Defect-Mediated Threshold Voltage Tuning in β-Ga₂O₃ MOSFETs via Fluorine Plasma Treatment

Abstract

We demonstrate high-performance MOSFETs on β-Ga₂O₃ films grown by plasma-assisted molecular beam epitaxy (PA-MBE). The high crystalline quality of the β-Ga₂O₃ epilayer was confirmed by X-ray diffraction and atomic force microscopy. An optimized CF₄-plasma treatment was employed to introduce fluorine (F) into the near-surface region, effectively suppressing donor-like states. The resulting devices exhibit an ultralow off-state current of 1 × 10⁻⁹ mA/mm and a stable on/off ratio of 10⁵. A controllable positive threshold voltage shift up to +12.4 V was achieved by adjusting the plasma duration. X-ray photoelectron spectroscopy indicates that incorporated F atoms form F–Ga-related bonds and compensate oxygen-related donor defects. Sentaurus TCAD simulations reveal reduced near-surface charge and a widened depletion region, providing a physical explanation for the experimentally observed increase in breakdown voltage from 453 V to 859 V. These results clarify the role of fluorine in modulating surface defect states in PA-MBE β-Ga₂O₃ and demonstrate an effective route for threshold-voltage engineering and leakage suppression in Ga₂O₃ power devices.

1. Introduction

With the continuous advancement of power-device scaling driven by demands for high power density and miniaturization, silicon (Si)-based power devices are confronting fundamental physical limitations. The relatively low critical electric field of Si (≤0.3 MV/cm) restricts further reduction in device dimensions while maintaining breakdown robustness, motivating the development of wide-bandgap semiconductor technologies [1,2,3]. Materials such as GaN, SiC, and Ga₂O₃ offer distinct advantages in this regard [4]. These materials exhibit superior Baliga’s figure of merit (BFOM), enabling the development of smaller, faster, more efficient, and reliable electronic devices [5,6].

Among these materials, β-Ga₂O₃ has attracted considerable attention due to its ultrawide bandgap (4.9 eV), extremely high critical breakdown field (8 MV/cm), and exceptional theoretical BFOM (3200) [7,8,9]. To improve the electrical performance of β-Ga₂O₃ MOSFETs, various device and material engineering techniques have been investigated, including field-plate structures, trench gate designs, ion implantation, and heterojunction engineering [10,11,12]. Post-growth surface treatments represent another effective approach for modulating interface states and improving device performance [13,14,15].

Fluorine-based treatments in particular have been widely studied. Prior reports have demonstrated that fluorine incorporation can deplete channel carriers and tune the threshold voltage in various semiconductor platforms, such as AlGaN/GaN high-electron-mobility transistors (HEMTs), amorphous InGaZnO (a-IGZO) thin-film transistors, and β-Ga₂O₃ MOSFETs [16,17,18]. Related studies on fluorine-treated Schottky diodes and edge terminations have shown leakage current reduction and breakdown voltage improvement through defect control and electric field redistribution [19,20,21]. Meanwhile, both theoretical and experimental studies have demonstrated that fluorine may act as a shallow donor and increase the carrier concentration in Ga₂O₃ films grown by different methods and in different crystal phases [22,23,24,25]. These diverse observations highlight the complex and system-dependent role of fluorine in β-Ga₂O₃, underscoring the need for a more detailed understanding of F-induced surface and interface modifications.

In this work, we apply fluorine plasma surface engineering to plasma-assisted molecular beam epitaxy (PA-MBE) grown β-Ga₂O₃ films on 2-inch sapphire substrates and demonstrate its simultaneous impact on threshold voltage tuning and breakdown enhancement in lateral MOSFETs. The F treatment effectively reduces near-surface donor-like states, enabling an ultralow $I_{off}$ of 1 × 10⁻⁹ mA/mm and a controllable positive $V_{th}$ shift up to +12.4 V by adjusting the plasma duration. XPS analysis reveals F incorporation and the formation of F–Ga-related bonding, while TCAD simulations suggest that the suppressed near-surface charge and the widened depletion region are responsible for the experimentally observed increase in breakdown voltage. These results clarify the role of fluorine in modifying surface defect states in PA-MBE β-Ga₂O₃ and provide an effective route for engineering threshold voltage and leakage in Ga₂O₃-based power devices.

2. Experimental Section

2.1. Growth of β-Ga₂O₃ Thin Films

The β-phase Ga₂O₃ films were grown using PA-MBE (SVT-35, SVT Associates, Eden Prairie, MN, USA). Firstly, the cleaned sapphire substrates were degassed in the buffer chamber at 170 °C for 4 h to remove residual moisture and surface contaminants. The substrates were then transferred into the growth chamber and annealed at 750 °C under an oxygen-rich atmosphere for 15 min to further desorb surface impurities. Subsequently, β-Ga₂O₃ films were grown for 2 h at 700 °C with a Ga source temperature of 950 °C and an oxygen flow of 1.6 sccm. The O-to-Ga flux ratio was maintained above the stoichiometric value to ensure oxygen-rich growth conditions. Such conditions have been demonstrated to suppress suboxide (Ga₂O) formation and GaO_x decomposition, leading to improved growth rate and crystalline quality of films [26,27]. The resulting β-Ga₂O₃ films exhibited high crystalline quality with an average thickness of ~300 nm, as confirmed by subsequent structural characterization.

2.2. Device Fabrication

The fabrication of β-Ga₂O₃ circular transmission line model (CTLM) and MOSFET structures was carried out using high-resolution ultraviolet lithography. Prior to device fabrication, the as-grown β-Ga₂O₃ films were cleaned by sequential ultrasonic treatment in acetone, ethanol, and deionized water (5 min each) to remove surface organic contaminants. F-plasma treatment was conducted in an inductively coupled plasma reactive ion etching (ICP-RIE) chamber using CF₄/O₂ (20/5 sccm) at 5.0 mTorr, with an ICP coil power of 200 W and an RF platen power of 100 W at a chuck temperature of 300 K. The treatment was applied across the entire wafer area. The treatment duration was varied to investigate the time-dependent effects. For MOSFETs fabrication, a 120 nm thick SiO₂ gate dielectric was subsequently deposited by ICP chemical vapor deposition (ICP-CVD) using N₂O/SiH₄ gas flows of 20/5 sccm at 2.0 mTorr. The deposition was carried out at an ICP power of 1000 W, an RF power of 20 W, and a DC bias of 80 V at 575 K. No post-deposition annealing was performed. The source/drain (S/D) contact windows were subsequently defined by maskless lithography and opened using RIE. Ti/Au (10/100 nm) S/D electrodes were then formed through thermal evaporation followed by a lift-off process. Finally, the Ni/Au (30/100 nm) gate (G) electrodes were deposited using the same lithography and lift-off procedures.

Each β-Ga₂O₃ MOSFET featured a circular geometry, with the source electrode radius designed as 200 μm. The gate adopted an annular configuration with inner and outer radii of 210 μm and 220 μm, respectively, corresponding to a gate length ($L_G$) of 10 μm and an equivalent gate-source spacing ($L_{GS}$). The drain electrode was designed in a semi-annular layout with an inner radius of 240 μm, yielding a drain-source spacing ($L_{DS}$) of 40 μm. This symmetric configuration ensured uniform electric-field distribution and stable gate control across the active region.

2.3. Characterization

The morphology of the β-Ga₂O₃ films was examined using field-emission scanning electron microscopy (FE-SEM, S4800, Hitachi, Tokyo, Japan) and atomic force microscopy (AFM, nanoIR2-s, Anasys Instruments, Billerica, USA) to evaluate the surface topography and roughness. The crystalline structure and orientation were characterized by X-ray diffraction (XRD, Empyrean, Malvern Panalytical, Great Malvern, UK), while the chemical states and elemental composition were analyzed using X-ray photoelectron spectroscopy (XPS, Nexsa, Thermo Fisher Scientific, Waltham, MA, USA). The XPS measurements were performed with an Al Kα excitation source (hν = 1486.6 eV). The analyzer operated in constant analyzer energy (CAE) mode with a pass energy of 40 eV and a step size of 0.05 eV. Spectra were collected at a take-off angle of 90°. A dual-beam charge neutralization system, comprising low-energy electrons and low-energy ions, was employed to compensate for surface charging during the measurements.

All electrical measurements of the β-Ga₂O₃ MOSFETs were performed at room temperature (RT) under ambient conditions. The transfer and output characteristics were obtained using a semiconductor parameter analyzer (4200A-SCS, Keithley Instruments, Beaverton, OR, USA), and the breakdown voltage was determined with a power device analyzer (B1500A, Agilent Technologies, Santa Clara, CA, USA). Breakdown voltage measurements were conducted on multiple devices, and the values reported correspond to representative characteristics that reflect the consistent behavior observed among the tested samples under the same measurement conditions.

2.4. XPS Analysis

All XPS spectra were charge-corrected to the C 1s peak at 284.8 eV. Peak fitting was performed using XPS Peak software (ver. 4.1, Cabit Information Technology, Shanghai, China), while the peak identification and the charge correction were verified using Thermo Avantage software (ver. 5.9921, Thermo Fisher Scientific, Waltham, USA). Standard fitting constraints commonly applied to metal-oxide systems were used. Mixed Gaussian-Lorentzian (GL(20)) line shapes were applied to all components. Binding energies were initialized from literature values and allowed to vary within ±0.2–0.3 eV. The FWHM was constrained to 1.0–2.0 eV, with differences between related chemical states limited to <0.3 eV.

3. Results and Discussion

The as-grown 2-inch β-Ga₂O₃ wafer exhibited a transparent and optically uniform film surface on the c-plane sapphire substrate, as shown in Figure 1a. AFM and planar-view SEM images revealed a smooth surface morphology with a root-mean-square (RMS) roughness of 5.8 nm over a 10 × 10 μm² area (Figure 1b,c). The cross-sectional SEM image further confirmed a uniform film thickness of approximately 300 nm (inset in Figure 1b). XRD analysis was conducted to evaluate the crystalline quality of the β-Ga₂O₃ film (Figure 1d). Distinct diffraction peaks located at 18.9°, 38.4°, and 59.0° correspond to the (−201), (−402), and (−603) planes of β-Ga₂O₃, respectively [28]. The absence of any additional peaks indicated a single-phase β-Ga₂O₃ epilayer with well-defined crystallographic orientation. Collectively, the excellent optical uniformity, smooth surface morphology, and phase-pure crystal structure verified by XRD confirmed the successful heteroepitaxial growth of a high-quality β-Ga₂O₃ thin film on sapphire substrate via PA-MBE.

The schematic structure and fabrication process of the β-Ga₂O₃ circular transmission line model (CTLM) patterns are illustrated in Figure 2a. The optical microscopy image of the fabricated CTLM structures is shown in Figure 2b, where each device featured a fixed inner-circle radius ($r_0$ = 200 μm). The outer-circle radii ($r_i, i=1~9$) varied from 220 to 300 μm in 10 μm increments, defining the electrode gap spacing $d=r_i-r_0$ (20 to 100 μm, in steps of 10 μm). Figure 2c presents the current–voltage (I–V) characteristics of CTLMs fabricated on β-Ga₂O₃ films subjected to a 3 min F-plasma treatment. For comparison, additional I–V curves corresponding to 5 min and 7 min treatments are provided in Supplementary Material, Figure S1. In contrast to the as-grown films (Supplementary Material, Figure S2), the S/D contacts formed on F-plasma treated β-Ga₂O₃ exhibited clear ohmic behavior, indicating a substantial reduction in contact barrier height and improved carrier injection at the metal/semiconductor interface. This improvement can be attributed to the fluorine-passivation effect, which effectively reduced surface states and improved the interfacial quality between the metal and β-Ga₂O₃. As further supported by AFM measurements (Supplementary Material, Figure S3), the surface roughness of β-Ga₂O₃ continuously decreased with increasing plasma treatment time, indicating smoother morphology and reduced surface contamination. The smoother surface mitigated impurity scattering and consequently enhanced carrier mobility. To exclude the possibility of plasma-induced structural degradation, the F-plasma etching rate was also evaluated and found to be approximately 5.4 nm/min (Supplementary Material, Figure S4). Given the 300 nm thickness of the β-Ga₂O₃ layer, this slow etching rate ensured that the plasma treatment exerted a negligible impact on the overall film thickness and structural integrity.

The total resistance ($R_T$) as a function of $d$ exhibited a clear linear dependence (Figure 2d), consistent with the behavior expected from the CTLM structure. The relationship can be expressed as [29]:

$$R_T={\displaystyle \frac{R_{sh}}{2\pi r_0}}\left(d+{2L}_T\right)$$

(1)

where $R_{sh}$ represents the sheet resistance and $L_T$ is the transfer length. From the linear fitting, $R_{sh}$ and $L_T$ were extracted to be 14.3 MΩ/square and 3.9 μm, respectively. The specific contact resistivity (${\rho }_c$) was subsequently calculated according to [30]:

$${\rho }_c=R_{sh}{L}_T^2$$

(2)

resulting in a value of 2.2 Ω·cm². This relatively high ${\rho }_c$ value, compared with previously reported values, can be primarily attributed to the unintentional doping level of the Ga₂O₃ thin film, which limits carrier concentration near the contact interface and thereby increases contact resistance.

The fabrication flow of the β-Ga₂O₃ MOSFETs is summarized in Figure 3a, and an optical micrograph of a representative device is shown in Figure 3b. A circular source electrode was adopted to alleviate electric-field crowding commonly observed in rectangular layouts, while an annular gate electrode was employed to suppress lateral electric-field spreading and ensure electrical isolation from adjacent structures. For clarity, MOSFETs without F-plasma treatment are denoted as F0, while devices subjected to 3, 5, 7 min treatments are referred to as F3, F5, and F7, respectively.

The transfer characteristics of F0 and F3 under a drain bias ($V_{DS}$) of 10 V are presented in Figure 3c,e. The untreated device (F0) exhibited a relatively large off-state current ($I_{off}$) of ~1 × 10⁻⁵ mA/mm, resulting in a limited on/off current ratio ($I_{on}$/$I_{off}$) of ~10¹. In contrast, F3 demonstrated clear gate-voltage modulation with an $I_{off}$ suppressed to 1 × 10⁻⁹ mA/mm. The $I_{off}$ was reduced by nearly four orders of magnitude, leading to an $I_{on}$/$I_{off}$ exceeding 10⁵. The subthreshold swing ($SS$) was extracted to be 0.50 V/dec according to [31]:

$$SS= {\left({\displaystyle \frac{d log\left(I_{DS}\right)}{d{V}_{GS}}}\right)}^{-1}$$

(3)

where $I_{DS}$ and $V_{GS}$ represent the drain-to-source current and gate-to-source voltage, respectively. The relatively small $SS$ value indicates efficient gate electrostatic control and a reduced interfacial trap density, consistent with the surface passivation effect induced by F incorporation.

The output characteristics of F0 and F3 are shown in Figure 3d,f. For F0, both the saturation current ($I_{D,sat}$) and linear region were severely limited, and a noticeable negative current appeared at low drain biases ($V_{DS}$ < 5 V), which may arise from drain-to-gate electron transport through defective interface states. By comparison, F3 exhibited typical field-effect transistor behavior, with linear $I_{DS}$–$V_{DS}$ curves and minimal current crowding at $V_{DS}$ < 5 V, confirming the improved ohmic contact achieved after F-plasma treatment.

The influence of F-plasma treatment duration on the transfer and breakdown characteristics of β-Ga₂O₃ MOSFETs was systematically examined to clarify the role of F-induced surface modification. As shown in Figure 4a, the $I_{off}$ remained consistently at the ultralow level of 10⁻⁹ mA/mm with increasing treatment time, demonstrating the robustness of the surface passivation. It is noted that such a low current is close to the measurement limit of the parameter analyzer; therefore, the slight fluctuations observed in $I_{off}$ are attributed to instrument-induced noise.

Meanwhile, the $V_{th}$ exhibited a monotonic positive shift with increasing F-plasma treatment time, which can be attributed to the progressive charge compensation at the SiO₂ interface and the depletion of near-surface donor-like defects. Despite the significant $V_{th}$ shift, the devices have not yet achieved true enhancement-mode (E-mode) operation, implying that further improvements in epitaxial quality and device architecture will be required to achieve reliable E-mode behavior.

In addition, the dependence of the breakdown voltage ($V_{br}$) on F-plasma treatment duration is summarized in Figure 4b. For the unintentionally doped β-Ga₂O₃ films used in this study, $V_{br}$ was defined as the $V_{DS}$ at which the $I_{DS}$ reached 0.1 mA/mm. The measured $V_{br}$ increased significantly from 453 V (F0) to 859 V (F7), representing a nearly twofold enhancement. This improvement in blocking capability is attributed to the widening of the depletion region and the reduction of leakage-related defect states at the β-Ga₂O₃/SiO₂ interface induced by optimized F incorporation. A qualitative assessment of device robustness was also performed (Supplementary Material, Figure S4). The fluorinated devices maintained stable $I_{off}$ and $I_{on}$/$I_{off}$ for up to 90 days in air, with only small $V_{th}$ variations, while more detailed trends are discussed in the Supplementary Material.

Table 1. Performance of β-Ga₂O₃ MOSFETs with varied F-plasma treatment durations.

Performance	F0	F3	F5	F7
$I_{on}/I_{off}$	10	105	105	105
$SS$ (V/dec)	11	0.50	0.66	0.63
$V_{th}$ (V)	-	−25.8	−22.8	−13.4
$V_{br}$ (V)	453	667	778	859

summarizes the extracted electrical parameters of β-Ga₂O₃ MOSFETs with different F-plasma treatment durations, highlighting the pronounced improvements in subthreshold swing, threshold voltage stability, off-state leakage, and breakdown voltage. These results collectively confirm that controlled F incorporation effectively passivates donor-like surface defects and improves the interface quality, leading to enhanced device performance.

To elucidate the chemical states and bonding configurations induced by F-plasma treatment, comprehensive XPS measurements were performed on β-Ga₂O₃ films. All spectra were calibrated to the adventitious C 1s peak at 284.8 eV [32,33] (Supplementary Material, Figure S5). The survey spectra revealed a distinct F 1s peak at 686.1 eV and an F KLL peak at 833 eV in the fluorinated samples, confirming successful fluorine incorporation (Figure 5a) [34]. Figure 5b summarizes the evolution of F incorporation, including the F 1s peak intensity and its atomic percentage. While the absolute F 1s intensity increases monotonically with exposure time, the F atomic fraction reaches a small maximum at an intermediate exposure duration (5 min) and decreases slightly after prolonged treatment (7 min). Figure 5b further presents the corresponding Ga/O and Ga/F ratios, both of which vary inversely with the F atomic fraction. As F atomic percentage increases, Ga/O and Ga/F decrease accordingly. This behavior indicates that fluorine substitutes a portion of lattice oxygen or occupies oxygen-vacancy sites [35], effectively altering the local chemical environment. As for the slight reduction in the F atomic fraction at longer exposure, it may originate from plasma-induced surface etching or from increased adsorption of –OH groups that partially dilute the detected F concentration. These possibilities require further investigation.

To clarify the detailed chemical states associated with fluorine incorporation, we next examine the high-resolution F 1s, O 1s, Ga 3d, and Ga 2p spectra. As shown in Figure 6a, the pristine β-Ga₂O₃ film exhibits only a barely detectable F 1s signal near 685.0 eV. After a 3 min exposure (Figure 6b), the F 1s peak becomes prominent and can be deconvoluted into two components centered at 685.2 (F–Ga bonds) and 687.0 eV (F–OH interactions), respectively [36]. This observation indicates simultaneous fluorine incorporation into the lattice and modification of the surface hydroxyl environment.

Consistent trends are observed in the O 1s spectra (Figure 6c,d). Both pristine and F-plasma treated samples show two components at ~530.6 and 532.3 eV, corresponding to O_I (Ga–O bonds) and O_II (surface –OH groups), respectively [36]. Following fluorination, the –OH-related peak becomes stronger, consistent with the enhanced –OH adsorption inferred from Figure 5. This increase likely results from the high electronegativity of F atoms that promote F–HO bond formation [37]. The Ga 3d spectra (Figure 6e,f) further corroborate the defect passivation behavior implied by the decreasing Ga/O and Ga/F ratios. In the untreated film, the Ga¹⁺ component at 19.6 eV contributes significantly to the total signal. After F-plasma exposure, this Ga¹⁺ component is substantially suppressed (47.5% to 23.8%), while Ga³⁺ peak near 20.4 eV (stoichiometric Ga₂O₃) becomes dominant [38]. The transition indicates that incorporated F atoms effectively passivate oxygen vacancies within the Ga₂O₃ lattice, consistent with the observed suppression of donor-like defect states [39]. Because potential Ga–F features cannot be resolved separately in the Ga 3d region, analysis of deeper core levels is required to distinguish Ga–F contributions.

To further clarify the contribution of Ga–F bonding and assess the effective depth of fluorine incorporation, Ga 2p spectra were examined (Figure 7). As shown in Figure 7a, both the Ga 2p_3/2 and Ga 2p_1/2 peak positions gradually shift toward higher binding energies with increasing F-plasma treatment time. This observation is consistent with F incorporation, as F is known to induce pronounced chemical shifts in neighboring elements [40,41]. Deconvolution of the Ga 2p_3/2 peak (Figure 7b–d) reveals a Ga–O component at ~1118.1 eV and a Ga–F component at ~1118.8 eV [42], whose relative fraction increases with plasma duration (from 37.9% to 40.1%). Given the larger probing depth of Ga 2p relative to Ga 3d, the emergence and growth of the Ga–F component indicate that F incorporation is not confined to the extreme surface but extends into the near-surface region of the β-Ga₂O₃ lattice.

Together with the elemental-ratio evolution shown in Figure 5b, these spectral features establish a consistent chemical picture: fluorine atoms substitute for oxygen or occupy vacancy sites, passivating donor-like defects while simultaneously modifying the surface hydroxyl environment. This multichannel chemical evolution directly underpins the enhanced electrical performance observed in fluorinated β-Ga₂O₃ MOSFETs.

To further elucidate the charge-modulation mechanism in β-Ga₂O₃ MOSFETs, Sentaurus TCAD simulations were carried out based on the experimentally derived device geometry (Supplementary Material, Figure S6). Donor-like trap states associated with oxygen-vacancy-related defects were introduced into the model, with their spatial distribution adjusted to reflect the experimentally observed higher defect density near the surface region. It should be noted that the specific values used in the simulation were chosen solely for qualitative analysis, and the simulation outcomes were robust against reasonable variations in these parameters.

The simulated space-charge distributions under a bias condition of $V_{GS}$ = −30 V and $V_{DS}$ = 10 V revealed distinct differences between the pristine and F-treated devices (Figure 8a,b). Cross-sectional charge density profiles extracted along the AA’ and BB’ cutlines (dashed lines in Figure 8a,b) demonstrated a substantial reduction in peak charge density after fluorine incorporation (Figure 8c), indicating effective neutralization of donor-like defects and suppression of interfacial charge trapping. The $V_{th}$ can be expressed as [43]:

$$V_{th}={\psi }_{bi}-{\displaystyle \frac{q{N}_i{d}^2}{2{\epsilon }_s}}$$

(4)

where ${\psi }_{bi}$, $N_i$, $q$, $d$, and ${\epsilon }_s$ denote the built-in potential, trap concentration, electronic charge, depletion width, and permittivity of the channel material, respectively. According to this relationship, the reduction of $N_i$ induced by F passivation leads to a positive shift in $V_{th}$. Energy band diagrams further illustrate these effects (Figure 8d,e): in the pristine device, the potential drop is primarily confined within the SiO₂ dielectric, causing localized electric-field crowding and premature breakdown. In contrast, the F-plasma-treated device exhibits a broadened depletion region and a more uniform potential distribution across both SiO₂ and β-Ga₂O₃ layers, effectively mitigating field localization and enhancing the $V_{br}$. Finally, the simulated transfer characteristics show excellent agreement with the experimental data (Figure 8f), validating the proposed charge-modulation and field-redistribution mechanism.

To facilitate a clearer understanding of the device performance,

Table 2. Summary of some key parameters of β-Ga₂O₃ FETs fabricated on native and sapphire substrates.

${\mathit{V}}_{\mathit{b}\mathit{r}}$ (V)	${\mathit{L}}_{\mathit{D}\mathit{S}}$ (μm)	${\mathit{V}}_{\mathit{t}\mathit{h}}$ (V)	${\mathit{R}}_{\mathit{o}\mathit{n},\mathit{s}\mathit{p}}$ (mΩ·cm2)	${\mathit{I}}_{\mathit{D}\mathit{S},\mathit{m}\mathit{a}\mathit{x}}$ (mA/mm)	${\mathit{I}}_{\mathit{o}\mathit{f}\mathit{f}}$ (mA/mm)	${\mathit{I}}_{\mathit{o}\mathit{n}}$ / ${\mathit{I}}_{\mathit{o}\mathit{f}\mathit{f}}$	$\mathit{S}\mathit{S}$ (mV/dec)	Ref.
370	-	−15	-	39	10−9	1010	-	[44] *
2700	40	−12	78 kΩ·mm	0.2	10−5	105	-	[11] *
800	11.4	−18	7.41	238.1	10−9	108	86	[45]
400	40	−32	-	100	10−9	1011	210	[46]
-	70	−18	117.9 kΩ	10−1	10−7	106	-	[25]
650	15	−8	-	10−1	10−8	107	-	[47]
770	15	1.5	-	10−4	10−8	104	-
910	20	1.5	223	10−3	10−7	104	-	[48]
240	20	1.5	752	0.7	10−5	105	-
1085	15	−7.5	600	3.71	10−7	107	313	[49]
1250	15	−3.9	5820	0.5	10−8	107	351
859	40	−13.4	106	10−4	10−9	105	630	This work

*: β-Ga₂O₃ substrate.

summarizes representative parameters of β-Ga₂O₃ MOSFETs fabricated on both native and sapphire substrates. Overall, devices grown and processed on single-crystal native substrates typically exhibit superior performance, benefiting from higher film quality and lower defect densities. In contrast, MOSFETs on sapphire generally face inherent trade-offs due to heteroepitaxial constraints, which degrade material quality and limit electrical performance.

Compared with previously reported sapphire-based devices, our MOSFETs demonstrate an ultralow $I_{off}$ and achieve a notable $V_{br}$ even without employing field-plate structures. However, the presence of unintentional doping and residual defects restricts the on-state current and weakens gate electrostatic control, leading to a relatively modest $I_{on}$/$I_{off}$ and a larger $SS$. Further improvements in epitaxial film quality, intentional doping control, gate oxide engineering, and the incorporation of field-plate designs will be essential for pushing the device performance toward its full potential.

4. Conclusions

In conclusion, we successfully demonstrated β-Ga₂O₃ MOSFETs featuring ultralow off-state current and enhanced subthreshold characteristics through F-plasma treatment. The optimized devices achieved a low $I_{off}$ of 1 × 10⁻⁹ mA/mm and a positive $V_{th}$ shift of 12.4 V. Furthermore, the breakdown voltage was significantly improved from 453 V to 859 V. XPS analysis confirms that fluorine incorporation suppresses near-surface donor-like defects through the formation of Ga–F bonds and reduction of oxygen-vacancy-related states. Complementary TCAD simulations qualitatively reproduce the resulting charge-density reduction and electric-field redistribution, consistent with the experimentally observed positive $V_{th}$ shift and enhanced $V_{br}$. These findings not only provide fundamental insights into the mechanisms of fluorine incorporation in β-Ga₂O₃, but also establish a viable pathway for advancing the performance and reliability of Ga₂O₃-based power electronic devices.