried-gate MoS2 FET, with Vgs varying from −2 to 6 V. When Vgs is higher than 3 V, the current increases significantly as Vds increases. It is noted that Ids increases nonlinearly due to the unordered defective interface between the MoS2 membrane and the metal electrodes. The linearity could be improved by inserting a very thin TiO2 interlayer into the interface.[13] It can be seen that, under a certain Vds, Ids increases with an increase in Vgs, indicating a good gate-control ability.

Figure 2

Electrical properties of the buried-gate MoS2FET. (A) Transfer characteristics of the device at a source-drain voltage (Vds) of 1 V. (B) Output characteristics of the device.

The photoelectrical characterization of the buried-gate MoS2 FET was carried out by using the experimental setup shown in Figure A. A LED with a wavelength of 395 nm illuminated the FET. A semiconductor parameter analyzer was used to supply the source-drain and gate bias voltages and measure the Ids values of the device. Figure B,C shows the temporal photoresponses of the FET under different incident light power densities and source-drain voltages (Vds), where the full power density is 219.1 mW/cm2. It can be seen from both graphs that, when the light is switched on, Ids increases immediately and significantly and gradually reaches a saturation state. When the light is switched off, Ids decreases promptly and markedly and then gradually returns to its original level. These behaviors were also observed in other reported back-gate MoS2 FET photodetectors and could be ascribed to a photoconductive effect (or a photovoltaic effect).[24,27−29] The photocurrent (Iph) is defined as Iph = Ion – Ioff, where Ion and Ioff are the values of Ids when the light is on and off, respectively. Furthermore, it can be seen that Iph increases when the incident light power intensity increases from 2.57 to 219.06 mW/cm2. Under the conditions of maximum incident light power intensity, zero gate bias voltage, and 1 V source-drain voltage, the value of Iph reached 10.56 μA (as shown in Figure B), which is 3500 times higher than that of a reported back-gate monolayer MoS2 FET (3 nA at 580 nm, Vds = 1 V, Vgs = −7 V, 0.1 mW/cm2).[26] This value is also significantly higher than that in the previous work of our group (0.87 μA at 395 nm, Vds = 1 V, Vgs = 0 V, 219.06 mW/cm2).[31] This phenomenon may be due to the fact that the electrode could not only collect the photogenerated carriers between the channels but also collect the photogenerated carriers on the other side. Excess photoinduced carriers cause an increase in the free carrier concentration, leading to further reduced resistance of the MoS2.[7] When Vds is increased to 8 V, Iph is increased to 101.86 μA (as shown in Figure C), which is 10 times higher than that at Vds = 1 V. It is also observed that Ion fluctuates significantly as the light power intensity increases, especially when Vds is high.

Figure 3

Temporal photoresponse of the buried-gate MoS2 FET photodetector under the illumination of different light power densities and source-drain voltages. (A) Schematic of the experimental setup with LED illuminaiton. (B, C) Temporal photoresponses of the FET under various incident light power intensity, at source-drain voltages of 1 and 8 V, respectively. The LED light was switched on and off periodically, and the gate voltage was set to zero.

In order to comprehensively characterize the photodetection performance of the buried-gate MoS2 FET photodetectors, the photoresponsivity (R) and detectivity (D*) are evaluated in Figure . The photoresponsivity is defined as R = Iph/(PS), where P is the light power intensity and S is the area of the MoS2 conductive channel. Hence, R decreases as the power intensity increases, as indicated in Figure A, due to the photogating effect, and can be ascribed to the saturated absorption under high light power.[30] Moreover, it is clearly seen that at Vds = 8 V the values of R are significantly higher than those at Vds = 1 V. A photoresponsivity value as high as 6.86 A/W is obtained, under the conditions of zero gate bias voltage, Vds = 8 V, and P = 2.57 mW/cm2, which is higher than those of most of the recently reported MoS2 photodetectors without integration or combination with additional light harvesters.[13,15,34] The detectivity D* is expressed as D* = RS1/2/(2eIoff)1/2, where e is the electron charge with a value of 1.6 × 10–19 C. From Figure A, the value of D* is calculated to be 0.72 × 1012 jones, which is also superior to those of most of the MoS2 photodetectors given in Table .

Figure 4

Photoresponsivity and speed of the buried-gate MoS2 FET photodetector. (A) The light power intensity dependent photoresponsivity of the device under different source-drain voltages. (B) Rise (tr) and decay times (tf) of the photocurrent at Vds = 1 V and Vgs = 0 V.

Table 1

Photoelectrical Characteristics of Typical Photodetectors Based on MoS2

description	wavelength (nm)	responsivity	detectivity (jones)	rise time (s)	ref
single-layer MoS2 phototransistor	550	7.5 mA/W (80 μW)		0.05	(15)
monolayer MoS2 photodetector	561	880 A/W (150 pW) @ 8 V		4	(16)
pristine MoS2 photodetector	360	16.9 A/W @ 5 V	2.3 × 1012	31	(17)
dye-sensitized MoS2 photodetector	520	1.17 A/W (1 μW) @ 5 V	1.5 × 107	5.1 × 10–6	(33)
large-scale two-dimensional MoS2 photodetector	850	1.8 A/W @ 5 V	∼5 × 108	0.3	(34)
flowerlike MoS2 flexible broad-band photodetector	405	0.963 A/W @ 1 V	2.9 × 1010	9.33	(35)
MoS2/PGS photodetector	460	0.25 A/W (1.373 μW) @ 20 V	5.6 × 108	2.19	(36)
photodetector based on CVD-grown 2D MoS2	405	0.2907 A/W	1014.84	0.0778	(37)
this work	395	6.86 A/W @ 8 V	0.72 × 1012	0.35

Photoresponse speed is another important feature for a photodetector. Figure B shows a complete on/off cycle in which the photocurrent exhibits rise/decay and reaches a steady saturation between. The rise time (tr) and decay time (td) follow the definitions used in the MoS2 photodetectors reported by Roqan et al.,[13] where they are defined as the times required to reach 70% and to attenuate to 50% of the peak current, respectively. The tr and td values of the buried-gate MoS2 photodetectors are 350 and 370 ms, respectively, which are relatively faster than those of recently reported MoS2 photodetectors with photoresponsivities of more than 1 A/W.[13−15] What is more, a 2D alloying strategy could be used to improve the photoresponse of the MoS2 device.[32] A comparison of the performance of the buried-gate MoS2 photodetectors with those of other reported MoS2 photodetectors is shown in Table . It can be seen that the comprehensive performance of our devices is rather good.

Systematic measurements demonstrated that the normalized photoresponsivity of the buried-gate MoS2 photodetector increases approximately linearly with an increase in the source-drain voltage (Vds), as shown in Figure A, due to an increase in the carrier drift velocity. In contrast, the carrier transit time Tt (defined as Tt = l2/μVds, where l is the device length, μ is the carrier mobility, and Vds is the bias voltage) decreases as Vds increases.[15,38] The photoresponsvity can also be enhanced by applying a proper gate bias voltage (Vgs), as shown in Figure B. Due to the buried-gate structure, a small Vgs value as low as 8 V can enhance the photoresponsivity by nearly 10 times at Vds = 1 V, in comparison to that at Vgs = 0 V. On consideration that at Vgs = 0 V the photoresponsvity of the buried gate MoS2 photodetector could reach 6.86 A/W (Figure A), the maximum photoresponsivity of the device could be very high. Most of the reported MoS2 photodetectors adopt a back-gate structure, which always require much higher gate voltages of more than 50 V. For example, the ultrasensitive monolayer MoS2 photodetectors demonstrated by Kis et al. required a back-gate voltage of ∼70 V.[16] Therefore, the buried-gate MoS2 photodetectors proposed in this paper have a promising potential for use in low-voltage and energy-saving photoelectric fields.

Figure 5

Normalized responsivities of the buried-gate MoS2 FET photodetector as functions of source-drain and gate voltages. (A) Source-drain-voltage-dependent normalized photoresponsivity, where R0 is the value of R at Vgs = 0 V. (B) Gate-voltage-dependent photoresponsivity at Vds = 1 V.

Conclusion

Sensitive buried-gate FET photodetectors based on CVD-grown monolayer MoS2 were fabricated and demonstrated. A high photoresponsivity of 6.86 A/W was obtained under zero gate bias voltage, on illumination by a LED light with a wavelength of 395 nm and a power intensity of 2.57 mW/cm2. The photoresponsivity could be enhanced nearly 10 times by applying an 8 V voltage to the buried-gate electrode. Furthermore, the response speed of the buried-gate MoS2 FET phototransistors is about 350 ms. This study offers a simple way to obtain high-performance and low-power MoS2-based photodetector arrays.