Continuous Si doping in (010) and (001) β-Ga₂O₃ films by plasma-assisted molecular beam epitaxy

β-Ga₂O₃ has attracted enormous attention due to its ultra-wide bandgap (∼4.8 eV)¹ and high breakdown field (∼8 MV/cm).² These exceptional properties make this material a promising candidate for next-generation power electronics. Compared to other ultra-wide bandgap semiconductors such as AlN and diamond, β-Ga₂O₃ has shallow hydrogenic donors³ such as Ge,^4–6 Sn,^7–9 and Si.^10–12 In addition, β-Ga₂O₃ is the only ultra-wide bandgap semiconductor that can be mass-produced as high-purity bulk substrates by the edge-defined fed growth (EFG) method¹³ and the Czochralski method.¹⁴ The availability of cost-effective bulk substrates allows the growth of β-Ga₂O₃ epitaxial films with superior crystal quality,^15–17 which leads to an exceptionally low background carrier concentration.^18,19 To measure the potential of a material for power applications, Baliga’s Figure of Merit (BFOM) is widely used.²⁰ While BFOM is based on the assumption that the donors are hydrogenic and the carrier concentrations are not affected by background compensation, this FOM can be modified to reflect the real activation energies of donors and background carrier concentrations. As a result, β-Ga₂O₃ significantly outperforms other wide-bandgap semiconductor materials for power applications when the doping abilities and materials purity limitation are taken into consideration.²¹ The crystal orientation of β-Ga₂O₃ bulk substrates for epitaxial growth is also critical toward the application of β-Ga₂O₃-based power electronics. Vertical power electron devices are typically large and, therefore,^22,23 the scalability of the β-Ga₂O₃ bulk substrates is the key to determining which crystal orientation is most efficient for the epitaxial growth. While the [010] zone is a scalable orientation in β-Ga₂O₃ crystal system, both the (100) and (001) orientations are in the [010] zone, which means they are suitable for large-scale bulk substrates. Moreover, even 100 mm diameter of (001) β-Ga₂O₃ oriented wafers are commercially available. Therefore, the (001) orientation is promising for future large-size power electronics applications.²⁴ 

Among all the shallow hydrogenic donors of β-Ga₂O₃, the importance of Si doping in β-Ga₂O₃ has been proven by its outstanding electrical properties and, therefore, its application to lateral and vertical electronic devices.^2,25–27 High-performance enhancement-mode metal-oxide semiconductor field effect transistors (MOSFETs)²⁸ and current aperture vertical electron transistors (CAVETs)^29,30 have been achieved by Si-doped β-Ga₂O₃ films. While ion implantation is a well-established technique to obtain Si-doping at certain areas in β-Ga₂O₃ films, damage from ion radiation is inevitable during this process.³¹ The Si-doping for β-Ga₂O₃ epitaxial film has been widely studied in different growth methods such as metal-organic chemical vapor deposition (MOCVD),^11,32 halide vapor phase epitaxy (HVPE),³³and pulsed laser deposition (PLD).³⁴ Moreover, for the application of β-(Al_xGa_1−x)₂O₃/β-Ga₂O₃ heterostructures, Si-doping in the β-(Al_xGa_1−x)₂O₃ has revealed its importance by showing superior conductivity³⁵ and record-low contact resistance.³⁶ On the other hand, the rapid oxidation of the Si source in molecular beam epitaxy (MBE) has limited the ability of Si doping to δ-doping.³⁷ Effusion cells with an endplate³⁸ and Si suboxide (SiO) source³⁹ have been used to achieve controllable Si-doping in β-Ga₂O₃ epitaxial films grown by MBE. Due to the existence of a high background O₂ partial pressure, typically >10⁻⁵ Torr, and the presence of active oxygen (presumably atomic oxygen O) inside the MBE growth environment, the Si source material in the effusion cell needs to be protected from oxidation. In this work, we report the continuous doping of Si in β-Ga₂O₃ epitaxial films on the (010) and (001) substrates by utilizing a valved effusion cell.

β-Ga₂O₃ epitaxial films were grown on Fe-doped (010) and (001) semi-insulating β-Ga₂O₃ substrates from Novel Crystal Technology by plasma-assisted MBE (PAMBE). The growth temperature was varied from 700 to 800 °C. The MBE system was equipped with a Veeco Unibulb radio-frequency plasma source running at 200 W with a foreline pressure of 60 Torr for the generation of O plasma to produce 1%–2% atomic O from pure O₂ gas. Prior to the epitaxial growth, the substrate was Ga polished and O polished at a substrate temperature of 800 °C to remove the impurities on the surface of β-Ga₂O₃ substrates. Ga polishing is proven to be capable of effectively removing impurities and chemical-mechanical polish (CMP)-related damage from the surface and smoothing the surface of both the (010) and (001) substrates.^4,40 O polishing by O plasma can further remove impurities to make sure the surface is ready for epitaxial growth. For Si doping, the temperature of the Si effusion cell was varied from 900 to 1200 °C. Figure 1 shows the schematic of a valved effusion cell used in this study. The cell uses a needle valve to control the flux into the growth chamber. The effusion cell of the Si source is designed to prevent the exposure of Si to the high oxygen background in the chamber while providing Si flux at the same time. The needle valve on the top of the crucible minimizes the exposure of the crucible to the environment in the main growth chamber while sustaining the Si flux from the source material. When the needle valve is fully open (100% open), the needle and the body seats are separated so that the Si flux can flow out of the cell through the openings. When the valve is fully closed (0% open), the needle and the body seats have no gap between each other, which results in no flux between the effusion cell and the growth. The opening of the needle valve is controlled by a programmable motor with a precision valve opening of ∼2.5 µm level to further control the Si doping concentration. While the Si flux can be controlled by changing the cell temperature, this requires time for cooling down/heating up and stabilizing the Si flux, whereas the needle valve can instantly change the Si flux to achieve controllable Si doping. During the growth, the Ga flux was varied from 1.0 × 10⁻⁷ to 3.0 × 10⁻⁷ Torr for conventional growth. We further performed metal-oxide catalyzed epitaxy (MOCATAXY) to improve the crystal quality of Si-doped (010) and (001) β-Ga₂O₃ epitaxial films.^41–43 During the MOCATAXY growth, In flux was provided simultaneously with Ga flux to serve as a catalyst for the β-Ga₂O₃ epitaxial growth.⁴⁴ Previously, the MOCATAXY growth showed that the growth rate of β-Ga₂O₃ epitaxial films was significantly enhanced by introducing an In flux into the MBE growth environment.⁴² This improvement in the growth rate of β-Ga₂O₃ epitaxial films is due to the catalytic mechanism of In during the growth of β-Ga₂O₃. In conventional MBE growth, Ga can only access atomic O as the active oxygen flux. On the other hand, in the MOCATAXY growth environment, In catalyst layer can access more oxygen, presumably molecular O₂. Subsequently, a metal-exchange mechanism between Ga and In then allows Ga to preferentially incorporate in the β-Ga₂O₃ epitaxial film over In. As all the In is replaced by Ga during the growth, pure β-Ga₂O₃ epitaxial films with high growth rates can be grown at sufficiently high Ga fluxes. In addition, the maximum growth temperature can also be enhanced due to the suppression of Ga₂O suboxide desorption by the presence of In in the MBE growth environment.⁴⁵ Here, the Ga flux was fixed at 2.5 × 10⁻⁷ Torr with In flux at 4.0 × 10⁻⁷ Torr for catalyzed growth. Prior to the growth of the Si-doped β-Ga₂O₃ epitaxial films, an unintentionally doped (UID) buffer layer was grown to prevent the impurities on the surface of the substrate from incorporating into the Si-doped layer. Followed by the UID buffer layer, a 2-nm-thick β-(Al_1−xGa_x)₂O₃ spacer layer with a target al composition of 2% to 3% was grown to enable the observation of thickness fringes in high resolution x-ray diffraction (HRXRD) ω-2θ scans, which helps determine the thickness of the Si-doped layer and characterize the crystal quality. Reflection high energy electron diffraction (RHEED) was used to observe the surface structure during the MBE growth. Surface morphologies of the epitaxial films were characterized by atomic force microscopy (AFM). For the determination of Si concentrations in the β-Ga₂O₃ epitaxial films, secondary ion mass spectrometry (SIMS) was used with a CAMECA IMS 7f tool. Hall measurements were performed with Ti/Au electrodes deposited on the corners of the samples in Van der Pauw geometry.

During the growth, RHEED patterns were monitored to confirm the surface structure of Si-doed β-Ga₂O₃ epitaxial films. Figures 2(a) and 2(b) show the RHEED patterns of conventional MBE growth and (c), (d) the MOCATAXY growth of Si-doped β-Ga₂O₃ epitaxial films. The thicknesses of the UID layers and Si-doped layers grown by both growth methods were 100 and 200 nm, respectively. The RHEED images of conventional MBE growth of the UID buffer layer and Si-doped β-Ga₂O₃ epitaxial films did not show clear streaky patterns indicative of rough surface morphologies. The pattern of the Si-doped film was even more blurry than that of the UID buffer layer, which indicates surface roughening during Si doping of β-Ga₂O₃ by conventional MBE growth. This suggests that Si can be an anti-surfactant in the β-Ga₂O₃ epitaxial growth, which is also confirmed by the following HRXRD and AFM results. On the other hand, both the UID buffer layer and the Si-doped film grown by MOCATAXY growth showed clear streaky patterns during the whole growth.

Figure 3(a) shows the SIMS result of a stack of Si-doped β-Ga₂O₃ layers separated by UID layers grown by MOCATAXY on a (010) β-Ga₂O₃ substrate. The spike in Si concentration at the film/substrate interface was due to the residual impurities on the surface of the β-Ga₂O₃ substrate though Ga polishing has reduced the Si concentration by more than one order of magnitude.⁴⁶ Continuous Si doping in β-Ga₂O₃ epitaxial films with flat doping profiles and sharp turn-on was achieved by exploiting the valved effusion cell. The Si doping concentration was able to be either controlled by the aperture of the needle valve or the temperature of the effusion cell with an upper limit of ∼1.0 × 10²⁰ cm⁻³ and a lower limit below the background impurity level of ∼5 × 10¹⁶ cm⁻³, which is the detection limit of SIMS measurement. The temperature of the Si effusion cell was varied from 900 to 1200 °C. At the effusion cell temperature of 1000 °C, the Si concentration was ∼4.0 × 10¹⁷ cm⁻³ with a needle valve opening of 20%. When the needle valve was fully open (100%), the Si concentration increased to ∼2.0 × 10¹⁸ cm⁻³, which is five times larger than that of the needle opening at 20%. This result demonstrates that the linear control of Si concentration in β-Ga₂O₃ epitaxial film was realized by utilizing the valved effusion cell. On the other hand, some surface segregation effect was observed in the turn-off of the doping profiles, which indicates possible surface riding of Si or a “memory” effect during the growth of β-Ga₂O₃ epitaxial films on (010) β-Ga₂O₃ substrates.

An additional Si-doped (010) β-Ga₂O₃ SIMS-stack sample was prepared to investigate the Si doping behaviors at high temperatures (>1000 °C) range. Each Si-doped layer was grown for 240 nm with a growth rate of 4 nm/min to confirm the stability of Si doping during the growth. The SIMS profile of Si-doped (010) β-Ga₂O₃ film shown in Fig. 3(b) reveals the cell temperature dependence of Si doping concentrations. The Si-doped layers with Si effusion cell temperatures at 1200, 1100, and 1000 °C showed flat Si doping profiles and sharp turn-on/off. Moreover, two doping regions were observed from a single Si-doped layer. The first region (Region A) of the doping has a slightly larger Si concentration than the following region (Region B). Considering the needle valve is closed during the growth of the UID layer (prior to the Si-doped layers), excessive gas species, including Si and suboxide SiO, that could be active dopants after being incorporated into the β-Ga₂O₃ epitaxial films, were accumulated inside the Si effusion cell. Since this phenomenon is more obvious when the Si effusion cell is at a higher temperature, this suggests that the continuous Si doping by the valved cell is conducted by gas species that are able to accumulate in an effusion cell.^47,48 Figure 3(c) shows the apparent activation energy extracted from the Arrhenius plot of the cell temperatures and Si doping concentrations. The activation energy showed a value of about 1.3 eV, which was lower than that of the element Si flux (∼5.0 eV)⁴⁹ or SiO flux (∼3.6 eV).³⁹ However, the identification of the exact gas species coming out of the Si effusion cell needs to be further investigated at actual MBE growth conditions and is beyond the scope of this study.

The crystal quality of the Si-doped (010) β-Ga₂O₃ epitaxial films was investigated by HRXRD, as shown in Fig. 4. While the film with a film thickness of 200 nm and Si concentration under 1 × 10¹⁷ cm⁻³ show clear thickness fringes, indicating a high quality film and an abrupt substrate–film interface. Thickness fringes were not observed from films with Si concentrations beyond 6 × 10¹⁷ cm⁻³, either by MOCATAXY (film thickness: 200 nm) or by conventional PAMBE (film thickness: 150 nm). In addition, a peak appears next to the (020) β-Ga₂O₃ peak on the higher angle side at Si concentrations larger than 6 × 10¹⁷ cm⁻³. This can be attributed to the strain introduction while Si is doped into (010) β-Ga₂O₃ films. The coherency strain calculated from the peak separation⁵⁰ from the β-Ga₂O₃ (020) peak showed a value of −0.023%, which indicates Si doping might introduce tensile stress into β-Ga₂O₃ epitaxial films during the growth. Further work is ongoing to understand the peak separation. Figure 5 shows the AFM images of Si-doped β-Ga₂O₃ films grown by conventional MBE and MOCATAXY. The surface morphologies of Si-doped β-Ga₂O₃ films with Si concentrations above the mid-10¹⁷ cm⁻³ range have larger RMS roughness than films with Si concentrations under 10¹⁶ cm⁻³. Hall measurement was performed on Si-doped β-Ga₂O₃ epitaxial films grown on (010) substrates by conventional MBE growth and MOCATAXY growth. While MOCATAXY-grown Si-doped (010) films showed 1.8 × 10¹⁸ cm⁻³ Hall concentration and 72 cm²/Vs mobility, conventional MBE-grown Si-doped (010) films showed mobility under 10 or even unmeasurable results. This suggests that MOCATAXY growth can enhance the electronic properties by improving the crystal quality of β-Ga₂O₃ epitaxial films. Therefore, the HRXRD, AFM, and Hall results suggest that Si is potentially an anti-surfactant during the growth of β-Ga₂O₃ epitaxial films causing microstructure degradation and surface roughening along with the previous studies showing Si is known as an anti-surfactant in GaN growth.^51,52 On the other hand, MOCATAXY growth showed its potential to act as a surfactant to counteract the anti-surfactant effect of Si in β-Ga₂O₃ epitaxial growth.

The Si doping availability of β-Ga₂O₃ was further investigated on the scalable (001) substrates. Since the (001) substrates were co-loaded with the (010) substrates in the MBE system, the growth conditions of the (001) Si-doped films were identical to those on the (010) substrates, which were shown in Sec. II. By growing with MOCATAXY, co-loaded (010) and (001) β-Ga₂O₃ films exhibited the best crystal qualities at the same growth conditions, which were confirmed by Mauze et al.^42, Figure 6 shows the HRXRD and AFM results of continuous Si-doped β-Ga₂O₃ films grown on (001) substrates by MOCATAXY growth. Clear thickness fringes in the XRD patterns indicate the Si-doped (001) β-Ga₂O₃ epitaxial films have superior crystal quality and smooth surface morphology even at the high Si concentration of 10¹⁸ cm⁻³ range. The growth rates on (001) substrates were 5.0 nm/min, which corresponds to the results from previous MOCATAXY-grown UID films.^42,43 The RHEED patterns during the MOCATAXY growth of Si-doped β-Ga₂O₃ epitaxial films showed clear streaky patterns, along with the AFM image showing the smooth surface morphology. The results of Hall measurements of the Si-doped films on (001) substrates are shown in the figures. For the 200-nm-thick Si-doped film with a Hall electron concentration of 3 × 10¹⁸ cm⁻³, the electron mobility was 67 cm²/Vs. The electron mobility of this Si-doped film is superior to previous studies on Sn-doped and Ge-doped β-Ga₂O₃ epitaxial films grown on (001) substrates.^24,42,53

Figure 7 shows the SIMS depth profile of this 100-nm-thick Si-doped β-Ga₂O₃ epitaxial film grown on the (001) substrate with the Hall measured electron concentration of 9 × 10¹⁸ cm⁻³ and electron mobility of 49 cm²/Vs. This result shows that a flat Si doping profile of 8.3 × 10¹⁸ cm⁻³ with sharp turn-on was achieved throughout the Si-doped (001) β-Ga₂O₃ epitaxial film. Additionally, this indicates that the doped Si has a sufficiently high ionization efficiency, which is also proven by previous studies suggesting that Si is a shallow donor in the β-Ga₂O₃ materials system.³ Therefore, the continuous Si-doping capability in β-Ga₂O₃ epitaxial films grown on (001) substrates was achieved in the range from below the background impurity level to as high as ∼1.0 × 10²⁰ cm⁻³.

In conclusion, continuous Si doping in β-Ga₂O₃ epitaxial films on (010) and (001) substrates grown by PAMBE was achieved by utilizing the valved effusion cell for the Si source material. By varying the effusion cell temperatures and valve aperture, the Si doping concentration can be precisely controlled on a linear scale corresponding to the opening of the needle valve. The clear thickness fringes confirmed from the HRXRD results prove that the Si-doped MOCATAXY-grown β-Ga₂O₃ epitaxial films exhibited high crystal quality. In addition, AFM images showed that the surface morphology of Si-doped films was smooth with a value of RMS roughness of around 1.0 nm. Controllable continuous doping is crucial to the β-Ga₂O₃ toward the application of power electronics, especially vertical devices that require a large uniform donor doping area.

The authors acknowledge funding from AFOSR through the GAME MURI Program, Award No. FA9550-18-1-0479, under project manager Dr. Ali Sayir.

The authors have no conflicts to disclose.

Takeki Itoh: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead). Akhil Mauze: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting). Yuewei Zhang: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Supervision (supporting); Validation (supporting). James S. Speck: Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (lead); Writing – review & editing (lead).

The data that support the findings of this study are available within the article.