Research ArticleENGINEERING

Crystal orientation dictated epitaxy of ultrawide-bandgap 5.4- to 8.6-eV α-(AlGa)2O3 on m-plane sapphire

See allHide authors and affiliations

Science Advances  08 Jan 2021:
Vol. 7, no. 2, eabd5891
DOI: 10.1126/sciadv.abd5891

Abstract

Ultrawide-bandgap semiconductors are ushering in the next generation of high-power electronics. The correct crystal orientation can make or break successful epitaxy of such semiconductors. Here, it is found that single-crystalline layers of α-(AlGa)2O3 alloys spanning bandgaps of 5.4 to 8.6 eV can be grown by molecular beam epitaxy. The key step is found to be the use of m-plane sapphire crystal. The phase transition of the epitaxial layers from the α- to the narrower bandgap β-phase is catalyzed by the c-plane of the crystal. Because the c-plane is orthogonal to the growth front of the m-plane surface of the crystal, the narrower bandgap pathways are eliminated, revealing a route to much wider bandgap materials with structural purity. The resulting energy bandgaps of the epitaxial layers span a broad range, heralding the successful epitaxial stabilization of the largest bandgap materials family to date.

INTRODUCTION

Since the middle of the last century, semiconductor materials have made a steady climb to the top of the periodic table to wider energy bandgaps. Starting from the early days of Ge to the Si revolution and the current wide-bandgap SiC and GaN materials, each new generation of semiconductors has enabled electronic, photonic, and sensing and actuation functions in regimes that were considered impossible before. Next-generation electronics and photonics need much larger bandgaps beyond the established material families. Ultrawide-bandgap (UWBG) semiconductor materials for power and microwave electronics and deep–ultraviolet (UV) photonics include Aluminum Nitride (AlN), Boron Nitride (BN), diamond, and Ga2O3 (14). The recent availability of single-crystal β-Ga2O3 substrates has generated substantial interest in this wide-bandgap semiconductor material family for high-voltage electronics and UV photonics (5, 6). However, efforts to create β-(AlGa)2O3 with high Al contents to obtain UWBGs to exceed the bandgap of ~6 eV available in the nitride material family in AlN have met a roadblock. Because β-(AlGa)2O3 is not the energetically favored crystalline phase for large Al compositions, the crystal converts to competing structural phases when grown on β-Ga2O3 substrates (710).

This leads to the intriguing question: Can single-phase and highly crystalline α-Al2O3 of bandgap Eg ~ 8.8 eV (11) and its alloys and heterostructures with α-Ga2O3 of bandgap Eg ~ 5.3 eV (1214) be grown directly by molecular beam epitaxy (MBE) on α-Al2O3 (sapphire) substrates? If this can be done, and if these layers can be controllably doped, it will open new application arenas. The UWBG energies (~5.3 to 8.8 eV), a large portion of which is unachievable in the nitride semiconductor family, and the attractive dielectric constants (~11 for || c axis and ~ 9 for ⊥ c axis for α-Al2O3) (15, 16) offer the possibility to take semiconductor electronics and photonics into regimes that currently remain out of reach. In this work, we report that the first of these challenges can be met: High–crystalline quality, single-phase α-(AlGa)2O3 can be grown by MBE directly on m-plane sapphire, spanning bandgaps of ~5.3 to 8.8 eV, i.e., the entire Al composition of the alloy from 0 to 1. Although the conductivity control of all compositions is not achieved over the entire range of bandgaps yet, experimental evidence of n-type α-Ga2O3 and theoretical prediction for (AlGa)2O3 indicate that this may be feasible (1719). The bandgaps and sharp heterostructures achieved by MBE put this material family beyond those that are available today, and the cost-effective, large-area substrates for achieving this are already available.

Significance of crystal orientation

The orientation of the substrate crystal determines the subsequent crystal phases in unique ways in this UWBG material family. Homoepitaxial growth of α-Al2O3 films has been reported on c-, a-, and r-plane substrates using MBE (20, 21) and pulsed laser deposition (PLD) (22, 23), but no study of their optical properties exist. On the other end of the α-(AlxGa1−x)2O3 alloy system, single-crystalline α-Ga2O3 films and conductive, n-type doped films are successfully grown on c- and m-plane sapphire substrates by mist chemical vapor deposition (CVD) (12, 17) and halide vapor-phase epitaxy (HVPE) (24, 25). α-(AlGa)2O3 alloys have also been grown using MBE on a-plane (0 ≤ x ≤ 1) (26), mist-CVD on c-plane (0 ≤ x ≤ 0.81) (2730), and PLD on r-plane sapphire substrates (0.92 ≤ x ≤ 1) (23). Among the growth methods, MBE and MOCVD offer the capability to maintain sharp interfaces across layers in quantum heterostructures of different alloy compositions (8, 21, 31, 32), as well as sharply controlled doping profiles, which are problematic with HVPE or mist-CVD techniques. MBE growth of α-(AlxGa1−x)2O3 of high Ga content is hampered because metastable α-Ga2O3 in the rhombohedral corundum structure has a propensity to revert to the thermodynamically most stable monoclinic β-Ga2O3, especially whenever a corundum c-plane becomes available. This can occur either on the sapphire substrate itself or in the underlying corundum structure of the epitaxial layers. The growth of β-Ga2O3 on c-plane sapphire has been obtained using MOVCD and PLD (33), while mist-CVD and HVPE (12, 24, 30) have been used for the growth of pure α-(AlGa)2O3 on c-plane sapphire. However, as described below, the stability of the α-phase grown by MBE, MOCVD, and PLD differs significantly from those grown by mist-CVD and HVPE (12, 24, 33), which currently is a scientifically intriguing phenomena. In addition, preventing this phase transformation in MBE and MOCVD holds the key to access desired abrupt heterostructures and sharp doping profiles in this UWBG family, as indicated in Fig. 1.

Fig. 1 Schematics of growth behaviors of Ga2O3 grown on sapphire.

Schematics of growth behaviors of Ga2O3 grown on (A) c-plane, (B) r-plane, and (C) m-plane sapphire substrates by MBE and (D) unit cells of sapphire, α-Ga2O3, and β-Ga2O3. β-Ga2O3 is grown on c-plane sapphire substrates with the coherent α-Ga2O3 interlayer. (33) The epitaxial relationship is (2¯01) β-Ga2O3 || (0001) α-Ga2O3/sapphire. On r-plane sapphire, after c-plane facets are formed on the surface of the α-Ga2O3 layer, β-Ga2O3 appears with the same epitaxial relationship of (2¯01) β-Ga2O3 || (0001) α-Ga2O3 (14). On the other hand, m-plane, which is perpendicular to the c-plane, is expected to have less chance to form the c-plane facets and allow the growth of phase-pure α-Ga2O3.

The growth of β-Ga2O3 on c-plane sapphire substrates by MBE, MOCVD, and PLD has been found to have the epitaxial relationship of (2¯01) β-Ga2O3 || (0001) sapphire (33, 34). Recent studies by scanning transmission electron microscopy (STEM) have revealed that when grown on c-plane sapphire, a three-monolayer-thick coherent α-Ga2O3 first forms, and the subsequent epitaxial layers undergo a crystalline phase transition to the β-phase (33), as shown in Fig. 1A. Also shown in the figure are the planes (2¯01) β-Ga2O3 || (0001) sapphire, which cause this transition. As shown in Fig. 1A, if the hexagonal sapphire crystal plane is rotated from the c-plane to the r-plane, the c-plane is now at an angle and can potentially stabilize the growth of α-Ga2O3 by avoiding the crystalline phase transition. In 2018, Kracht et al. attempted this strategy: They reported the growth of α-Ga2O3 on r-plane sapphire substrate with a film thickness of 217 nm (14). They observed that after c-plane facets were formed on the surface of the α-Ga2O3 film, the β-Ga2O3 appeared with the same epitaxial relationship of (2¯01) β-Ga2O3 || (0001) α-Ga2O3/sapphire, as shown in Fig. 1B, indicating that the c-plane facet enhanced the growth of β-Ga2O3 in MBE. These results suggest that a sapphire crystal plane perpendicular to the c-plane, such as a- or m-planes, could potentially allow the growth of phase-pure α-Ga2O3 by avoiding facets, as shown in Fig. 1C. Nd-doped α-(AlxGa1−x)2O3 and α-Ga2O3 with a film thickness of 14 nm have been reported on a-plane sapphire in previous studies (26, 34). These prior studies provided x-ray diffraction (XRD) study of the layers, but the atomic crystal structure and optical properties remain unknown.

The MBE growth of α-(AlxGa1−x)2O3 material family on m-plane sapphire has not been investigated in detail. However, Sn-doped α-Ga2O3 of bandgap ~5.3 eV on m-plane sapphire grown by mist-CVD exhibits a room temperature electron mobility of 65 cm2/V·s at a high carrier density of n = 1.2 × 1018 cm−3 (18), suggesting promising electronic properties of α-(AlxGa1−x)2O3 on m-plane sapphire. High room temperature electron mobility of ~600 cm2/V·s was concluded for optically generated electrons in α-Al2O3, increasing to ~4000 cm2/V·s at low temperatures (35). A recent first-principles study predicts that Si can be an efficient shallow donor for high–Al content (AlxGa1−x)2O3 (19), to achieve n-type conductivity of (AlxGa1−x)2O3.

Motivated by the above reasons, in this study, we explore growth of this UWBG semiconductor family on m-plane sapphire substrates. We find that the MBE growth of α-(AlxGa1−x)2O3 on m-plane sapphire solves the faceting problem completely, allowing its bandgap engineering over the entire composition range to be achieved. A series of epitaxial layers of α-(AlxGa1−x)2O3 with x = 0.0 to 1.0 and thicknesses ranging from ~50 to 85 nm were grown by oxygen plasma-assisted MBE on m-plane sapphire substrates. The growth and subsequent structural, chemical, and optical characterization methods are described in Materials and Methods.

RESULTS AND DISCUSSION

Single-crystal UWBG epitaxial layers

Symmetric XRD 2θ/ω scans of the samples are displayed in a logarithmic intensity scale in Fig. 2A. The profiles reveal that the diffraction peak is from the α-(AlxGa1−x)2O3 303¯0 reflection at angles slightly lower than that of the sapphire substrate. There are no peaks originating from other crystal phases or other planes. With decreasing Al composition x, the peaks from α-(AlxGa1−x)2O3 303¯0 monotonically shift to lower angles without any compositional segregation, suggesting a higher Ga composition in the film. No additional peaks or fringes are observed for the MBE-grown α-Al2O3 epitaxial layer, which is identical to the substrate despite the ~60-nm thickness, as verified by secondary ion mass spectrometry (SIMS) discussed later. From these results, we conclude that single-phase α-(AlxGa1−x)2O3 films are epitaxially stabilized successfully on m-plane sapphire substrates for the entire range of Al compositions without phase or composition separation/segregation.

Fig. 2 XRD scans of α-(AlxGa1−x)2O3 films and SIMS profiles of the α-Al2O3 epitaxial film.

(A) Symmetric XRD 2θ/ω scans of α-(AlxGa1−x)2O3 films grown on m-plane sapphire substrates by plasma-assisted MBE. Diffraction peaks from the α-(AlxGa1−x)2O3 are denoted by the filled circles. Note that no diffraction peaks from other phases are observed. (B) Atomic densities of typical impurities in the α-Al2O3 epitaxial film grown on m-plane sapphire measured by SIMS. The film/substrate interface is marked by the clear peak in the B profile, indicating that the epitaxial film thickness is ~60 nm. The concentrations of H and C are at background levels. Secondary ion intensities of Al and O are also shown on the side axis on the right, indicating stoichiometric epitaxial growth.

Figure 2B shows the SIMS profiles measured on the homoepitaxial 100% α-Al2O3 layer grown on m-plane sapphire. The atomic concentrations of typical impurities (H, B, C, Na, and Si) are indicated by the axis on the left, while secondary intensities of Al and O are shown on the axis on the right. The intensities of Al and O in the epitaxial layer are observed to be identical to that in the m-plane sapphire substrate, suggesting growth of stoichiometric Al2O3. Compared to the sapphire substrate, a large B peak is seen at the growth interface, which is a marker indicating that the film thickness of the MBE-grown α-Al2O3 epitaxial layer is ~60 nm. There is an increase in the Si concentration at the growth interface, whereas the H and C concentrations in the epitaxial layer remain identical to the m-plane sapphire bulk substrate. The profiles for H and C represent the background levels of the measurements since densities of ~1018 cm−3 are not expected in sapphire substrates. The Na concentration is ~1016 cm−3. At present, we assume that the source of B is the pBN crucibles used for the MBE source materials, that of Si is from the quartz tube used as a component of the O plasma source, and that of Na is from the sapphire substrate. Si could accumulate during the substrate treatment using O plasma before growth as discussed in Materials and Methods.

To evaluate the strain in the epitaxial films, asymmetrical reciprocal space maps (RSMs) for 224¯0 reflections were taken. Figure 3A shows the relation between the (224¯0) and (101¯0) m-plane of the corundum structure. These results indicate that the α-(AlxGa1−x)2O3 films (0.37 ≦ x ≦ 0.74) consist of coherent and relaxed layers, and the relaxed layers are subject to slight in-plane compressive strain (Fig. 3, C to F). On the other hand, the α-(AlxGa1−x)2O3 films (x < 0.37) were completely lattice-relaxed because of the lattice mismatch with the sapphire substrate (fig. S1). Because 224¯0 decomposes into 303¯0 + 1¯21¯0, the directions of Qz and Qx are along 101¯0 (the growth direction) and 1¯21¯0 (one of the a axes), respectively. The dashed red line intersects the theoretical positions of sapphire [(Qx, Qz) = (−0.420, 0.728)] and α-Ga2O3 224¯0 reflection [(Qx, Qz) = (−0.401, 0.695)]. The α-Al2O3 homoepitaxial film shows a streak along Qx = −0.42 Å−1 (Fig. 3B), which is also observed for the reflections at Qx = −0.42 Å−1 in the samples whose x is larger than 0.37. The vertical streak arises from a small film thickness. Two reflection peaks of α-(AlxGa1−x)2O3 are obtained at x between 0.37 and 0.74: One is on the same Qx line (Qx = −0.42) as that of the sapphire substrate and the other is located between the dashed line and Qx = −0.42 Å−1 (Fig. 3, C to F).

Fig. 3 RSMs (Qx, Qz) around the 224¯0 reflections of the α-(AlxGa1−x)2O3 epitaxial films.

(A) The relation between the (224¯0) and (101¯0) m-plane of the corundum structure. The directions of Qz and Qx are along 101¯0 and 1¯21¯0, respectively. RSMs for x = (B) 0, (C) 0.74, (D) 0.59, (E) 0.54, and (F) 0.37.

Figure 4 shows the surface atomic force microscopy (AFM) images of the m-plane sapphire substrate and the epitaxial α-(AlxGa1−x)2O3 films. The surface morphology of the as-received substrate was extremely smooth with a root mean square (RMS) roughness as small as 0.079 nm (Fig. 4A). The α-Al2O3 epitaxial film shows a small RMS roughness of 0.32 nm (Fig. 4B) but does not exhibit atomic steps, which are not observed on the starting substrate either. Curiotto et al. reported a step-terrace morphology for m-plane sapphire substrates by annealing under Ar-O2 at 1235 K for 78 hours (36), suggesting thermal treatment for potential improvement of the surface morphology of α-Al2O3 films in the future. When the Al mole fraction x is smaller than 0.54, the RMS roughness remains smaller than 1.1 nm (Fig. 4, E to H). For x = 0.59 and x = 0.74, α-(AlxGa1−x)2O3 films have rougher surface morphologies of 3.5 and 1.42 nm RMS values (Fig. 4, C and D), probably due to the low substrate temperature of 650°C.

Fig. 4 Surface morphology of α-(AlxGa1−x)2O3 films on m-plane sapphire substrates.

(A) Sapphire substrate. (B) to (H) are α-(AlxGa1−x)2O3 films with x = 1, 0.74, 0.59, 0.54, 0.37, 0.22, and 0, respectively. The RMS roughness of the α-(AlxGa1−x)2O3 films for x ≤ 0.54 is less than 1.1 nm.

Electron microscopy reveals single-crystalline UWBG layers

The XRD spectra showed the growth of single-phase α-(AlxGa1−x)2O3 films on m-plane sapphire substrates. To investigate the atomic details of the crystal structures of the epitaxial layers, high-angle annular dark-field STEM (HAADF-STEM) images were taken on the cross-sectional α-(AlxGa1−x)2O3 samples for x = 1, x = 0.37, x = 0.22, and x = 0. All the images were viewed along the 〈0001〉 zone axis. Figure 5 shows the cross-sectional images of the α-(AlxGa1−x)2O3 film series grown on m-plane sapphire substrates and the corresponding enlarged images at the interfaces. As shown in Fig. 5A, since the epitaxial α-Al2O3 layer and the substrate have the same contrast, the exact location of the interface is invisible. On the basis of the corroborating SIMS measurement of Fig. 2B, an enlarged image was taken at the interface that was formed 60 nm below the surface. The uniformity of the crystal structure indicates that epitaxial α-Al2O3 films grow homoepitaxially on m-plane sapphire substrate with no visible structural defects.

Fig. 5 HAADF-STEM images showing an overview of α-(AlxGa1−x)2O3 film grown on m-plane sapphire and the sequentially enlarged interfaces from the boxes in the overview images.

(A) to (D) correspond to x = 1, x = 0.37, x = 0.22, and x = 0, respectively. (A) The film and the substrate show identical contrast, as expected for a homoepitaxial film. The enlarged area shows no differences in contrast and lattice structure. (B) to (D) show film thickness of 66, 55, and 60 nm, respectively. They all show sharp interfaces, while the films show contrast variations from strain relaxation resulting from misfit dislocations near the interface. Further details of the defects are explained in the Supplementary Materials. Enlarged atomic resolution image in (D) shows that α-Ga2O3 film is relaxed at the interface; however, no other phases are observed.

On the other hand, a clear contrast between the epitaxial film and the substrate is seen for the images of the α-(Al0.37Ga0.63)2O3, α-(Al0.22Ga0.78)2O3, and α-Ga2O3 samples (Fig. 5, B to D). The low-magnification images reveal that these films grow uniformly on m-plane sapphire substrates. Contrast variations inside the film are due to strain relaxation by misfit dislocations near the substrate interface. The edge dislocation density is estimated at ~5 × 105 cm−1 (or ~1011 cm−2) from the cross-sectional TEM sample of α-(Al0.37Ga0.63)2O3, in the range also corroborated by x-ray rocking curves as shown in fig. S2 and table S2 discussed in the Supplementary Materials. This is consistent with the incoherent nature of the α-(Al0.37Ga0.63)2O3 epitaxial layer indicated by RSM of Fig. 3F compared to the coherent α-Al2O3 epitaxial layer in Fig. 3B. No notable phase separation or compositional segregation is observed in the epitaxial layers: They are phase pure. The film thicknesses of the α-(Al0.37Ga0.63)2O3, α-(Al0.22Ga0.78)2O3, and α-Ga2O3 films are measured to be 66, 57, and 60 nm, respectively, which agree well with the thicknesses (d) estimated from x-ray reflectivity (XRR) measurements (table S1). The enlarged images show that sharp heterojunction interfaces are formed between the epitaxial films and the substrates and that the α-(AlxGa1−x)2O3 (x ≤ 0.37) films have an identical crystal structure and orientation to the substrate. α-Ga2O3 films show dislocations at the interface due to lattice mismatch (Fig. 5D and fig. S3); however, no secondary phases were observed. The combination of the STEM images with XRD thus indicates that single-crystalline, single-phase epitaxial films are successfully grown on the m-plane sapphire substrate over the full 0 ≤ x ≤ 1 composition range.

Deep-UV spectroscopy reveals UWBGs

Figure 6A shows the optical transmittance spectra of the α-(AlxGa1−x)2O3 films (T) and a bare m-plane sapphire substrate control sample (Ts) as a function of the photon energy (). Each epitaxial film shows the same high transparency in the visible and UV spectral ranges as the substrate. The bandgaps of the epitaxial α-(AlxGa1−x)2O3 layers are clearly observed despite their small thicknesses, indicating strong photon absorption. Note that the transmittance of α-(Al0.37Ga0.63)2O3 appears lower than the other samples because of a smaller sample size of this particular sample than the optical aperture. Each absorption coefficient (α) was calculated using the equation T/Ts = exp(−αd), and (α)2 is plotted as a function of in Fig. 6B. When x = 0 and 0.22, the absorption edge exhibits a double step-like onset, while the α-(AlxGa1−x)2O3 (x ≥ 0.37) has a single step onset. The observation of double step-like onset in α-Ga2O3 thin films has been reported earlier, suggesting two allowed direct optical interband transitions (13, 14). The Ga-rich α-(AlxGa1−x)2O3 films (0 ≤ x ≤ 0.37) show broader absorption onsets than the Al-rich samples, which is related to the slightly indirect bandgap character of α-Ga2O3 (13).

Fig. 6 Optical transmittance spectra and bandgap energies of the α-(AlxGa1−x)2O3 films.

(A) Optical transmittance of the α-(AlxGa1−x)2O3 films and substrate. (B) (αhν)2- plot as a function of the Al composition. The α-(Al0.37Ga0.63)2O3 film exhibited a lower transmittance than the other samples since the sample was smaller than the aperture for the transmittance measurement. The appearance of higher transparency of the samples with x = 0 and x = 0.22 is due to different measurement systems. Compared to the absorption spectra of the bulk α-Al2O3 substrate, the transmittance and absorption spectra of the α-Al2O3 epitaxial sample shows absorption around 7 to 8.5 eV and what appears as a slight red shift in the fundamental absorption edge. Both are attributed to absorption by sub-bandgap defect states (likely O and/or Al interstitials or vacancies) since the TEM image in Fig. 5A does not show extended defects, nor does the SIMS profile (Fig. 2B). Direct bandgaps (solid symbols) for the α-(AlxGa1−x)2O3 films as a function of the Al composition are shown in (C) the wide range from 0 to 10 eV and (D) the UWBG region (Eg > 4 eV). The red dashed line is a quadratic fit to the direct bandgaps. The black solid line is a quadratic fit to the computed direct bandgap for the corundum structure as reported in (37). The bandgaps for AlGaN and AlGaAs are also shown for comparison in (C) (38).

Direct optical bandgap energies of the α-(AlxGa1−x)2O3 films were estimated from the relationship of (α)2 versus (-Eg). As shown by the red filled circles in Fig. 6C, the experimental direct bandgaps monotonically increase as the Al composition x increases. In Fig. 6D, the black solid line shows the theoretical direct bandgap energies calculated by Peelaers et al. (37) using hybrid density functional theory. The experimentally measured bandgaps of the α-(AlxGa1−x)2O3 epitaxial layers are, for the most part, in excellent agreement. A ~0.2 eV lower than the predicted values from the simple Tauc plots of (α)2 versus (-Eg) may be due to an underestimation of the bandgap energy owing to excitonic absorption with low-energy tails, related to the slightly indirect character of α-Ga2O3 and to as-yet unknown defects (13). The bowing parameter (b) is obtained from Eg(x) = (1 − x)Eg[Ga2O3] + xEg[Al2O3] − bx(1 − x), where Eg[Ga2O3] and Eg[Al2O3] are the bandgap energies of α-Ga2O3 and α-Al2O3, respectively, by fitting the plot as depicted by the red dashed line. The value of b is estimated to be 1.1 eV. This experimental value agrees well with the theoretically calculated value of 1.37 eV and proves the UWBG nature of the epitaxial layers available for bandgap engineering from 5.4 to 8.6 eV.

In conclusion, we have demonstrated the successful epitaxial stabilization of UWBG single-phase α-(AlxGa1−x)2O3 films over the entire composition range on m-plane sapphire substrates by MBE by avoiding phase transformation with the choice of crystal orientation for epitaxy. Cross-sectional HAADF-STEM images reveal that single-phase, single-crystalline α-(AlxGa1−x)2O3 epitaxial films are formed, by avoiding the formation of c-plane facets. By varying the alloy composition, bandgap energies from ~5.4 eV up to 8.6 eV with a bowing parameter of 1.1 eV are achieved, making α-(AlxGa1−x)2O3 the largest bandgap epitaxial material family to date. If these layers can be controllably doped, it would pave the way for α-(AlxGa1−x)2O3–based high-power heterostructure electronic and photonic devices at bandgaps far beyond all materials available today.

MATERIALS AND METHODS

Materials

All films reported in this work were grown on double-side polished m-plane (101¯0) sapphire substrates in the Veeco Gen930 MBE System equipped with standard effusion cells for elemental Ga and Al and a radio frequency (RF) plasma source for active oxygen species. The pressure in the growth chamber was ~10−5 to 10−6 torr during the growth runs. For the growth of α-(AlxGa1−x)2O3 films (x ≥ 0.37), the input RF power and oxygen flow rate were fixed at 250 W and 0.50 sccm (standard cubic centimeters per minute), respectively. The α-Al2O3 and α-(AlxGa1−x)2O3 films (1 > x ≥ 0.37) were grown for 2 hours at thermocouple substrate temperatures (Tsub) of 750°C and 650°C, respectively. The α-(AlxGa1−x)2O3 (x = 0.22 and 0) samples were grown at Tsub = 650°C using the input RF power of 260 W and the oxygen flow rate of 0.23 sccm. The growth times of α-(Al0.22Ga0.78)2O3 and α-Ga2O3 were 3 hours and 3 hours 13 min, respectively. The Al concentration (x) in the films was varied from 0 to 1 by regulating the beam equivalent pressures (BEPs) of Ga [(0 to 1.1) × 10−8 torr] and Al [(0 to 1.6) × 10−8 torr]. Before each growth, the m-plane sapphire substrates were subjected to an oxygen plasma treatment in the growth chamber for 10 min at Tsub = 800°C.

Characterization of materials

Structural properties and surface morphology of the samples were characterized by XRD measurements, AFM, and STEM. The film thicknesses of the samples (x < 1) were determined by XRR measurements, while SIMS was used to determine the chemical impurity species as well as the thickness of the α-Al2O3 epitaxial layer, since there were no thickness fringes observed from the XRR profile for the α-Al2O3 film due to homoepitaxial growth. The film thicknesses were in the range of 56 to 84.3 nm. Table S1 summarizes the Al and Ga BEPs and film thicknesses (d) of the samples. The Al composition was estimated by XRD and x-ray photoelectron spectroscopy (XPS). Optical bandgaps of the films were estimated by analyzing the optical transmission spectra; an m-plane sapphire substrate was used as a reference sample. The optical transmittance spectra were measured at room temperature using a Varian Cary 50 UV-Vis spectrometer with a wavelength range of 190 to 800 nm for x ≤ 0.22 and using a Bunkoukeiki KV-201 vacuum UV spectrophotometer with a wavelength range of 140 to 300 nm for x ≥ 0.37. The Bunkoukeiki KV-201 unit had a 20-cm focal-length Czerny-Turner monochromator equipped with a 1200 groove/mm grating and a 30-W deuterium lamp that were purged with nitrogen gas during operation. A separate α-Al2O3 sample used for transmittance measurement was grown at Tsub = 750°C for 2 hours using an input RF power of 250 W, an oxygen flow rate of 0.50 sccm, and an Al BEP of 1.5 × 10−8 torr. By inserting a thin α-Ga2O3 layer to intentionally introduce thickness fringes, the film thickness of 78.2 nm was measured by XRR. We used the sample directly grown on sapphire in the same growth condition for the transmittance measurement.

Transmission electron microscopy measurement

Cross-sectional TEM specimens were prepared using a FEI Strata 400 Focused Ion Beam with a final milling step of 5 keV to reduce damage. Carbon and platinum protective layers were deposited before milling to minimize ion-beam damage. The samples were then examined by STEM, using an aberration-corrected Titan Themis operating at 300 keV.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/7/2/eabd5891/DC1

https://creativecommons.org/licenses/by-nc/4.0/

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: We acknowledge useful discussions with M. Thompson, F. Rana, J. McCandless, Y.-T. Shao, and N. Tanen of Cornell University and K. Chabak, A. Neal, A. Green, S. Mou, and T. Asel of the Air Force Research Laboratories. Funding: This work was, in part, supported by JSPS Overseas Challenge Program for Young Researchers 1080033 and by the Air-Force/Cornell Center for Epitaxial Solutions (ACCESS) center of excellence monitored by A. Sayir. C.S.C. acknowledges support from the Air Force Office of Scientific Research (FA9550-18-1-0529). Author contributions: R.J., Y.C., and D.J. conceived the study and designed the experimental study. R.J., C.S.C., T.O., S.-T.H., D.R., M.C.C., and K.L. performed the experiments. R.J., C.S.C., T.O., and Y.C. conducted the subsequent data analysis. R.J., C.S.C., and D.J. wrote the manuscript. V.P., D.G.S., D.A.M., and H.G.X. contributed ideas and feedback on the analysis and edited the manuscript. All authors edited the manuscript and approved its final version. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article