Gamma/Beta-Ga2O3 Polymorph Heterostructures

• Gamma/Beta-Ga2O3 polymorph heterostructures are homo-interfaces that join metastable cubic and stable monoclinic Ga2O3 phases without changing stoichiometry.
• Formation methods such as ion irradiation and MOCVD allow precise control over layer thickness, interface abruptness, and defect profiles to tailor functional properties.
• Atomic-scale and thermal studies reveal strain-sensitive band offsets with type-II alignment, offering design principles for high-power electronic and optoelectronic devices.

Searching arXiv for papers on γ/β-Ga2O3 polymorph heterostructures to ground the article in the cited literature. Gamma/Beta-Ga$_2$O$_3$ polymorph heterostructures are homo-interfaces formed between two polymorphs of the same compound, namely metastable cubic defective-spinel $\gamma$-Ga$_2$O$_3$ and stable monoclinic $\beta$-Ga$_2$O$_3$. Unlike conventional semiconductor heterostructures, where interfacial functionality is produced by changing chemical composition, these structures modulate properties across an interface while retaining identical chemistry on both sides. In the recent literature, such stacks are generated either by disorder-induced ordering in $\beta$-Ga$_2$O$_3$0 under ion irradiation or, in a distinct epitaxial context, by exploiting a narrow MOCVD window for $_3$1-Ga$_3$2O$_3$3 on MgAl$_3$4O$_3$5, with $_3$6 inclusions or conversion emerging under thermal treatment (Abdullaev et al., 2024, Azarov et al., 2023, Tang et al., 2023).

1. Definition, structural motif, and distinction from conventional heterostructures

The defining feature of a $_3$7-Ga$_3$8O$_3$9 polymorph heterostructure is that it joins two crystal structures of Ga$\gamma$0O$\gamma$1 across a sharply defined interface without changing stoichiometry. In the ion-irradiation-derived form, a top layer of $\gamma$2-Ga$\gamma$3O$\gamma$4 transforms into $\gamma$5-Ga$\gamma$6O$\gamma$7 once a disorder threshold is reached, leaving a crystalline $\gamma$8 film atop the remaining $\gamma$9 substrate and thereby creating a homo-interface between polymorphs of the same oxide (Abdullaev et al., 2024, Azarov et al., 2023). This configuration differs from conventional heterostructures because the interfacial contrast arises from symmetry, strain, and defect topology rather than from a compositional discontinuity.

The two constituent phases are structurally distinct. $_2$0-Ga$_2$1O$_2$2 is monoclinic and anisotropic, whereas $_2$3-Ga$_2$4O$_2$5 is a cubic defective spinel with a well-ordered oxygen FCC or cubic close-packed sublattice and a partially occupied Ga sublattice (Tang et al., 2023, Abdullaev et al., 9 Sep 2025). In diffraction, the $_2$6 phase is associated with a broad peak near $_2$7 assigned to the $_2$8 reflection of the cubic spinel structure in irradiation-derived stacks, and with an out-of-plane $_2$9 peak at $_3$0 in MOCVD-grown films on MgAl$_3$1O$_3$2(100) (Abdullaev et al., 2024, Tang et al., 2023).

A central point in the literature is interface abruptness. Rutherford backscattering spectroscopy in channeling mode shows a “box-like” signal characteristic of the disorder-induced $_3$3 transition, while prior TEM work and modeling support a sharply defined $_3$4 interface; in thermal-transport modeling, an interface thickness of $_3$5 nm was adopted conservatively (Abdullaev et al., 2024). High-resolution transmission electron microscopy likewise confirms abruptness and crystallographic registry in the band-offset study, where the interface is treated as atomically abrupt and formed with preservation of the oxygen sublattice and Ga migration (Liu et al., 2024).

This structural paradigm has broader implications. Because the interface is chemically homogeneous yet crystallographically discontinuous, it provides a platform for tuning thermal transport, optical signatures, radiation response, and band alignment through orientation, strain state, thickness, and disorder history rather than by alloy design. A plausible implication is that $_3$6-Ga$_3$7O$_3$8 occupies an intermediate conceptual position between classical heteroepitaxy and phase-boundary engineering in functional oxides.

2. Formation routes and crystallographic realization

The most established route is disorder-induced ordering in $_3$9-Ga$\beta$0O$\beta$1. In this approach, ion irradiation at room temperature converts the near-surface region of $\beta$2-Ga$\beta$3O$\beta$4 into $\beta$5-Ga$\beta$6O$\beta$7 once sufficient disorder has accumulated (Azarov et al., 2023). In thermal-transport studies, Ga$\beta$8 irradiation at a fluence of $\beta$9 cm$_2$0 and energies of 0.5, 1.0, and 1.7 MeV yielded $_2$1-layer thicknesses of approximately 350, 650, and 1000 nm, respectively, on $_2$2, $_2$3, and $_2$4 $_2$5-Ga$_2$6O$_2$7 substrates (Abdullaev et al., 2024). In optical studies, $_2$8 layers of approximately $_2$9m were generated from $_3$0-Ga$_3$1O$_3$2 films on sapphire using 1.5 MeV $_3$3Ni$_3$4 at a dose of $_3$5 cm$_3$6, and 300 nm auxiliary $_3$7 layers were formed on $_3$8 and $_3$9 $\beta$0-Ga$\beta$1O$\beta$2 wafers to produce double-polymorph bilayers (Galeckas et al., 2024). A related annealing study used 400 keV Ni$\beta$3 at $\beta$4 cm$\beta$5, with 7° off-axis incidence, to fabricate a $\beta$6 nm $\beta$7 film on a $\beta$8 $\beta$9-Ga$_2$0O$_2$1 substrate (Azarov et al., 2024).

These irradiation-generated structures exhibit orientation relationships and low-mismatch registry. High-resolution ADF-STEM in the radiation-tolerance work showed $_2$2 alignment across a sharp interface (Azarov et al., 2023). In the band-offset study, experimentally observed or modeled interface orientations included $_2$3, $_2$4, and $_2$5, with detailed modeling focused on $_2$6 and $_2$7 (Liu et al., 2024).

A distinct realization is epitaxial $_2$8-Ga$_2$9O$_3$00 growth on MgAl$_3$01O$_3$02(100) by MOCVD. There, nominally phase-pure $_3$03-Ga$_3$04O$_3$05 is obtained only within a narrow temperature window centered near $_3$06C; at $_3$07C mixed $_3$08 occurs, and at $_3$09C nominally phase-pure $_3$10-Ga$_3$11O$_3$12 is obtained (Tang et al., 2023). The epitaxy is cube-on-cube, with $_3$13 and $_3$14 out of plane. In that system, $_3$15-Ga$_3$16O$_3$17 inclusions can form inside the $_3$18 matrix, and annealing at $_3$19C promotes $_3$20 recrystallization while preserving the ordered oxygen sublattice of the $_3$21 matrix (Tang et al., 2023).

The two formation routes are complementary rather than interchangeable. Ion irradiation directly generates $_3$22 homo-interfaces in bulk or wafer substrates, whereas MOCVD on MgAl$_3$23O$_3$24 exposes atomic-scale defect physics, epitaxial relationships, and thermal conversion pathways in an epitaxial $_3$25-dominated environment. This suggests that the field uses “heterostructure” in two closely related senses: a deliberately formed $_3$26-on-$_3$27 stack and a $_3$28 matrix containing oriented $_3$29 domains or inclusions.

3. Atomic-scale structure, defects, and interface topology

At the atomic scale, $_3$30-Ga$_3$31O$_3$32 is treated as a defective spinel with a fully ordered cubic close-packed oxygen sublattice and a partially occupied cation sublattice (Tang et al., 2023). In STEM along $_3$33, image contrast scales with Ga column density, and the ideal $_3$34 motif appears as hexagonal arrays with six tetrahedrally coordinated Ga columns at the edges and five octahedral Ga columns, including a central octahedral column with approximately double the Ga column density and therefore higher HAADF intensity (Tang et al., 2023). This ideal motif is especially visible in APB-free regions near the interface.

A high density of antiphase boundaries is a defining feature of relaxed $_3$35-Ga$_3$36O$_3$37 grown on MgAl$_3$38O$_3$39(100). These APBs arise from lattice shifts in the Ga sublattice while preserving oxygen close packing. A $_3$40 nm fully strained transition layer at the film/substrate interface is APB-free, after which APBs nucleate as the film relaxes (Tang et al., 2023). The planar-defect model extended from $_3$41-Al$_3$42O$_3$43 explains these APBs as glide on oxygen-preserving $_3$44 planes with cation sublattice shifts that reproduce the diagonal high-intensity stripes, tri-column low-intensity patterns, and alternating octahedral/tetrahedral intensities seen in HAADF-STEM (Tang et al., 2023).

Within $_3$45, $_3$46-Ga$_3$47O$_3$48 inclusions exhibit specific epitaxial and rotational-domain relationships. XRD $_3$49-scans after annealing reveal four in-plane rotational domains of $_3$50, with the relationship $_3$51, and 180° rotational $_3$52 domains form via twinning on the $_3$53 plane (Tang et al., 2023). The prevalence of 180° domain pairs is rationalized by the comparatively low twin-boundary formation energy on $_3$54, while 90° $_3$55 rotations are discussed as likely facilitated by the higher symmetry of $_3$56 (Tang et al., 2023).

In the irradiation-derived $_3$57 stacks, the interface is also structurally distinctive. Rutherford backscattering/channeling never reaches the random level, indicating that the transformation does not proceed through full amorphization during $_3$58-layer formation (Abdullaev et al., 2024). In the band-offset work, nanoscale strain maps recorded along $_3$59 and $_3$60 yielded $_3$61 in-plane and $_3$62 out-of-plane, with $_3$63 strained relative to $_3$64 by $_3$65 and $_3$66 at the interface (Liu et al., 2024).

A common misconception is that disorder-induced formation of $_3$67-Ga$_3$68O$_3$69 implies an amorphous or poorly crystallized layer. The combined diffraction, channeling, and microscopy evidence does not support that interpretation. Rather, the $_3$70 layer is crystalline, exhibits a defective-spinel topology, and can support either abrupt homo-interfaces or coherent inclusion relationships, even though its cation sublattice is intrinsically disordered (Abdullaev et al., 2024, Azarov et al., 2023, Tang et al., 2023).

4. Thermal transport and anisotropic heat-flow engineering

Thermal transport is one of the clearest experimentally quantified functionalities of $_3$71-Ga$_3$72O$_3$73 polymorph heterostructures. The cross-plane thermal conductivity was measured by femtosecond laser-based time-domain thermoreflectance with modulation frequencies from 0.73 to 10 MHz, using the thermal penetration depth

$_3$74

which enabled depth profiling across $_3$75 stacks (Abdullaev et al., 2024). The spot size was approximately $_3$76m, larger than the thermal diffusion length, so the measurements were restricted to cross-plane conductivity.

The room-temperature conductivity contrast between the two phases is strong. For $_3$77-Ga$_3$78O$_3$79, cross-plane thermal conductivity was measured as $_3$80–$_3$81 W m$_3$82 K$_3$83, independent of the initial $_3$84-substrate orientation (Abdullaev et al., 2024). For pristine $_3$85-Ga$_3$86O$_3$87, the measured cross-plane thermal conductivity was anisotropic: $_3$88 W m$_3$89 K$_3$90 for $_3$91, $_3$92 W m$_3$93 K$_3$94 for $_3$95, and $_3$96 W m$_3$97 K$_3$98 for $_3$99 (Abdullaev et al., 2024). The $\gamma$00 values exceed the amorphous Ga$\gamma$01O$\gamma$02 minimum, reported as the Cahill limit of approximately $\gamma$03 W m$\gamma$04 K$\gamma$05, which confirms the crystalline character of the $\gamma$06 phase (Abdullaev et al., 2024).

For $\gamma$07 layers formed by 0.5, 1.0, and 1.7 MeV Ga$\gamma$08 irradiation, three-layer fits yielded $\gamma$09, $\gamma$10, and $\gamma$11 W m$\gamma$12 K$\gamma$13, respectively, with corresponding $\gamma$14 values of $\gamma$15, $\gamma$16, and $\gamma$17 MW m$\gamma$18 K$\gamma$19 (Abdullaev et al., 2024). In the model, the $\gamma$20 thermal boundary conductance was fixed at 250 MW m$\gamma$21 K$\gamma$22, consistent with the assumption of an abrupt interface and limited sensitivity to this parameter within the measurement window (Abdullaev et al., 2024).

The most striking thermal result is the order-of-magnitude conductivity step across the interface when $\gamma$23 is formed on $\gamma$24-$\gamma$25. There, $\gamma$26–$\gamma$27 W m$\gamma$28 K$\gamma$29 contrasts with $\gamma$30–20 W m$\gamma$31 K$\gamma$32, producing a step-like change in cross-interface heat flow (Abdullaev et al., 2024). Depth profiling showed that for the $\gamma$33 nm $\gamma$34 layer, the effective conductivity began to increase near $\gamma$35 MHz, where $\gamma$36 nm, matching the $\gamma$37-layer thickness extracted from RBS-C. For thicker $\gamma$38 layers of $\gamma$39 and $\gamma$40 nm, no frequency dependence was observed because the thermal penetration depth at the lowest modulation frequency did not exceed $\gamma$41 nm, so the measurement probed only the $\gamma$42 region (Abdullaev et al., 2024).

Molecular-dynamics simulations provide a partially convergent but not identical picture. For $\gamma$43-Ga$\gamma$44O$\gamma$45, equilibrium MD gave $\gamma$46 W m$\gamma$47 K$\gamma$48 with the Born–Mayer–Huggins potential and $\gamma$49 W m$\gamma$50 K$\gamma$51 with tabGAP, while reverse NEMD extrapolation yielded bulk $\gamma$52 W m$\gamma$53 K$\gamma$54 and a phonon mean free path $\gamma$55 nm (Abdullaev et al., 2024). The approximately twofold discrepancy relative to experiment was attributed to residual biaxial strain after $\gamma$56 transformation and/or ion-induced defects; applying strain $\gamma$57, $\gamma$58 in MD reduced $\gamma$59 to values comparable to experiment (Abdullaev et al., 2024).

These results establish a design principle already emphasized in the literature: select $\gamma$60-$\gamma$61 to maximize the conductivity contrast and tune the $\gamma$62-layer thickness via irradiation energy to control the depth at which heat encounters the high-$\gamma$63 $\gamma$64 substrate (Abdullaev et al., 2024). This suggests an avenue for thermal energy conversion and phononic functionality within a chemically uniform oxide platform.

5. Optical signatures, phase identification, and annealing-induced transformation

Optical characterization has been used both to build a comparative library of Ga$\gamma$65O$\gamma$66 polymorph signatures and to monitor the evolution of $\gamma$67 bilayers. In the optical library study, diffuse-reflectance spectroscopy with Kubelka–Munk analysis, transmittance spectroscopy, low-temperature photoluminescence, and nano-FTIR were applied under uniform analysis conditions across polymorphs and heterostructures (Galeckas et al., 2024). For band-gap extraction, the work used

$\gamma$68

with $\gamma$69 for direct-allowed and $\gamma$70 for indirect-allowed transitions, and in practice followed the conventional direct-gap analysis across polymorphs (Galeckas et al., 2024). The diffuse-reflectance form was

$\gamma$71

while the Urbach tail was analyzed using

$\gamma$72

For standalone $\gamma$73-Ga$\gamma$74O$\gamma$75 thin films, the reported room-temperature values were $\gamma$76 eV and $\gamma$77 meV from DRS, and $\gamma$78 eV and $\gamma$79 meV from transmittance (Galeckas et al., 2024). For $\gamma$80-Ga$\gamma$81O$\gamma$82 thin film $\gamma$83, the values were $\gamma$84 eV and $\gamma$85 meV by DRS, and $\gamma$86 eV and $\gamma$87 meV by transmittance (Galeckas et al., 2024). For the $\gamma$88 double-polymorph bilayer, DRS yielded $\gamma$89 eV and $\gamma$90 meV (Galeckas et al., 2024). The study explicitly noted that the bilayer value is lower than those of standalone $\gamma$91 and $\gamma$92 films, reflecting near-surface probing depth, possible spectral mixing across the interface, and the tendency of DRS to underestimate $\gamma$93 relative to transmittance in thick or stacked structures (Galeckas et al., 2024).

Near-field nano-FTIR provided spatially resolved identification of the heterointerface. A cross-sectional line scan over 1500 nm with 50 nm step size showed distinct near-field amplitude spectra in $\gamma$94 and $\gamma$95 domains, enabling phase identification with $\gamma$96 nm spatial resolution (Galeckas et al., 2024). For $\gamma$97-Ga$\gamma$98O$\gamma$99, near-field modes were observed around 666, 670, 696, and 728 cm$_2$00, and the $_2$01 phase exhibited a distinct near-field spectral envelope sufficient for domain discrimination (Galeckas et al., 2024). This confirms that the $_2$02 interface is optically recognizable on the nanoscale even where detailed atomic-scale interface characterization is not the focus.

Low-temperature photoluminescence is dominated by self-trapped-hole emission. Across crystalline polymorphs, including $_2$03 and $_2$04, intrinsic UV emission in the 2.9–3.3 eV range is attributed to recombination of free electrons with self-trapped holes (Galeckas et al., 2024). Orientation-dependent $_2$05-Ga$_2$06O$_2$07 spectra show a main UV peak at 3.28 eV with FWHM 0.54 eV for $_2$08, and 3.22 eV with FWHM $_2$09–0.64 eV for $_2$10 (Galeckas et al., 2024). Optical anisotropy in $_2$11-Ga$_2$12O$_2$13 is negligible for $_2$14, with $_2$15 eV, but strong for $_2$16, where polarization along $_2$17 produces a $_2$18 eV blue shift of the absorption edge (Galeckas et al., 2024).

Annealing studies connect these optical signatures to structural transformation. In $_2$19 bilayers with $_2$20 nm $_2$21 on $_2$22 $_2$23-Ga$_2$24O$_2$25, isochronal air anneals for 30 min from 300 to $_2$26C revealed a two-stage kinetic behavior (Azarov et al., 2024). Stage I, below $_2$27C, consists of epitaxial $_2$28 regrowth at the buried interface; the $_2$29 layer shrinks, for example from $_2$30 nm to $_2$31 nm after $_2$32C, while its XRD $_2$33 peak shifts closer to the tabulated $_2$34, indicating improved crystallinity (Azarov et al., 2024). Stage II, beginning at $_2$35–$_2$36C, is a non-planar transformation driven by misoriented $_2$37 nucleation near the surface; the $_2$38 peak vanishes by $_2$39C and is replaced by $_2$40 and $_2$41 reflections at approximately $_2$42 and $_2$43, respectively (Azarov et al., 2024).

The optical changes mirror the structural transition. DRS Tauc analysis of the bilayers gave absorption edges around 4.38–4.43 eV at low-to-moderate annealing temperatures, with a threshold temperature range beyond which the bilayer becomes optically single-phase $_2$44-like (Azarov et al., 2024). PL at 10 K showed the self-trapped-hole band near 3.2 eV, implantation-induced green luminescence near 2.5 eV, red luminescence near 2.0 eV, and an extrinsic Cr$_2$45 R-line doublet near 1.78 eV (Azarov et al., 2024). The red band intensified up to $_2$46C and then shut off rapidly at higher temperatures, tracking the disappearance of $_2$47 and indicating that the RL band is associated with defective spinel structure and its defects (Azarov et al., 2024).

These observations make phase identification in $_2$48 heterostructures explicitly multimodal. XRD and RBS-C establish structure and thickness, DRS gives near-surface optical band-edge behavior, PL tracks intrinsic and defect-related recombination, and nano-FTIR resolves polymorph domains at the $_2$49 nm scale (Galeckas et al., 2024, Azarov et al., 2024).

6. Radiation tolerance, disorder recovery, and polymorphic transformation under irradiation

Radiation response is a defining theme of this material system. The irradiation-derived $_2$50 heterostructure was introduced as a crystalline product of disorder accumulation in $_2$51-Ga$_2$52O$_2$53, rather than an endpoint of amorphization (Azarov et al., 2023). In room-temperature implantation experiments, $_2$54 double-polymorph Ga$_2$55O$_2$56 structures tolerated disorder equivalent to hundreds of displacements per atom without severe degradation of crystallinity, with the $_2$57-Ga$_2$58O$_2$59 channeling fingerprint persisting up to 265 dpa for 400 keV $_2$60Ni$_2$61 implantation (Azarov et al., 2023). Classical MD correlated the $_2$62 transformation with approximately 200 Ga Frenkel pairs per simulation cell, corresponding to $_2$63 dpa (Azarov et al., 2023).

The mechanistic explanation is sublattice-specific. In $_2$64-Ga$_2$65O$_2$66, the Ga sublattice is intrinsically defective and therefore relatively insensitive to added point defects, while the oxygen sublattice exhibits a strong tendency to recover FCC stacking after collision cascades (Azarov et al., 2023). Pearson-correlation analysis of partial radial distribution functions and O–O bond-angle distributions showed that the O sublattice retains fcc-characteristic peaks under Ni or Au excess but loses them for Ga excess, which drives local amorphization through charge transfer and disruption of ionic cohesion (Azarov et al., 2023). This establishes an important limitation: the exceptional radiation tolerance is not chemically unconditional.

Under swift heavy ion irradiation, the divergence between $_2$67 and $_2$68 becomes even more striking. In $_2$69 heterostructures with $_2$70 thicknesses of 350, 650, and 1000 nm on $_2$71 $_2$72-Ga$_2$73O$_2$74, irradiation with Kr$_2$75 ions at 50, 100, and 147 MeV and fluences of $_2$76 and $_2$77 cm$_2$78 produced crystalline $_2$79-phase tracks inside $_2$80-Ga$_2$81O$_2$82, but no observable tracks in $_2$83-Ga$_2$84O$_2$85 itself under identical conditions (Abdullaev et al., 9 Sep 2025). For 147 MeV Kr, the electronic stopping powers were essentially the same in the two phases, $_2$86 eV/\AA\ in $_2$87 and 2150 eV/\AA\ in $_2$88, so the difference is not attributable to markedly different deposited electronic energy density (Abdullaev et al., 9 Sep 2025).

In $_2$89-Ga$_2$90O$_2$91, tracks appear as crystalline $_2$92-phase cylinders of approximately 3–4 nm diameter by STEM; MD showed initial amorphous cores of $_2$93 nm at 200 ps that recrystallized into final tracks of $_2$94 nm after long-term evolution (Abdullaev et al., 9 Sep 2025). In $_2$95-Ga$_2$96O$_2$97, MD showed initial amorphous regions of $_2$98 nm at 200 ps, but these were erased upon long-term evolution, leaving only residual point defects below STEM detection; even larger initial amorphous cores of $_2$99 nm recovered (Abdullaev et al., 9 Sep 2025). Arrhenius extrapolation of oxygen-sublattice recovery yielded room-temperature half-lives of $_3$00 s for $_3$01 and $_3$02 s for $_3$03, with similar activation barriers of $_3$04 and $_3$05 eV, respectively (Abdullaev et al., 9 Sep 2025).

The thermal-transport signature is consistent with the microscopy. In irradiated bulk $_3$06-Ga$_3$07O$_3$08, normalized thermal conductivity $_3$09 dropped markedly and by more than half at the highest energy, whereas in $_3$10-Ga$_3$11O$_3$12 $_3$13 remained nearly constant, indicating maintained homogeneity and the absence of extended defects (Abdullaev et al., 9 Sep 2025). This is counter to the usual correlation between low thermal conductivity and track formation, since $_3$14-Ga$_3$15O$_3$16 has the lower thermal conductivity yet shows no observable tracks (Abdullaev et al., 9 Sep 2025).

A frequent oversimplification is to attribute radiation hardness here purely to reduced heat transport or to a generic “disordered structure.” The published interpretation is more specific: $_3$17-Ga$_3$18O$_3$19 rapidly recovers disorder because its defective spinel cation sublattice offers multiple energetically accessible configurations, while the oxygen FCC backbone supplies a robust template for reordering (Abdullaev et al., 9 Sep 2025, Azarov et al., 2023). This suggests that radiation tolerance in these heterostructures is inseparable from polymorphic topology and sublattice dynamics.

7. Band alignment, strain sensitivity, and device-relevant considerations

Electronic band alignment at the $_3$20-Ga$_3$21O$_3$22 interface has been addressed by density functional theory informed by experimental atomistic interface analysis (Liu et al., 2024). Because both sides consist of the same compound, core-level XPS alignment is not applicable, so the study used potential-lineup and vacuum-alignment methods. The valence-band offset in the lineup method is

$_3$23

and the conduction-band offset is obtained from the band gaps: $_3$24 In vacuum alignment, absolute band edges are referenced to vacuum through the slab average potential (Liu et al., 2024).

The principal conclusion is that the offsets are small, type-II, and highly sensitive to strain and orientation. For Interface I, $_3$25, the lineup was $_3$26 eV; for Interface II, $_3$27, $_3$28 eV (Liu et al., 2024). In terms of conduction-band offsets $_3$29, Interface I gave approximately $_3$30 meV for strained $_3$31 and $_3$32 meV for unstrained $_3$33, while Interface II gave approximately $_3$34 meV for strained $_3$35 and $_3$36 meV for unstrained $_3$37 (Liu et al., 2024). The corresponding valence-band offsets were also orientation- and strain-dependent, but in all computed cases electrons and holes preferred opposite sides of the interface, yielding type-II alignment (Liu et al., 2024).

Strain is decisive because the $_3$38-phase conduction-band minimum has a strong deformation potential. The study reported $_3$39 eV for the $_3$40-oriented slab and cited $_3$41 eV for $_3$42 (Liu et al., 2024). When strain is applied to $_3$43 only,

$_3$44

so compressive volumetric strain in $_3$45 can move the CBO by several tenths of an eV (Liu et al., 2024). For the experimentally GPA-informed interfacial state, $_3$46, and the extrapolated CBO shift was approximately $_3$47 eV relative to the unstrained case, implying a net $_3$48 eV for Interface I under that strain state (Liu et al., 2024). The authors cautioned that this value is indicative because strain relaxation and defects may modify both the average electrostatic potential and the band edges.

These results motivate discussion of a possible two-dimensional electron gas at the interface. The work states that 2DEG formation is theoretically possible, especially when $_3$49 is sufficiently positive so that electrons localize on the $_3$50 side, but that gradual strain relaxation with distance from the interface may shift the location of any electron gas or reduce its overall probability of formation (Liu et al., 2024). The study does not solve a Poisson–Schrödinger problem, but provides the standard triangular-well expression for subband energies as a context for confinement (Liu et al., 2024).

From a device standpoint, several implications recur across the literature. Type-II alignment and small offsets in the few-hundred-meV range suggest potential use in HEMTs, modulation-doped structures, tunnel barriers, or resonant transport elements within a single-chemistry platform, provided orientation and strain can be controlled (Liu et al., 2024). At the same time, the thermal and structural studies impose constraints. Annealing above $_3$51–$_3$52C drives non-planar $_3$53 conversion and texture evolution in bilayers (Azarov et al., 2024), while the radiation studies indicate that $_3$54 is best placed on the exposed surface if track-free behavior under electronic-energy-dominant irradiation is desired (Abdullaev et al., 9 Sep 2025). The optical study further indicates that DRS with focal-plane control and nano-FTIR provide practical metrology for phase identification in such stacks (Galeckas et al., 2024).

A plausible synthesis of these findings is that $_3$55-Ga$_3$56O$_3$57 heterostructures are most technically distinctive when their polymorphic contrast is used deliberately rather than incidentally: $_3$58 can provide low thermal conductivity, nanoscale optical identifiability, and exceptional radiation tolerance, whereas $_3$59 supplies anisotropic heat transport, structural stability, and the conventional electronic backbone. The outstanding challenge, as the band-offset study makes explicit, is that interfacial electronic functionality is not fixed by chemistry and therefore depends critically on local strain, orientation, and defect topology (Liu et al., 2024).