Ohmic contact structures on β-Ga₂O₃ with n+ β-Ga₂O₃ pulsed laser deposition layers

Beta-gallium oxide (β-Ga₂O₃) is a wide bandgap (E_G ∼ 4.8 eV) semiconductor that has attracted growing interest due to the availability of bulk crystals and a variety of shallow n-type dopants (Sn, Si, and Ge)¹ covering a broad doping range. It is considered a promising candidate for next-generation power electronics,^2,3 in part due to the fact that it has a higher theoretical breakdown field and power figure-of-merit than either GaN or SiC.⁴ 

For power devices, it is necessary to minimize the ohmic contact resistance, which contributes to the on-resistance of the device and increases the conduction losses in the circuit. Reducing the contact resistance is often achieved by doping the semiconductor surface heavily to decrease the depletion width, making it thin enough to allow carriers to the tunnel.⁵ 

The standard metallization for Ga₂O₃ ohmic contacts consists of annealed Ti/Au.^6–9 Various processes have been reported to reduce the contact resistance of Ti/Au contacts to Ga₂O₃ (Table I). Values range from ∼10⁻³ to ∼10⁻⁷ Ω cm² and depend on doping concentration and the type of process. Many processes incorporate a highly doped (n+), 100–180 nm-thick Ga₂O₃ layer produced using Si-ion implantation with high-temperature (925–975 °C) activation anneal.^1–4 Several of the aforementioned processes also included reactive ion etching using BCl₃/Ar.^10–13 A specific contact resistance of 4.6 × 10⁻⁶ Ω cm² was achieved using Si-ion implantation (implant dosage = 5 × 10¹⁹ cm⁻²) and a postimplant anneal at 950 °C.¹⁴ In a separate study, contacts on the (010) orientation without implantation were pseudo-ohmic with a specific contact resistivity of $ρ C ∼ 1 0 − 3 Ω c m 2$⁠; Si-ion implantation and RIE treatment (MBE)^15,16 and metal organic vapor phase epitaxy (MOVPE) have also been used to create highly doped Ga₂O₃ layers for ohmic contacts.^10,17,18 Low-resistance (8.3 × 10⁻⁷ Ω cm²) ohmic contacts were achieved on a Si-doped β-Ga₂O₃ (1.4 × 10²⁰ cm⁻³) regrown MOVPE layer.¹⁰ The ∼100 nm-thick n+-Ga₂O₃ growth step was preceded by a heavy Si delta doping at the etched Ga₂O₃ surface in the contact region. The lowest specific contact resistance (1.62 × 10⁻⁷ Ω cm²) was achieved on heavily doped (3.23 × 10²⁰ cm⁻³) MOVPE layers.¹⁹ 

In this study, we investigated the use of thin (∼40–150 nm), highly doped (n+) Ga₂O₃ layers deposited using pulsed laser deposition (PLD) as an approach to reduce the contact resistance of Ti/Au ohmic contacts on (001) and (010) β-Ga₂O₃ substrates. PLD has been demonstrated to readily produce highly conductive β-Ga₂O₃ films.^20–24 The use of PLD could simplify processing and avoid the high temperatures (900–1000 °C) associated with Si-ion implantation.^11–14 

Single-crystalline, Sn-doped (n-type) β-Ga₂O₃ wafers produced by the edge-defined film-growth (EFG) method were purchased from Novel Crystal Technology, Japan. Two orientations with slightly different doping concentrations were used: (001) with N_D = 5.10 × 10¹⁸ cm⁻³ and (010) with N_D = 4.80 × 10¹⁸ cm⁻³ or N_D = 2.50 × 10¹⁸ cm⁻³. A commercial pulsed laser deposition system with a KrF excimer laser, located at the Air Force Research Laboratory, was used to deposit n+ layers from a Ga₂O₃–1 wt. % SiO₂ target. The base pressure of the PLD chamber was 2.66 × 10⁻⁶ Pa, and deposition was performed in Ar at 13.3 Pa. The substrate temperature was 550 °C, and the laser was operated at a pulse rate of 10 Hz with an energy density of 3 J cm⁻² measured at the target. The thicknesses of the PLD layers were 40, 80, and 150 ± 10 nm, based on a growth rate of 96 nm/h.²⁰ 

The substrates were ultrasonicated in acetone, isopropyl alcohol, and de-ionized (DI) water for 10 min each and blown dry in nitrogen after each step.⁶ Ti/Au (20 nm/80 nm) ohmic contact metal stacks were deposited onto the Ga₂O₃ substrates (with or without PLD layers) via electron-beam evaporation using a Kurt Lesker PVD 75 deposition system having a base pressure of ∼3.3 × 10⁻⁵ Pa. Circular transfer length method (CTLM) patterns for measuring contact resistance were fabricated using photolithography as described below. First, a negative resist was used to pattern the CTLM mask, followed by UV exposure and development to expose the desired areas. CTLM structures comprised contact spacings, d = 10, 20, 30, 40, 50, and 60 μm, and the inner circle radius was fixed at 100 μm [Fig. 1(b)]. The contacts were etched using a Commonwealth Scientific Ion Mill (base pressure = ∼10⁻⁵ Pa) with Ar ions at 30 A and 450 V. It is noted that the ion milling step can produce damage to the contact sidewalls and to the Ga₂O₃ surface area between the contacts but does not damage the Ga₂O₃ surface directly underneath the contacts since these regions are covered and, therefore, not exposed to the ions. To ensure that the Au/Ti/PLD layers were completely etched, each device was over-etched 25–33 ± 4 nm (no PLD, 40, and 80 nm samples) and 40–45 ± 5 nm (150 nm samples) into the substrate, as determined from multiple measurements for each sample using a KLA Tencor P-15 profilometer. To complete the ohmic contact processing, the samples were subjected to a rapid thermal anneal (RTA) at 470 °C for 1 min in N₂.⁸ A schematic illustration of the fabrication steps and resulting cross-sectional structure is given in Fig. 1.

I–V characteristics of the fabricated CTLM devices were collected with an Agilent 4155C semiconductor parameter analyzer with a Signatone S-1160-4 N probe station. A voltage ranging from −1 to 1 V was applied between a probe placed on the inner metal circle and one on the outer metal area. Five to eight sets of CTLM structures were measured for each sample, and the specific contact resistivity value was taken from the average of these sets.

Hall measurements were conducted at room temperature using a Nanometrics HL5500 Hall System with a magnetic field strength of 0.5 T. Ti/Al/Ni/Au (20/100/50/50 nm) ohmic contacts were deposited via electron-beam deposition in a van der Pauw pattern followed by an RTA at 470 °C for 1 min in N₂. Ohmic contacts were placed on bare (001) and (010) Sn-doped β-Ga₂O₃ substrates to obtain the carrier concentration of each substrate and on (001) and (010) Fe-doped β-Ga₂O₃ substrates coated with Si-doped β-Ga₂O₃ PLD layers to obtain the carrier concentrations of the PLD layers.

Selected samples were prepared for characterization using transmission electron microscopy (TEM). The samples were capped with platinum to prevent damage during the sample preparation process. Liftout and thinning were achieved using a FEI Helios 650 Nanolab focused ion beam (FIB) instrument. After liftout, the samples were attached to a Cu TEM grid and thinned to ∼65 nm. Scanning transmission electron microscopy (STEM), high-angle annular dark field (HAADF) imaging, and energy-dispersive x-ray (EDX) analysis were conducted using a Themis 200 TEM. High resolution TEM (HRTEM) images were obtained using a Tecnai F20 TEM.

For contacts on semiconductor layers that are thicker than the contact spacing d, the current flows primarily in the topmost layer of the semiconductor. The thickness of this layer and its effective sheet resistance were calculated for a similar contact structure and similar doping concentration in the Ga₂O₃ substrate to be ∼40 μm and ∼5 Ω/sq, respectively.²⁵ For these conditions, the CTLM equations should be valid for specific contact resistivity values > 10⁻⁶ Ω cm².²⁶ 

The quality of the linear fit is measured by the Pearson product (regression coefficient, R²). In the ideal situation, the Pearson product equals unity (R²= 1). R-squared values for these data are typically, R² = 0.99. However, in some cases, there was significant scatter, e.g., 40 nm PLD layer on the (001) substrate.

Table II and Fig. 4 show the specific contact resistivity values calculated from the data shown in Figs. 2 and 3. The average specific contact resistivities for contacts without a PLD layer were 2.64 × 10⁻⁴ and 8.03 × 10⁻⁴ Ω cm², on the (001) and (010) substrates, respectively. The different values can be attributed to the anisotropic properties of different β-Ga₂O₃ surfaces and to the slightly higher doping concentration in the (001) substrate. Different thermal, optical, and electrical properties along different β-Ga₂O₃ crystal directions have been reported.^29–32 

β-Ga₂O₃ has two strong cleavage planes of (100) and (001) with low surface energies potentially susceptible to the growth of off-axis planes. Substrate orientations for the epitaxial growth of β-Ga₂O₃ films with high electrical quality are typically (010) or (100) with miscuts. Homoepitaxial growth on (001) by MBE revealed an additional $( 4 ¯ 01)$ reflection by x-ray diffraction and depositions in this work by PLD on (001) also indicated a minor $( 4 ¯ 01)$ contribution.³³ The additional planes could possibly generate point defects in the lattice causing Ga vacancies, which are dominant acceptors. Lower carrier concentrations in films deposited on (001) substrates are, thus, attributed to elevated acceptor levels.

For the (010) substrate, the contact resistance values decreased as the PLD layer thickness increased. The lowest average value of 4.79 × 10⁻⁵ Ω cm² on the (010) substrate was achieved for a PLD layer thickness of 150 nm. For the (001) substrate, the contact resistivity initially increased for contacts with a 40 nm PLD layer. This increase could be associated with the nonuniformity of the thin PLD layer (e.g., in terms of thickness, doping, and/or stoichiometry); as discussed above, the TEM results indicate that the PLD layers grow differently on different surface orientations. However, the contact resistivity decreased for (001) contacts with thicker (80 and 150 nm) PLD layers. The (001) substrate with the 150 nm PLD layer had the lowest average value of specific contact resistivity: $ρ c=3.48 ×1 0 − 5 Ω c m 2$⁠. The lowest contact resistance values for both (010) and (001) substrates were achieved for contact structures that comprised PLD layers having a deposited thickness of 150 nm, which is in the range of thicknesses of reported n+ layers used to reduce the contact resistance in Ga₂O₃ ohmic contacts (see Table I). However, further investigation is required to precisely determine the optimal thickness of the n+ PLD layer.

We note here that the contact resistivity values were typically lower for the (001) substrate than the (010) substrate even though the carrier concentration was approximately an order of magnitude lower in the PLD layers grown on the (001) substrates. This is further evidence that the PLD layers in the gaps between contacts were completely etched during the ion milling step, indicating that the n+ PLD layers provide a low-resistance tunneling barrier for current transport into the β-Ga₂O₃ substrate.

Contact structures were characterized in the cross section using TEM to investigate the microstructure, morphology, and chemical compositions. The thickness of the as-deposited PLD layers in the samples selected for analysis was 80 nm. HAADF STEM images of annealed contact structures on (001) and (010) β-Ga₂O₃ substrates are shown in Figs. 5(a) and 6(a), respectively. It is apparent from the images that, after annealing, the thickness of the PLD layers was reduced by ∼20 nm. The PLD film grown on the (001) substrate appears to comprise multiple grains, whereas the film grown on the (010) substrate appears epitaxial such that the interface between the substrate and film is difficult to discern. Additional TEM images and FFT diffraction patterns are included in the supplementary material.³⁴ The layers on both substrates, however, appear continuous and without voids. According to EDX analysis [Figs. 5(b), 5(c), 6(b), and 6(c)], a similar layered structure is present on both substrates after annealing. From bottom to top, the layers can be described as $Ti− TiO$⁠, intermixed Au–Ti, Au, intermixed Au–Ti, and another layer of $Ti− Ti O x$⁠.

The results of this study show that thin n+ Ga₂O₃ PLD layers can be used to reduce the contact resistance of Ti/Au ohmic contacts on (001) and (010) β-Ga₂O₃ substrates. The specific contact resistance was reduced by up to 8× and 16× on (001) and (010) substrates, respectively. The results provide a point of reference for the further optimization of highly doped Ga₂O₃ PLD layers to minimize the contact resistance in Ga₂O₃ devices.

This material is based upon work supported by the Air Force Office of Scientific Research under Award No. FA9550-21-1-0360 (Program Manager Ali Sayir) and Air Force Research Laboratory under Award No. FA807518D0015. The use of the Materials Characterization Facility at Carnegie Mellon University was supported by Grant No. MCF-677785.

The authors have no conflicts to disclose.

Elizabeth V. Favela: Conceptualization (equal); Data curation (equal); Formal analysis (lead); Investigation (equal); Methodology (equal); Validation (equal); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Hyung Min Jeon: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Kevin D. Leedy: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Kun Zhang: Data curation (equal); Formal analysis (equal); Investigation (equal); Validation (equal); Writing – review & editing (equal). Szu-Wei Tung: Investigation (equal); Writing – review & editing (equal). Francelia Sanchez Escobar: Investigation (equal); Validation (equal); Writing – review & editing (equal). C. V. Ramana: Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Lisa M. Porter: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.