Recently, as device processing enters the nanometer scale, device development based on nano-materials has received increasing attention ^[1-4]. Among various technical approaches, the "bottom-up" method emphasizes the rational construction of nano-functional devices based on atoms and molecules, which is regarded as the core of nanotechnology. Nanowire-based devices have shown great application potential in Field Effect Transistors (FETs), light-emitting diodes, sensors, nano-lasers and nano-generators, among others ^{[3, 5-10]}. In particular, III-V semiconductor nanowires are attracting great attention due to their excellent electrical and optical properties ^[11-12]. Among the many potential nanowire materials, In₂O₃ nanowires possess a wide bandgap of 2.8 eV and high field-effect mobility ^[13]. In₂O₃ nanowires are easy to prepare and can be produced in various ways including physical vapor deposition ^[14], hydrothermal methods ^[15], chemical vapor deposition ^[16], pulsed laser deposition ^[17] and molecular beam epitaxy ^[1]. Many previous studies mainly focused on the preparation and electrical properties of nanowires due to their beneficial properties such as chemical stability, excellent crystallinity, easily doping, low-cost synthesis, quantum confinement effect, and high quantum efficiency ^[1]. However, few studies have focused on their optoelectronic properties since the first report on the ultraviolet (UV) photodetection capabilities of In₂O₃ nanowires ^[18]. Despite recent advances, the performance of photodetectors still cannot meet practical demand.

This work presents a strategy to fabricate and suspend In₂O₃ nanowire devices. Through its unique design and relatively simple manufacturing process, an In₂O₃ nanowire channel can be separated from its substrate. The source and drain patterns are also defined by only one Electron Beam Lithography (EBL) step. Utilizing this device’s structure, we conducted an intrinsic study on the electrical characteristics of In₂O₃ nanowire FETs, exhibiting an on/off current ratio larger than 10³, a Subthreshold Swing (SS) of 241.5 mV/dec and a low dark current of ~10⁻¹³ A. The calculated carrier mobility is approximately 54.6 cm²V⁻¹s⁻¹. The In₂O₃ nanowire photodetector exhibits a high photoconductive gain of 4.3 × 10³, a responsivity of 1.6 × 10⁵ A•W⁻¹, and a specific detectivity of D^* up to 9.3 × 10¹⁰ Jones at a low power density of 0.06 mW·cm⁻²(λ = 450 nm). The response time (rise time τ_r), which is defined as the time for the photocurrent to increase from 10% to 90% I_peak, is about 150 ms. It is faster than In₂O₃ nanowire-based photodetectors that have been reported previously ^[19-20].

The In₂O₃ nanowires used in this study were synthesized by Chemical Vapor Deposition (CVD) through a Vapor-Liquid-Solid (VLS) growth mechanism. A mixture of high-purity In₂O₃ powder and graphite powder with a weight ratio of 5:1 was put into a quartz boat. A silicon substrate with a pre-deposition of 1-nm and an Au catalyst was placed downstream in the quartz tube to collect the products. A furnace was set to 1100 °C in 22 min and maintained for 1 hour under a constant flow of gas (argon/oxygen = 100:1, flow rate = 200 sccm). Finally, the system was left to naturally cool to room temperature. The TEM image in Fig. 1(a) shows that a large quantity of wire-like In₂O₃ microcrystals with good uniformity was fabricated by the CVD method. The diameter of the nanowire is approximately 50 nm. Fig. 1(b) and (c) show a dark-field microscopy image of the In₂O₃ nanowire FET and three-dimensional schematic view of the suspended single In₂O₃ nanowire photodetector. In₂O₃ nanowires used in this research were characterized by field-emission scanning electron microscopy and high-resolution transmission electron microscopy (HR-TEM, HT7700). The SEM result is shown in Fig. 1(d), confirming that the In₂O₃ nanowire is free of the substrate.

Figure 1.  Design and characterization of the single In₂O₃ nanowire device. (a) TEM of a single In₂O₃ nanowire. (b) Dark-field microscopy image of an In₂O₃ nanowire FET. (c) Three-dimensional schematic view of a suspended single In₂O₃ nanowire photodetector. (d) SEM of the suspended single In₂O₃ nanowire photodetector

To fabricate suspended nanowire devices, the key part is to find a feasible way to fabricate a structure that is both suspended and stable. We found inspiration from previous work, which wrote about electron beam lithography on both poly-methyl-methacrylate (PMMA) and hydrogen silsesquioxane (HSQ) ^[21]. The process flow for preparing the suspended In₂O₃ nanowire FET is shown in Fig.2 (Color online). Specific steps are as follows: firstly, a layer of 280-nm-thick PMMA is spin-coated onto the Si/SiO₂ substrate. Secondly, In₂O₃ nanowires are drop-casted onto the temporary layer of the spin-coated photoresist. The N-doped In₂O₃ nanowires produced by CVD have a typical length of 10 μm and a diameter of less than 100 nm. Thirdly another photoresist layer is spin-coated. Standard e-beam lithography is performed to define the electrodes on a marked In₂O₃ nanowire. Then, 15 nm chromium (Cr) and 300 nm gold (Au) are deposited using thermal evaporation. Finally, the fabrication of the device is finished after the lift-off process. To improve ohmic contacts, devices were annealed at 200 °C in Ar atmosphere for 2 hours.

Figure 2.  Fabrication process of suspended nanowire devices

The electrical properties of fabricated suspended In₂O₃ transistors are investigated with a Lake Shore Probe Station together with an Agilent B1500 (Agilent Technologies, Santa Clara, CA, USA) semiconductor parameter analyzer. The I–V curves of the nanowire-based photodetector are shown in Fig. 3(a) (Color online) and indicate that different atmospheres can affect electrical conductivity. The annealing process can effectively improve the contact of metal electrodes. Under dark conditions, oxygen molecules tend to chemisorb onto the surface of In₂O₃ nanowire by trapping free electrons from the conduction band and then becoming negative ions O₂^–, which reduces the system’s free electron density. The larger current in a vacuum is a result of the lack of oxygen in such a test environment. Fig. 3(b) (Color online) shows the typical output characteristics of the device. The liner behavior suggests good ohmic contact between the nanowires and electrodes. As shown in Figs. 3(c) and 3(d) (Color online), low dark current (approximately 10⁻¹³ A) in both transfer curves can be attributed to the design of the suspended structure, which can effectively avoid the limitation of gate leakage current. Transfer curves of the same device in different atmospheres show a similar clockwise hysteresis performance. It can be confirmed that the curve does not significantly shift with an increase in drain bias for a static V_gs sweep range, which suggests ideal transistor characteristics for these suspended devices in a vacuum ^[22]. Compared with the sharp shift of transfer curves and higher turn-off voltage in air, this suspended design suggests applications in vacuum transistors are possible. The electron mobility μ_FE of a single In₂O₃ nanowire device can be calculated by using the expression μ_FE = g_mL²/(C_gV_ds), where the channel length L = 2.0 μm, and g_m = dI_ds/dV_gs is the transconductance of the nanowire device. C_g is the back-gate capacitance including the two parts that can be deduced based on the cylinder on-plane model: C_g1 = 2πε₁ε_rL/[ln(4h₁/d)] ^[22], where ε₁ε_r is the dielectric constant of the air, h₁(200 nm) is the height of suspension, and d(50 nm) is the In₂O₃ nanowire’s diameter. C_g2 =2πε₂ε_rL/[ln(4h/d)], where ε₂ is the permittivity of the SiO₂, and h₂ (280 nm) is the thickness of the SiO₂ substrate. V_gs is the voltage between drain and source. The calculated carrier mobilities of the device in air and a vacuum are 12.4 cm² V⁻¹ s⁻¹and 54.6 cm² V⁻¹ s⁻¹, respectively. The system may be influenced by contact between nanowires and metal electrodes. The SS can be expressed by the equation SS = ∂V_g/∂(lgI_d), where V_g is the applied gate voltage and I_d is the drain current. The SS value calculated in the different atmospheres is 324.7 mV/dec and 241.5 mV/dec, respectively. Previous work found that performance degradation is mainly attributed to the charge accumulation in the In₂O₃/SiO₂ interface, and the gate coupling in the non-suspended structure is not as desirable as that in suspended couplings ^[13]. It is necessary to design gate dielectric and electrode systems to make a breakthrough in SS value limitation (60 mV/dec). The device shows a high on/off ratio that is larger than 10³, which implies that this suspended device has immense potential in high-performance FETs.

Figure 3.  Electric performance of the single In₂O₃ nanowire devices under dark conditions. (a) I-V curves of the nanowire-based photodetector in different atmospheres. (b) Output characteristics of the nanowire-based photodetector under different gate voltages. Orange arrow represent that V_g is from 20 V to −50 V with −10 V steps. Transfer curve switch at different source-drain biases ranging from 0.1 V to 1 V (c) in the air and (d) in vacuum state. Black arrows represent the scanning direction of gate voltage.

The spectral responsivity is measured using a series of laser sources (250, 375, 450, 520, 637 nm; Thorlabs, Inc) combined with Agilent B2902. The I_ds-V_ds characteristics of the photodetector were measured in the dark and different wavelengths of light, as shown in Fig. 4(a) (Color online). The dark current of the device is 0.12 μA. The light wavelengths of 637 nm and 520 nm (~1.10 mW·cm⁻²) have little influence on the device’s photocurrent compared to the dark current. Contrarily, as shown in Fig. 4(b) (Color online), the photocurrent increases drastically when the wavelength of incident light reaches to 450 nm, which has comparable energy to the bandgap of In₂O₃ (2.8 eV). The responsivity (R) can be defined as I_ph/(PA), where I_ph is the photocurrent, P is the incident power density, and A is the effective irradiated area on the nanowire. Fig. 4(b) also shows the dramatic responsivity difference (about three orders of magnitude) on either side of the wavelength corresponding to In₂O₃ bandgap. The value of R is as large as 1.6 × 10⁵ A·W⁻¹ when incident light reaches the UV band with an applied voltage of 0.1 V. The results demonstrate the excellent photoresponsivity and wavelength selectivity of the In₂O₃ nanowire-based photodetector. The photoconductive gain (G), responsivity (R), and specific detectivity (D^*) are the key parameters for evaluating the sensitivity of the photodetectors ^[23].

Figure 4.  Photoresponse properties of the single In₂O₃ nanowire devices. (a) I_ds-V_ds characteristics of the photodetector in the dark and different wavelengths of light. (b) Photocurrent and responsivity of the photodetector under different wavelengths of light. (c) I_ds-V_ds characteristics of the photodetector in the dark and different intensities of light (450 nm). (d) Photoresponsivity and detectivity of the photodetector under the different intensities of light (450 nm).

The corresponding external quantum efficiency (EQE) is defined as the ratio between the number of collected electrons and the number of incident photons ^[19]. It can be expressed by the following equation, EQE =Rhc/(eλ). The EQE of the device is calculated to be as high as (9.9 × 10⁷)% with irradiation of UV light (~ 200 nm). The high EQE is related to the exposure of the nanowire surface via the chemisorption/photodesorption of oxygen. It has been mentioned above that by trapping free electrons from the conduction band to becoming negative O₂^– ions, oxygen molecules tend to chemisorb onto the surface of an In₂O₃ nanowire in dark conditions, thereby reducing the free electron density O₂(g) + e^–→ O₂^–(ad) ^[19]. Electron–hole pairs are created in the bulk when illuminated with short-wavelength light. Due to the existence of new valence subbands, the photogenerated holes move from the bulk to the surface and these accumulated holes will react with the adsorbed oxygen ions, then release a captured free electron, h⁺ + O₂⁻ (ad) → O₂ (g) ^[19]. Furthermore, separation of the structure between In₂O₃ nanowire and substrate caused ultrahigh surface-to-volume ratios and a free-standing interface, which played an important role in high R and EQE. Figure 4(c) (Color online) shows the I_ds-V_ds characteristics for different power intensities at a wavelength of 450 nm. The I_ds-V_ds curve shows a linear regime with increasing V_ds. It could be predicted that there will be photon-generated carrier saturation and electron-hole recombination under strong light illumination if the bias and light intensity is large enough ^[24-26].

The specific detectivity is an important figure-of-merit characterizing the capability of the smallest detectable signal for a photodetector. Considering that the shot noise from the dark current is the major factor limiting the detectivity, the specific detectivity can be given by D^* = RA′^1/2/(2eI_dark)^1/2, where R is the responsivity, A′ is the effective area of the detector (A′ = L × d, L is the channel length, d is the nanowire’s diameter), e is the unit charge, and I_darkis the dark current. Figure 4(d) (Color online) shows the calculated values of responsivity and detectivity at different power intensities. The incident light’s power is calibrated with a PM100D power meter. It shows that R and D^* increase dramatically with a decreasing light intensity, which can be caused by the trap states of the In₂O₃ nanowires. The R and D^* of the photodetector are up to 1.6 × 10⁵ A·W⁻¹ and 9.2 × 10¹⁰ Jones, respectively, under a low light intensity of 0.06 mW·cm⁻². The highest responsivity is two orders of magnitude larger than that of commercial photodetectors based on Si, GaAs, and InGaAs ^[27]. Considering the small depletion of the intrinsic carriers in the nanowire channel, D^* could be enhanced to up to 10¹⁵ Jones if further gate dielectric and electrode structures are built (to drive dark current down from 10⁻⁶ μA to 10⁻¹⁰ μA). The photo-conductive gain (G) is defined as the ratio between the number of charges collected by the electrodes per unit of time and the number of photons absorbed by the nanowire per unit of time (G = N_e/N_ph). It can be calculated as G = (I_ph/e)/(PA/hν), where I_ph is the photocurrent, e is the electronic charge, P is the incident power density, A is the effective irradiated area, and hν is the energy of an incident photon. The G of the photodetector is up to 4.34 × 10⁵ under a low light intensity of 0.062 mW/cm⁻², which is due to the long photon-generated carrier lifetime in the nanowire compared to the short carrier transit time between the electrodes ^{[24, 26, 28]}. High gain shows that large photocurrent output signals can be achieved with relatively low optical input.

The repeatability and response speed are also important parameters for photodetection ^[29]. Time-resolved photoresponse measurements were performed to characterize the response speed of the photodetector when switching a green light on and off repeatedly (520 nm). The photocurrent responses under 520 nm of illumination with a back gate voltage of −30 V are shown in Fig. 5(a). The response time (rise time τ_r) and the recovery time (fall time τ_f), defined as the time for the photocurrent increasing (decreasing) from 10% to 90% (90% to 10%) of I_peak (photocurrent peak), is 0.912 s and 8 s, respectively. This data is extracted with a Tektronix MDO3014. It shows high stability and reliability with the on/off photoswitching behavior at V_ds = 1 V. As for the response speed characterization of UV light (375 nm) ^[25], a high-speed oscilloscope was used to monitor the fast-varying optical signal. Its longer recovery time may be related to the influence of the surface states ^[30]. The recombination of electrons and holes may be affected by the surface trap state of the nanowire after high-energy ultraviolet radiation. The rise time τ_r drops sharply to 150 ms, which is faster than previous reports for In₂O₃ nanowires photodetectors, to our best knowledge ^{[19-20, 31]}. This can be attributed to the reduction of the interaction effect between the nanowire and the substrate.

Figure 5.  Time-response characterizations of the In₂O₃ single-nanowire photodetector. (a) Photocurrent response of the device, where the laser light is switched on/off at an interval of 20 s (520 nm, 2.6 mW·cm⁻² at V_ds = 1 V. (b) Enlargement of the curve in the 53~72 s range outlined in (a). (c) A single UV photocurrent response of the device (375 nm, 1.6 mW·cm⁻²). (d) Time-resolved photoresponse of the device showing the rise and fall time of the photocurrent at V_ds = 0.1 V.

In summary, we have fabricated a suspended In₂O₃ nanowire transistor based on one-step EBL micro-nano processing technology that can stable and complete separation between a nanowire and a substrate. The intrinsic electrical transportation and photoresponse properties of In₂O₃ nanowire FETs are studied. The In₂O₃ nanowire transistors achieve a stable electrical property in a vacuum. Furthermore, this type of photodetector exhibits excellent optoelectronic performance including broad spectral responsivity, good repeatability, and fast response. Suspended nanowire structures provide an ideal way to investigate the intrinsic properties of nano-materials and broadens their applications in flexible and thermal isolation devices.