Tin-assisted heteroepitaxial PLD-growth of $κ$-Ga₂O₃ thin films with high crystalline quality

The wide bandgap semiconductor Ga₂O₃ has interesting material properties for potential applications such as high power transistors and rectifiers, deep-UV and quantum infrared photo detectors, and sensors. It was stabilized in at least five polymorphs; the monoclinic beta gallia structure is the thermodynamically stable structure under standard conditions. β-Ga₂O₃ bulk single crystals can be grown by, e.g., the floating zone method, the Czochralski method, and edge-defined film fed growth, and substrate materials for homoepitaxial growth are commercially available. This triggered enormous research efforts in the last decade; promising results on monoclinic gallium oxide can be reviewed in Refs. 1–4. Rhombohedral α-Ga₂O₃ is isostructural to rhombohedral In₂O₃ and Al₂O₃ being beneficial for bandgap engineering. It was successfully fabricated by, e.g., mist chemical vapor deposition (CVD),⁵ and high power devices using binary material have been demonstrated.⁶ The orthorhombic polymorph of Ga₂O₃ has recently attracted research interests due to the predicted high value of its spontaneous polarization of 0.23 C/m², being a factor of three higher than that of AlN.^7–9 Heteroepitaxy of orthorhombic Ga₂O₃, mostly referred to as ε-Ga₂O₃ in the literature, was reported for pulsed-laser deposition^10–13 (PLD), halide vapor phase epitaxy^14,15 (HVPE), metal-organic chemical vapor deposition^16–20 (MOCVD), metal-organic vapor phase epitaxy²¹ (MOVPE), atomic layer deposition,¹⁷ molecular beam epitaxy,²² plasma-assisted molecular beam epitaxy²³ (PMBE), and mist CVD.^24–28 Typically, α-Al₂O₃(00.1) is used as a substrate,^{10–12,17,18,20,22,29,30} but growth on 3C-SiC(001) and (111),¹⁷ 6H-SiC(001),¹⁶ GGG(111),³¹ MgO(111) and YSZ(111),²⁶ or GaN(0001), AlN(0001), and β-Ga₂O₃(⁠$2¯01$⁠)¹⁴ was reported as well. For PLD and MBE, a tin-assisted growth facilitates the stabilization of the orthorhombic polymorph;^10,11,22 however, room temperature electrical conduction was not affected by the potential tin donors. Furthermore, similarity of the structure of heteroepitaxial, orthorhombic Ga₂O₃ thin films to κ-Al₂O₃ was suggested by Matsuzaki et al.¹² for PLD layers and proven for MOCVD layers by Cora et al.²¹ First reports on bandgap engineering have been published,^25,27 and a first device in the form of a photoconductive detector was demonstrated.³⁰ 

In this report, we present a detailed analysis of the structural and morphological properties of orthorhombic Ga₂O₃ thin films grown heteroepitaxially by pulsed-laser deposition on Al₂O₃(00.1), yttria-stabilized ZrO₂(111) (YSZ(111)), SrTiO₃(111) (STO(111)), and MgO(111).

The Ga₂O₃ thin film samples were grown via pulsed-laser deposition (PLD) either simultaneously on 5 × 5 mm² c-sapphire, MgO(111), YSZ(111), and STO(111) substrates (Crystec) or on single 10 × 10 mm² c-sapphire substrates. Simultaneous growth on four 5 × 5 mm² substrates was on the same position in the substrate holder as for single 10 × 10 mm² substrates. A 248 nm KrF excimer laser (Coherent LPX Pro 305 F) was employed to ablate ceramic target pellets sintered from ball-milled and homogenized powders of (i) pure β-Ga₂O₃ (99.999%, Alfa Aesar), labeled target A from now on, or (ii) β-Ga₂O₃ with additional 1 wt. % SnO₂ (99.999%, Alfa Aesar), labeled target B from now on, in air at 1350 °C for 72 h. The first 300 laser pulses were applied at a frequency of 1 Hz and provided a nucleation layer for the subsequent thin film growth, while for the remaining pulses (15 000 or 50 000), a repetition rate of 10 Hz was chosen. The laser fluence was adjusted to ≈2 J/cm². Both the oxygen partial pressure p(O₂) and the growth temperature T_g were varied for the different samples investigated. The substrate holder was further rotated during deposition to ensure homogeneous thin film growth. For more details on the used PLD setup, see, for example, Refs. 32 and 33. The STO(111) substrates were etched and annealed in oxygen atmosphere prior to deposition. MgO(111) substrates were annealed in oxygen atmosphere to reduce surface degradation due to the formation of hydroxide during storage. The crystalline structure of the samples was investigated by means of 2θ-ω scans, rocking curve measurements, ϕ scans, and reciprocal space maps (RSMs) employing a PANalytical X’pert PRO MRD diffractometer with Cu K_α radiation using a parabolic mirror and a PIXcel^3D detector. The surface morphology was evaluated using atomic force microscopy (AFM), where a Park Systems XE-150 or NX10 was employed in non-contact or tapping mode. The thickness of the samples (15 000 pulses) was in the range of ≈200–600 nm and was determined by spectroscopic ellipsometry (RC2 dual rotating compensator ellipsometer by J.A. Woollam). The thicknesses and growth rates of the thin films on c-sapphire as a function of oxygen pressure can be found in the supplementary material. The optical properties of the samples were investigated by transmission spectroscopy with a PerkinElmer Lambda 19 spectrometer. The tin content of the thin films was determined by energy dispersive X-ray spectroscopy (EDX) employing a Nova NanoLab 200 by FEI company. Depth-resolved XPS measurements were performed to determine the Sn, Ga, and O content of the samples as a function of distance from the sample surface. A JEOL JPS-9030 system with a Mg K_α X-ray source was used for excitation. Argon with a pressure of 3 × 10⁻⁴ mbar was used as a sputter gas for depth profiling, with a low acceleration voltage and ion current (300 V and 3.5 mA, respectively) to avoid sputter damage. The resulting sputter rate was ≈0.2 nm/s. For determination of atomic concentrations, sensitivity factors listed in the JEOL system were used.

Figure 1(a) shows wide-angle X-ray diffraction (XRD) 2θ-ω scans of a sample series of Ga₂O₃ thin films deposited on α-Al₂O₃(00.1) substrates for various values of the oxygen partial pressure using target B and T_g = 670 °C. A drastic difference in the XRD patterns for p(O₂) above and below 0.03 mbar can be observed. Above this pressure, the thin films exhibit a typical diffraction pattern of monoclinic β-Ga₂O₃ thin films grown by PLD on c-sapphire substrates; see, for example, Refs. 10–12 and 34–36. Apart from the α-Al₂O₃(00.6) and (00.12) substrate reflections, rather broad peaks at approximately 18.8°, 30.2°, 38.2°, 58.9°, 59.9°, and 110.13° can be indexed as reflections corresponding to the (⁠$2¯$01), (110), (⁠$4¯$02), (⁠$6¯$03), (113), and (⁠$10¯$05) lattice planes of the monoclinic β-phase of Ga₂O₃ (JCPDS card no. 76-0573), respectively. The layers are mainly (⁠$2¯$01)-oriented which is typical for PLD grown thin films on c-sapphire substrates.^{10–12,34–36} Below 0.03 mbar, only 5 reflections with strongly increased intensity and slightly higher angular positions of 19.1°, 38.8°, 59.8°, 83.3°, and 112.4° are observed which are marked with dashed gray vertical lines in Figs. 1(a)–1(c). These reflections belong to the (002), (004), (006), (008), and (0010) lattice planes of Ga₂O₃ in a different crystal modification, the hexagonal ε-Ga₂O₃ (symmetry group P6₃mc³⁷), or the closely related orthorhombic κ-Ga₂O₃ (symmetry group Pna2₁^21,37,38), respectively. This phase is metastable and will convert to the thermodynamically stable β-phase for temperatures of $≥$870 °C.^19,39 As mentioned before, Cora et al. showed by the evaluation of TEM measurements²¹ that this crystal modification indeed has orthorhombic symmetry and also other studies proved by X-ray diffraction measurements that reflections occurring for the orthorhombic but not the hexagonal polymorph can be measured and identified.^22,25 We therefore refer to this phase as κ-Ga₂O₃ and will also provide unambiguous evidence for the growth in the orthorhombic structure below.

These orthorhombic thin films are grown with an (001)-out-of-plane orientation as previously reported for κ-Ga₂O₃ thin films on c-sapphire in the literature.^{10–12,17,18,20,22,29} Figure 1(b) shows the (002), (004), and (008) reflections of the thin films in a magnified view, which makes the differences between the κ- and β-modification apparent. The reflections of the thin films in the κ-phase show much higher intensities, higher angular positions, and are much narrower in comparison to those of the β-phase. This is most apparent for the (004) peak whose intensity even exceeds that of the α-Al₂O₃(00.6) substrate peak. Moreover, the (004) peak has a similar broadening as the substrate reflection showing the high crystalline quality of our layers. The (008) peak is strongly visible in contrast to the (⁠$8¯$04) reflection of β-Ga₂O₃, which is missing for the films in the β-modification due to very low intensities. This is in agreement with intensity ratio calculations using the VESTA program,⁴⁰ where the unit cell by Cora et al.²¹ and by Åhman et al.⁴¹ was used for κ-Ga₂O₃ and for β-Ga₂O₃, respectively.

The tin in the target is expected to facilitate the growth in the κ-phase as was already reported for PLD-grown layers^10–12 as well as MBE-grown thin films²² in the literature. Consequently, Kracht et al.²² proposed a model for tin acting as a catalyst for the growth of the κ-phase in MBE. Indeed, nominally undoped reference samples grown in the present study with nominally identical process parameters as for the thin films in the κ-phase crystallized solely in the monoclinic β-phase. The XRD 2θ-ω scans of two representative samples for p(O₂) = 0.016 mbar and 0.1 mbar are shown in Fig. 1(c) in comparison to the thin films grown from target B with identical process parameters. The peaks for these samples are at the expected positions for a mainly (⁠$2¯$01)-oriented growth in the β-phase. Furthermore, they are much less intense and much broader than those of the film in the κ-phase. XRD patterns similar to the samples from target A are obtained for the thin films from target B when the critical pressure for κ-phase growth is exceeded as evident for layers deposited at p(O₂) = 0.1 mbar. κ-Ga₂O₃ thin films grown by PLD therefore have superior crystalline quality compared to β-Ga₂O₃ thin films deposited at similar process parameters. As a remark, it should be mentioned that the thicknesses of the thin films are similar for nominally identically process parameters and hence not a cause of different peak intensities. Figure 1(d) shows the c-lattice constant of the κ-Ga₂O₃ thin films in dependence on p(O₂) and for the different substrates used. The c-lattice constants were determined by a fitting of the film peaks in the XRD 2θ-ω scans using pseudo Voigt functions. The resulting c-lattice constants calculated from the position of the (004), (006), (008), and (0010) reflections were then extrapolated to θ = 90° using the function $c=f0.5tan(θ)−1+cos(θ)tan(θ)−1$⁴² to minimize the goniometer error (see the supplementary material for the demonstration of a fit). The resulting c-lattice constants shown in Fig. 1(d) correspond well to those reported in the literature ranging from 9.24 Å to 9.41 Å.^{9,11,12,18,21,22,25,27,29,37,38} The lattice constants for the thin films decrease slightly with increasing p(O₂). For samples grown on c-sapphire, c lies closest to the value of 9.2554 Å given by Playford et al.³⁷ for the ε-phase, while for the films on YSZ(111) and STO(111), c is rather close to the value of 9.2833 Å reported by Cora et al.²¹ or 9.2759 Å reported by Playford et al.³⁷ for the κ-phase. The lattice constants for the films on MgO(111) show even higher values. However, these films also exhibit an inferior crystalline quality compared to those on the other substrates [see also Figs. 3(a) and 6(c)].

The surface morphology of the thin films was studied using atomic force microscopy (AFM). Figures 2(a)–2(d) depict the surface morphology of orthorhombic (green κ) and monoclinic (blue β) thin films deposited with identical growth parameters (except for the tin content in the target). These samples were deposited on c-sapphire substrates, and thin films on the other substrates showed similar surface properties. Obviously, the growth in the κ-phase improved the surface morphology considerably. While the thin films in the β-phase typically exhibit rough surfaces with large and irregularly shaped grains, films in the κ-phase show much smoother surfaces consisting of small crystallites. Figure 2(e) shows the root-mean-squared roughness R_q in dependence on p(O₂) for the thin films on the different substrates. The thin films in the κ-phase [green ellipse in Fig. 2(e)] reveal low roughness values between 3 nm and even below 1 nm, while being in the thickness range of 250-600 nm. The roughness further decreases with increasing p(O₂). At the same time, the thickness of the thin films increases with p(O₂), which might also be a reason for the decreased surface roughness. For p(O₂) ≥ 0.03 mbar, the thin films begin to crystallize in the β-phase for targets A and B and the roughness increases again [small blue ellipse in Fig. 2(e)]. The largest roughness of up to 10 nm is found for the monoclinic reference samples grown with target A shown in gray in Fig. 2(e). These β-phase thin films exhibit similar thicknesses and growth rates as the κ-phase thin films for nominally identical process parameters; see also Fig. S1 in the supplementary material. However, in contrast to the thin films in the κ-phase, the roughness increases with increasing p(O₂) for the same pressure window. The κ-Ga₂O₃ thin films also show a superior surface morphology compared to the films in the β-phase. Furthermore, the surfaces of our thin films are qualitatively sufficient for the growth of heterostructures, which can utilize electrical polarization differences at the interface for polarization doping. Moreover, the lowest roughness of 0.5 nm is in the same range as that of homoepitaxial β-Ga₂O₃ thin films already used for heterostructures or device applications.^1,43–45 Up until today, it is also the lowest reported roughness value for κ-Ga₂O₃ in general (for comparison, see Refs. 16, 17, 22, and 24).

To further quantify the crystalline quality of our samples, we determined the broadening of the κ-Ga₂O₃(004) reflection in 2θ by using the full width at half maximum (FWHM) from the fitting of the peaks of the 2θ-ω scans for the estimation of the c-lattice constants. Figure 3(a) shows the determined FWHM values for the κ-Ga₂O₃(004) peaks for different substrate materials in dependence on p(O₂) (green background). For comparison, the FWHM of the β-Ga₂O₃(⁠$4¯$02) peak for thin films grown in the β-phase are shown (blue background) as well. The films on c-sapphire, YSZ(111), and STO(111) substrates exhibit the lowest FWHM values of the κ-Ga₂O₃ thin films. These are slightly decreasing with increasing pressure and are almost as small as the FWHM of the Al₂O₃(00.6) substrate reflection shown as the red dashed line (0.06° at 0.016 mbar vs. 0.05° for the substrate reflection), which marks the resolution of the experimental setup. The thin films grown on MgO(111) show higher FWHM, in general, which could be due to an inferior substrate film interface. The surface quality of MgO substrates is known to degrade in ambient atmosphere due to the formation of hydroxides.⁴⁶ The initial step terrace structure can be restored by annealing the substrates in oxygen. Although the substrates were annealed in 800 mbar oxygen atmosphere for 2 h at 800 °C, this may not have been enough in temperature or annealing time to completely restore the surface properties, since typically 900–1000 °C are used for this annealing procedure.⁴⁷ This, in turn, could disturb the crystal growth and can cause a lower crystallinity of the thin film layer. Nevertheless, even the samples with the highest FWHM still show lower values than all films grown in the β-phase. These are generally higher than 0.225°, independent of process parameters. Also apparent is a sudden increase in FWHM by an order of magnitude for the samples grown with target B when the critical pressure for κ-phase growth is exceeded. This further confirms the superior crystalline quality of the films in the κ-phase. Moreover, the FWHM values reported in the present study are lower than for any other report on high-quality κ-Ga₂O₃ thin films grown by methods such as MOCVD¹⁶ or PLD.¹³ κ-Ga₂O₃ thin films are therefore highly promising for demanding device applications such as field-effect transistors, power devices, quantum well photodetectors, or laser LEDs.

For the quantification of the out-of-plane mosaicity of the thin films, XRD rocking curve measurements were performed around the κ-Ga₂O₃(004) reflection on the thin film samples grown on c-sapphire. A typical rocking curve is shown in Fig. 3(b) for a thin film grown at p(O₂) = 0.016 mbar and T_g = 670 °C. The determined FWHM value of the rocking curve is approximately 0.59° and is similar to other reports on κ-Ga₂O₃ thin films grown by PLD,^11,12 MOCVD,¹⁶ mist CVD,^24,25 or HVPE.¹⁴ Figure 3(c) shows the FWHM of the κ-Ga₂O₃(004) reflection of all measured rocking curves in dependence on p(O₂) during film growth. For all pressures, the FWHM is in the range of 0.55°-0.61° indicating low out-of-plane mosaicity and high crystallinity for all κ-Ga₂O₃ thin films. To the best of our knowledge, there are only two reports for thin films on c-sapphire, where a lower FWHM of 0.29° or 0.42° was measured for MBE²² or MOCVD,⁴⁸ respectively, and two reports with a significantly lower FWHM in the (004) rocking curve of 0.21° for a film grown by HVPE on β-Ga₂O₃(⁠$2¯$01) single crystals¹⁴ and of 0.24° for a PMBE grown thin film on a β-Ga₂O₃(⁠$2¯$01) buffer layer,²³ for which the lattice mismatch to κ-Ga₂O₃ is lower than that for c-sapphire.

The influence of the growth temperature T_g was studied for c-sapphire substrates at p(O₂) = 0.006 mbar but varying substrate temperatures between 670 °C and 410 °C. Figure 4(a) shows the wide-angle XRD 2θ-ω scans for these thin films. The films are phase pure and (001)-oriented in the κ-phase for T_g ≥ 550 °C. At 500 °C, additionally a (⁠$2¯$01)-oriented β-phase is observable and the thin film grown at 470 °C shows crystallization mainly in (⁠$2¯$01)-oriented β-phase. For lower T_g, the thin films are X-ray amorphous. Orita et al.¹⁰ observed the same phase transition in their PLD setup at a temperature of ≈435 °C. This behavior could be due to the fact that the suboxide desorption is temperature-dependent and decreases with decreasing T_g as shown for MBE-growth of Ga₂O₃.⁴⁹ This corresponds to growth at higher oxygen pressure. At lower temperature, the suboxide formation therefore may not be sufficient for the tin to facilitate the κ-phase growth. Figure 4(b) shows the calculated c-lattice constants for the thin films in dependence on T_g. It increases only slightly with decreasing temperature for the phase pure thin films and is in agreement with the ones of the pressure series shown in Fig. 1(d). The thin film exhibiting the mixed phase shows a higher lattice constant, which could be due to an influence of the additional β-phase on the crystallinity. A corresponding behavior is also observed for the T_g-dependent FWHM of the κ-Ga₂O₃(004) reflection shown in Fig. 4(c). The FWHM only slightly increases until the mixed phase appears at 500 °C for which it doubles in value. The FWHM of the β-Ga₂O₃(⁠$4¯$02) reflection of the phase pure β-Ga₂O₃ thin film is again approximately 5 times higher than the lowest FWHM of the (004) reflection of the κ-phase. Therefore, higher growth temperatures seem to be beneficial for the crystalline quality of the κ-Ga₂O₃ layers.

For further evaluation of the crystalline structure and lattice constants, we have grown a sample with a larger thickness of 2 μm for further XRD RSMs, as well as rocking curves and 2θ-ω scans of asymmetrical reflections. The sample was deposited with optimized growth parameters at p(O₂) = 0.016 mbar and T_g = 670 °C. A wide-angle 2θ-ω scan confirmed (001)-oriented growth in the κ-phase with similar c-lattice constant as for the thinner films. Figure 5 shows the RSMs around (a) the symmetrical α-Al₂O₃(00.6) substrate and (004) film reflection, (b) the symmetrical α-Al₂O₃(00.12) substrate and (008) film reflection, and (c) the asymmetrical α-Al₂O₃(11.12) substrate and (138) as well as (139) film reflection. The α-Al₂O₃(11.9) and the (206), (347) and (015) film reflections were measured as well (not shown). The symmetric reflections are perfectly aligned with the corresponding substrate peaks in the q_∥-direction indicating a negligible tilt of the c-axis of the thin film with respect to the substrate c-axis. The $Kα1$/$Kα2$ splitting observed for the film reflections and the low broadening in the q_⊥-direction indicate a high film quality. The higher broadening in the q_∥-direction reflects the slightly increased mosaicity observed in the rocking curves. Similarly, also the (138) and (139) reflections show a low broadening with $Kα1$/$Kα2$ splitting in the q_⊥-direction corroborating the high crystalline quality. However, here a slightly increased broadening in the q_∥-direction can also be observed which indicates lower elongation of regions with coherent scattering in the lateral direction, e.g., lower lateral grain sizes. Nevertheless, the presence of the (138) and (139) reflections at the correct positions with respect to the substrate reflection and the similar q_∥ values confirm the epitaxial growth of the thin film. Furthermore, the q_∥ and q_⊥ values for these reflections are not matching with the corresponding substrate peak position. Therefore, pseudomorphic growth can be excluded and the thin film can be considered completely relaxed. This is also expected for a thin film with a thickness of 2 μm. Rocking curve measurements of the (004) and (131) reflections revealed FWHM values of 0.57° and 1.19°, respectively. The FWHM of the symmetric reflection is similar to the one of the thinner films, while the FWHM of the skew symmetric reflection indicates an increased twist mosaicity. Nevertheless, it is similar to other thin films grown by PLD¹² or HVPE on GaN and AlN growth templates.¹⁴ As for the symmetric rocking curves, significantly lower values for asymmetric reflections were only reported for HVPE on β-Ga₂O₃(⁠$2¯$01) single crystals¹⁴ (0.84°) and MBE growth on β-Ga₂O₃(⁠$2¯$01) buffer layers²³ (0.96°). In-plane lattice constants were evaluated by the peak positions of additional 2θ-ω scans of the (327) and (347) reflections as well as with the RSMs of the (347), (015) and (139) reflection. The results are summarized in Table I in comparison to selected literature values. Within the experimental error, both methods yield the same lattice constant. Our values for the c-lattice constant are slightly lower than those reported by Cora et al., while the ones for the a-lattice constant are slightly higher. All values are however lower than the ones calculated by Yoshioka et al.³⁸ 

To confirm that the κ-Ga₂O₃ thin films grew epitaxially on the used substrates [c-sapphire, YSZ(111), STO(111), and MgO(111)] and to determine the in-plane epitaxial relationships, XRD ϕ scans were performed on samples grown at optimized process parameters [p(O₂) = 0.016 mbar and T_g = 670 °C]. The samples show all (001)-oriented growth in the κ-phase as confirmed by XRD 2θ-ω scans around the (004) reflections shown in Fig. 6(c).

The XRD ϕ scans were measured for several asymmetric κ-Ga₂O₃ reflections to further distinguish the different possible structures (hexagonal or orthorhombic). Figures 6(a) and 6(b) show these ϕ scans for the (131), (122), (206), and (212) reflections of κ-Ga₂O₃ for thin films on (a) c-sapphire and (b) YSZ (111), which are representative for all cubic substrates. Skew symmetric substrate reflections measured are the (10.2) reflection for c-sapphire and the (200) reflection for YSZ(111). The positions of these reflections in 2θ and χ are summarized in Table II.

The (131) and (206) reflections in the XRD ϕ scans shown in Figs. 6(a) and 6(b) reveal for all samples a six-fold symmetry with a separation of 60°. If the κ-phase has an orthorhombic structure, this implies a growth in 3 rotational domains rotated by an angle of 120° with respect to each other. However, these reflections also occur in the hexagonal structure, where they are differently indexed, but are expected to be at the sameangular positions. Several reports^{11,12,15,18,21,22,27,48} already reported heteroepitaxial growth of κ-Ga₂O₃ on c-sapphire substrates, both with an interpretation in the hexagonal as well as the orthorhombic structure. In most reports on the epitaxial growth of hexagonal or orthorhombic Ga₂O₃, these six-fold lattice planes are measured in ϕ scans and pole figures.^{11,14,16,24,27} To confirm the orthorhombic structure, additional measurements of reflections, which are only expected in the orthorhombic system, are required. We, therefore, performed measurements of the (122) and (212) reflections, which occur 12-fold corresponding to two mirror planes in each rotational domain which are not overlapping in direction with the ones of other domains. In a similar way, the orthorhombic structure was also confirmed by Kracht et al.²² and Tahara et al.²⁵ However, in the work of Kracht et al., the (131) reflection was incorrectly indexed as (211) reflection. We indeed obtain 12 peaks for the (122) and (212) lattice planes in Figs. 6(a) and 6(b), confirming the orthorhombic structure of the crystal lattice. From the position of the substrate reflections with respect to those of the thin films, we could determine the in-plane and out-of-plane epitaxial relationships for c-sapphire as α-Al₂O₃ $⟨1¯0.0⟩∥⟨010⟩$ κ-Ga₂O₃, $α−Al2O3⟨1¯2.0⟩∥⟨100⟩$ κ-Ga₂O₃, and α-Al₂O₃ [00.1]∥[001] κ-Ga₂O₃. These relations are the same as those reported for κ-Ga₂O₃ thin films grown by PLD,¹² mist CVD,²⁷ or MBE.²² In the same way, the in-plane and out-of-plane relationships for all investigated cubic substrates can be determined as cubic $⟨2¯11⟩∥⟨100⟩$ κ-Ga₂O₃, cubic $⟨01¯1⟩∥⟨010⟩$ κ-Ga₂O₃, and cubic [111]∥[001] κ-Ga₂O₃. These relationships and the possible orientation of the three rotational domains of the orthorhombic unit cell on (111)-oriented cubic substrates are depicted in Fig. 6(d). The given relations hold for all investigated cubic substrates (YSZ(111), STO(111), MgO(111)) since the positions of the film reflections in the XRD ϕ scans with respect to the (200) substrate reflections were the same for all these materials. XRD ϕ scans for the other substrates not shown in Fig. 6 can be found in the supplementary material. The a-lattice constant for the orthorhombic Ga₂O₃ system matches reasonably well with the lattice constant of the cubic substrates in $⟨2¯11⟩$-direction. Correspondingly, the b-lattice constant matches quite well with 3/2 times the lattice constant of the cubic substrate in $⟨01¯1⟩$-direction. With this configuration, the lattice mismatches in the a- and b-direction are minimized. The corresponding mismatches are given in Table III for the substrates used in the present study. The lowest mismatch is given for the MgO(111) substrates, which, in contrast to this, showed the lowest crystalline quality of the corresponding thin films. However, the formation of hydroxides as already discussed is still an issue here. The epitaxial growth on YSZ(111) and MgO(111) was already reported for κ-Ga₂O₃ thin films grown by mist CVD by Nishinaka et al.,²⁶ although the epitaxial relations were given there for the hexagonal system.

The optical properties of κ-Ga₂O₃ thin films were investigated by means of RT transmission measurements depicted in Fig. 7(a) for samples deposited at 670 °C as a function of p(O₂). Independent of the growth pressure, sub-bandgap absorption was not observed. The data were evaluated (assuming that κ-Ga₂O₃ is a direct bandgap semiconductor) by plotting the square of the absorption coefficient α versus the energy of the incident photons shown in the inset of Fig. 7(a). The steepest absorption onset was observed for the layer grown at 0.016 mbar that also has superior structural properties. The optical bandgap can be estimated to about 4.91 eV for this layer. Films grown at lower pressure exhibit somewhat higher absorption edges of about 4.95 eV and 5.04 eV for growth at 2 × 10⁻³ and 3 × 10⁻⁴ mbar, respectively. These values correspond well to literature data.^{11,14,20,25,26}

The orthorhombic thin films discussed above are electrically insulating independent of the choice of substrate and growth conditions. Orita et al. described already that tin does not generate free electrons in orthorhombic Ga₂O₃:Sn PLD thin films and concluded that either the transition level of tin donors is too deep to release electrons at room temperature or Sn⁴⁺ donors are compensated by Sn²⁺ acceptor states.¹⁰ Nevertheless, they concluded that the presence of tin is indispensable for obtaining orthorhombic Ga₂O₃ layers by PLD. Concerning MBE, Kracht et al. confirmed that the presence of tin induces the formation of κ-Ga₂O₃.²² They demonstrated further that a thin monoclinic interlayer with reduced tin content is formed on the substrate and that the thickness of this layer depends on the Ga/Sn flux ratio. Within the orthorhombic phase, the amount of incorporated tin is higher than in the interlayer and it scales with the amount of Sn offered. For our PLD thin films, we observe that the tin content in the samples with monoclinic structure corresponds to the tin content in the target; however, an increase from about 0.6 at. % to about 0.8 at. % is observed as p(O₂) is increased from 0.03 to 0.1 mbar as depicted in Fig. 8(a). For orthorhombic layers, the tin content is in the order of the detection limit of the device and with that considerably lower than the tin content in the PLD target. This is indicated by error bars for κ-Ga₂O₃ being much larger than for β-Ga₂O₃. Furthermore, charging effects occur since tin is not electrically active in κ-Ga₂O₃.¹⁰ We analyzed the tin concentration near the surface by ex situ depth-resolved XPS measurements shown in Fig. 8(b). Detailed Sn 3d, O 1s, and Ga 3d core level spectra for different sputtering times can be found in the supplementary material. The data reveal a decrease in the Ga content towards the surface on the one hand and an increase in oxygen and tin content on the other hand. The tin content in the bulk part of the layer is similar to the values of Fig. 8(a). The considerably higher tin content at the surface is evidence of the tin-assisted growth mechanism and indicates that tin acts as a surfactant during deposition of κ-Ga₂O₃. This is supported by (i) the considerably lower surface roughness and (ii) higher crystallinity and with that improved layer quality of our κ-Ga₂O₃ compared to our β-Ga₂O₃ films. Surfactant-mediated epitaxy (SME) with tin was observed for Si on Si(111).⁵⁰ Concerning gallium oxide, SME was reported for MOCVD growth of β-(In,Ga)₂O₃ on Ga₂O₃(100) where indium acts as a surfactant.⁵¹ For MBE heteroepitaxy of (In,Ga)₂O₃ on Al₂O₃(00.1), a metal-exchange catalysis was reported to extend the growth window to regimes not accessible without offering In²³ and allowed to stabilize ε-Ga₂O₃ on a β-Ga₂O₃ buffer layer on Al₂O₃(00.1). This metal-exchange catalysis involves the reduction of In₂O₃ according to $2Ga(ads)+In2O3(sol)→Ga2O3(sol)+2In(ads)$⁠, where sol (ads) means solid phase (adsorbate). Metal-exchange by reduction of SnO₂ according to 2Ga(ads) + 3SnO₂(sol) → 2Ga₂O₃(sol) + 3Sn(ads) seems unlikely due to the higher dissociation energy of Sn–O compared to Ga–O bonds.⁵² Concerning PLD growth of κ-Ga₂O₃, additional experiments and theoretical studies are required to clearly identify the influence and kinetics of tin during heteroepitaxy.

In summary, we have grown high-quality Ga₂O₃ thin films in the orthorhombic κ-phase by means of pulsed-laser deposition. A detailed analysis of the structural and morphological properties in dependence of the growth conditions was performed. The thin films exhibit superior crystalline quality in comparison to β-Ga₂O₃ thin films deposited at nominally identical growth parameters as well as other κ-Ga₂O₃ thin films reported in the literature. Furthermore, the broadening of X-ray reflections and therefore the crystalline quality of our thin films, as revealed by reciprocal space maps, is less than previously reported for κ-Ga₂O₃.^12,25,27 The surface roughness of our layers, which can be adjusted by optimal growth parameters as low as 0.5 nm, is on par with high-quality homoepitaxial β-Ga₂O₃ thin films and represents the lowest value for κ-Ga₂O₃ at present. We were further able to unambiguously confirm the growth in the orthorhombic crystal system as well as the epitaxy of our layers on c-sapphire, YSZ(111), STO(111), and MgO(111) substrates. Optical transmission measurements confirmed the transparency of our layers up to the deep UV-regime with bandgaps of around 4.9 eV consistent with the literature. Additionally, we revealed the presence of a tin-enriched surface layer on our κ-Ga₂O₃ thin films and propose surfactant-mediated epitaxy as a possible tin-assisted growth mechanism in the deposition of κ-Ga₂O₃ via PLD. However, for a further clarification of the growth mechanism, additional studies are necessary. Our findings corroborate the promising properties of κ-Ga₂O₃ as a thin film layer for high-performance wide-bandgap electronic and optoelectronic devices. The challenge of efficient doping of this metastable phase however remains an issue.

See supplementary material for the thickness and growth rate of the Ga₂O₃ thin films as a function of p(O₂), XRD 2θ-ω scans of asymmetric reflections and rocking curves of symmetric and asymmetric reflections of the ≈2 μm thick κ-Ga₂O₃ sample, an exemplary linear extrapolation of the (001)-lattice plane distances calculated from several κ-Ga₂O₃(00N) reflections, XRD ϕ scans of κ-Ga₂O₃ thin films deposited on STO(111) and MgO(111) substrates, and depth-resolved XPS measurements of a κ-Ga₂O₃ sample.

We are indebted to Gabriele Ramm and Monika Hahn for PLD target fabrication and to Jörg Lenzner for EDX measurements. We also thank Ulrike Teschner for transmission measurements. This work was supported by the European Social Fund within the Young Investigator Group “Oxide Heterostructures” (SAB 100310460). M.K. and A.H. also acknowledge the Leipzig School for Natural Sciences BuildMoNa. T.S. and N.K. acknowledge support from the DFG (SFB951). We acknowledge support from the German Research Foundation (DFG) and Leipzig University within the program of Open Access Publishing.