1. INTRODUCTION

Monoclinic β-Ga₂O₃ has attracted much attention as a promising material for high-power devices [¹-³] due to its outstanding properties, such as an ultra-wide band gap (~4.8 eV) [⁴] and a high theoretical breakdown electric field (8 MV/cm) [⁵], which surpass those of GaN and SiC. In addition, a significant advantage over other wide-bandgap materials such as GaN and SiC is the availability of large-sized and high-quality native β-Ga₂O₃ substrates grown by cost-effective melt-based crystal growth methods like edgedefined film-fed growth (EFG) [⁶], Czochralski growth (CZ) [⁷], and floating zone (FZ) growth [⁸,⁹].

Although the outstanding properties and the available substrates make β-Ga₂O₃ a promising candidate for highpower devices, it is known that intrinsic β-Ga₂O₃ is an insulating material [¹⁰]. Therefore, fabricating high-quality doped β-Ga₂O₃ films with controllable conductivity is essential. Growth of the n-type doped β-Ga₂O₃ films using dopants such as Si [¹¹-¹⁶], Sn [¹⁷,¹⁸], and Ge [¹⁹,²⁰] has been reported. Among these dopants, Si is mostly employed as the best dopant for the n-type doping of β-Ga₂O₃ [¹⁷,²¹]. The growth of Si doping in β-Ga₂O₃ has been reported by investigating its electrical properties and enabling applications to lateral [²²] and vertical electronic devices [²³].

Si doping into the β-Ga₂O₃ epitaxial films has been widely studied using various growth methods, such as pulsed laser deposition (PLD) [¹¹], low-pressure chemical vapor deposition (LPCVD) [¹²], metal-organic chemical vapor deposition (MOCVD)[²⁴], metalorganic vapor-phase epitaxy (MOVPE) [²⁵], and molecular beam epitaxy (MBE) [¹³-¹⁶,²⁶,²⁷].

MBE offers advantages of high-purity β-Ga₂O₃ film growth and a precise doping control by adjusting the effusion cell temperature of dopant. However, challenges remain in controlling the Si dopant concentration in β-Ga₂O₃ films grown by MBE using a solid Si source under the oxygen gas environment in the growth chamber due to the easy oxidation of pure Si solid source, which is used for the dopant in MBE [¹³]. Therefore, finding solutions to this issue is a critical matter that needs to be addressed.

Several procedures have been proposed to mitigate this problem, aiming to achieve controllable Si doping over a wide range [¹⁵,²⁶,²⁷]. McCandless et al. modified a conventional Si effusion cell by inserting an endplate with small inclined holes into the opening of the Si crucible to prevent oxidation of the Si source [²⁶]. T. Itoh et al. inserted a needle valve on top of the crucible of the conventional Si effusion in order to minimize the exposure of Si from the oxidizing environment, thereby reducing the oxidation of the Si source [¹⁵]. J. Zhan et al. addressed the oxidation of the Si source by employing an electron beam evaporator modude containing Si, instead of the conventional Si effusion cell [²⁷].

In this work, we employed a simple and effective method to reduce the Si source oxidation without requiring complex modifications of growth system, which is aiming to achieve controlled Si doping in β-Ga₂O₃ films grown by plasmaassisted MBE (PAMBE). Specifically, we made a Si source to be melted fully by heating the Si source to 1450°C followed by cooling to room temperature to form a solidified bulk Si source. By this simple procedure the initial Si solid source of the small pieces as chip-shaped with an exposed large total surface area to the oxidizing environment could be packed well inside the con-shaped PbN crucible, which could reduce the exposed surface area of the solid Si source. The small exposed area of the Si source in the crucible effectively reduced the oxidation problem of the Si source by decreasing the exposed surface area to the oxidizing environment. We demonstrated the effectiveness of this method by growing homoepitaxial undoped and Si-doped β-Ga₂O₃ films at various Si effusion cell temperatures and investigated the effects of Si doping on the structural and electrical properties of the β-Ga₂O₃ films.

2. EXPERIMENTAL

Two sets of β-Ga₂O₃ films were grown on β-Ga₂O₃ substrates by PAMBE. The first set of samples was a multilayer β-Ga₂O₃ film grown on (001) Sn-doped β-Ga₂O₃ substrate consisting of alternating four undoped β-Ga₂O₃ layers and four Si-doped β-Ga₂O₃ layers with varying Si effusion cell temperatures. This film was used to evaluate the incorporation of Si into the β-Ga₂O₃ films. The second set of samples includes one undoped β-Ga₂O₃ film and three Si-doped β-Ga₂O₃ films with different Si effusion cell temperatures grown on semi-insulating (010) Fe-doped β-Ga₂O₃ substrates to investigate their crystal characteristics, surface morphology, and electrical properties. Each Si-doped β-Ga₂O₃ sample consists of double-layer structure; The first layer is an undoped layer with a thickness of 50 nm and the Si-doped second layer with a thickness of 200 nm. The first undoped layer was purposed for a barrier to inhibit Fe diffusion from the substrate, which leads to a decrease in doping efficiency, to the Si-doped layer. The second layer is the main Si-doped layer to control the conductivity of the β-Ga₂O₃ films. The undoped single-layer structure sample was grown to a thickness of 250 nm being equal to the total thickness of the double-layer structure.

Commercial (001) Sn-doped and (010) Fe-doped β-Ga₂O₃ substrates grown by EFG method were used in this study. The substrates were cleaned by ultrasonic agitation in methanol, acetone, and deionized (DI) water for 3 mins, respectively, and then the substrates were flowed in DI water for 5 mins. Next, the substrates were cleaned in a chemical solution of DI water: H₂O₂: H₂SO₄ = 1:1:4 (vol.%) for 5 mins. Finally, the substrates were flowed in DI water for 15 mins and then dried by nitrogen gas blowing. The chemically cleaned substrtaes were loaded into the growth chamber and thermally cleaned at 800°C for 30 mins under the radiofrequency (RF) plasma oxygen exposure.

All the β-Ga₂O₃ films were grown at 800°C for 250 mins. The RF plasma power and oxygen gas flow rate were set to 300W and 2 sccm, respectively. The Ga source was supplied by using a double filament effusion cell equipped with the Sumo crucible with elemental Ga and its flux was set to 6 Å/min. The Si flux was produced by a solid elemental Si source which was solidified after melting in the crucible contained in a standard high-temperature effusion cell. The Si flux was controlled by changing the temperature of effusion cell. The base pressure was about 1.35×10⁻⁷ Pa with a liquid nitrogen supply. The pressure during thermal cleaning and growth processes was maintained in the range of 5.31 to 6.75×10⁻³ Pa by supplying the oxygen gas for the RF oxygen plasma.

Growth evolution during the growth was monitored by in-situ reflection high-energy electron diffraction (RHEED). The surface morphology of the grown films was characterized by atomic force microscopy (AFM) on 2×2 μm² area. Si concentrations incorporated into the β-Ga₂O₃ films were investigated by secondary ion mass spectrometry (SIMS). Carrier concentration, mobility, and resistivity of the doped β-Ga₂O₃ were characterized by Hall-effect measurements.

3. RESULTS AND DISCUSSION

To investigate the incorporation of Si into the homoepitaxial β-Ga₂O₃ films, SIMS measurement was performed on the homoepitaxial (001) β-Ga₂O₃ film with a multilayer structure consisting of alternating undoped β-Ga₂O₃ layers and Si-doped β-Ga₂O₃ layers with different Si effusion cell temperatures. Figure 1(a) shows a profile of Si concentration measured by SIMS from the multilayer (001) β-Ga₂O₃ sample with the Si effusion cell temperature from 900 to 975°C. The film/substrate interface is identified clearly by the presence of the Si stack, which is attributed to residual impurities on the surface of the β-Ga₂O₃ substrate [²⁸]. The background of Si concentration observed from the SIMS profile in the β-Ga₂O₃ substrate region is around 4.3×10¹⁶ cm^-3. This may be attributed either to unintentional Si doping or to the detection limit of the SIMS measurement. Similar results, including an unintentional background Si doping density of ~ 5×10¹⁶ cm^-3 and the lower detection limit of SIMS measurement have been reported [¹⁵,²⁸]. The Si atom concentrations in the Si-doped Ga₂O₃ layers grown with the Si cell temperatures of 975 and 950°C were clearly appeared and distinguished from the undoped interlayers as shown in Figure 1(a). However, the Si atom concentrations were difficult to be distinguished from the undoped interlayer for the Ga₂O₃ layers grown with the lower Si cell temperatures of 925 and 900°C. This might be due to the diffusion of Si from the Si substrate, which is comparable to the Si concentrations of intentionally Si doped Ga₂O₃ layers with the low Si cell temperatures.

Although the Si atom concentrations of the Si-doped Ga₂O₃ layers with the Si effusion cell temperatures of 900 and 925°C are not distinguished clearly from undoped layers, the Si concentration increased with the Si effusion cell temperatures. The average Si concentrations of Si-doped layers with the Si effusion cell temperatures of 900, 925, 950, and 975°C are determined to be 7.7×10¹⁶, 3.8×10¹⁷, 5.4×10¹⁸, and 1.6×10¹⁹ cm^-3, respectively. The relationship between the average Si concentration in the Si-doped layers and the Si effusion cell temperature was plotted in Figure 1(b). This plot indicates that the average Si concentration increases almost linearly with the Si effusion cell temperatures in our experiments. The observed linear trend implied possible precise control Si doping over the moderate doping range.

Figure 2 shows RHEED patterns observed along the [001] azimuth from the (010) Fe-doped β-Ga₂O₃ s ub strates, a nd undoped and Si-doped β-Ga₂O₃ films grown with the different Si effusion cell temperatures. The RHEED patterns of the substrates shown in Figure 2(a-d) are sharp streaky consistent for the all samples, which implied the uniform quality, smoothness, and cleanness of the substrates for the growth of homoepitaxial β-Ga₂O₃ films. The RHEED patterns of the homoepitaxial undoped film and Si-doped films are shown in Figure 2(e-h). These RHEED patterns from the grown Ga₂O₃ films also showed sharp streaky features with more clear additional reflections. This indicates that both the undoped and the Si-doped films were grown to the single crystals with smooth surfaces.

Figure 3 shows AFM images obtained from (2×2) μm² scan areas of undoped and Si-doped Ga₂O₃ films. Surface morphology of the undoped film showed typical surface features of homoepitaxial (010) Ga₂O₃ films as shown in Figure 3(a), consisting of alternating stripes and grooves along the [001] crystal orientation as indicated by the arrow. The root mean square (RMS) roughness value of the undoped Ga₂O₃ film was 0.63 nm, which meant the grown film surface is very smooth. The surface morphologies of the Si-doped β-Ga₂O₃ films also feature the elongated stripes and grooves along the [001] crystal orientation as shown in Figure 3 (b)-(d). However, stripes and grooves surface features were more clarified for Figure 3(c) and 3(d) compared to the undoped film. The surface morphology changed to the stripes and grooves with the enlarged width for Figure 3(d). The change of surface features with increasing the Si cell temperature resulted in the changes of the RMS roughness values. The RMS roughness values of the Si-doped β-Ga₂O₃ with the Si effusion cell temperatures of 910°C and 920°C showed almost similar and determined to 1.34 and 1.35, respectively. In case of the Ga₂O₃ film with the increase Si cell temperature of 930°C, it was decreased to 1.01 nm, but larger than that of the undoped film. From the AFM observations, it is obvious that Si doping causes a changes of the surface morphology and RMS roughness. Despite increased RMS roughness with the Si doping, the doped Ga₂O₃ films showed smooth RMS roughness values smaller than 1.4 nm, which implied that the grown films are well-suited for device fabrications.

Similar surface morphological feature of the [001] crystal orientation-elongated stripes and grooves on the surface of homoepitaxial (010) β-Ga₂O₃ films resulting in the increase of the surface RMS roughness values was also observed in H₂O vapor-assisted growth of the homoepitaxial (010) β-Ga₂O₃ films by MOCVD [²⁹]. The authors reported that the surface morphologies were characterized by the presence of parallel grooves with the (110) and ( 1 ¯ 10) facets elongated along the [001] crystal orientation, with the lateral spacing between the grooves and the peak-to-valley height of the films grown with H₂O vapor is greater than these of the film grown without H₂O vapor [²⁹]. The surface morphology, which forms the grooves with the presence of the (110) and ( 1 ¯ 10)-facets on the nominally (010) surface of Ga₂O₃ substrate after Ga-treatment, has been reported to be related to the free energies E₁₁₀ and E₀₁₀ of (110) and of (010) surfaces, where E₁₁₀/cos(14°) is less than E₀₁₀, and 14° is angle between (110), ( 1 ¯ 10) facets and (010) plane [³⁰]. Therefore, we believe the change in the surface morphology, which is the expansion of the [001] crystal orientation and the elongated stripes and grooves, as well as the increase of surface RMS roughness of our β-Ga₂O₃ films may be attributed to changes in surface free energy upon Si doping as explained in the literatures [³¹]. Additionally, changes in the surface morphologies and the increase in surface RMS values may be due to the distortion of the β-Ga₂O₃ crystal lattice caused by differences in the radius of the Si and Ga atoms.

Hall effect measurements at room temperature were conducted to examine the electrical properties of the homoepitaxial undoped and Si-doped (010) β-Ga₂O₃ samples. For Hall measurements, four electrodes were fabricated at the corners of each sample. Indium (In) was used as the ohmic contact and applied by melting through a soldering process, the I-V characteristics of (010) β-Ga₂O₃ samples with the Si temperatures at 910°C, 920°C, 930°C shown in Figure 4 (a). The undoped β-Ga₂O₃ sample exhibits extremely low electrical conductivity, below the detection threshold of the Hall effect measurement [³²]. Figure 4 (b), 4 (c) show plots of the carrier concentration, carrier mobility, and resistivity of the Si-doped Ga₂O₃ samples versus the Si effusion cell temperatures. As the Si effusion cell temperature increases, the carrier concentration increased, while the carrier mobility decreased as shown in Figure 4 (b). The increase in Si effusion cell temperature leads to a higher incorporation of Si into β-Ga₂O₃ crystal lattice, where the incorporated Si⁴⁺ can substitute for Ga³⁺ lattice sites, thereby increasing the carrier concentration. However, the incorporation of Si dopant in the β-Ga₂O₃ creates scattering centers in the lattice, which results in the decrease in the carrier mobility. The doping leads to a decrease in resistivity as shown in Figure 4 (c). Specific values of the carrier concentration, the carrier mobility, and resistivity of the homoepitaxial (010) Si-doped β-Ga₂O₃ samples are shown in Table 1.

4. CONCLUSIONS

Homoepitaxial undoped and Si-doped β-Ga₂O₃ films were grown by PAMBE with the different Si effusion cell temperatures. The Si concentration in the Si-doped β-Ga₂O₃ films could be controlled by employing the bulk solid Si source prepared through a simple solidification after the melting of the pieces of Si chip source in the crucible of the effusion cell. By this simple method the exposed Si source surface to the oxidizing growth environment was significantly reduced. All the β-Ga₂O₃ films grown on the (010) β-Ga₂O₃ substrates exhibited smooth surface and quality addressed from appearance of the strong streak RHEED patterns during the growth. The surface morphology of the Si-doped (010) β-Ga₂O₃ films retained the typical surface morphology features of homoepitaxial (010) β-Ga₂O₃ consisting of the [001] crystal orientation-elongated stripes and grooves. Although the features of stripes and grooves are clarified and enlarged, resulting in the increase of RMS roughness values compared to that of the undoped film, the homoepitaxial (010) Si-doped β-Ga₂O₃ films showed smooth surfaces with the RMS roughness values less than 1.4 nm. By changing the Si effusion cell temperature, the carrier concentrations could be achieved in the ranges from 8.25×10¹⁸ to 2.18×10²⁰ cm^-3 resulting in resistivities from 1.01×10^-2 to 5.79×10^-4 Ω·cm.