Optimization of CuO_x/Ga₂O₃ Heterojunction Diodes for High-Voltage Power Electronics

Abstract

This study optimizes the CuO_x/Ga₂O₃ heterojunction diodes (HJDs) by tailoring the structural parameters of CuO_x layers. The hole concentration in the sputtered CuO_x was precisely controlled by adjusting the Ar/O₂ gas ratio. Experimental investigations and TCAD simulations were employed to systematically evaluate the impact of the CuO_x layer dimension and hole concentration on the electrical performance of HJDs. The results indicate that increasing the diameter dimension of the CuO_x layer or tuning the hole concentration to optimal values significantly enhances the breakdown voltage (V_B) of single-layer HJDs by mitigating the electric field crowing effects. Additionally, a double-layer CuO_x structure (p⁺ CuO_x/p⁻ CuO_x) was designed and optimized to achieve an ideal balance between the V_B and specific on-resistance (R_on,sp). This double-layer HJD demonstrated a high V_B of 2780 V and a low R_on,sp of 6.46 mΩ·cm², further yielding a power figure of merit of 1.2 GW/cm². These findings present a promising strategy for advancing the performance of Ga₂O₃ devices in power electronics applications.

1. Introduction

Beta-gallium oxide (β-Ga₂O₃) has emerged as a highly promising ultra-wide bandgap material with great potential in the field of power electronics. Due to its large bandgap of 4.8 eV, high critical electric field of 8 MV/cm, and high Baliga’s figure of merit, Ga₂O₃ is an exceptional candidate for high-voltage applications, outperforming materials like silicon (Si), gallium nitride (GaN), and silicon carbide (SiC) [1,2,3]. In addition, the availability of melt-grown substrate production also makes it more economically desirable [4,5]. Over the past decade, the high-performance Ga₂O₃ devices, such as Schottky barrier diodes (SBDs), have been successfully demonstrated [6,7]. However, the absence of p-type conductivity remains a significant obstacle to the development of Ga₂O₃ bipolar power electronics [8]. To address this challenge, the p-n heterojunction diodes (HJDs) that integrate n-type Ga₂O₃ with alternative p-type materials have been explored as a promising solution.

In this regard, various p-type oxide semiconductors, including cuprous oxide (Cu₂O) [9], nickel oxide (NiO) [10,11], tin oxide (SnO) [12], and chromic oxide (Cr₂O₃) [13,14], have been investigated for developing Ga₂O₃ HJDs. Among these, Cu₂O stands out due to its non-toxic, low-cost, and bandgap energy of 2.1 eV. Cu₂O films are particularly promising as p-type active layers in heterostructures, benefiting from their high hole mobility [15]. It has been reported that the p-type behavior of Cu₂O is attributed to the negatively charged copper vacancies or nitrogen doping [16]. To date, the p-Cu₂O/n-Ga₂O₃ HJDs with a breakdown voltage (V_B) of 1490 V and a specific on-resistance (R_on,sp) of 8.2 mΩ·cm² have been successfully demonstrated, firstly making Cu₂O as a viable candidate for integration with Ga₂O₃ in power devices [9]. Subsequent studies have explored the electrical characteristics of Cu₂O/Ga₂O₃ HJDs, including those prepared by sputtering Cu₂O films at 600 °C, which achieved a V_B of 1015 V and an R_on,sp of 8.32 mΩ·cm² [17]. Additionally, the ampere-class Cu₂O/Ga₂O₃ trench heterojunction barrier Schottky diodes have shown potential in high-voltage, high-current applications, demonstrating a maximum current up to 3.5 A, a leakage current of 1.3 × 10⁻³ A/cm², and a V_B of 986 V [18]. Whereas Cu₂O/Ga₂O₃ HJDs still suffer from poor device performance and require more comprehensive guidelines for design optimization.

Previous studies have shown that the fabrication process of p-type layers plays a crucial role in determining the heterojunction quality and, consequently, the electrical performance of the HJDs. The design optimization of NiO/Ga₂O₃ HJDs has been extensively explored [19]. The hole concentration in NiO layers was controlled by adjusting the flow rate of argon/oxygen (Ar/O₂). Additionally, the impact of NiO layer geometry on the HJDs’ performance was investigated. The improvements in device performance were achieved by either enlarging the diameter dimensions of the NiO layer or tuning the hole concentration to alleviate the electric field crowding effect [20]. However, a precise design strategy for the Cu₂O/Ga₂O₃ heterojunctions is still unclear. Therefore, achieving further enhancements in the performance of Cu₂O/Ga₂O₃ HJDs requires an elaborate and detailed optimization approach.

In this work, we present an optimization approach for the CuO_x/Ga₂O₃ HJDs by adjusting the structural parameters of the CuO_x layer, including its hole concentration and geometric dimensions. A combination of experimental studies and technology computer-aided design (TCAD) simulations was conducted to evaluate the effects of these parameters on the performance of HJDs. The results indicate that adjusting the hole concentration and expanding the diameter dimensions of the CuO_x layer contribute to a lower R_on,sp and a higher V_B. Moreover, CuO_x/Ga₂O₃ HJDs featuring a double-layer CuO_x structure were fabricated and optimized. This double-layer CuO_x structure design achieved a great balance between the V_B and R_on,sp. The optimized double-layer CuO_x/Ga₂O₃ HJDs exhibited a high V_B of 2780 V and a low R_on,sp of 6.46 mΩ·cm², further resulting in a power figure of merit (PFOM) of 1.2 GW/cm². These findings suggest that the optimized p-type CuO_x layers offer an effective strategy for achieving high-performance Ga₂O₃-based bipolar device in power electronics applications.

2. Device Structure and Fabrication Process

To investigate the hole concentration in Cu₂O films, a series of Cu₂O films were deposited on sapphire substrates using radio frequency (RF) magnetron sputtering technique. The sapphire substrates were pre-cleaned by acetone, isopropyl alcohol, and deionized water to remove surface impurities and organic residues. The target material is high-purity Cu₂O (99.99%). The RF sputtering power is 100 W, and the sputtering time sets at 30 min. The gas flow rate of Ar is 50 sccm, and the flow rate of O₂ is varied from 0 sccm to 4 sccm to modulate the hole concentration in the CuO_x films. The deposition conditions and results are summarized in

Table 1. The results of CuO_x films deposited under various sputtering conditions.

Power (W)	Ar (sccm)	O2 (sccm)	Hole Concentration (/cm3)	Mobility (cm2/V·s)	Thickness (nm)
100	50	0	1.8 × 1016	3.29	85.02
		1	3.7 × 1017	0.94	89.68
		2	2.7 × 1019	0.28	76.70
		3	3.9 × 1019	0.12	68.94
		4	1.4 × 1020	0.05	60.09

. In addition, an untreated sapphire sheet served as a control sample for comparison. The hole concentration and mobility of the CuO_x films were measured using an HL5500PC resistor Hall Effect Measurement System. The surface morphology and root mean square (RMS) surface roughness were detected by Atomic Force Microscope (AFM, MFP-3D SA, Asylum Research, Santa Barbara, CA, USA). Qualitative analysis of CuO_x films using Raman spectroscopy (LabRAM HR Evolution, HORIBA, Kyoto, Japan).

The Ga₂O₃ epitaxial wafer used in this study was purchased from Novel Crystal Technology, Inc, Japan. It features a 10 μm Si-doped n-type drift layer and 650 μm Sn-doped bulk substrate. Mesa isolation was performed using inductively coupled plasma (ICP) etching with BCl₃ as the etchant and followed by a 10 min piranha solution treatment (H₂SO₄:H₂O₂ = 4:1) to remove surface impurities. The depth of the mesa isolation is 650 nm. The distance between the mesa edge and the anode or p⁻ CuO_x layer is 5 μm. A Ti/Au (20/100 nm) metal stack was deposited on the backside of the wafer by e-beam evaporation and annealed at 510 °C for 1 min in an N₂ ambient to create the backside ohmic contacts. Then, the double-layer CuO_x films were then sputtered onto the Ga₂O₃ drift layer, followed by a lift-off process at room temperature. Finally, a Ni/Au (50/50 nm) metal stack was subsequently deposited on the CuO_x layer. Moreover, all the Ga₂O₃ HJDs were fabricated with the identical anode metal (radius = 50 μm) to ensure the consistent forward current. In this study, the room temperature forward current–voltage and breakdown voltage measurements were conducted using a Keithley-4200-SCS (TEKTRONIX, INC., Beaverton, ON, USA) and Agilent B1505A (Keysight Technologies, Santa Rosa, CA, USA), respectively.

3. Results and Discussion

3.1. Characterization of CuO_x Films

Table 1 summarizes the deposition parameters for CuO_x films, which were deposited for 30 min under varying deposition conditions. The thickness of the CuO_x films initially shows an increase but then subsequently decreases, suggesting that an appropriate quantity of O₂ content can boost the deposition rate of the film, while an excessive amount of O₂ can suppress it. The hole concentration in the CuO_x films rises with increased O₂ content, likely due to a reduction in the Cu⁺/Cu²⁺ ratio. Since CuO has a lower formation energy for Cu vacancies compared to that of Cu₂O, it exhibits a higher intrinsic carrier density than Cu₂O [21]. This result implies that adjusting the Cu⁺/Cu²⁺ ratio could be an effective approach to modulating the hole concentration in CuO_x films. The mobility significantly decreases with the increase in hole concentration, primarily due to the enhanced ionized impurity scattering and electron-hole scattering at a high hole concentration [22].

The morphology of the CuO_x films was measured by an AFM over a 5 × 5 μm² scanning area. Figure 1 compares the three-dimensional surface morphology of the CuO_x films deposited under varying O₂ fluxes. Figure 1a shows the bare sapphire substrate with a root mean square (RMS) surface roughness of 0.08 nm. The RMS values for films in Figure 1b,c are 0.11 nm and 0.12 nm, respectively, indicating that the introduction of a small amount of O₂ can enhance the deposition rate of CuO_x films without significantly impacting their surface morphology. As the O₂ content continuous to increase, however, the deposition rate of CuO_x film gradually declines. It is evident that excessive O₂ can impede film growth. Additionally, the surface roughness of CuO_x films increases notably with higher O₂ fluxes, with RMS values in Figure 1d–f of 0.25 nm, 0.26 nm, and 0.31 nm, respectively. These results highlight that while a controlled quantity of O₂ can be beneficial, excessive O₂ adversely affects both the deposition rate and surface smoothness of the films.

Figure 2 depicts the Raman spectra of various CuO_x films sputtered on sapphire substrates, with the spectrum of the bare sapphire substrate serving as a reference. At Ar/O₂ ratio of 50:0 and 50:1, the characteristic peaks were observed at 150 cm⁻¹, 210 cm⁻¹, and 630 cm⁻¹, corresponding to the Cu₂O. Additionally, the peak at 300 cm⁻¹, characteristic of CuO, indicates that Cu₂O is gradually oxidized to CuO as O₂ flow increases. Meanwhile, the peak at 540 cm⁻¹ is attributed to the Cu₄O₃, a phase containing copper in both +1 and +2 valence states [23]. As the O₂ introduced increases, the disappearance of the Cu₄O₃ peak suggests that the films are nearly completely oxidized to CuO [24]. These results demonstrate that samples with Ar/O₂ ratios of 50:0 and 50:1 contain a mixture of Cu₂O, CuO, and Cu₄O₃, whereas the samples with higher oxygen content consist predominantly of CuO. These findings align with the AFM analysis, which showed changes in surface morphology with the increasing O₂ content.

3.2. Ga₂O₃ HJDs with a Single-Layer CuO_x Structure

In this section, we would mainly discuss the effects of the hole concentrations and geometric dimensions of the CuO_x layer on the performance of HJDs. Figure 3 shows various vertical CuO_x/Ga₂O₃ HJDs with a single-layer CuO_x structure on the same wafer.

Table 2. Device parameters for the Ga₂O₃ HJDs with a single-layer CuO_x structure.

Sample ID	CuOx Structure	Hole Concentration (/cm3)	Radius of p+ CuOx (μm)	Radius of p− CuOx (μm)
${\mathrm{p}}_1^{+}$	single-layer	3.9 × 1019	60	/
${\mathrm{p}}_2^{+}$	single-layer	3.9 × 1019	70	/
${\mathrm{p}}_3^{+}$	single-layer	3.9 × 1019	80	/
${\mathrm{p}}_1^{-}$	single-layer	3.7 × 1017	/	60
${\mathrm{p}}_2^{-}$	single-layer	3.7 × 1017	/	70
${\mathrm{p}}_3^{-}$	single-layer	3.7 × 1017	/	80

summarizes the key parameters for these Ga₂O₃ HJDs with single-layer CuO_x. The p⁺ CuO_x layer had a high hole concentration of 3.9 × 10¹⁹ cm⁻³ (Ar/O₂ = 50:3), contrasting with the p⁻ CuO_x O layer, which has a lower hole concentration of 3.7 × 10¹⁷ cm⁻³ (Ar/O₂ = 50:1). Figure 4a illustrates the forward characteristics of all single-layer p⁺ and p⁻ CuO_x HJDs with various dimensions. The ideal factor of p⁺ and p⁻ CuO_x HJDs are around 1.8 and 1.6, respectively, indicating that Shockley–Read–Hall (SRH) recombination may be the predominant mechanism. Due to the higher hole concentration and improved electrical conductivity, the HJDs with the p⁺ CuO_x layer demonstrate lower R_on,sp than those with the p⁻ CuO_x layer. Additionally, as the diameter dimensions of the CuO_x layer increase, the R_on,sp tends to decrease. This is attributed to the larger junction area that facilitates carrier transport [25].

Figure 4b depicts the semi-log plots of the forward I–V characteristics, indicating that HJDs with p⁺ CuO_x layers have a higher turn-on voltage (V_ON) than those with p⁻ CuO_x layers. The increased hole concentration in the CuO_x layer narrows the depletion region, resulting in a stronger built-in electric field and requiring a higher external voltage to initiate conduction, thereby increasing the V_ON [26]. The reverse I–V characteristics in Figure 4c further reveal that HJDs with p⁺ CuO_x layers have a significantly higher V_B compared to those with p⁻ CuO_x layers. Furthermore, the V_B value of HJDs increases with the size of the CuO_x layer. The highest V_B observed in the single-layer HJD, featuring a p⁺ CuO_x layer with 80 μm radius, achieves a PFOM value of 0.71 GW/cm².

To comprehensively investigate the impact of geometric dimensions in the p⁺ and p⁻ CuO_x layers on the V_B performance of HJDs, we analyzed the simulated electric field distributions for single-layer HJDs with varying CuO_x layer diameter dimensions under a reverse bias voltage of 1000 V, as shown in Figure 5. The extracted electric field profiles of the HJDs with p⁺ and p⁻ CuO_x layers along the cutline at the Ga₂O₃ surface are summarized in Figure 5a,d, respectively. In the HJD with p⁻ CuO_x, the peak electric field is observed beneath the edge of the anode foot, whereas in the HJD with p⁺ CuO_x, it shifts to the edge of the CuO_x foot. Moreover, the HJDs with p⁺ CuO_x layers present a significantly higher peak electric field compared to those with p⁻ CuO_x layers. This is attributed to the narrowing of the PN junction depletion region as the hole concentration increases, resulting in a stronger electric field intensity concentrated within the junction region [27]. These findings also suggest that optimizing the electric field distribution in the CuO_x/Ga₂O₃ HJDs can be effectively achieved by tuning the hole concentration in the CuO_x layer. Furthermore, the size of the CuO_x also has an effect on the electric field distribution. As the CuO_x layer size increases, the peak electric field within the HJD decreases, leading to a notable enhancement in the V_B. This improvement is due to the expansion of the PN junction area, which enlarges the depletion region under reverse bias conditions.

3.3. Ga₂O₃ HJDs with a Double-Layer CuO_x Structure

To fully optimize the p⁺ and p⁻ CuO_x layers, we fabricated the CuO_x/Ga₂O₃ HJDs featuring a double-layer CuO_x. This section focuses on the impact of the geometric dimensions in the bilayer CuO_x on the performance of the HJDs via a comparative analysis of devices fabricated on the samples D₁–D₅.

Table 3. Device parameters for the Ga₂O₃ HJDs with a double-layer CuO_x structure.

Sample ID	CuOx Structure	Hole Concentration (/cm3)	Radius of p+ CuOx (μm)	Radius of p− CuOx (μm)
D1	double-layer	p+/p− (~e19/~e17)	80	85
D2	double-layer	p+/p− (~e19/~e17)	80	90
D3	double-layer	p+/p− (~e19/~e17)	80	100
D4	double-layer	p+/p− (~e19/~e17)	80	110
D5	double-layer	p+/p− (~e19/~e17)	80	120

lists the geometric parameters for these Ga₂O₃ HJDs with a double-layer CuO_x. The radial dimension of the p⁺ CuO_x layer was fixed at 80 μm, while the radii of the p⁻ CuO_x layer in the samples D₁–D₅ were set to 85, 90, 100, 110, and 120 μm, respectively. Figure 6a,b show the device structure and top view of the optical image of the CuO_x/Ga₂O₃ HJD (D₄). Figure 6c presents the TEM image and EDS analysis of HJD with a double-layer CuO_x structure, revealing the polycrystalline nature of the sputtered CuO_x. It also highlights the sharp and well-defined heterojunction interface, demonstrating the high-quality formation of the CuO_x/Ga₂O₃ HJDs [28].

Figure 7a presents the forward linear-scale I–V characteristics and the extracted differential R_on,sp of the Ga₂O₃ HJDs. With increasing the diameter dimensions of the p⁻ CuO_x layer size, the R_on,sp value of the double-layer HJDs increases slightly from 6.43 mΩ·cm² to 6.47 mΩ·cm², indicating that the expanded p⁻ CuO_x layer has minimal impact on the R_on,sp and suggesting negligible current diffusion through the bottom p⁻ CuO_x layer. The ideal factors of the double-layer HJDs are around 1.7 to 1.9. Figure 7b shows the log-scale I–V characteristics, showing a similar V_ON across all the devices at a forward current density of 1 A/cm². All the devices demonstrated high rectification ratios exceeding 10⁸. Figure 7c illustrates the measured V_B for the double-layer HJDs. The V_B increased from 2360 V in D₁ to 2780 V in D₄ with a larger p⁻ CuO_x layer. This improvement in the V_B is mainly attributed to the enhanced depletion depth of the PN junction achieved by the double-layer CuO_x structure with varying hole concentrations [27]. This design effectively suppresses the peak electric field at the anode edge and redistributes it within the devices. However, sample D₅ showed a decreased V_B of 2540 V. It indicates that a moderate size distance between the p⁺ and p⁻ CuO_x layers is beneficial for the enhancement of V_B, while an excessive distance (over 30 μm) negatively impacts performance. Figure 7d presents the benchmark plots of the R_on,sp versus V_B for state-of-the-art Ga₂O₃ diodes. The fabricated Ga₂O₃ HJDs with a double-layer CuO_x structure (D₄) achieves a PFOM of 1.20 GW/cm². The performance of this HJD is comparable to that of the extensively studied NiO/Ga₂O₃ HJDs, demonstrating its competitive potential for high-performance power electronics applications [29,30].

Figure 8a–e show electric field distribution for double-layer HJDs at reverse voltage of 1000 V, while Figure 8f presents the extracted electric field profiles along the cutline at the Ga₂O₃ surface. Notably, as the radius difference between the p⁺ and p⁻ CuO_x layers increased from 5 to 40 μm, the peak electric field of the double-layer HJDs gradually reduced from 2.80 MV/cm to 2.56 MV/cm. These results are consistent with the V_B results discussed previously.

Table 4. Values of the R_on,sp and V_B for the Ga₂O₃ HJDs with a double-layer CuO_x structure.

Sample ID	Ron,sp (mΩ·cm2)	VB (V)	PFOM (GW/cm2)
D1	6.43	2360	0.87
D2	6.45	2510	0.98
D3	6.45	2650	1.09
D4	6.46	2780	1.20
D5	6.47	2540	1.00

summarizes the values of the R_on,sp and V_B for the Ga₂O₃ HJDs with a double-layer CuO_x structure, along with the calculated PFOM values. The data clearly demonstrate the influence of the p⁻ CuO_x layer’s structural parameters on the electrical performance of CuO_x/Ga₂O₃ HJDs. As the radius difference between the p⁺ and p⁻ CuO_x layers reached 30 μm, the PFOM value achieved 1.2 GW/cm², representing the highest reported value for CuO_x/Ga₂O₃ HJDs.

4. Conclusions

In conclusion, we successfully prepared high-quality CuO_x layers and investigated the effects of their geometric dimensions and hole concentration on the electrical characteristics of the CuO_x/Ga₂O₃ HJDs via both experimental analysis and TCAD simulations. These results indicated that tuning the hole concentration and expanding the diameter dimensions of the CuO_x layer effectively lowered the R_on,sp and raised the V_B. Moreover, the high-performance CuO_x/Ga₂O₃ HJDs with a double-layer CuO_x structure were demonstrated by optimizing these CuO_x parameters. It exhibits a high V_B of 2780 V and a low R_on,sp of 6.46 mΩ·cm², further yielding a record PFOM value of 1.2 GW/cm². This work provides an effective strategy for improving the performance of Ga₂O₃-based bipolar power electrons.