Low-pressure CVD-grown β-Ga₂O₃ bevel-field-plated Schottky barrier diodes

β-Ga₂O₃ is an ultra-wide bandgap semiconductor with a room-temperature bandgap of 4.5–4.9 eV^1,2) and an estimated breakdown field of 6–8 MV/cm.³⁾ Its high breakdown field leads to large theoretical figures of merit for power switching,⁴⁾ with a lower expected on-resistance (R_ON)^3–5) for a given breakdown voltage (V_BR) than those of incumbent wide bandgap materials such as SiC and GaN. The availability of large-area, high-quality native Ga₂O₃ substrates prepared by scalable and low-cost melt-grown techniques^6–10) make Ga₂O₃-based power devices promising for technological insertion in high-power systems. Lateral devices such as field-effect transistors^11–18) and vertical devices including Schottky rectifiers^19–25) and trench MOSFETs^26–28) have been recently demonstrated by various groups using both β- and α-Ga₂O₃.²⁹⁾

Vertical devices are preferred over lateral geometries for high-power applications, but vertical device topologies require an epitaxial growth technology capable of growing thick and low-doped drift layers. Several growth techniques, including molecular beam epitaxy^30–32) (MBE), pulsed layer deposition³³⁾ (PLD), mist-CVD,²⁹⁾ atmospheric-pressure CVD³⁴⁾ (APCVD), low-pressure CVD^35,36) (LPCVD), halide vapor phase epitaxy³⁷⁾ (HVPE), and metal organic chemical vapor deposition^38,39) (MOCVD), have been demonstrated for the growth of β-Ga₂O₃. HVPE with a growth rate of 5 µm/h on (001)³⁷⁾ and LPCVD with a growth rate of ∼2 µm/h on (010) substrates³⁶⁾ are two demonstrated homoepitaxial growth technologies with relatively high growth rates. In addition, LPCVD provides a low-cost solution for growing high-quality epitaxial Ga₂O₃ with controllable doping⁴⁰⁾ and high electron mobility compared with other present techniques. Using HVPE, Schottky diodes with planar field-plate termination²³⁾ exhibiting breakdown voltage greater than 1 kV have already been demonstrated. However, there are no reports to date on Schottky diodes fabricated using LPCVD-grown β-Ga₂O₃ films. In this work, we have fabricated Schottky diodes to verify the material quality and investigate device performance in order to demonstrate the viability of LPCVD as a growth technique for vertical power electronic devices. The fabricated diode exhibits a breakdown field (F_BR) of 4.2 MV/cm, extrinsic R_ON of 3.9 mΩ·cm², and extracted intrinsic R_ON of 0.023 mΩ·cm². The extracted F_BR is higher than the theoretical F_BR values for 4H-SiC (2.2 MV/cm)⁴¹⁾ and GaN (3.3 MV/cm).⁴¹⁾ The high-performance devices obtained on LPCVD-grown layers demonstrate the potential of this growth method for future β-Ga₂O₃ technology.

A three-dimensional representation of the fabricated beveled field-plate Schottky diode is shown in Fig. 1(a), along with its cross-sectional schematic in Fig. 1(b). The epitaxial stack consists of ∼2 µm β-Ga₂O₃ thin film grown by LPCVD on commercially available Sn-doped (010) β-Ga₂O₃ substrates.⁴³⁾ Growth was carried out in a horizontal furnace with programmable temperature and precise pressure controllers. A growth temperature of 900 °C and a pressure of 4 Torr were used, leading to a nominal growth rate of 2 µm/h. Pregrowth sample preparation involved solvent cleaning using acetone, toluene, and isopropyl alcohol, followed by an N₂ blow-dry. Prior to the growth, the samples were annealed in situ at 900 °C for 30 min in O₂ ambient. High-purity gallium pellets (Alfa Aesar, 99.99999%) and O₂ were used as the source materials and argon (Ar) was used as the carrier gas. Silicon tetrachloride (SiCl₄) was used as the n-type dopant source. Atomic force microscopy (AFM) on the sample surface after growth showed a root-mean-square (rms) surface roughness (t_rms) of 4.86 nm, as seen in Fig. 1(c).

Zoom In Zoom Out Reset image size

i-Line stepper lithography was used in the processing of the diodes. The diode fabrication commenced with the deposition of the anode (Pt/Au/Ni) using an e-beam evaporator with Pt as the work function metal. The anode electrode in the devices discussed here is a stripe [Fig. 1(a)] of 50 µm length and 1 µm width designed with rounded corners to avoid breakdown voltage degradation due to the formation of spherical junctions.⁴²⁾ The cathode ohmic contact was deposited on the back of the substrate using e-beam-deposited Ti/Au/Ni. Bevel/trench patterns were formed, which covered the anode metal and extended nominally 0.25 µm from the anode edge on all sides. The device breakdown just after the formation of the anode and cathode was achieved at a depletion width of ∼0.8 µm; hence, the anode-bevel extension (L_AB) was chosen as 0.25 µm (<0.8 µm) to leverage the advantages of a bevel design.⁴²⁾ Further details of the device simulation and design are provided later. Subsequently, dry etching to form the trench was carried out in a Plasma-Therm SLR 770 ICP-RIE system using BCl₃/Ar gas flow of 35/5 sccm, 30 W RIE, 200 W ICP, and 5 mTorr chamber pressure. In previous studies, dry etching using BCl₃ was shown to result in an inclined profile.⁴⁴⁾ AFM was used to measure the etch depth (t) and the bevel angle (φ) as ∼800 nm and ∼56°, respectively (Fig. 2). Thereafter, 300 nm of SiO₂ was deposited conformally by plasma-enhanced CVD (PECVD) at 250 °C as a surface passivation layer. A SiO₂ thickness of 300 nm was chosen on the basis of the results of two-dimensional simulations using Silvaco ATLAS to target F_BR ∼5 MV/cm. To form the metal field-plate, the oxide layer on top of the anode metal was etched away to form a contact between the anode and the metal plate deposited by conformal sputtering of Ti. Finally, the field plates with a field-plate overlap (L_FP) of ∼4 µm were patterned with the remaining metal being etched using dilute hydrochloric acid (HCl) at 65 °C. Figure 3(a) shows the scanning electron microscopy (SEM) images of the diode after field-plate integration where the beveled profile is distinct. Figure 3(b) shows the SEM image for one end of the diode along its length with the rounded pattern designed to avoid sharp corners.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Electrical characteristics were measured using an Agilent B1500A semiconductor device analyzer. Room-temperature current–voltage (J–V) characteristics of the device are shown in Fig. 4(a). The anode current density vs voltage curves were plotted and compared for the following three structures: (a) after the formation of anode and cathode electrodes and before trench etching, (b) after trench etching and before the deposition of SiO₂, and (c) after the final fabrication step. An ideality factor (η) of 1.03 ± 0.02, a high I_ON/I_OFF ratio of ∼10¹⁰, and a specific on-resistance (R_ON) of 3.6 mΩ·cm² were calculated before trench etching. I_ON for the diode decreases after BCl₃/Ar etching, accompanied by increases in R_ON to 6.7 mΩ·cm² and η to 1.1 ± 0.03. We believe that the increase in resistance is related to surface damage that is subsequently repaired during the passivation. The mechanism of this is not understood at present. This is similar to previous reports on BCl₃/Ar etching being unfavorable for Schottky diodes.⁴⁵⁾ With the formation of the field plate, R_ON decreased to 3.9 mΩ·cm², and similar characteristics to those before trench etching were obtained, suggesting that the surface damage caused by BCl₃/Ar etching was passivated by the SiO₂. Figure 4(b) shows the charge profiles before and after trench etching as a function of the depletion width. A doping density (N_D) of 2.5 × 10¹⁷ cm⁻³ was obtained from capacitance–voltage (C–V) measurements (the C–V characteristics from which the corresponding charge profiles were extracted are shown in the inset). The measured charge density was similar before and after trench etching, suggesting that while BCl₃/Ar etching results in damage of the sidewall surface, it does not cause significant charge depletion. The C–V plot after the formation of the field plate includes contributions from the Ti/SiO₂/Ga₂O₃ (metal oxide semiconductor) sidewall and bond-pad regions, and is therefore not shown here.

Zoom In Zoom Out Reset image size

Figure 5(a) shows the two-terminal reverse current breakdown characteristics for identical, fresh devices at various stages of the process, with immersion in Fluorinert⁴⁶⁾ to prevent air breakdown. An anode current compliance of 0.2 A/cm² (3–4 orders of magnitude lower than I_ON) is seen to cause destructive breakdown with a breakdown voltage (V_BR) of −138 V for diodes before trench etching and −74 V after trench etching. This indicates that the expected enhancement in breakdown voltage using a bevel edge-termination design was not obtained after BCl₃/Ar etching. Further analysis is needed to understand this, but one possible reason could be the surface damage caused by BCl₃/Ar etching,⁴⁵⁾ which leads to surface-related leakage. By fabricating the field-plate structure on top of the "after trench" device, V_BR was increased from −74 to −190 V. Field-plated Schottky diodes were also found to withstand one order higher current density (>1 A/cm²) before breaking down catastrophically.

Zoom In Zoom Out Reset image size

The reverse current seen in the curve after −130 V was deduced to be leakage current through SiO₂, which adds to the overall current. Setting a current compliance of 0.2 A/cm² for the field plates resulted in repeatable breakdown [Fig. 5(b)] with V_BR = −129 V. The measured V_BR values, Schottky contact barrier height (1.5 eV), and doping density (both extracted from the C–V data) were used to calculate the depletion width (W_dep) and peak electric field (F_max) using one-dimensional Poisson's equation. W_dep of 0.8 µm, which corresponds to F_max of 3.5 MV/cm, was estimated for the device before trench etching. Since W_dep of 0.8 µm is less than the epitaxial layer thickness (∼2 µm), this corresponds to non-punch-through breakdown of the devices. Two-dimensional device simulations using Silvaco ATLAS were used to estimate the electric field at the center of the diode, as shown in Fig. 6(b). The foot edge of the anode shows a peak field higher than the theoretical Ga₂O₃ breakdown field (8 MV/cm), similar to a previous report.²³⁾ The corresponding W_dep and F_max values for the device after trench etching decreased to 0.65 µm and 2.4 MV/cm, respectively. With the field plate, F_max of 3.4 MV/cm for repeatable breakdown and 4.2 MV/cm for destructive breakdown were calculated (i.e., F_max increased from 2.4 MV/cm after trench etching to 4.2 MV/cm after field-plate insertion). 2D Silvaco simulations for destructive breakdown on the field plate were seen to match the calculated 4.2 MV/cm field at the center of the diode with the peak field of 5.9 MV/cm at the corner. The calculated values of W_dep and F_max are tabulated in Table I. The last column in the table lists the intrinsic R_ON values for all N_D and W_dep values estimated using the equation

$$\begin{equation*} R_{\text{ON}} = \frac{W_{\text{dep}}}{qN_{\text{D}}\mu}, \end{equation*}$$

where q = 1.6 × 10⁻¹⁹ C is the elementary electric charge and μ is the mobility (∼100 cm² V⁻¹ s⁻¹, estimated from the Hall measurement of in-plane mobility). As expected for a back contact, the measured device resistance is dominated by the extrinsic resistance of the substrate.

Zoom In Zoom Out Reset image size

In summary, we demonstrated a field-plate bevel mesa Schottky diode using LPCVD-grown β-Ga₂O₃ thin film. A high breakdown field of 4.2 MV/cm, extrinsic R_ON of 3.9 mΩ·cm², and intrinsic R_ON of 0.023 mΩ·cm² were estimated for the devices, indicating the high quality of the LPCVD-grown layers. Device performance degradation due to BCl₃/Ar was mitigated by SiO₂ passivation. Future developments in epitaxial growth for lower background doping and field management will enable β-Ga₂O₃ power devices with lower conduction and switching losses. The reported work shows the promise of LPCVD β-Ga₂O₃ for low-cost high-power devices.

Acknowledgments