Empowered by an ultrawide bandgap (Eg = 4.5–4.9 eV), beta gallium oxide (β-Ga2O3) crystal is an ideal material for solar-blind ultraviolet (SBUV, λ < 280 nm) detection. Here, we report on the first demonstration of dual-modality SBUV light sensing integrated in the same device enabled by multi-physics coupling across photo-electrical and photo-thermo-mechanical domains. The specially designed suspended β-Ga2O3 nanoelectromechanical systems (NEMS) transducer reveals dual-modality responses, with a photocurrent responsivity of 4 mA/W and a frequency shift responsivity of 250 Hz/nW, upon SBUV light exposure. An additional demonstration of a β-Ga2O3 photo-field-effect transistor exhibits a boosted responsivity of 63 A/W. Analysis on the device suggests that reducing the thickness and length of the transducer could further improve the SBUV light sensing responsivities for both modalities. The demonstration could pave the way for future realization of SBUV detectors with dual modalities for enhanced detection fidelity, or respectively optimized for different sensing scenarios.

Beta gallium oxide (β-Ga2O3), an ultrawide bandgap (UWBG) material with a bandgap of 4.5–4.9 eV,1,2 has spurred extensive research interest as an emerging candidate for future power electronics,3–6 thanks to its high internal critical breakdown strength (Ebr,β-Ga2O3 = 8 MV cm−1 predicted7 and Ebr,β-Ga2O3 = 5.2 MV cm−1 measured8). In addition, its UWBG leads to a photon absorption edge (270–280 nm) right at the cutoff wavelength (280 nm) of the solar-blind regime, promising for next generation solar-blind ultraviolet (SBUV) optoelectronics (e.g., photodetector),9–14 which may enable applications such as missile plume tracking, flame detection, and environmental monitoring without suffering from strong solar background.15,16 While photodetectors (PDs) made by UWBG III-nitrides, such as h-BN and AlN, have been demonstrated,17–20 the bandgap could be too wide that the detectable light of the PDs is strongly absorbed in air. Instead, a III-nitride material with an engineered narrower bandgap, AlGaN,21–23 has been explored for SBUV detection. However, the sophisticated alloying and lack of a high quality native substrate have impeded its practical use. A high quality β-Ga2O3 bulk crystal can be grown from melt, providing a potentially more economical alternative for SBUV detection with compelling performance. Motivated by these advantages, β-Ga2O3 photoconductive and Schottky diode (SD) PDs have been fabricated and characterized toward SBUV detection.9–14 Beyond promises as an optoelectronic material, β-Ga2O3 crystal possesses excellent mechanical properties, including high Young’s modulus of EY = 261 GPa,24 for making mechanical devices. Resonance characteristics of β-Ga2O3 micro- and nano-electromechanical systems (MEMS and NEMS) have been examined, showing robust mechanical oscillations and various vibrational modes.24,25 Owing to their ultrasensitive detection capability of external stimuli (e.g., force, mass, and temperature) based on the resonance frequency shift, resonant MEMS and NEMS can be promising platforms for light sensing.26 Given the properties, we envision that β-Ga2O3 resonant MEMS and NEMS, as transducers for SBUV detection, may exhibit similar or even higher sensitivity and detection speed compared to those of their optoelectronic counterparts. Furthermore, optoelectronics and resonant NEMS light sensing modalities can be integrated into the same device for a full-fledged versatile solution to SBUV light detection. For example, the optoelectronic modality can be engineered with ultralow power consumption and act as a photoswitch to trigger the operation of the NEMS modality sensing for fast and high responsivity SBUV light detection.

In this work, we demonstrate, for the first time, a single-crystal β-Ga2O3 transducer for SBUV detection based on dual sensing modalities. The dual sensing modalities are enabled by multi-physics coupling across photo-electrical (opto-electronic) and photo-thermo-mechanical domains. Specifically, the device can respond to SBUV based on (i) photocurrent modulation caused by the photoelectric effect (Modality I: optoelectronic modality) and (ii) resonance frequency shift induced by the photothermal effect (Modality II: NEMS modality). The transducer is fashioned into a metal-semiconductor-metal (MSM) structure with a mechanically suspended portion in the β-Ga2O3 channel [Fig. 1(a)]. We first examine the optoelectronic modality by using a β-Ga2O3 MSM PD and a β-Ga2O3 photo-field-effect transistor (photo-FET) upon different types of illumination. We then characterize the light sensing capability based on the NEMS modality while monitoring the open loop frequency response to the UV irradiation by using the suspended β-Ga2O3 channel as a resonator. Finally, we build a feedback oscillator using the β-Ga2O3 resonator to demonstrate real-time light detection based on the oscillator’s frequency response. Along with the experimental demonstrations, we analyze the sensing mechanisms for both sensing modalities and delineate guidelines for improving the SBUV sensing responsivity.

FIG. 1.

Schematic illustration of the β-Ga2O3 transducer for dual-modality SBUV radiation detection. (a) A β-Ga2O3 transducer in the form of a suspended doubly clamped beam with electrical contacts for two modalities of SBUV light detection, Modality I and Modality II. PD: photodetector; BS: beam splitter; LNA: low-noise amplifier; PS: phase shifter; BPF: band-pass filter; SA: spectrum analyzer; FC: frequency counter; and NA: network analyzer. Signal transduction and the mechanism of SBUV detection by (b) photocurrent modulation based on the photoelectric effect (Modality I) and (c) resonance frequency shift induced by the photothermal effect (Modality II).

FIG. 1.

Schematic illustration of the β-Ga2O3 transducer for dual-modality SBUV radiation detection. (a) A β-Ga2O3 transducer in the form of a suspended doubly clamped beam with electrical contacts for two modalities of SBUV light detection, Modality I and Modality II. PD: photodetector; BS: beam splitter; LNA: low-noise amplifier; PS: phase shifter; BPF: band-pass filter; SA: spectrum analyzer; FC: frequency counter; and NA: network analyzer. Signal transduction and the mechanism of SBUV detection by (b) photocurrent modulation based on the photoelectric effect (Modality I) and (c) resonance frequency shift induced by the photothermal effect (Modality II).

Close modal

Figure 1(a) illustrates the device structure and optical and electrical circuitries where SBUV light can be detected in two sensing modalities. For Modality I [Fig. 1(b)], the electrical conductivity of the β-Ga2O3 channel could be enhanced by photons in two different processes: modulation of channel conductance and lowering of Schottky barrier height at the β-Ga2O3-metal contact. Since β-Ga2O3 crystal is prone to oxygen (O2) adsorption,25,27 the chemisorbed and physisorbed O2 on the crystal surface can capture electrons and create oxygen ions. These trapped oxygen ions will induce strong upward band bending through Fermi-level pinning at the surface of the β-Ga2O3 flake, thus generating a finite, non-conductive depletion region near the surface of the crystal. Therefore, if the size of the nanocrystal is smaller than a critical value, i.e., a critical diameter for a nanowire or a critical thickness for a nanoflake, the crystal is fully depleted. If the size is larger than the critical dimension, a conductive channel will be formed in the nanocrystal. Such an effect is well studied in GaN and ZnO nanowires,28–30 while the thickness dependence of threshold voltage in β-Ga2O3 FETs indicates that similar surface depletion is present in β-Ga2O3 nanoflakes.6 An ideal case for light sensing could be the β-Ga2O3 nanoflake that has a thickness (b) right at its critical value. The crystal is fully depleted, where depletion region width (W) is half of the β-Ga2O3 thickness, W = 0.5b [Fig. 2(b)]. By shining light with photon energy larger than the bandgap of the β-Ga2O3 crystal, electron-hole pairs will be generated [Fig. 2(c)]. Due to the band bending, the excess holes generated by photo illumination prefer to move to the surface to neutralize the adsorbed oxygen ions and induce O2 desorption, which will modify the Fermi-level pinning. Accordingly, the upward band bending will be reduced so that the recombination barrier (Φ) is lowered. Thus, the depletion region width (W) is narrowed and an electron conducting channel is created and further broadened as photon absorption increases, which lowers the channel resistance [Fig. 2(d)].28–30 In addition, since the electrons are confined in the conductive channel, a spatial separation between electrons and holes is created, inducing an increased electron carrier lifetime. This can contribute to a high responsivity of the PD. Furthermore, photo illumination can also lower the Schottky barriers between β-Ga2O3 and metal electrodes, thus reducing the electrical contact resistance. The combined effects on resistance reduction result in an increase in photocurrent with a constant bias. By measuring the current under a static electrical bias between the drain and source electrodes, the power of incident SBUV light can be determined.

FIG. 2.

Schematic of photon induced channel conductance modulation. (a) Device schematic indicating the relation between the device structure and energy band diagrams in panels (b-d). (b) Fully depleted β-Ga2O3 channel before photon radiation. (c) Electron-hole pair generation by photon with energy higher than the bandgap of the β-Ga2O3 crystal. The electrons tend to stay inside the crystal, while the holes tend to migrate to the surface due to the band bending. Holes will neutralize the adsorbed oxygen ions on the surface, resulting in desorption of O2 molecules, and modifying the Fermi-level pinning on the surface to reduce the band bending. (d) Opening of a conductive channel in the crystal due to the reduced band bending by photon radiation.

FIG. 2.

Schematic of photon induced channel conductance modulation. (a) Device schematic indicating the relation between the device structure and energy band diagrams in panels (b-d). (b) Fully depleted β-Ga2O3 channel before photon radiation. (c) Electron-hole pair generation by photon with energy higher than the bandgap of the β-Ga2O3 crystal. The electrons tend to stay inside the crystal, while the holes tend to migrate to the surface due to the band bending. Holes will neutralize the adsorbed oxygen ions on the surface, resulting in desorption of O2 molecules, and modifying the Fermi-level pinning on the surface to reduce the band bending. (d) Opening of a conductive channel in the crystal due to the reduced band bending by photon radiation.

Close modal

In the resonant NEMS modality [Modality II, see Fig. 1(c)], the SBUV photon radiation will photothermally heat up the suspended nanostructure. The elevated temperature will expand the doubly clamped β-Ga2O3 beam due to its positive thermal expansion coefficient, lowering the stress in the device. The photothermally lowered stress level will downshift the mechanical resonance frequency of the β-Ga2O3 resonator. Therefore, by monitoring the resonance frequency of the NEMS transducer, the incident SBUV irradiation can be resolved.

We fabricate the β-Ga2O3 transducer with dual-modality SBUV sensing using mechanically exfoliated β-Ga2O3 flakes. Apart from using β-Ga2O3 nanostructures directly grown by low pressure chemical vapor deposition (LPCVD) demonstrated in our previous studies,24,25,31 we use mechanically exfoliated flakes from the bulk crystal synthesized by the edge-defined film-fed growth (EFG) method.32 Thanks to high-quality bulk β-Ga2O3 with controllable doping level and thus electrical properties, these flakes can be used to form high performance β-Ga2O3 electronics.4,6 Furthermore, exfoliated β-Ga2O3 flakes show well-defined crystal orientation due to the anisotropic crystal structure. The mechanically cleaved β-Ga2O3 flakes are usually in belt shapes, where the large surfaces are coincident with the (100) plane and the long sides are in parallel with the [010] axis of the crystal.32 After identifying a β-Ga2O3 flake with desired size and thickness, we transfer the flake exfoliated on the polydimethylsiloxane (PDMS) stamp to a 290 nm-SiO2-on-Si substrate with pre-defined microtrenches using a dry transfer technique.33 The transferred flake is suspended over a 900 nm deep microtrench, forming a doubly clamped beam structure. Subsequently, we deposit metal electrodes (40 nm Au on 150 nm Ti) on both sides of the microtrench to electrically access the β-Ga2O3 device. The metal deposition is performed through a high-precision stencil mask (shadow mask) without using photoresists and wet chemicals; thus, the pristine nature of the fabricated β-Ga2O3 device is ensured. The final device has a 5 μm-wide, 32 μm-long, and 310 nm-thick β-Ga2O3 channel defined by two metal electrodes with a 20 μm-long portion of the channel suspended over the microtrench [Fig. 3(a)].

FIG. 3.

SBUV radiation detection based on photocurrent modulation by the photoelectric effect. (a) Optical image of the device with labels of source (S) and drain (D) electrodes. (b) Transport characteristics of the PD in dark and upon illumination from SBUV (255 nm), blue (460 nm), green (518 nm), and red (646 nm) LEDs. (c) Transport characteristics of the PD in dark and upon illumination from a mercury (Hg) lamp and a white lamp. (d) Light sensing responsivity of the PD at VD = 10 V for different wavelengths of LED light.

FIG. 3.

SBUV radiation detection based on photocurrent modulation by the photoelectric effect. (a) Optical image of the device with labels of source (S) and drain (D) electrodes. (b) Transport characteristics of the PD in dark and upon illumination from SBUV (255 nm), blue (460 nm), green (518 nm), and red (646 nm) LEDs. (c) Transport characteristics of the PD in dark and upon illumination from a mercury (Hg) lamp and a white lamp. (d) Light sensing responsivity of the PD at VD = 10 V for different wavelengths of LED light.

Close modal

Here we introduce the analysis of the optoelectronic sensing modality. The incident SBUV illumination is transduced into photocurrent, which can be expressed as

Iph=eNe/t,
(1)

where e is the elementary charge, Ne is the number of collected charge, and t is the time. We can also write the equation for the rate of photon absorption by the device using

Nν/t=ηPihc/λ,
(2)

where Nν is the number of absorbed photons, η is the optical absorbance of the β-Ga2O3 material, Pi is the power of the incident light on the device, h is Plank’s constant, c is the speed of light, and λ is the wavelength of the photon. The photoconductive gain of the photodetector, which is the ratio between the number of collected charges and absorbed photons, can be determined using Eqs. (1) and (2),

G=NeNν=Iph/eηPi/hc/λ=1ηehcλIphPi.
(3)

Since the responsivity of the photodetector is the photocurrent over the incident light power, ℜI = Iph/Pi, we can modify Eq. (3) to express the responsivity,17,34

RI=Gηehc/λ.
(4)

The photoconductive gain G can also be expressed by the ratio between the carrier recombination lifetime (τlife) and carrier transit time (τtrans) in the device channel, G = τlife/τtrans. The transit time can be further determined by τtrans = L2/(µV), where L is the channel length, µ is the electron mobility, and V is the applied voltage across the channel. Since responsivity of the PD is proportional to its photoconductive gain, it can be enhanced by decreasing τtrans or increasing τlife. To reduce τtrans, we can either reduce the channel length by making electrodes closer or optimize the Schottky barrier at the semiconductor-metal contact to modulate the lateral electrical field in the channel through modification of the contact resistance. We can also enhance the responsivity by manipulating the effects illustrated in Fig. 2, which can provide pathways to increase the carrier lifetime τlife. By laying the device at the edge of opening of a conductive channel (the ideal case illustrated in Fig. 2), weak SBUV photon illumination on the device can significantly modulate the channel conductance and induce a high current response. The methods to achieve this include (i) selecting a β-Ga2O3 nanoflake with an ideal thickness and (ii) band modulation of the channel using gating.

To experimentally demonstrate the optoelectronic sensing modality, first, we measure the photocurrent response (Modality I) of the β-Ga2O3 PD using a source measure unit (SMU), as illustrated in Fig. 1(a). Figure 3 shows the channel current of the β-Ga2O3 under variable biases upon different illumination conditions. While illumination by blue (460 nm), green (518 nm), and red (646 nm) light emitting diodes (LEDs) only induces negligible change in current, the device is sensitive to SBUV LED illumination with a wavelength at about 255 nm [Fig. 3(b)]. Upon 12 nW of SBUV light irradiated on the active area of the β-Ga2O3 device (5 × 32 μm2 area of the flake between electrodes), the device exhibits a current increase from 20.53 pA to 67.71 pA at 10 V bias, which represents a responsivity ℜI of ∼4 mA/W. The relatively low responsivity of the PD is first limited by the relatively long distance (32 μm) between the two electrodes. A high carrier recombination rate occurs with the long source-drain distance, which is currently restricted by the complexity in fabricating a dual-modality device and can be shortened in the future. The non-ideal flake thickness could also contribute to the low responsivity since the device may not operate in the ideal case, as illustrated in Fig. 2. With a dark current of Idark ≈ 21 pA, the PD has a noise equivalent power NEPI = (2eIdark)1/2/ℜI = 6.3 × 10−13 W/Hz1/2. In addition to LEDs, we characterize the current response of the PD using a mercury lamp (with the illumination spectrum mainly in 200–600 nm range) and a white light lamp [Fig. 3(c)]. While the device shows no response to the white lamp, the current increases upon mercury lamp illumination, part of whose illumination spectrum lies in the SBUV range. The response of PD to 750 nW of mercury lamp illumination is close to its response to 255 nm LED (∼12 nW power on device), indicating that the power of the SBUV portion of the mercury lamp illumination could be more than an order of magnitude lower than its full-spectrum illumination power.

Furthermore, we demonstrate SBUV light sensing based on a β-Ga2O3 FET using another device (Fig. 4). This device shares the same spacing (32 μm) between electrodes, while the β-Ga2O3 flake is narrower (∼3 μm in width) and thicker (446 nm in thickness). By back gating the device through the silicon (p++ doped) substrate, the β-Ga2O3 FET shows an on current of ∼0.4 μA and a threshold voltage around −25 V. By applying a gate voltage VG = −27 V, the dark current of the FET is quenched from ID = 384 nA for VG = 0 V to ID = 22 nA at a bias of VD = 10 V [Fig. 4(b)]. Since the FET is gated at the point of just switched off [the ideal situation shown in Fig. 2(b)], it is easy to open the conductive channel through photon illumination. The inset in Fig. 4(c) shows that the device has the highest photocurrent Iph = ItotalIdark = 227 nA at VG = −27 V, which corresponds to a much boosted responsivity ℜI of ∼63 A/W for 255 nm light. The β-Ga2O3 FET has an external quantum efficiency of EQE = hcI/() = 307 at VG = −27 V. The device also shows good SBUV sensing selectivity—the responsivity of the PD in the visible range is much eliminated [Fig. 4(f)]. The results from the β-Ga2O3 FET confirm that the gain of the β-Ga2O3 PD can be enhanced by gating the semiconductor channel to a just fully depleted regime, which also elucidate the SBUV sensing mechanism illustrated in Fig. 2. Given the negative threshold voltage of the FET for an n-type semiconductor, to eliminate the requirement of applying gate voltage, the β-Ga2O3 flake needs to be thinner to reach the just fully depleted scenario.

FIG. 4.

Light detection based on a β-Ga2O3 field-effect transistor (FET). (a) The transfer curve of the β-Ga2O3 FET in the log and linear (inset) scale with a VD of 10 V. (b) The transport curve of the FET without photon illumination. (c) The transport curve of the FET with ∼3.6 nW of 255 nm light illumination on the device. (d) Optical image of the β-Ga2O3 FET. (e) Photocurrent response of the FET to LED light illumination with different wavelengths. (f) Responsivity of the β-Ga2O3 photo-FET at different wavelengths.

FIG. 4.

Light detection based on a β-Ga2O3 field-effect transistor (FET). (a) The transfer curve of the β-Ga2O3 FET in the log and linear (inset) scale with a VD of 10 V. (b) The transport curve of the FET without photon illumination. (c) The transport curve of the FET with ∼3.6 nW of 255 nm light illumination on the device. (d) Optical image of the β-Ga2O3 FET. (e) Photocurrent response of the FET to LED light illumination with different wavelengths. (f) Responsivity of the β-Ga2O3 photo-FET at different wavelengths.

Close modal

Figure 1(c) illustrates the signal transduction of SBUV sensing using the photothermally induced mechanical resonance frequency shift of the β-Ga2O3 nanostructure (Modality II). Upon SBUV illumination, the suspended β-Ga2O3 flake absorbs photons with energy higher than the bandgap. The photon energy transfers to the crystal lattice and generates heat, thus the temperature of the suspended structure rises. The generated heat diffuses along the suspended β-Ga2O3 flake and then dissipates to the substrate. Accordingly, thermal resistance of the β-Ga2O3 doubly clamped beam and the contact thermal resistance between β-Ga2O3 and the substrate dictate device’s temperature distribution. The temperature distribution along the beam can be expressed using the following equation:

ΔTx=ΔTBx+ΔTC,
(5)

where ∆TB(x) is the temperature distribution in the β-Ga2O3 beam and ∆TC is the temperature difference between β-Ga2O3 and the substrate at contacts. For ∆TB(x), we have a differential equation for heat diffusion,

d2Tdx2=q̇κ,whereq̇=qwdxbwdx=qb,
(6)

κ = 27 W/(m K) is the thermal conductivity of the β-Ga2O3 crystal in the [010] direction (along the beam),35 w and b are the width and thickness of the beam, respectively, x denotes the location along the beam, and q is the heat flux caused by the photon absorption. Therefore, we have q = ηPi/(wl), where η is the optical absorbance, Pi is the power of the incident SBUV light, and l is the length of the suspended resonator. By integrating Eq. (6) and considering the boundary conditions, T (x = 0) = TC and T (x = l) = TC, we obtain

ΔTBx=q2κblxx.
(7)

For ∆TC, we can assume the thermal contact resistance to be RC. Then, we have

ΔTC=RCηPi.
(8)

With elevated temperature, the β-Ga2O3 flake will expand because of the positive thermal expansion coefficient of β-Ga2O3, and it lowers the built-in stress of the device. The stress change can be determined by

Δσ=αEY0lΔTxdxl=αEYηPil12κwb+RC,
(9)

where α = 3.37 × 10−6 K−1 is the thermal expansion coefficient in the [010] direction of the β-Ga2O3 crystal36 and EY is Young’s modulus of the β-Ga2O3 crystal. The total stress in the device is the combination of the initial stress σ0 and the stress change induced by the thermal expansion ∆σ,

σ=σ0+Δσ.
(10)

The resonance frequency of a tensioned, doubly clamped beam is governed by the following equation:37 

fn=Anbl2EYρ1+σl23.4EYb2,
(11)

where An is the eigenvalue for the resonance mode and ρ = 5950 kg/m3 is the volume mass density of the β-Ga2O3 crystal. From Eq. (11), it can be seen that the built-in stress of the device can alter its mechanical resonance frequency.

Since the light sensing responsivity of the resonator ℜf is the differential ratio between the resonance frequency shift and the incident irradiation power, ℜf = df/dPi, by combining Eqs. (9)(11), we can express the responsivity of the SBUV sensing resonator using

Rf=dfndPi=αηAn6.8bEYρl12κwb+RC1+σl23.4EYb21/2.
(12)

Although the responsivity is related to multiple variables, Eq. (12) suggests that a higher thermal resistance RC can improve the responsivity of the resonance-based SBUV sensor.

We detect the UV light irradiation by measuring the resonance frequencies of the β-Ga2O3 doubly clamped beam [Fig. 3(a)] using an ultrasensitive laser interferometry system [Fig. 1(a), Modality II]. The motion of the mechanical structure is driven by an amplitude modulated blue (405 nm) laser. By also shining a red (633 nm) laser light to the device, the driven motion of the device interferometrically modulates the intensity of the reflected 633 nm laser light, which is detected by using a low noise PD. For the open loop measurement [dark blue path in Fig. 1(a), Modality II], a network analyzer is used to resolve the driven resonance spectrum of the β-Ga2O3 resonator. We measure the first 5 resonance modes of the β-Ga2O3 resonator with resonance frequencies from 5 to 28 MHz and quality factors (Qs) from 800 to 1700 (Fig. 5). We focus on the third resonance mode (f ≈ 14.320 MHz with a Q ≈ 1170) of the device for the light sensing experiments. After shining light from the mercury lamp to the resonator, the resonator exhibits a 55 kHz frequency downshift upon 230 nW illumination on the device and a 32 kHz downshift upon 115 nW illumination. Note here that the active areas of the device for different modalities are different. For Modality I, the active area is the β-Ga2O3 flake area between the electrodes (5 × 32 μm2). For Modality II, the active area is the suspended area of the β-Ga2O3 flake (5 × 20 μm2). In addition, for Modality II, since the measurements are performed with the device in vacuum, an optical window attenuates (92% transmission) the light toward the device. Moreover, in order to track the resonance frequency in real-time, a closed-loop feedback oscillator should be built to enhance the detection speed.

FIG. 5.

Photon irradiation detection based on photothermally induced resonance frequency shift. (a) Wide driven resonance spectrum of a doubly clamped β-Ga2O3 resonator with up to the 5th resonance mode. Insets show validations of mode shapes by finite element modeling (FEM) simulations. [(b)–(e)] Resonance spectra of the 1st, 2nd, 4th, and 5th modes. (f) The frequency response spectra of the resonator’s third resonance mode upon mercury lamp illumination.

FIG. 5.

Photon irradiation detection based on photothermally induced resonance frequency shift. (a) Wide driven resonance spectrum of a doubly clamped β-Ga2O3 resonator with up to the 5th resonance mode. Insets show validations of mode shapes by finite element modeling (FEM) simulations. [(b)–(e)] Resonance spectra of the 1st, 2nd, 4th, and 5th modes. (f) The frequency response spectra of the resonator’s third resonance mode upon mercury lamp illumination.

Close modal

We then build a self-sustained oscillator based on the β-Ga2O3 doubly clamped beam resonator by adding a feedback loop to the laser interferometry system [dark green path in Fig. 1(a), Modality II]. The signal detected by using the PD is amplified and phase-shifted to satisfy the Barkhausen criterion. Then, the signal goes through a proper band-pass filter to remove the unwanted resonance modes and higher harmonics. Afterwards, the signal is fed back to the resonator using the blue laser. Upon completion of the feedback, linewidth of the frequency-domain spectrum can be significantly narrowed compared to that of the passive resonator. We measure the spectrum of the feedback oscillator using a spectrum analyzer. The full-width at half maximum (FWHM) of the oscillator is ∆osc ≈ 0.3 kHz. While the Lorentzian linewidth of the passive resonator is ∆res = f3/Qres ≈ 12 kHz, the oscillator provides an around 40-fold improvement in Q, with an effective quality factor of the oscillator of Qosc,eff ≈ 48 000 [Fig. 6(a)].

FIG. 6.

β-Ga2O3 oscillator for real-time light detection. (a) Oscillation spectrum of the oscillator (using the third mode of the β-Ga2O3 resonator) without light illumination. (b) Allan deviation of the feedback oscillator. The magenta curve shows the Allan deviation limited by thermomechanical noise which is depicted by using Eq. (14). [(c) and (d)] Frequency response of the oscillator upon cyclic mercury lamp light irradiation.

FIG. 6.

β-Ga2O3 oscillator for real-time light detection. (a) Oscillation spectrum of the oscillator (using the third mode of the β-Ga2O3 resonator) without light illumination. (b) Allan deviation of the feedback oscillator. The magenta curve shows the Allan deviation limited by thermomechanical noise which is depicted by using Eq. (14). [(c) and (d)] Frequency response of the oscillator upon cyclic mercury lamp light irradiation.

Close modal

It is of crucial importance to measure the frequency stability of the oscillator since for a resonance-frequency-shift-based sensor, the frequency-fluctuation noise determines the ultimate detection sensitivity. We characterize the frequency stability of the β-Ga2O3 oscillator by measuring its Allan deviation σA. The defining expression of the Allan deviation is38 

σAτ=12fc21N1i=2Nf¯if¯i121/2,
(13)

where f¯i is the averaged frequency in the ith time interval of τ and fc is the nominal carrier frequency, which is the resonance frequency of the third mode of the resonator in this case. As shown in Fig. 6(b), the oscillator achieves an Allan deviation of σA = 2.6 × 10−5 for an averaging time of τ = 0.1 s and σA < 4 × 10−5 for an averaging time up to τ = 50 s. The Allan deviation represents the frequency stability of the oscillator and can be translated into the frequency fluctuation δf of the oscillator δfτ=0.1s=2σAfc = 530 Hz.

We use the oscillator to detect light irradiation in real time. By illuminating periodic light from the mercury lamp to the β-Ga2O3 resonator, the real-time tracking of the oscillator’s frequency shows frequency downshifts upon illumination [Figs. 6(c) and 6(d)]. The 230 nW and 115 nW mercury lamp illumination on the suspended device exhibits −52 kHz and −30 kHz frequency shifts of the oscillator, respectively. With the results of both open-loop spectra and closed-loop feedback oscillation, we can extract an average frequency responsivity of the β-Ga2O3 resonator ℜf = ∆f/Pi ≈ 250 Hz/nW. The extracted responsivity could be an underestimation of the responsivity for SBUV irradiation. As indicated from an earlier observation, the power of the SBUV portion of the mercury lamp could be more than an order of magnitude lower than its total illumination power. The responsivity of the β-Ga2O3 resonator to SBUV light could be much higher than the currently extracted value. Furthermore, we can calculate the minimum detectable power (MDP) of the UV sensing oscillator δPmin = δf/ℜf ≈ 2 nW.

The thermomechanical noise limited Allan deviation is38 

σA,thτ=πkBTPcτQ2,
(14)

where kB is the Boltzmann constant, T = 300 K is the temperature, τ is the averaging time, and Pc is the power handling of the mechanical resonator which is determined by39,40

Pc=2πfcEc,1dBQ,
(15)

where Ec,1dB=12keffac,1dB2 is the mechanical energy stored in the system, keff is the effective stiffness, which can be determined using the resonance frequency and effective mass (Meff) of the resonator, (2πfc)2 = keff/Meff, and ac,1dB = 0.745ac is the critical amplitude (ac) at the 1 dB compression point. The effective mass of the resonator can be expressed using

Meff=0lϕn2xdxlϕn,max2M,
(16)

where ϕn(x) is the mode shape of the nth resonance mode, ϕn,max(x) is the maximum displacement of the nth resonance mode, and M is the total mass of the mechanical resonator. In addition, the critical amplitude (ac) of the resonator for the nth mode can be determined using39 

ac=4πfc323αnQ,
(17)

where αn is the nonlinear coefficient of the nth mode, which can be expressed by39 

αn=EY2ρl0lϕnx2dx20lϕn2xdx.
(18)

Given the geometry of the β-Ga2O3 doubly clamped beam resonator, we conduct eigenfrequency simulation by finite element modeling (FEM) using COMSOL Multiphysics to extract the mode shapes of the resonator. To match the measured resonance frequencies of the resonator, a pre-stress of 10 MPa is applied along the beam direction in the simulation. For Young’s modulus, we use EY = 240 GPa, which is extracted separately from a thicker (b = 430 nm) and shorter (l = 10.2 μm) β-Ga2O3 doubly clamped beam resonator (where pre-stress in the device has a minimal effect on the resonance frequency) fabricated using the same EFG-grown bulk β-Ga2O3 crystal. Thus, we have ac = 3.64 nm, Meff = 79.5 pg, and Pc = 0.18 nW for the third resonance mode [right curve in Fig. 5(f)] of the β-Ga2O3 resonator using the simulated mode shape from FEM and Eqs. (15)(18). Based on this analysis, we plot the thermomechanical-noise-limited Allan deviation using Eq. (14) for the third mode of the resonator in Fig. 6(b). At an averaging time of 0.1 s, we have an Allan deviation limited by thermomechanical noise: σA,th (τ = 0.1 s) = 2.29 × 10−8. The thermomechanical-noise-limited Allan deviation is much smaller than the measured Allan deviation, meaning that the noises from the electrical-optical feedback circuit play a major role and they compromise the measured Allan deviation of the oscillator. The noises from the feedback circuit include the noise of the amplifier, the noise of the laser, the noise of the photodetector, and the noise of other electrical circuit components.

The thermomechanical frequency noise spectral density can be extracted by38 

Sf,th1/2fc=2fcπτσA,th,
(19)

and with an averaging time of 0.1 s, the thermomechanically limited frequency noise of the device is 0.38 Hz/Hz1/2. Thus, we can calculate the thermomechanical-noise-limited noise equivalent power (NEP) of our device, NEPth=Sf,th1/2fc/2πf = 2.4 × 10−13 W/Hz1/2.

Table I illustrates our device performance compared to that of previously reported SBUV detectors based on the β-Ga2O3 crystal. Comparing the values to our previous work,31 the device in this work has the lower responsivity and higher NEP and MDP for NEMS modality. These could be attributed to the upscaling of the resonator in size to accommodate a more complex device structure for realizing dual-modality sensing.

TABLE I.

Comparison with β-Ga2O3-based SBUV detectors reported earlier.

Device typeSource materialActive areaResponsivity |ℜ|MDPaδPminNEPb (W/Hz1/2)References
Vertical back-to-back Bulk wafer 0.8 cm2 39.3 A/W 28.0 nW 1.5 × 10−14 9  
Schottky diode 
Vertical diode Gallium evaporation ∼7 mm2 0.07 A/W 1–10 nW ∼8 × 10−14 10  
in oxygen plasma 
Vertical Schottky Bulk wafer by floating ∼0.8 mm2 8.7 A/W 1–10 nW ∼7 × 10−15 11  
diode zone method 
Vertical Schottky MOCVD on p-GaN 0.24 cm2 12.8 A/W ∼1 nW 3.8 × 10−14 12  
diode 
Lateral Schottky MBE growth on ∼0.08 mm2 1.14 A/W <10 nW 1.4 × 10−14 13  
diode c-plane sapphire 
Lateral back-to-back RF magnetron sputtering NA 96.13 A/W 15 fW ∼7 × 10−18 14  
Schottky diodes on p-Si 
Lateral back-to-back Mechanical exfoliation 160 μm2 0.004 A/W 5.5 nW 6.3 × 10−13 This work 
Schottky diodes from EFG grown bulk 
Lateral photo-field Mechanical exfoliation 96 μm2 63 A/W 0.35 nW 2.7 × 10−15 This work 
-effect transistor from EFG grown bulk 
Resonant LPCVD 21.2 μm2 3.1 Hz/pW 0.53 nW 8.2 × 10−14 31  
transducer nanoflakes 
Resonant Mechanical exfoliation 100 μm2 250 Hz/nW 2 nW 2.4 × 10−13 This work 
transducer from EFG grown bulk 
Device typeSource materialActive areaResponsivity |ℜ|MDPaδPminNEPb (W/Hz1/2)References
Vertical back-to-back Bulk wafer 0.8 cm2 39.3 A/W 28.0 nW 1.5 × 10−14 9  
Schottky diode 
Vertical diode Gallium evaporation ∼7 mm2 0.07 A/W 1–10 nW ∼8 × 10−14 10  
in oxygen plasma 
Vertical Schottky Bulk wafer by floating ∼0.8 mm2 8.7 A/W 1–10 nW ∼7 × 10−15 11  
diode zone method 
Vertical Schottky MOCVD on p-GaN 0.24 cm2 12.8 A/W ∼1 nW 3.8 × 10−14 12  
diode 
Lateral Schottky MBE growth on ∼0.08 mm2 1.14 A/W <10 nW 1.4 × 10−14 13  
diode c-plane sapphire 
Lateral back-to-back RF magnetron sputtering NA 96.13 A/W 15 fW ∼7 × 10−18 14  
Schottky diodes on p-Si 
Lateral back-to-back Mechanical exfoliation 160 μm2 0.004 A/W 5.5 nW 6.3 × 10−13 This work 
Schottky diodes from EFG grown bulk 
Lateral photo-field Mechanical exfoliation 96 μm2 63 A/W 0.35 nW 2.7 × 10−15 This work 
-effect transistor from EFG grown bulk 
Resonant LPCVD 21.2 μm2 3.1 Hz/pW 0.53 nW 8.2 × 10−14 31  
transducer nanoflakes 
Resonant Mechanical exfoliation 100 μm2 250 Hz/nW 2 nW 2.4 × 10−13 This work 
transducer from EFG grown bulk 
a

MDPs for optoelectronic PDs are all calculated using δPmin = Idark/ℜ, where Idark is the dark current.

b

NEPs for some of the optoelectronic PDs are calculated here, by using NEP = (2eIdark)1/2/ℜ, where e is the electronic charge, ℜ is the responsivity, and 2eIdark is the shot noise of the PD.

In order to fully understand NEMS sensing modality and give guidelines to designing resonators with higher SBUV sensing responsivity, we use Eq. (12) to scale the responsivity with respect to different device parameters (Fig. 7). By using the geometries of the measured device, we can scale the thermal contact resistance to match the device responsivity, where the responsivity improves linearly with respect to the thermal contact resistance. We also notice that the thermal resistance of the beam plays a minimal role in the responsivity since the responsivity is approaching zero when contact resistance is zero. In addition, we can shrink the thickness of β-Ga2O3 crystal to improve the responsivity. Figure 7(b) shows the responsivity of the resonator with thickness of the device ranging from 50 nm to 500 nm using a thermal contact resistance of 8.9 K/μW. From the plot, the responsivity can be much improved by making the device thinner. By changing the length of the suspended resonator, the responsivity can only be slightly improved by making the device length shorter [Fig. 7(c)]. Furthermore, we reduce the thickness and the length of the resonator together by keeping b/l at constant [Fig. 7(b)inset]. In this way, the responsivity of the device can be improved and the scaling shows much less dependence on the initial stress in the device. Note that limited by the absorption coefficient of ∼10−5 cm−1 in the [100] direction at a wavelength of ∼250 nm,41 which corresponds to a penetration depth of 100 nm, a special design is needed to improve the light absorption with devices thinner than 100 nm. Given the modeling and scaling, we could attribute the reduced responsivity to a possibly lower contact resistance because of the addition of the metal electrodes and larger device footprint compared to our previous work.31 To design a better SBUV sensing device using NEMS modality in the future, two methods could be taken to improve the responsivity, including (i) increasing the thermal contact resistance from the β-Ga2O3 flake to the substrate and (ii) reducing the size (length and thickness) of the resonator.

FIG. 7.

Modeling and scaling of responsivity of NEMS modality for devices with different initial stresses. The geometry of the measured device is used for scaling. (a) Responsivity scaling with respect to the contact resistance. (b) Responsivity scaling for β-Ga2O3 resonators with thicknesses in 20–500 nm range. Inset: responsivity scaling for thicknesses in 20–500 nm range with constant b/l. (c) Responsivity scaling for devices with lengths in 1–50 μm range.

FIG. 7.

Modeling and scaling of responsivity of NEMS modality for devices with different initial stresses. The geometry of the measured device is used for scaling. (a) Responsivity scaling with respect to the contact resistance. (b) Responsivity scaling for β-Ga2O3 resonators with thicknesses in 20–500 nm range. Inset: responsivity scaling for thicknesses in 20–500 nm range with constant b/l. (c) Responsivity scaling for devices with lengths in 1–50 μm range.

Close modal

In conclusion, we have demonstrated, for the first time, the β-Ga2O3 SBUV transducer with dual sensing modalities, including photocurrent modulation caused by the photoelectric effect and resonance frequency shift induced by the photothermal effect. We find that the transducer has a photocurrent responsivity of 4 mA/W and a resonance frequency shift responsivity of 250 Hz/nW to UV illumination. We also demonstrate a β-Ga2O3 photo-FET with a boosted photocurrent responsivity of 63 A/W. Furthermore, through analysis, we find that a thinner device with shorter channel length could improve the SBUV light sensing responsivity for both modalities, which provides a guideline to optimize the dual-modality sensing performance. Finally, the transducer provides a promising platform with dual sensing modalities, which, because of the different sensing mechanisms, can each be designed for different applications. For example, the photocurrent-based detection can be engineered to consume ultralow power, i.e., zero bias,13,14 that can work as a trigger to turn on the high speed, high responsivity, yet higher power consuming resonance-frequency-based SBUV sensor for real-time activity monitoring.

We thank the Army Research Office (ARO, Grant No. W911NF-16-1-0340), the US Department of Energy (DOE) EERE (Grant No. DE-EE0006719), the National Science Foundation (NSF) SNM Award (Grant No. CMMI-1246715), and the ThinkEnergy Fellowship (X.-Q. Zheng) for financial support.

1.
M. R.
Lorenz
,
J. F.
Woods
, and
R. J.
Gambino
, “
Some electrical properties of the semiconductor β-Ga2O3
,”
J. Phys. Chem. Solids
28
,
403
(
1967
).
2.
N.
Ueda
,
H.
Hosono
,
R.
Waseda
, and
H.
Kawazoe
, “
Anisotropy of electrical and optical properties in β-Ga2O3 single crystals
,”
Appl. Phys. Lett.
71
,
933
(
1997
).
3.
M.
Higashiwaki
,
K.
Sasaki
,
H.
Murakami
,
Y.
Kumagai
,
A.
Koukitu
,
A.
Kuramata
,
T.
Masui
, and
S.
Yamakoshi
, “
Recent progress in Ga2O3 power devices
,”
Semicond. Sci. Technol.
31
,
034001
(
2016
).
4.
W. S.
Hwang
,
A.
Verma
,
H.
Peelaers
,
V.
Protasenko
,
S.
Rouvimov
,
H.
Xing
,
A.
Seabaugh
,
W.
Haensch
,
C.
van de Walle
,
Z.
Galazka
,
M.
Albrecht
,
R.
Fornari
, and
D.
Jena
, “
High-voltage field effect transistors with wide-bandgap β-Ga2O3 nanomembranes
,”
Appl. Phys. Lett.
104
,
203111
(
2014
).
5.
A. J.
Green
,
K. D.
Chabak
,
E. R.
Heller
,
R. C.
Fitch
,
M.
Baldini
,
A.
Fiedler
,
K.
Irmscher
,
G.
Wagner
,
Z.
Galazka
,
S. E.
Tetlak
,
A.
Crespo
,
K.
Leedy
, and
G. H.
Jessen
, “
3.8-MV/cm breakdown strength of MOVPE-grown Sn-doped β-Ga2O3 MOSFETs
,”
IEEE Electron Device Lett.
37
,
902
(
2016
).
6.
H.
Zhou
,
M.
Si
,
S.
Alghamdi
,
G.
Qiu
,
L.
Yang
, and
P. D.
Ye
, “
High-performance depletion/enhancement-mode β-Ga2O3 on insulator (GOOI) field-effect transistors with record drain currents of 600/450 mA/mm
,”
IEEE Electron Device Lett.
38
,
103
(
2016
).
7.
M.
Higashiwaki
,
K.
Sasaki
,
A.
Kuramata
,
T.
Masui
, and
S.
Yamakoshi
, “
Gallium oxide (Ga2O3) metal-semiconductor field-effect transistors on single-crystal β-Ga2O3 (010) substrates
,”
Appl. Phys. Lett.
100
,
013504
(
2012
).
8.
X.
Yan
,
I. S.
Esqueda
,
J.
Ma
,
J.
Tice
, and
H.
Wang
, “
High breakdown electric field in β-Ga2O3/graphene vertical barrister heterostructure
,”
Appl. Phys. Lett.
112
,
032101
(
2018
).
9.
W.-Y.
Kong
,
G.-A.
Wu
,
K.-Y.
Wang
,
T.-F.
Zhang
,
Y.-F.
Zou
,
D.-D.
Wang
, and
L.-B.
Luo
, “
Graphene-β-Ga2O3 heterojunction for highly sensitive deep UV photodetector application
,”
Adv. Mater.
28
,
10725
(
2016
).
10.
S.
Nakagomi
,
T.
Momo
,
S.
Takahashi
, and
Y.
Kokubun
, “
Deep ultraviolet photodiodes based on β-Ga2O3/SiC heterojunction
,”
Appl. Phys. Lett.
103
,
072105
(
2013
).
11.
T.
Oshima
,
T.
Okuno
,
N.
Arai
,
N.
Suzuki
,
S.
Ohira
, and
S.
Fujita
, “
Vertical solar-blind deep-ultraviolet Schottky photodetectors based on β-Ga2O3 substrates
,”
Appl. Phys. Express
1
,
011202
(
2008
).
12.
R.
Lin
,
W.
Zheng
,
D.
Zhang
,
Z.
Zhang
,
Q.
Liao
,
L.
Yang
, and
F.
Huang
, “
High-performance graphene/β-Ga2O3 heterojunction deep-ultraviolet photodetector with hot-electron excited carrier multiplication
,”
ACS Appl. Mater. Interfaces
10
,
22419
(
2018
).
13.
A. S.
Pratiyush
,
S.
Krishnamoorthy
,
S.
Kumar
,
Z.
Xia
,
R.
Muralidharan
,
S.
Rajan
, and
D. N.
Nath
, “
Demonstration of zero bias responsivity in MBE grown β-Ga2O3 lateral deep-UV photodetector
,”
Jpn. J. Appl. Phys., Part 1
57
,
060313
(
2018
).
14.
K.
Arora
,
N.
Goel
,
M.
Kumar
, and
M.
Kumar
, “
Ultrahigh performance of self-powered β-Ga2O3 thin film solar-blind photodetector grown on cost-effective Si substrate using high-temperature seed layer
,”
ACS Photonics
5
,
2391
(
2018
).
15.
M.
Razeghi
and
A.
Rogalski
, “
Semiconductor ultraviolet detectors
,”
J. Appl. Phys.
79
,
7433
(
1996
).
16.
M.
Razeghi
, “
Short-wavelength solar-blind detectors–status, prospects, and markets
,”
Proc. IEEE
90
,
1006
(
2002
).
17.
W.
Zheng
,
R.
Lin
,
Z.
Zhang
, and
F.
Huang
, “
Vacuum-ultraviolet photodetection in few-layered h-BN
,”
ACS Appl. Mater. Interfaces
10
,
27116
(
2018
).
18.
W.
Zheng
,
F.
Huang
,
R.
Zheng
, and
H.
Wu
, “
Low-dimensional structure vacuum-ultraviolet-sensitive (λ < 200 nm) photodetector with fast-response speed based on high-quality AlN micro/nanowire
,”
Adv. Mater.
27
,
3921
(
2015
).
19.
W.
Zheng
,
R.
Lin
,
J.
Ran
,
Z.
Zhang
,
X.
Ji
, and
F.
Huang
, “
Vacuum-ultraviolet photovoltaic detector
,”
ACS Nano
12
,
425
(
2018
).
20.
W.
Zheng
,
R.
Lin
,
D.
Zhang
,
L.
Jia
,
X.
Ji
, and
F.
Huang
, “
Vacuum-ultraviolet photovoltaic detector with improved response speed and responsivity via heating annihilation trap state mechanism
,”
Adv. Opt. Mater.
6
,
1800697
(
2018
).
21.
N.
Biyikli
,
O.
Aytur
,
I.
Kimukin
,
T.
Tut
, and
E.
Ozbay
, “
Solar-blind AlGaN-based Schottky photodiodes with low noise and high detectivity
,”
Appl. Phys. Lett.
81
,
3272
(
2002
).
22.
R.
McClintock
,
A.
Yasan
,
K.
Mayes
,
D.
Shiell
,
S. R.
Darvish
,
P.
Kung
, and
M.
Razeghi
, “
High quantum efficiency AlGaN solar-blind p-i-n photodiodes
,”
Appl. Phys. Lett.
84
,
1248
(
2004
).
23.
H.
Jiang
and
T.
Egawa
, “
Low-dark-current high-performance AlGaN solar-blind p–i–n photodiodes
,”
Jpn. J. Appl. Phys.
47
,
1541
(
2008
).
24.
X.-Q.
Zheng
,
J.
Lee
,
S.
Rafique
,
L.
Han
,
C. A.
Zorman
,
H.
Zhao
, and
P. X.-L.
Feng
, “
Ultrawide band gap β-Ga2O3 nanomechanical resonators with spatially visualized multimode motion
,”
ACS Appl. Mater. Interfaces
9
,
43090
(
2017
).
25.
X.-Q.
Zheng
,
J.
Lee
,
S.
Rafique
,
L.
Han
,
C. A.
Zorman
,
H.
Zhao
, and
P. X.-L.
Feng
, “
Free-standing β-Ga2O3 thin diaphragms
,”
J. Electron. Mater.
47
,
973
(
2018
).
26.
X. C.
Zhang
,
E. B.
Myers
,
J. E.
Sader
, and
M. L.
Roukes
, “
Nanomechanical torsional resonators for frequency-shift infrared thermal sensing
,”
Nano Lett.
13
,
1528
(
2013
).
27.
M.
Ogita
,
K.
Higo
,
Y.
Nakanishi
, and
Y.
Hatanaka
, “
Ga2O3 thin film for oxygen sensor at high temperature
,”
Appl. Surf. Sci.
175-176
,
721
(
2001
).
28.
R.
Calarco
,
M.
Marso
,
T.
Richter
,
A. I.
Aykanat
,
R.
Meijers
,
A. v.d.
Hart
,
T.
Stoica
, and
H.
Lüth
, “
Size-dependent photoconductivity in MBE-grown GaN-nanowires
,”
Nano Lett.
5
,
981
(
2005
).
29.
C.
Soci
,
A.
Zhang
,
B.
Xiang
,
S. A.
Dayeh
,
D. P. R.
Aplin
,
J.
Park
,
X. Y.
Bao
,
Y. H.
Lo
, and
D.
Wang
, “
ZnO nanowire UV photodetectors with high internal gain
,”
Nano Lett.
7
,
1003
(
2007
).
30.
M.
Madel
,
F.
Huber
,
R.
Mueller
,
B.
Amann
,
M.
Dickel
,
Y.
Xie
, and
K.
Thonke
, “
Persistent photoconductivity in ZnO nanowires: Influence of oxygen and argon ambient
,”
J. Appl. Phys.
121
,
124301
(
2017
).
31.
X.-Q.
Zheng
,
J.
Lee
,
S.
Rafique
,
M. R.
Karim
,
L.
Han
,
H.
Zhao
,
C. A.
Zorman
, and
P. X.-L.
Feng
, “
β-Ga2O3 NEMS oscillator for real-time middle ultraviolet (MUV) light detection
,”
IEEE Electron Device Lett.
39
,
1230
(
2018
).
32.
W.
Mu
,
Z.
Jia
,
Y.
Yin
,
Q.
Hu
,
J.
Zhang
,
Q.
Feng
,
Y.
Hao
, and
X.
Tao
, “
One-step exfoliation of ultra-smooth β-Ga2O3 wafers from bulk crystal for photodetectors
,”
CrystEngComm
19
,
5122
(
2017
).
33.
R.
Yang
,
X.-Q.
Zheng
,
Z.
Wang
,
C. J.
Miller
, and
P. X.-L.
Feng
, “
Multilayer MoS2 transistors enabled by a facile dry-transfer technique and thermal annealing
,”
J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom.
32
,
061203
(
2014
).
34.
W.
Zheng
,
R.
Lin
,
Y.
Zhu
,
Z.
Zhang
,
X.
Ji
, and
F.
Huang
, “
Vacuum ultraviolet photodetection in two-dimensional oxides
,”
ACS Appl. Mater. Interfaces
10
,
20696
(
2018
).
35.
Z.
Guo
,
A.
Verma
,
X.
Wu
,
F.
Sun
,
A.
Hickman
,
T.
Masui
,
A.
Kuramata
,
M.
Higashiwaki
,
D.
Jena
, and
T.
Luo
, “
Anisotropic thermal conductivity in single crystal β-gallium oxide
,”
Appl. Phys. Lett.
106
,
111909
(
2015
).
36.
F.
Orlandi
,
F.
Mezzadri
,
G.
Calestani
,
F.
Boschi
, and
R.
Fornari
, “
Thermal expansion coefficients of β-Ga2O3 single crystals
,”
Appl. Phys. Express
8
,
111101
(
2015
).
37.
S. C.
Jun
,
X. M. H.
Huang
,
M.
Manolidis
,
C. A.
Zorman
,
M.
Mehregany
, and
J.
Hone
, “
Electrothermal tuning of Al-SiC nanomechanical resonators
,”
Nanotechnology
17
,
1506
(
2006
).
38.
A. N.
Cleland
and
M. L.
Roukes
, “
Noise processes in nanomechanical resonators
,”
J. Appl. Phys.
92
,
2758
(
2002
).
39.
I.
Kozinsky
, “
Nonlinear nanoelectromechanical systems
,” Ph.D. dissertation (
Department at the Kavli Nanoscience Institute Condensed Matter Physics, California Institute of Technology
,
Pasadena, CA, USA
,
2007
).
40.
J.
Lee
,
Z.
Wang
,
K.
He
,
R.
Yang
,
J.
Shan
, and
P. X.-L.
Feng
, “
Electrically tunable single- and few-layer MoS2 nanoelectromechanical systems with broad dynamic range
,”
Sci. Adv.
4
,
eaao6653
(
2018
).
41.
T.
Onuma
,
S.
Saito
,
K.
Sasaki
,
T.
Masui
,
T.
Yamaguchi
,
T.
Honda
,
A.
Kuramata
, and
M.
Higashiwaki
, “
Spectroscopic ellipsometry studies on β-Ga2O3 films and single crystal
,”
Jpn. J. Appl. Phys., Part 1
55
,
1202B2
(
2016
).