Enhanced UVC Responsivity of Heteroepitaxial α-Ga₂O₃ Photodetector with Ultra-Thin HfO₂ Interlayer

Abstract

In this study, the influence of HfO₂ interlayer thickness on the performance of heteroepitaxial α-Ga₂O₃ layer-based metal–insulator–semiconductor–insulator–metal (MISIM) ultraviolet photodetectors is examined. A thin HfO₂ interlayer enhances the interface quality and reduces the density of interface traps, thereby improving the performance of UVC photodetectors. The fabricated device with a 1 nm HfO₂ interlayer exhibited a significantly reduced dark current and higher photocurrent than a conventional metal–semiconductor–metal (MSM). Specifically, the 1 nm HfO₂ MISIM device demonstrated a photocurrent of 2.3 μA and a dark current of 6.61 pA at 20 V, whereas the MSM device exhibited a photocurrent of 1.1 μA and a dark current of 73.3 pA. Furthermore, the photodetector performance was comprehensively evaluated in terms of responsivity, response speed, and high-temperature operation. These results suggest that the proposed ultra-thin HfO₂ interlayer is an effective strategy for enhancing the performance of α-Ga₂O₃-based UVC photodetectors by simultaneously suppressing dark currents and increasing photocurrents and ultimately demonstrate its potential for stable operation under extreme environmental conditions.

1. Introduction

Ultraviolet (UV) detection is crucial for a wide range of applications, including fire detection, sterilization, environmental monitoring, aerospace, and defense systems. Within the UV spectrum, high-energy UVC radiation (wavelength < 280 nm) is particularly significant in extreme environments [1,2,3,4,5]. However, its carcinogenic nature and capacity to cause severe DNA damage pose significant risks to biological systems, including human skin and eyes [6,7]. Although the stratospheric ozone layer effectively absorbs most UVC radiation, the selective detection of weak UVC signals under visible or solar illumination remains a considerable technical challenge. Therefore, the development of highly sensitive and selective UVC detectors capable of operating in solar environments is essential for advancing these critical applications.

Conventional wide-bandgap semiconductors, such as aluminum gallium nitride (AlGaN) and diamond, have long been considered for UVC detection. Yet, AlGaN faces challenges in achieving high-quality thin films because of the complexities of its ternary alloy system, whereas diamond, despite its exceptional intrinsic properties, is hindered by its high fabrication cost and limited scalability for large-area integration [8,9,10,11,12].

These limitations have spurred interest in gallium oxide (Ga₂O₃), an ultra-wide-bandgap material (4.6–5.3 eV) emerging as a next-generation candidate for UVC photodetectors. Ga₂O₃ exists in five different crystal phases (α, β, γ, δ, and ε), among which the α-phase exhibits the widest bandgap (~5.2 eV) and a nearly direct band-to-band transition. In addition, α-Ga₂O₃ can be grown at relatively low temperatures (~500 °C), which is advantageous for industrial-scale manufacturing. Combined with its visible-blind nature that effectively suppresses noise from visible and near-UV light, α-Ga₂O₃ also offers low-noise and high-sensitivity performance, making it a highly promising material for reliable and highly selective UVC photodetectors, particularly in extreme environments, such as flame detection, where precision and stability are essential [13,14,15,16,17]. Furthermore, α-Ga₂O₃ offers excellent compatibility with large-area sapphire substrates and can be grown using various techniques, including hydride vapor phase epitaxy (HVPE), mist chemical vapor deposition (mist-CVD), and molecular beam epitaxy (MBE). Among these, HVPE is particularly favorable for scalable production due to its extremely high growth rate, while MBE, though capable of producing highly crystalline films, is constrained by a slow deposition rate that limits its suitability for mass production [18,19,20,21]. These attributes make α-Ga₂O₃ a promising material for high-reliability, selective UVC photodetectors, particularly in demanding applications, such as flame detection, where precision and stability are paramount [22,23,24].

In this study, α-Ga₂O₃-based metal–insulator–semiconductor–insulator–metal (MISIM) UVC photodetectors were fabricated and systematically characterized. We fabricated devices with different HfO₂ thicknesses to explore their impact, systematically evaluating performance through current–voltage (I–V) measurements under dark and UVC-illuminated conditions. These tests revealed significant dark current suppression and photocurrent enhancement. A spectral responsivity analysis at 235 nm confirmed the high wavelength selectivity and rejection ratio of the fabricated device. In addition, time-dependent photocurrent measurements were used to assess the response speed, and high-temperature measurements up to 300 °C were conducted to evaluate thermal stability. These results demonstrate that the incorporation of a thin HfO₂ interlayer effectively enhances the photoresponse and reliability of α-Ga₂O₃-based UVC photodetectors, thereby supporting their practical applicability in extreme and high-temperature environments.

2. Experimental Section

Heteroepitaxial α-Ga₂O₃ layers were grown on a 2-inch (0001) sapphire substrate using hydride vapor phase epitaxy (HVPE) at 470 °C, with GaCl and O₂ as precursors. The undoped α-Ga₂O₃ films were grown to a thickness of 700 nm for 6 min. Following surface cleaning with acetone and isopropyl alcohol (IPA), HfO₂ layers were deposited via atomic layer deposition (ALD) at 360 °C, using tetrakis (ethylmethylamino) hafnium (TEMAhf) as the precursor and ozone as the reactant. HfO₂ thicknesses of 1 nm and 2 nm were achieved, with a growth rate of approximately 0.74 Å/cycle. Interdigitated Ti/Au (20 nm/80 nm) electrodes were then fabricated using electron-beam evaporation and a lift-off process, covering an active area of 0.5 mm², with electrode widths of 5 μm and an interelectrode spacing of 12.5 μm. Two device configurations were prepared: a metal–insulator–semiconductor–insulator–metal (MISIM) structure with the HfO₂ insulating layer and a metal–semiconductor–metal (MSM) structure without it.

For UVC illumination, a monochromator (CS 130B-1-MC, Newport Corporation, Irvine, CA, USA) paired with a xenon arc lamp (6269, Newport Corporation, USA) was used, and the wavelength-specific intensity was measured with an optical power and energy meter (2938-R, Newport Corporation, USA). Current–voltage (I–V) characteristics were assessed using a precision source/measurement unit (B2912B, Keysight, Santa Rosa, CA, USA). The crystal quality was analyzed by high-resolution X-ray diffraction (HR-XRD) on a SmartLAB system (Rigaku), and the surface chemistry was characterized via X-ray photoelectron spectroscopy (XPS) using a VersaProbe III system (ULVAC-PHI). A transmission electron microscope (TEM) was used for the crystal structure (EM-ARM200F, JEOL, Ltd., Tokyo, Japan) at 200 keV. The selected area electron diffraction (SAED) pattern was analyzed using the Gatan Digital Micrograph^® software (version 3.6.0, AMETEK, Inc., Berwyn, PA, USA).

3. Results

Figure 1a illustrates a schematic of the fabricated device and a 3D schematic diagram of the α-Ga₂O₃ lattice structure. Figure 1b presents the optical transmittance spectrum and Tauc plot of the heteroepitaxial α-Ga₂O₃ layer, used to determine its optical bandgap. The results show that the α-Ga₂O₃ film fully absorbs light with wavelengths below 250 nm, and its optical bandgap was calculated to be approximately 5.22 eV, derived from the Tauc plot using Equation (1).

(αhν)² = A(hν − Eg)

(1)

This value is consistent with the previously reported bandgaps of α-Ga₂O₃ [18,20,22]. In addition, Figure 1c presents the XRD measurements, revealing two distinct diffraction peaks corresponding to the Al₂O₃ (0006) and α-Ga₂O₃ (0006) planes. These findings confirm that the α-Ga₂O₃ film grew epitaxially along the (0006) direction on the sapphire (Al₂O₃) substrate. In Figure 1d, the rocking curve of α-Ga₂O₃ exhibits a full width at half maximum (FWHM) of 35.6 arcseconds, which is consistent with previously reported values for α-Ga₂O₃ [22,25].

The cross-view TEM image in Figure 1e shows that α-Ga₂O₃ was preferentially grown in the (0006) direction along the c-axis of the sapphire substrate, consistent with the XRD results. The ideal d-spacing of α-Ga₂O₃ (0006) is 2.238 Å, whereas the measured d-spacing from the SAED pattern is 2.215 Å, showing evidence of compressive strain. This has been attributed to the residual stresses caused by the lattice mismatch between the sapphire substrate and heteroepitaxial α-Ga₂O₃ layer, which is generally observed in a heteroepitaxial growth of α-Ga₂O₃ on a sapphire substrate [26].

Figure 2 investigates the deposition of HfO₂ via X-ray photoelectron spectroscopy (XPS) and assesses the impact of ozone supplied during atomic layer deposition (ALD) on oxygen vacancies. Figure 2a presents the Ga 3d spectrum for the MSM structure, showing a peak at a binding energy of 20.1 eV. In contrast, Figure 2b reveals additional Hf 4f peaks at 16.9 eV and 18.5 eV in the MISIM structure, confirming successful HfO₂ deposition [27,28]. Figure 2c,d display the O 1s spectra, deconvoluted into two peaks at 530.6 eV (Ga-O bonds) and 532.1 eV (oxygen vacancies). The oxygen vacancy ratio, calculated from the peak areas, was 28.8% for the MSM structure and 23.1% for the MISIM structure, indicating a 5.7% reduction attributed to the oxygen vacancy treatment effect from the ozone during ALD. Oxygen vacancies are known to impair photodetector performance by increasing the dark current [29]. While oxygen annealing typically requires temperatures above 600 °C for a significant effect, the ALD process at 360 °C yielded only modest vacancy reduction [30,31].

Figure 3a presents the I–V characteristics of the α-Ga₂O₃ MSM and MISIM devices under dark and UVC illumination (235 nm) conditions. At 20 V, the dark currents were 73.3 pA for the MSM device, 6.61 pA for the MISIM device with 1 nm HfO₂, and 5.27 pA for the 2 nm HfO₂ device, indicating that HfO₂ layers effectively reduce the dark current. The decrease in the dark current is attributed to suppressed electron injections due to reduced tunneling through the additional insulating HfO₂ layer. Under UVC illumination at 20 V, the photocurrents were 1.14 μA (MSM), 2.3 μA (1 nm HfO₂ MISIM), and 79.4 nA (2 nm HfO₂ MISIM), with the 1 nm HfO₂ MISIM device surpassing the MSM device. The observed enhanced photocurrent is presumably attributed to reduced trap states at the HfO₂/α–Ga₂O₃ interface, not due to a change in Schottky barrier height. The 2 nm HfO₂ MISIM device showed an unstable photocurrent below 8 V, attributed to trap states at the HfO₂/Ga₂O₃ interface, which immobilize electrons in the Ga₂O₃ layer at low voltages. However, at voltages above 8 V, the traps become sufficiently filled, allowing electrons to escape under UVC illumination, resulting in an increased current [32,33]. Figure 3b compares the responsivity rejection ratio as a function of the applied voltage at wavelengths of 235 and 400 nm, highlighting the peak voltages. Figure 3c compares these wavelengths under the peak voltage, calculated using Equation (2):

$$R=(I_p-I_d)/ P\times S_{\mathit{device}}$$

(2)

where I_p is the channel current under illumination, I_d is the dark current, P is the incident light intensity, and S_device is the effective irradiated area. The rejection ratios (R₂₃₅/R₄₀₀) were 1.1 × 10⁴ (MSM), 1.8 × 10⁴ (1 nm HfO₂ MISIM), and 7.6 × 10³ (2 nm HfO₂ MISIM), with the 1 nm HfO₂ MISIM device showing high responsivity at 235 nm and low responsivity at 400 nm, confirming enhanced UVC selectivity. Figure 3d shows the response current during UVC switching on/off at 235 nm under a bias of 20 V. The rise time (τ_r) and decay time (τ_d), defined as the time required for the current to change from 10% to 90% and from 90% to 10% of its final value, respectively, were measured for the MSM, as well as the 1 and 2 nm HfO₂ MISIM devices as 4.51 s, 4.48 s, and 27.32 s for τ_r and 0.29 s, 0.15 s, and 0.45 s for τ_d, respectively. The 1 nm HfO₂ MISIM device’s faster τ_r and τ_d resulted from the reduced trapping of photogenerated holes at the surface states. Conversely, the 2 nm HfO₂ MISIM device’s thicker layer lowered the tunneling probability, slowing carrier transport and extending response times [34].

Figure 4a,b show the band diagrams of the MSM and MISIM structures, respectively. Typically, an insulating HfO₂ layer suppresses tunneling, reducing the current, as evidenced by the lower dark currents in the MISIM devices shown in Figure 3a. However, when a 1 nm HfO₂ layer is used under UVC illumination (energy exceeding the α-Ga₂O₃ bandgap), the MISIM device exhibits an increased photocurrent. As shown in Figure 4b, the HfO₂/Ga₂O₃ interface in the MISIM structure has fewer traps than the Ga₂O₃/Ti/Au interface in the MSM structure, enhancing the photocurrent by mitigating trap-related losses. This improvement is expected to outweigh the current reduction owing to tunneling suppression [35,36]. However, increasing the HfO₂ thickness to 2 nm reduces the photocurrent, as the lower tunneling probability dominates over the benefits of reduced traps. For a dark current, where no UVC energy is present, even a 1 nm HfO₂ layer effectively suppresses tunneling, resulting in a significantly lower dark current in the MISIM structure compared to the MSM structure.

Figure 5a presents the I–V measurements at 300 °C. Below 8 V, the 1 nm HfO₂ MISIM device exhibited a higher current than the MSM device, which is attributed to reduced interface traps and moderated tunneling suppression. Figure 5b compares the responsivity at 20 V as a function of temperature, showing an increase with temperature. This behavior can be attributed to thermal excitation, which affects both I_p and I_d. However, while the increased I_d is primarily attributed to thermally generated carriers, the increased I_p is ascribed not only to the thermally generated carriers but also to enhanced photoexcitation at elevated temperatures, which further facilitates the de-trapping of carriers from trap states. Consequently, more carriers become available for conduction under illumination, leading to an increased photocurrent and enhanced responsivity at an elevated temperature [37]. Figure 5c illustrates PDCR with variable temperatures for UV photodetectors based on Ga₂O₃ and various other materials [3,4,10,15,19,22,28,31,33,38,39,40]. Most previously reported works have included characterizations at room temperature, and characterizations at elevated temperatures are not investigated enough. Our work demonstrates insertion of the ultra-thin HfO₂ interlayer-enabled high PDCR even under high-temperature conditions. Figure 5d shows the transient response at 300 °C. The τ_r and τ_d for the MSM and 1 and 2 nm MISIM devices were 0.47 s, 1.21 s, and 1.39 s and 1.16 s, 1.05 s, and 1.67 s, respectively. At 300 °C, the increased thermal energy enabled electrons in the MSM device to escape interface traps and enter the conduction band more readily, resulting in a shorter τ_r compared with the 1 nm HfO₂ MISIM device. However, the decay times were longer than those at room temperature because of the increased persistent photoconductivity (PPC) effect. PPC refers to the phenomenon in which photoconductivity persists even after the light source has been removed. This effect is commonly observed in oxide semiconductors with oxygen vacancies and a high density of defects [3,37,41].

Table 1. Performance comparison of photodetectors fabricated on Ga₂O₃ thin films.

Phase	Grown	Detection	PDCR	Responsivity	Dark Current	Ref
Alpha	HVPE	235 nm	8.7 × 103 (@20 V)	1.38 A/W (@8 V)	73 pA (@20 V)	This work (w/o)
Alpha	HVPE	235 nm	7.3 × 104 (@20 V)	0.68 A/W (@4 V)	5.5 pA (@20 V)	This work (1 nm)
Alpha	HVPE	235 nm	5.2 × 102 (@20 V)	0.17 A/W (@13 V)	5.2 pA (@20 V)	This work (2 nm)
Beta	PLD	254 nm	14 (@10 V)	30.45 A/W (@10 V)	4.2 nA	[2]
Beta	PLD	255 nm	7.1 × 103 (@10 V)	0.74 A/W (@10 V)	0.32 nA	[3]
Beta	MBE	236 nm	103 (@20 V)	1.5 A/W (@4 V)	4 nA (@20 V)	[10]
Beta	RF sputtering	254 nm	105	303 A/W	10 pA (@20 V)	[11]
Amorphous	Sputtering	254 nm	2.6 × 107 (@5 V)	3.2 × 104 A/W (@ 5V)	51 pA (@5 V)	[42]
Epsilon	MOCVD	254 nm	2.4 × 107	639 A/W (@10 V)	4.6 pA	[43]
Epsilon	MOCVD	252 nm	1.0 × 103 (@15 V)	0.38 A/W (@15 V)	20 nA (@15 V)	[44]

summarizes the key performance metrics of Ga₂O₃ thin-film-based photodetectors, showing that the performance varies significantly depending on the Ga₂O₃ crystal phase and growth method [2,3,10,11,42,43,44]. The MISIM α-Ga₂O₃ device developed in this study demonstrated an enhanced photocurrent due to the reduced trap states at the HfO₂/α-Ga₂O₃ interface and a suppressed dark current by the insulating HfO₂ layer. These attributes resulted in both increased PDCR and rejection ratios that confirmed enhanced UVC selectivity. Furthermore, the use of the HVPE growth process presents strong potential for scalable production. Although further research is needed to improve film quality and device performance, this study offers a promising approach for the application of a heteroepitaxial α-Ga₂O₃ layer in UVC photodetectors.

4. Conclusions

In conclusion, this study investigated the impact of HfO₂-layer presence and thickness on the optical and electrical performance of heteroepitaxial α-Ga₂O₃ layer-based UVC photodetectors. The structural and compositional characterizations of α-Ga₂O₃ were thoroughly analyzed using HR-XRD, XPS, TEM, and SAED. I–V measurements revealed that MISIM devices with 1 nm and 2 nm HfO₂ layers achieved significantly lower dark currents than the MSM device, owing to the tunneling suppression induced by the HfO₂ layer. Under 235 nm UVC illumination, the 1 nm HfO₂ MISIM device delivered the highest photocurrent (2.3 μA at 20 V), driven by reduced surface traps, while the 2 nm HfO₂ MISIM device showed a lower photocurrent due to increased tunneling resistance. In addition, the 1 nm HfO₂ MISIM device exhibited the highest responsivity (1.8 × 10⁴ A/W), outperforming the MSM (1.1 × 10⁴ A/W) and 2 nm HfO₂ MISIM (7.6 × 10³ A/W) devices, indicating enhanced UVC selectivity and a reduced UVC on/off response time. At 300 °C, the 1 nm HfO₂ device exhibited a slight dark current increase from thermal-assisted tunneling but maintained superior responsivity compared to the MSM device. Overall, the 1 nm HfO₂ layer optimized the photoresponse by minimizing traps and ensured robust performance at high temperatures, establishing α-Ga₂O₃-based MISIM photodetectors as highly effective for UVC detection in demanding environments.