ScN in the rock salt structure is a well-investigated material due to its desirable properties like the high hardness or large thermal conductivity. Recent computations by Adamski et al. [Appl. Phys. Lett. 115, 232103 (2019)] showed that ScN/GaN heterostructures exhibit an outstanding polarization gradient which would be beneficial for polarization induced electron gases. The pseudobinary semiconductor Sc xAl 1 xN, when maintaining the cubic rock salt structure, could be beneficial for tailoring the polarization gradient using the Sc dependency of material properties. The structural properties of rs-Sc xAl 1 xN are not fully discovered yet, thus in this work, DC-magnetron sputtered cubic rock salt Sc xAl 1 xN thin films with 0.55 < x < 1.00 were grown and analyzed on ScN(111)/Si(111). The epitaxial relation of ScN(111) thin films on the Si(111) substrate is determined to be ScN[110]   Si[100]. Furthermore, concentration dependent properties like the lattice parameter of Sc xAl 1 xN were measured [a(ScN) = 4.50 Å, a(Sc0.55Al0.45N) = 4.30 Å] and the stress σ within the layers was determined. The crystal quality was evaluated using ω-scans, revealing FWHM = 1.14 ° for Sc0.95Al0.05N. The diameters of the columns were determined by atomic force microscopy and scanning electron microscopy and they are range from 34 to 59 nm for 0.55 < x < 1.00. At x = 0.55, Sc xAl 1 xN columns in the hexagonal wurtzite as well as cubic rock salt structure were detected. This information about the structural specifications of Sc xAl 1 xN in the rock salt structure forms the basis for further investigations and experimental confirmation of the electric properties of ScN/GaN heterostructures or even a Sc xAl 1 xN/GaN based approach for improved structures for high-electron-mobility transistors.

A new approach for high-electron-mobility transistors (HEMTs) was proposed by Adamski et al.1, stating that ScN(111)/GaN(0001) heterostructures exhibit a giant polarization discontinuity at the interface. Therefore, a carrier sheet charge at the interface of 8.5 × 10 14 cm 2 could be achieved. This is especially compelling because cubic ScN(111) and hexagonal GaN(0001) can be grown with a lattice mismatch of merely 0.02%. If the stack of the heterostructure is reversed [GaN(0001)/Sc xAl 1 xN(111)], the piezoelectric polarization of GaN only depends on the in-plane atomic distance dN-Sc of Sc xAl 1 xN, which is dependent on the Sc content x. Polarization induced electron accumulation with a density up to 2 × 10 13 cm 2 was already observed for lattice-matched wz-Sc0.18Al0.82N(0001)/GaN(0001) heterostructures. This led to the formation of a two-dimensional electron gas (2DEG).2 In order to produce heterostructures of GaN(0001) and Sc xAl 1 xN(111), sputter epitaxy is a desirable method due to it being a cost-efficient technique for industrial scale thin film production, besides other advantages like a large deposition area and compatibility with complementary metal-oxide-semiconductors (CMOSs),3 although the crystal quality and surface morphology still present a challenge.4 In order to use sputtered Sc xAl 1 xN thin films for microelectromechanical devices, the crystal quality needs to be improved. When thin films show a comparable quality to thin films grown by molecular beam epitaxy (MBE) or hydride vapor phase epitaxy (HVPE), they could also be used, e.g., for HEMT structures. In this work, an optimized pulsed DC-magnetron co-sputtering technique was used with a metal interlayer and NH3 as nitrogen source as described by Hörich et al.5 Along with low-cost Si(111) substrates, sputtering is a highly promising method for cost-efficient, heteroepitaxial Sc xAl 1 xN thin films.6 Cubic Sc xAl 1 xN has yet to be the focus of research and many of its properties have yet to be discovered, although the ternary semiconductor has numerous application possibilities due to enhanced properties compared to ScN and an often linear behavior dependent on the Sc concentration down to the alloy-induced phase transition to the wurtzite structure.7 

So far, only wurtzitic Sc xAl 1 xN was known for its prominent increase of the piezoelectric coefficient d33. It was revealed in 2009 that d33 can rise up to 28 pCN 1 at 400 °C for a Sc concentration of 38%.8 For comparison, AlN only shows a value of d 33 = 6 pCN 1 under same measurement conditions.9 The binary compounds exhibit a hexagonal wurtzite (wz) structure for AlN (Strukturbericht designation B4, space group P63mc) and a cubic rock salt (rs) structure for ScN (Strukturbericht designation B1, space group Fm3m), respectively.10 The ternary compound Sc xAl 1 xN undergoes a solid–solid phase transition with increasing Al content, unlike other group III-nitrides like Ga xAl 1 xN or In xAl 1 xN, which maintain the wurtzite structure for any alloy composition. An abrupt phase transition from hexagonal wurtzite to cubic rock salt structure is predicted to be within 0.3 < x < 0.6.7 The growth parameters and boundary conditions can be crucial in determining whether Sc xAl 1 xN appears cubic or hexagonal, like the selection of the seed layer,11 the growth temperature,12 the strain,10 and the thickness of the film.13 For example, Sc xAl 1 xN crystallites were stabilized in the rock salt structure within a TiN/Sc xAl 1 xN superlattice with x = 0.1614 or x = 0.2815 by DC-magnetron co-sputtering. Going beyond thermodynamic stability criteria can cause undesired spinodal decomposition,9,16,17 abnormally oriented18 or misoriented grains or columns.19 

Hitherto, epitaxial thin films of rs-Sc xAl 1 xN were grown by magnetron sputter epitaxy20–22 on diverse substrates, but the structural properties of the films have not been analyzed in detail. For instance, there is insufficient information about column diameters or crystal quality of Sc xAl 1 xN, but both properties are an important factor for material properties like thermal conductivity or electric conductivity. Electron mobility is decreased by dislocations and so is the device efficiency.23 This applies for dislocations as well as for other defects like grain boundaries. Therefore, samples with larger column diameters and, thus, a reduced number of grain boundaries are preferred for HEMT structures.7 Furthermore, the epitaxial relation of cubic Sc xAl 1 xN to a ScN seed layer and to a Si(111) substrate was unknown. In this paper, the structural specifications of DC-magnetron sputtered rs-Sc xAl 1 xN on Si(111) with a Sc content between 55% and 100% will be examined for thin film applications.

The series of samples investigated in this work was fabricated via DC-pulsed magnetron co-sputtering with power varying from 20 to 120 W on the Al target and a constant power on the Sc target with 100 W. p-Si(111) was used as a substrate, which was heated under H2 in the growth chamber in order to get an oxygen-free surface. A metallic interlayer was deposited on the substrate for 30 s with only 3 nm thickness. This is followed by few nm thick ScN grown with N2 as nitrogen supply. Then, the actual ScN buffer layer was grown with NH3 with around 100 nm thickness. On top, the Sc xAl 1 xN thin film with around 250 nm thickness was grown with NH3. The ammonia flow was between 3 and 6 cm3/min. The Sc content was controlled by the power on the Al target and determined by x-ray fluorescence spectroscopy (XRF), see Table I. A base pressure of 5 × 10 9 mbar and a working pressure of 10 2 mbar was applied. Both targets have a high purity with 5N (Sc) and 6N5 (Al) and the gases have a purity of 9N. The growth temperature was kept constant at 900 °C.

TABLE I.

Overview of sputtered ScxAl1−xN thin films and the correlation of the Sc content to the power P in W on the Al target. The Sc content was determined by XRF and correlates with the power applied to the Al target with steady power at the Sc target of 100 W.

x1.000.960.920.870.850.810.740.690.590.55
P in W  20 40 50 60 70 80 90 100 110 120 
x1.000.960.920.870.850.810.740.690.590.55
P in W  20 40 50 60 70 80 90 100 110 120 
X-ray diffractometry (XRD) is a powerful tool to determine lattice parameters. Due to the fact that the interplanar distance d0002 of the wurtzite and d111 of the cubic rock salt structure are similar, as well as that the reflections 1 1 ¯ 04 of the wz-lattice and 204 of the rs-lattice are at the same 2 θ-position, it is hard to distinguish them using only θ / 2 θ-scans. The atomic distance dhkl of the measured value of θ may be the same, but in order to determine the lattice parameter a of rs-Sc xAl 1 xN using reflection 111 or 204, Eq. (1) needs to be applied. The lattice parameter c of c-plane wz-Sc xAl 1 xN can be obtained directly by Bragg’s law: n λ = 2 d h k l sin θ. In order to calculate lattice parameter a of wz-Sc xAl 1 xN 1 1 ¯ 04, the lattice parameter c is required, following Eq. (2):19 
(1)
(2)
Problems may occur while distinguishing the wurtzite and the rock salt structure of Sc xAl 1 xN. For the purpose of confirming the presumed crystalline structure, φ χ-scans, so-called pole figures, are required. The threefold symmetry of 200, 020, and 002 reflections correlate with the cubic symmetry of (111)-oriented thin films, and the sixfold symmetry of reflection 1 1 ¯ 04 proves the hexagonal structure.19 In this work, an XPertMRD3 by Malvern Panalytical and CuK α 1 radiation with a hybrid monochromator [2xGe(220)] and a PIXcel3D-Medipix3 detector was used. The measured values are corrected with respect to the position of the Si reflection 111 to avoid errors in 2 θ caused by instrument offsets. A scanning electron microscope (SEM) from Thermofischer (Scios 2 DualBeam) was used for the characterization of the surface microstructure, and the transmission electron microscope (TEM) images were done utilizing a Talos F200X. For the atomic force microscopy (AFM) images, a JPK Instruments NanoWizard 3 NanoScience was used and the images were recorded in the AC mode.
In order to identify the epitaxial relation between the substrate, the seed layer, and the thin film, φ-scans are consulted. Figure 1 shows φ-scans of the Si(111) substrate and the ScN(111) seed layer. The positions of their peaks reveal their epitaxial relationship. The reflections of Si { 400 } planes show three peaks at φ = 115 °, φ = 235 °, and φ = 355 °, confirming the threefold rotation axis along [ 111 ]. Si 200 is extinguished by destructive interference; hence, the Si 400 reflection is compared to the reflections of ScN { 200 } planes. They show three peaks with high intensity at φ = 55 °, φ = 175 °, and φ = 295 ° and three peaks with lower intensity at φ = 115 °, φ = 235 °, and φ = 355 °. The latter peaks are at the exact same positions as the peaks of Si { 400 }. This shows that the majority of ScN(111) columns are rotated by 60 ° with respect to the substrate and they do not exhibit cube-on-cube epitaxy, which was described for rs-Sc0.27Al0.73N(111) on MgO(111).11 The minority of ScN(111) columns is oriented with a direct epitaxial relation to the Si(111) substrate within the basal plane. This growth behavior was described before for MBE-grown ScN(111) films on Si(111).24 The number of rotated columns can be associated with the intensity of the reflections. In order to ensure this epitaxial relationship, ScN 220 was measured and shows peaks at the same φ-angles as the substrate. Low-intensity peaks appear rotated by 60 °. This ensures the epitaxial relation of the majority of columns ScNmaj by 60 ° to the substrate and a cube-on-cube epitaxial relation of the minority of columns ScNmin to the substrate,
(3)
(4)
(5)
In order to fully understand the epitaxial relationship, transmission electron microscope (TEM) imaging was used. Figure 2(a) shows a TEM image of the interface of Si(111) and ScN(111). ScN shows its atoms as bright spots aligned in squares over the whole range of the image. This proves that the (111)-orientation of ScN starts right at the interface to Si(111). The horizontally aligned ScN atoms of the lattice plane (111) are parallel to horizontally aligned Si atoms. Due to the fact that Si is present in the diamond structure, the atoms are not aligned vertically as the Sc atoms in the ScN rock salt structure. The interface in Fig. 2(a) is not clearly visible because the Si structure is distorted. This can have multiple causes. The areas above and below were regarded in order to determine the relation of the interface as it is shown in Fig. 2(b). The atomic distance between two Si atoms is dSi–Si = 1.92 Å and inter-atomic distance in ScN dN-Sc = 1.52 Å were determined using the TEM images. They are in agreement with the calculated atomic distances which are calculated from lattice parameters determined by x-ray diffractometry (XRD) in Sec. III D: dSi–Si = aSi 2/4 = 1.92 Å and dN-Sc = aSi 2/4 = 1.58 Å. Therefore, a lattice mismatch of 17.5% results and ScN builds a coincidence lattice on Si(111) [Fig. 2(b)]. In this relation, 5 Si atoms match on 6 N (Sc) atoms, leaving one dangling bond. This epitaxial relation was determined using the atomic distances of the planes of ScN(111) compared to Si(111) which are extracted from Fig. 2(a). The selected area electron diffraction (SAED)-TEM diffraction pattern [Fig. 2(c)] of the zone axis [ 11 2 ¯ ] shows reflections 111 and 2 2 ¯ 0 of Si and ScN. The fact that the diffraction peaks of the substrate and ScN are oriented the same way leads to the conclusion that this is a column of ScNmin. ScNmaj would be rotated by 60 ° with respect to the Si(111) substrate. The relative distances between the reflections correlate with the lattice parameters obtained by XRD, see Sec. III D. Similar TEM images were published to prove the epitaxial relation of ScN(111) on GaN(0001)25 or on MgO(111),11 whereas the epitaxial relation to Si(111) was not revealed.
FIG. 1.

φ-scans of reflection 004 of Si(111) and reflections 002 and 220 of ScN(111) show the epitaxial relation ScNmaj[110] Si[100] and ScNmin[100] Si[100]. ScN(111) builds contact twins with a rotation by 60 °, which is indicated by the peaks with smaller intensity. The majority of ScN(111) columns is rotated by 60 ° with respect to the Si(111) substrate, the minority reveals a cube-on-cube epitaxy. These scans prove the in-plane orientation of the textured thin films.

FIG. 1.

φ-scans of reflection 004 of Si(111) and reflections 002 and 220 of ScN(111) show the epitaxial relation ScNmaj[110] Si[100] and ScNmin[100] Si[100]. ScN(111) builds contact twins with a rotation by 60 °, which is indicated by the peaks with smaller intensity. The majority of ScN(111) columns is rotated by 60 ° with respect to the Si(111) substrate, the minority reveals a cube-on-cube epitaxy. These scans prove the in-plane orientation of the textured thin films.

Close modal
FIG. 2.

(a) TEM image of the interface between the ScN(111) seed layer and the Si(111) substrate. (b) Epitaxial correlation between ScN(111) and Si(111) forming a coincidence lattice. One dangling bond appears every 5 Si atoms, matching to 6 N (Sc) atoms. (c) SAED-TEM diffraction pattern of the zone axis [ 11 2 ¯ ] revealing 111 and 2 2 ¯ 0 reflections of ScN and Si, respectively. They are in-plane aligned in the same way; therefore, they are related ScNmin[100]   Si[100].

FIG. 2.

(a) TEM image of the interface between the ScN(111) seed layer and the Si(111) substrate. (b) Epitaxial correlation between ScN(111) and Si(111) forming a coincidence lattice. One dangling bond appears every 5 Si atoms, matching to 6 N (Sc) atoms. (c) SAED-TEM diffraction pattern of the zone axis [ 11 2 ¯ ] revealing 111 and 2 2 ¯ 0 reflections of ScN and Si, respectively. They are in-plane aligned in the same way; therefore, they are related ScNmin[100]   Si[100].

Close modal

1. Structure of ScN(111)

The 100 nm thick ScN seed layers which were grown on Si(111) were investigated by XRD in order to extract the lattice parameters a and a . The θ / 2 θ-scan of ScN in Fig. 3 shows reflection 111 of ScN at 2 θ = 34.503 ° as an example. Therefore, the ScN thin film is (111)-oriented, which means the plane (111) is parallel to the sample’s surface. For the whole series of samples, the average peak position of reflection 111 of ScN is 2 θ = ( 34.485 ± 0.082 ) °. In order to calculate the mean lattice parameter of ScN(111) on Si(111), reflections 111, 222 [ 2 θ = ( 73.006 ± 0.209 ) °] and 333 [ 2 θ = ( 125.586 ± 0.455 ) °] were determined and the resulting mean lattice parameter is a ¯ = ( 4.497 ± 0.005 ) Å. In order to determine the in-plane lattice parameter, reflection 204 was determined with an asymmetric measurement geometry. Figure 3 shows reflection 204 at 2 θ = 99.795 ° as an example. For the whole set of samples, the reflection appears at 2 θ = ( 99.676 ± 0.858 ) ° which results in an averaged lattice parameter a ¯ = ( 4.5079 ± 0.0147 ) Å. Both lattice parameters are according to published values by Moram et al. with a = ( 4.5047 ± 0.0005 ) Å.24 The crystal quality of the ScN thin film is still challenging, with a FWHM = ( 1.28 ± 0.19 ) ° of a rocking curve (RC, ω-scan) of reflection 111. Burmistrova et al.26 proved the FWHM of a RC of the 111 reflection of ( 500 ± 20) nm thick ScN thin films to be 0.675 °; therefore, the thin film quality from this work could still be improved upon. Figure 4 shows a cross section of a ScN thin film which depicts that the material is not monocrystalline but grows in columns which are in-plane oriented. With this SEM image, the column diameters can be roughly determined to be ( 70 ± 20 ) nm. The width of the column vary from 50 up to over 90 nm and exact values are presented in Sec. III E 2. In order to prove that the films are textured and to define the in-plane orientation of the columns, a pole figure is required. Figure 5 shows three reflections (200, 020, and 002) with high intensity at χ = 54.7 ° and φ = [ 0 ° , 120 ° , 240 ° ], with a distance of 120 °, respectively. This threefold symmetry proves that the (111)-oriented columns are in-plane oriented. With a skew of 60 ° in φ, three further reflections with lower intensity show up at φ = [ 60 ° , 180 ° , 300 ° ]. These observations indicate contact twinning of the columns, which means that the columns are rotated by 60 ° to each other around [ 111 ]. These results ensure the epitaxial relation which was determined in Sec. III A. Contact twins with a distance of φ = 60 ° were observed earlier for sputtered ScN(111) thin films on GaN(0001)25 as well as on MBE-grown ScN(111) thin films on GaN(0001)27 with no preferred in-plane orientation detectable in both cases. Moram et al.24 observed a threefold symmetry of 200, 020, and 002 reflections as well as reflections rotated by φ = 60 ° with lower intensity.

FIG. 3.

θ/2 θ-scans of rock salt ScN(111) reflection 111 (symmetric) at 2 θ = 34.503 ° and 204 (asymmetric) at 2 θ = 99.795 °.

FIG. 3.

θ/2 θ-scans of rock salt ScN(111) reflection 111 (symmetric) at 2 θ = 34.503 ° and 204 (asymmetric) at 2 θ = 99.795 °.

Close modal
FIG. 4.

Cross-sectional SEM image of the samples. Monocrystalline Si(111) in dark gray on the bottom, then in bright gray columnar ScN(111) thin film with thickness t 1 = 132 nm and on top columnar Sc xAl 1 xN(111) with t 2 = 213 nm. The diameters of the columns are ( 70 ± 20 ) nm.

FIG. 4.

Cross-sectional SEM image of the samples. Monocrystalline Si(111) in dark gray on the bottom, then in bright gray columnar ScN(111) thin film with thickness t 1 = 132 nm and on top columnar Sc xAl 1 xN(111) with t 2 = 213 nm. The diameters of the columns are ( 70 ± 20 ) nm.

Close modal
FIG. 5.

Pole figure ( φ χ-scans) of rock salt ScN(111) reflections 200, 020, and 002. The rock salt (111)-oriented columns show a threefold symmetry of contact twins with a rotation by 60 °. The intensity values are adapted to relative units for better comparability.

FIG. 5.

Pole figure ( φ χ-scans) of rock salt ScN(111) reflections 200, 020, and 002. The rock salt (111)-oriented columns show a threefold symmetry of contact twins with a rotation by 60 °. The intensity values are adapted to relative units for better comparability.

Close modal

2. Stress state of ScN(111)

For the evaluation of stress state within the thin film, an unstrained lattice parameter is required. In this work, aScN = 4.5013 Å is used as the unstrained lattice parameter, as this value is confirmed by several publications.22,28–30 Therefore, the strain along [ 111 ] can be calculated to be ε = 0.0010 and the in-plane strain to be ε = 0.0015. The negative value indicates a compressive strain in the direction [ 111 ] and the positive value means the lattice has a tensile strain in-plane, e.g., parallel to the surface. Using the elastic coefficients ( C 11 = 369 GPa, C 12 = 100 GPa, and C 44 = 164 GPa) simulated by Wu et al.,20 the stress within the ScN thin film is calculated to be σ = 0.464 GPa and σ = 0.669 GPa. The in-plane biaxial stress of ScN(111) thin films was determined earlier to be below 1 GPa but did not reveal any value.24 Nonetheless, the ScN(111) thin film is grown in a very low stress state on Si(111).

The in-plane orientation of the columnar rs-Sc xAl 1 xN(111) thin films can be proven using pole figures. Figure 6 shows a threefold rotation symmetry of the 200, 020, and 002 reflections at χ = 54.7 ° for rs-Sc0.59Al0.41N exemplarily. Each pole figure of the whole series of rs-Sc xAl 1 xN thin films investigated in this work showed similar features. Comparable to the ScN seed layer, the Sc xAl 1 xN thin films exhibit contact twins, which is visible by a shift of 60 ° in φ of the reflections with lower intensity, maintaining strongly (111)-oriented columns. This is further proven by the fact that the 200, 020, and 002 reflections in Fig. 6 appear at the same χ and φ positions. Thin films of rs-Sc xAl 1 xN with 0.55 < x < 0.96 show the same in-plane orientation and comparable intensity differences between the contact twins as their ScN(111) seed layer. This relation is maintained over the whole thickness of 250 nm. Thus, the contact twins of the Sc xAl 1 xN(111) thin film are prescribed by the ScN seed layer and originate at the interface to the Si(111) substrate. The in-plane orientation is maintained over the interface from ScN(111) to Sc xAl 1 xN(111), keeping the same relation to Si(111) as described in the epitaxial relations [Eqs. (4) and (5)]. The properties of the 250 nm thick Sc xAl 1 xN(111) layer are described in the following section.

FIG. 6.

Pole figure ( φ χ-scans) of rs-Sc0.59Al0.41N(111) reflections 200, 020, and 002. The high-intensity reflections as well as the low-intensity reflections show a threefold symmetry and indicate the formation of contact twins with a rotation by 60 ° in φ, respectively. The intensity values are adapted to relative units for better comparability.

FIG. 6.

Pole figure ( φ χ-scans) of rs-Sc0.59Al0.41N(111) reflections 200, 020, and 002. The high-intensity reflections as well as the low-intensity reflections show a threefold symmetry and indicate the formation of contact twins with a rotation by 60 ° in φ, respectively. The intensity values are adapted to relative units for better comparability.

Close modal

1. Structure of ScxAl1−xN

XRD measurements are used in order to determine the lattice geometry of rs-Sc xAl 1 xN thin films. Figure 7 depicts the measured diffractograms of the Sc xAl 1 xN(111)/ScN(111) heterostructures with varying composition x ranging from 0.55 to 1. The correlation of the target power and the Sc content is given in Table I. The ScN 111 reflection of the seed layer occurs for each sample at 2 θ = ( 34.48 ± 0.09 ) °. The rs-Sc xAl 1 xN 111 reflection shifts to lower 2 θ-values from 2 θ = 36.13 ° for x = 0.55 to 2 θ = 34.63 ° for x = 0.96 with increasing Sc content which is caused by decreasing power on the Al target (from 120 to 40 W). The diffractogram also shows the ScN 200 reflection at 2 θ = ( 40.08 ± 0.20 ) ° which does not shift with varying power on the Al target. These {100}-oriented columns, consisting of pure ScN, appear in the ScN seed layer of the heterostructure. Al atoms show an increased affinity for incorporation compared to other group III metals;31 therefore, the growth kinetics could never lead to the growth of pure ScN while Al atoms are present. To the best of the authors knowledge, this is the first time that the growth of so-called abnormal oriented grains (AOG) has been detected by XRD because ScN(100) columns in the ScN(111) films are sufficiently abundant to be detectable in the graph. AOGs can be suppressed by reducing the Sc seed layer thickness below 1 nm, thus resulting in higher FWHM ω values for the reflections by { 111 }-planes. The out-of-plane lattice parameters a are calculated from the 111 reflections. The in-plane lattice parameters of the scans of reflection 204 were made using an asymmetric geometry, indicated by a . Figure 8 shows the calculated lattice parameters from the reflections of Fig. 7. The smallest lattice parameter is a = ( 4.30 ± 0.01 ) Å [ a = ( 4.29 ± 0.01 ) Å] of rs-Sc0.55Al0.45N and the largest lattice parameter is a = ( 4.50 ± 0.01 ) Å [ a = ( 4.49 ± 0.02 ) Å] for Sc0.96Al0.04N. The Sc dependency x on the lattice parameter a(x) of rs-Sc xAl 1 xN is described by Eq. (6),
(6)
This is in agreement with Vegard’s law and the reported measured lattice parameters of rs-Sc xAl 1 xN(111) ( 0.7 < x < 0.95) on Si(100) of Satoh et al.21 (Fig. 8) and the calculated lattice parameter by Berkok et al.32 [ a ( x = 0.75 ) = 4.341 Å]. The simulated data by Alling et al.33 slightly overestimate the lattice parameter with a value of aScN = 4.521 Å and reveal a bowed trend with increasing Al content. Yanagitani et al. reported lattice parameters of sputtered wz-Sc xAl 1 xN with c ( x = 0.47 ) = 5.08 Å up to c ( x = 0.63 ) = 5.14 Å,34 which could be interpreted as well as rs-Sc xAl 1 xN(111) using Bragg’s law and Eq. (1) with a ( x = 0.47 ) = 4.40 Å and a ( x = 0.63 ) = 4.46 Å. These lattice parameters of rs-Sc xAl 1 xN agree with the values reported in this work, whereas not many experimental data are published with a Sc content around or below 50%. For cubic crystals, the ratio of a to a for rs-Sc xAl 1 xN equals 1 for unstrained crystals. The ratio of the atomic distance in-plane vs out-of-plane is 235 for rock salt crystals and can be calculated by Eq. (7),
(7)
The results of the lattice ratio are shown along with the lattice ratios calculated by Satoh et al.21 and Wu et al.20 in Fig. 9. The samples investigated in this work are very close to the ideal lattice ratio, whereas all values increasingly deviate from the ideal ratio with increasing Al content. This means that a and a do not deviate significantly from each other. The lattice parameters of abnormally oriented grains [ScN(100)] are constant with a mean value of a ¯ = ( 4.50 ± 0.01 ) Å, extracted from reflection 200 (Fig. 7). The error bars of all measured lattice parameters are smaller than the data points. Figure 10 shows a reciprocal space map (RSM) of asymmetric reflection 204 of ScN [ a = ( 4.50 ± 0.01 ) Å] and rs-Sc0.95Al0.05N [ a = ( 4.49 ± 0.01 ) Å] as an example. The two reflections are not aligned with respect to q x, which was observed for all samples investigated in this work. This means that rs-Sc xAl 1 xN does not grow pseudomorphically on ScN. It is necessary to perform an exact measurement to distinguish the reflections for the thin films with high Sc content because they appear close to each other. The rs-Sc xAl 1 xN and ScN reflections 204 move away from each other in q x with increasing Al content.
FIG. 7.

θ/2 θ-scans of sputtered rs-Sc xAl 1 xN ( 0.55 < x < 1.00 ) thin films on a Si(111) substrate. The ScN 200 reflection appears at 40.08 °.

FIG. 7.

θ/2 θ-scans of sputtered rs-Sc xAl 1 xN ( 0.55 < x < 1.00 ) thin films on a Si(111) substrate. The ScN 200 reflection appears at 40.08 °.

Close modal
FIG. 8.

Lattice parameters a (rs-Sc xAl 1 xN reflection 111) and a (rs-Sc xAl 1 xN reflection 204) measured by XRD. A selection of experimentally determined lattice parameters c from wz-Sc xAl 1 xN21 and rs-Sc xAl 1 xN20,21 lattice parameters a are shown as full symbols to contextualize the measured data. The open symbols show simulated lattice parameters c17,36 and a.33 The straight line shows Vegard’s law between rs-AlN37 and rs-ScN.38 

FIG. 8.

Lattice parameters a (rs-Sc xAl 1 xN reflection 111) and a (rs-Sc xAl 1 xN reflection 204) measured by XRD. A selection of experimentally determined lattice parameters c from wz-Sc xAl 1 xN21 and rs-Sc xAl 1 xN20,21 lattice parameters a are shown as full symbols to contextualize the measured data. The open symbols show simulated lattice parameters c17,36 and a.33 The straight line shows Vegard’s law between rs-AlN37 and rs-ScN.38 

Close modal
FIG. 9.

Lattice ratio of rs-Sc xAl 1 xN thin films. Green triangles show the results of this work [rs-Sc xAl 1 xN/Si(111)], and blue diamonds and squares show the results of Satoh et al.21 and Wu et al.,20 respectively.

FIG. 9.

Lattice ratio of rs-Sc xAl 1 xN thin films. Green triangles show the results of this work [rs-Sc xAl 1 xN/Si(111)], and blue diamonds and squares show the results of Satoh et al.21 and Wu et al.,20 respectively.

Close modal
FIG. 10.

Reciprocal space map of ScN and rs-Sc0.95Al0.05N reflection 204, respectively. The lattice parameter of Sc0.95Al0.05N [ a = ( 4.49 ± 0.01 ) Å] is slightly smaller than the one of ScN [ a = ( 4.51 ± 0.01 ) Å]. The reflections are aligned with respect to q x, proving the pseudomorph growth of the rs-Sc xAl 1 xN thin films.

FIG. 10.

Reciprocal space map of ScN and rs-Sc0.95Al0.05N reflection 204, respectively. The lattice parameter of Sc0.95Al0.05N [ a = ( 4.49 ± 0.01 ) Å] is slightly smaller than the one of ScN [ a = ( 4.51 ± 0.01 ) Å]. The reflections are aligned with respect to q x, proving the pseudomorph growth of the rs-Sc xAl 1 xN thin films.

Close modal

2. Crystal quality of ScxAl1−xN

Many aspects are relevant when determining the crystal quality of a thin film semiconductor. In the following, ω-scans, so-called rocking curves, were measured by XRD and the full width at half maximum (FWHM) was determined. The FWHM of ω-scans is an indicator for the amount of crystal planes which are aligned in a way that they obey Bragg’s law at a certain 2 θ angle. Tilted lattice planes, caused by misoriented grains, defects, or dislocations, interfere constructively at a slightly different angle and broaden the peak in the diffractogram. The FWHM of the ω-scans of the ScN seed layer reflection 111 measures the value FWHM ω = ( 1.31 ± 0.19 ) ° and the FWHM ω of the rs-Sc xAl 1 xN reflection 111 are given in Table II. The values are between FWHM ω = 2.17 ° for rs-Sc0.55Al0.45N and FWHM ω = 1.14 ° for rs-Sc0.95Al0.05N. This means that the crystal quality increases with increasing Sc content. This can be explained by the incorporation of the rather small Al atoms, compared to Sc atoms in the rs-ScN lattice, leading to inconsistent bond lengths and strain in the lattice. The bond length of the rs-lattice is increasing with increasing Sc content and, thus, the strain is increasing as well. These results reveal a higher crystal quality than ScN(111) sputtered on MgO(111) at 700 °C by Le Febvrier et al.28 with FWHM = 2.1 °. These values of sputtered rs-Sc xAl 1 xN thin films are even lower than sputtered wz-Sc xAl 1 xN with FWHM = 3.06 ° of the 0002 reflection.12 The rs-Sc0.95Al0.05N 204 reflection shows a FWHM ω = 0.83 ° and for the thinner ScN seed layer FWHM ω = 1.05 ° (shown in Fig. 7). This is a remarkable crystal quality for sputtered rs-Sc xAl 1 xN(111), which was grown with a stabilized growth process and using NH3 as nitrogen source.5 Of course, the crystal quality is not as good as ScN thin films grown by MBE [FWHM(ScN(111)) =  0.55 °]39 or HVPE [FWHM(ScN(100)) =  0.45 °].40 

TABLE II.

The FWHMω of the 111 reflection is given for rs-ScxAl1−xN(111) 250 nm thick films with 0.55 < x < 1.00.

x0.960.920.870.850.810.740.690.590.55
FWHMω  1.14 1.50 1.11 1.30 1.63 1.63 1.64 1.71 2.17 
x0.960.920.870.850.810.740.690.590.55
FWHMω  1.14 1.50 1.11 1.30 1.63 1.63 1.64 1.71 2.17 

3. Stress state of ScxAl1−xN

Sc xAl 1 xN is growing strained on the ScN(111) seed layer. This in-plane deformation can be quantified by the strain ε using the lattice parameter a , whereas the out-of-plane strain along [111] is calculated by a and is denoted by ε . Figure 11 shows both ε and ε calculated for the range of 0.55 < x < 1.00. The values strongly depend on the unstrained lattice parameter used in the calculations. The ab initio simulated values for rs-Sc xAl 1 xN by Alling et al.33 were used as unstrained lattice parameter a0 and can be described by Eq. (8) with the bowing parameter b = 0.126,
(8)
Sc0.55Al0.45N and Sc0.59Al0.41N thin films show the largest strain along [111] with ε = 0.0124 and ε = 0.0125, respectively. The absolute value for strain increases with increasing Al content. The negative values show that Sc xAl 1 xN(111) grows compressively on ScN(111). Due to the low lattice mismatch, Sc xAl 1 xN grows fully strained on ScN. By using the elastic tensor Cij simulated by Wu et al.,20 the stress can be calculated in-plane ( σ ) and out-of-plane ( σ ) (Fig. 11). Sc0.55Al0.45N shows the highest stress state along [111] with σ = 6.126 GPa. Sc0.95Al0.05N has the lowest stress state parallel to the interface with ScN with σ = 1.442 GPa.
FIG. 11.

The stress σ and strain ε of Sc xAl 1 xN(111) thin films for 0.55 < x < 1.00. The out-of-plane values are indicated by and the in-plane values by . The dashed line shows a quadratic fit of the data.

FIG. 11.

The stress σ and strain ε of Sc xAl 1 xN(111) thin films for 0.55 < x < 1.00. The out-of-plane values are indicated by and the in-plane values by . The dashed line shows a quadratic fit of the data.

Close modal

1. Surface morphology of ScxAl1−xN

Figure 12 depicts a SEM image of a rs-Sc0.87Al0.13N(111) thin film. The vast majority of column surfaces show a triangular shape with a threefold symmetry along [ 111 ], showing the { 200 } facets. These columns are in-plane oriented, so they point to the top of the image, like the column highlighted in yellow. The minority of columns are pointing downward in the image, one of them highlighted with a red circle. They are in-plane oriented as well but build contact twins with a rotation of β = 60 °. Rotated domains were already examined for DC-magnetron sputtered ScN(111) on Si(111)24 or on GaN(0001).25 The amount of columns which point upward vs the amount of columns which point downward show the same relation as the reflections with higher intensity vs the reflections with lower intensity in the pole figure in Fig. 6. This image is exemplary for the whole set of samples with 0.55 < x < 1.00, whereas every sample showed triangularly shaped columns for most of the surface and few abnormally oriented grains (AOG). These AOG are 100 -oriented and have a fourfold rotation symmetry, given by 90 ° angles of the column edges. With diameters of 100–200 nm, AOGs are bigger than (111)-oriented Sc xAl 1 xN columns. These abnormal oriented grains build rotation twins with an angle α, which is inconsistent for each AOG. The Sc xAl 1 xN(100) grains are fiber-textured, which means that they are oriented along [100] with respect to the growth direction, but there is no preferred in-plane orientation. Also, the Sc xAl 1 xN(100) reflection cannot be detected by XRD. Sandu et al.18 reported AOGs in sputtered wz-Sc0.43Al0.57N(0001) thin films on Si(111) with larger diameters than the (0001)-oriented columns. Additionally, grains which are { 110 }-oriented or in other arbitrary angles with respect to the [ 111 ] direction are observable in the bottom right corner of Fig. 12, but there are insufficient grains to interfere constructively and be detectable in the x-ray diffractogram. Furthermore, two 110 -oriented grains are observed to be penetration twins (Fig. 14), growing through each other. ScN(110) was only reported to grow epitaxially on MgO(110) and m-plane α-Al2O3 so far.40 

FIG. 12.

SEM image of rs-Sc0.87Al0.13N(111) showing threefold, triangular-shaped columns, forming in-plane oriented contact twins with a rotation angle of 60 °. Abnormal oriented grains (AOG) align along 100 , 110 or arbitrary angles.

FIG. 12.

SEM image of rs-Sc0.87Al0.13N(111) showing threefold, triangular-shaped columns, forming in-plane oriented contact twins with a rotation angle of 60 °. Abnormal oriented grains (AOG) align along 100 , 110 or arbitrary angles.

Close modal
FIG. 13.

AFM images of rs-Sc0.95Al0.05N(111) [(a) and (b)] and rs-Sc0.55Al0.45N(111) [(c) and (d)], recorded in the AC mode. The height of the surface structure and the lock-in phase images are shown, respectively.

FIG. 13.

AFM images of rs-Sc0.95Al0.05N(111) [(a) and (b)] and rs-Sc0.55Al0.45N(111) [(c) and (d)], recorded in the AC mode. The height of the surface structure and the lock-in phase images are shown, respectively.

Close modal
FIG. 14.

Mean column diameter of 250 nm thick rs-Sc xAl 1 xN(111) films with 0.55 < x < 1.00 (squares) and ScN(111) seed layers with 100 nm thickness (triangle). The data were extracted from 1 × 1 μ m 2 AFM images recorded in the AC mode. Inset: (110)-oriented grains of rs-Sc0.95Al0.05N build penetration twins which are bigger than (111)-oriented columns. These abnormally oriented grains are bigger than 100 nm.

FIG. 14.

Mean column diameter of 250 nm thick rs-Sc xAl 1 xN(111) films with 0.55 < x < 1.00 (squares) and ScN(111) seed layers with 100 nm thickness (triangle). The data were extracted from 1 × 1 μ m 2 AFM images recorded in the AC mode. Inset: (110)-oriented grains of rs-Sc0.95Al0.05N build penetration twins which are bigger than (111)-oriented columns. These abnormally oriented grains are bigger than 100 nm.

Close modal

2. Column diameters

The column diameters on the surface of the rs-Sc xAl 1 xN(111) thin films were extracted from AFM images which were recorded in the AC mode. Figure 13 shows height images and lock-in phase images of rs-Sc xAl 1 xN(111) with x = 0.05 [Figs. 13(a) and 13(b)] and x = 0.55 [Figs. 13(c) and 13(d)] as an example. Due to the measurement mode, the triangular shape cannot be seen, but the dimensions of the roundly shaped columns can be determined in detail, especially in the lock-in phase images. This is advantageous compared to the extraction of the diameters from the SEM images. The diameters of the columns were extracted in an area of 1 × 1 μ m 2 and the mean column diameters and their variations were calculated. The results for rs-Sc xAl 1 xN(111) (squares) and ScN(111) (triangle) are given in Fig. 14. The range of column size varies a lot on each sample, nevertheless, an overall tendency of decreasing column diameters can be seen with increasing Al content from ( 78 ± 33 ) nm of pure ScN down to ( 34 ± 19 ) nm of rs-Sc0.55Al0.45N(111). These results were expected because the columns can grow larger when the thin film in less strained. As the strain increases with Al content, the column diameters decrease. These values exceed the previously reported values of rs-ScN with 60 nm size of twin domains38 and column diameters of ScN(111) with ( 24 ± 19 ) nm25 and even wz-Sc0.19Al0.81N(0001) with column diameters of 30–50 nm grown by MBE,41 although the column diameters differ by 30–60 nm on any reported samples. DC-magnetron co-sputtered wz-Sc0.43Al0.57N(0001) on Si(100) reveals larger column diameters than the samples measured in this work with ( 101 ± 36 ) nm.42 On the other hand, Casamento et al. showed wz-Sc0.18Al0.82N(0001) column diameters of ( 364 ± 300 ) nm.2 The reported values agree with the column diameters extracted from the SEM image (Fig. 12) and the intersection (Fig. 4). Considering the requirements of thin film properties for industrial applications, larger column diameters are favorable. For HEMT structures, for example, larger grains and, therefore, less grain boundaries as well as a reduced surface roughness reduce the probability of electron trapping.7 Therefore, rs-Sc xAl 1 xN(111) thin films with higher Sc content would be preferred for microelectromechanical applications.

A metastable, pseudobinary alloy like Sc xAl 1 xN can separate locally into Al- and Sc-rich regions by so-called spinodal decomposition. Another phenomenon to release the energy of the metastable compound is that it separates in two phases, namely, wurtzitic and rock salt which can co-exist with the same Sc content x. This can happen although the same boundary conditions like growth temperature, strain, etc., were applied during growth.12,41 It was reported that sputtered thin films with 90% Sc content grow as rs-Sc xAl 1 xN(111) in Sc-rich areas and wz-Sc xAl 1 xN(0001) domains in Al-rich areas.11 The SEM image in Fig. 15(a) shows two domains of the rs-Sc0.59Al0.41N thin film: triangularly shaped and arbitrary, round columns. The highlighted column in the center of the image shows a 120 ° angle between crystal edges, proving the existence of a two-phase alloy by a wz-domain. This is caused by a locally lowered Sc content which could not be quantified due to the small region of occurring domains. The wz-domains appear with smaller grain diameters, whereas the triangular-shaped rs-domains reveal column diameters according to Fig. 14. The surface morphology changes immensely from the Sc0.55Al0.45N thin film compared to films with larger Sc contents, where only arbitrarily shaped grains are found in Fig. 15(b). For instance, the highlighted column shows a hexagonally shaped surface. Similar grain shapes but with only 10–20 nm diameters were observed in 120 nm thick, sputtered wz-Sc0.16Al0.84N(0001) thin films.14 The rs-Sc0.55Al0.45N 111 reflection in Fig. 7 has a broadened peak compared to the other thin films with lower Al contents. This is due to the decrease of crystalline uniformity of the film caused by the presence of more than one crystal phase. The reflection can be interpreted as wz-Sc0.55Al0.45N with lattice parameters c = ( 4.97 ± 0.02 ) Å (triangle in Fig. 7) and a = ( 3.10 ± 0.03 ) Å. Energy dispersive x-ray spectroscopy (EDXS) reveals a local Sc content of x = 0.52 compared to the averaged Sc content of x = 0.55 determined by x-ray fluorescence (XRF) spectroscopy. The resulting c/a ratio of 1.601 proves that this thin film contains wz- and rs-domains with varying Sc content. For samples with x > 0.6, such arbitrarily or hexagonally shaped columns are not found.

FIG. 15.

(a) SEM image of a Sc0.59Al0.41N thin film with rs-domains (triangular) and wz-domains (arbitrarily shaped), revealing 120 ° angles between crystal edges. (b) Sc0.55Al0.45N thin films only showing arbitrary shaped grains with significantly smaller diameters than the rs-domains of other samples.

FIG. 15.

(a) SEM image of a Sc0.59Al0.41N thin film with rs-domains (triangular) and wz-domains (arbitrarily shaped), revealing 120 ° angles between crystal edges. (b) Sc0.55Al0.45N thin films only showing arbitrary shaped grains with significantly smaller diameters than the rs-domains of other samples.

Close modal

In this work, DC-magnetron sputtered cubic rock salt Sc xAl 1 xN thin films with 0.55 < x < 1.00 and a thickness of 250 nm were grown on a ScN(111) seed layer on Si(111). The epitaxial relation of ScN(111) thin films on the Si(111) substrate was determined to be ScN[110]   Si[100]. The lattice parameter [a(ScN) = 4.50 Å], crystal quality ( FWHM ω = 1.28 °), and the stress state ( σ ¯ = 0.5662 GPa) of the ScN(111) seed layer were determined. The epitaxial relation between ScN(111) and rs-Sc xAl 1 xN reveals a cube-on-cube epitaxy with rs-Sc xAl 1 xN[100]   ScN[100]. The lattice parameter of Sc xAl 1 xN ranges from a ( x = 0.55 ) = 4.29 Å to a ( x = 0.96 ) = 4.50 Å. The crystal quality decreases from FWHM ω ( x = 0.96 ) = 1.14 ° to FWHM ω ( x = 0.55 ) = 2.17 °. The stress state decreases from σ ( x = 0.55 ) = 6.126 GPa to σ ( x = 0.96 ) = 1.442 GPa. The surface morphology of the thin films was determined by SEM images and reveals triangularly shaped columns which build contact twins with 60 ° rotation. Abnormally oriented Sc xAl 1 xN(100) grains appear bigger [ d ¯ ( x = 0.87 ) = 150 nm] than (111)-oriented columns [ d ¯ ( x = 0.87 ) = 60 nm]. The diameters of the columns [ d ¯ ( x = 0.55 ) = 34 nm and d ¯ ( x = 0.96 ) = 59 nm] were determined depending on the Sc content by AFM and SEM. For x = 0.59, Sc xAl 1 xN thin films show hexagonal columns in addition to triangularly shaped rock salt columns. At even lower Sc content with x = 0.55, only arbitrarily and hexagonally shaped columns are present at the surface. This information about the structural specifications of Sc xAl 1 xN in the rock salt structure forms the basis for further investigations and experimental confirmation of the electric properties of Sc xAl 1 xN/GaN based heterostructures.

The authors would like to thank the Gips-Schüle Foundation, the Carl-Zeiss Foundation within the project SCHARF, and the German Science Foundation (DFG) who supported this work by the priority programme SPP 2312 (GaNius—Energy Efficient Power Electronics) within Projects Nos. 441885089 and 462722619 who made these scientific results possible.

The authors have no conflicts to disclose.

S. Mihalic: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Validation (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (equal). E. Wade: Formal analysis (equal); Investigation (equal); Validation (equal); Writing – review & editing (equal). C. Lüttich: Investigation (equal). F. Hörich: Investigation (equal). C. Sun: Investigation (equal); Writing – review & editing (supporting). Z. Fu: Investigation (equal). B. Christian: Data curation (equal); Validation (equal); Visualization (equal); Writing – review & editing (equal). A. Dadgar: Methodology (equal); Resources (equal); Supervision (supporting); Validation (equal); Writing – review & editing (equal). A. Strittmatter: Funding acquisition (equal); Resources (equal). O. Ambacher: Conceptualization (equal); Funding acquisition (equal); Resources (equal); Supervision (lead); Validation (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
N. L.
Adamski
,
C. E.
Dreyer
, and
C. G.
Van De Walle
,
Appl. Phys. Lett.
115
,
232103
(
2019
).
2.
J.
Casamento
,
T. S.
Nguyen
,
Y.
Cho
,
C.
Savant
,
T.
Vasen
,
S.
Afroz
,
D.
Hannan
,
H.
Xing
, and
D.
Jena
,
Appl. Phys. Lett.
121
,
192101
(
2022
).
3.
Y.
Liu
,
Y.
Cai
,
Y.
Zhang
,
A.
Tovstopyat
,
S.
Liu
, and
C.
Sun
,
Micromachines
11
,
630
(
2020
).
4.
M.
Yassine
,
A.
Nair
,
J.
Fammels
,
E.
Wade
,
Z.
Fu
,
A.
Yassine
,
L.
Kirste
, and
O.
Ambacher
,
J. Appl. Phys.
132
,
114101
(
2022
).
5.
F.
Hörich
,
R.
Borgmann
,
J.
Bläsing
,
G.
Schmidt
,
P.
Veit
,
F.
Bertram
,
J.
Christen
,
A.
Strittmatter
, and
A.
Dadgar
,
J. Cryst. Growth
571
,
126250
(
2021
).
6.
A.
Dadgar
,
F.
Hörich
,
R.
Borgmann
,
J.
Bläsing
,
G.
Schmidt
,
P.
Veit
,
J.
Christen
, and
A.
Strittmatter
,
Phys. Status Solidi A Appl. Mater. Sci.
220
,
2200609
(
2022
).
7.
O.
Ambacher
,
S.
Mihalic
,
E.
Wade
,
M.
Yassine
,
A.
Yassine
,
N. M.
Feil
, and
B.
Christian
,
J. Appl. Phys.
132
,
175101
(
2022
).
8.
M.
Akiyama
,
T.
Kamohara
,
K.
Kano
,
A.
Teshigahara
,
Y.
Takeuchi
, and
N.
Kawahara
,
Adv. Mater.
21
,
593
(
2009
).
9.
C.
Höglund
,
J.
Birch
,
B.
Alling
,
J.
Bareo
,
Z.
Czigány
,
P. O.
Persson
,
G.
Wingqvist
,
A.
Žukauskaitė
, and
L.
Hultman
,
J. Appl. Phys.
107
,
123515
(
2010
).
10.
Y.
Duan
,
L.
Qin
, and
H.
Liu
,
J. Phys.: Condens. Matter
28
,
205403
(
2016
).
11.
C.
Höglund
,
J.
Bareo
,
J.
Birch
,
B.
Alling
,
Z.
Czigány
, and
L.
Hultman
,
J. Appl. Phys.
105
,
113517
(
2009
).
12.
E.
Österlund
,
G.
Ross
,
M. A.
Caro
,
M.
Paulasto-Kröckel
,
A.
Hollmann
,
M.
Klaus
,
M.
Meixner
,
C.
Genzel
,
P.
Koppinen
,
T.
Pensala
,
A.
Žukauskaitė
, and
M.
Trebala
,
Phys. Rev. Mater.
5
,
035001
(
2021
).
13.
S.
Nayak
,
S.
Acharya
,
M.
Baral
,
M.
Garbrecht
,
T.
Ganguli
,
S. M.
Shivaprasad
, and
B.
Saha
,
Appl. Phys. Lett.
115
,
251901
(
2019
).
14.
B.
Saha
,
S.
Saber
,
E. A.
Stach
,
E. P.
Kvam
, and
T. D.
Sands
,
Appl. Phys. Lett.
109
,
172102
(
2016
).
15.
J. L.
Schroeder
,
B.
Saha
,
M.
Garbrecht
,
N.
Schell
,
T. D.
Sands
, and
J.
Birch
,
J. Mater. Sci.
50
,
3200
(
2015
).
16.
A.
Žukauskaitė
,
C.
Tholander
,
F.
Tasnádi
,
B.
Alling
,
J.
Palisaitis
,
J.
Lu
,
P. O.
Persson
,
L.
Hultman
, and
J.
Birch
,
Acta Mater.
94
,
101
(
2015
).
17.
S.
Zhang
,
D.
Holec
,
W. Y.
Fu
,
C. J.
Humphreys
, and
M. A.
Moram
,
J. Appl. Phys.
114
,
133510
(
2013
).
18.
C. S.
Sandu
,
F.
Parsapour
,
S.
Mertin
,
V.
Pashchenko
,
R.
Matloub
,
T.
LaGrange
,
B.
Heinz
, and
P.
Muralt
,
Phys. Status Solidi A Appl. Mater. Sci.
216
,
1800569
(
2019
).
19.
M. A.
Moram
and
M. E.
Vickers
,
Rep. Prog. Phys.
72
,
036502
(
2009
).
20.
D.
Wu
,
Y.
Chen
,
S.
Manna
,
K.
Talley
,
A.
Zakutayev
,
G. L.
Brennecka
,
C. V.
Ciobanu
,
P.
Constantine
, and
C. E.
Packard
,
IEEE Trans. Ultrason. Ferroelectr. Freq. Control
65
,
2167
(
2018
).
21.
S.
Satoh
,
K.
Ohtaka
,
T.
Shimatsu
, and
S.
Tanaka
,
J. Appl. Phys.
1322
,
1
(
2022
).
22.
R.
Deng
,
P. Y.
Zheng
, and
D.
Gall
,
J. Appl. Phys.
118
,
015706
(
2015
).
23.
S.
Zhang
,
W. Y.
Fu
,
D.
Holec
,
C. J.
Humphreys
, and
M. A.
Moram
,
J. Appl. Phys.
114
,
243516
(
2013
).
24.
M. A.
Moram
,
Z. H.
Barber
,
C. J.
Humphreys
,
T. B.
Joyce
, and
P. R.
Chalker
,
J. Appl. Phys.
100
,
023514
(
2006
).
25.
S.
Acharya
,
A.
Chatterjee
,
V.
Bhatia
,
A. I. K.
Pillai
,
M.
Garbrecht
, and
B.
Saha
,
Mater. Res. Bull.
143
,
111443
(
2021
).
26.
P. V.
Burmistrova
,
J.
Maassen
,
T.
Favaloro
,
B.
Saha
,
S.
Salamat
,
Y.
Rui Koh
,
M. S.
Lundstrom
,
A.
Shakouri
, and
T. D.
Sands
,
J. Appl. Phys.
113
,
153704
(
2013
).
27.
J.
Casamento
,
J.
Wright
,
R.
Chaudhuri
,
H.
Xing
, and
D.
Jena
,
Appl. Phys. Lett.
115
,
172101
(
2019
).
28.
A.
Le Febvrier
,
N.
Tureson
,
N.
Stilkerich
,
G.
Greczynski
, and
P.
Eklund
,
J. Phys. D: Appl. Phys.
52
,
035302
(
2019
).
29.
R.
Yagoub
,
H.
Rekkab Djabri
,
S.
Daoud
,
N.
Beloufa
,
M.
Belarbi
,
A.
Haichour
,
C.
Zegadi
, and
S.
Louhibi Fasla
,
Ukr. J. Phys.
66
,
699
(
2021
).
30.
N.
Takeuchi
,
Phys. Rev. B Condens. Matter Mater. Phys.
65
,
1
(
2002
).
31.
E.
Iliopoulos
and
T. D.
Moustakas
,
Appl. Phys. Lett.
81
,
295
(
2002
).
32.
H.
Berkok
,
A.
Tebboune
,
A.
Saim
, and
A.
Belbachir
,
Physica B
411
,
1
(
2013
).
33.
B.
Alling
,
A.
Karimi
, and
I. A.
Abrikosov
,
Surf. Coat. Technol.
203
,
883
(
2008
).
34.
T.
Yanagitani
and
M.
Suzuki
,
Appl. Phys. Lett.
105
,
122907
(
2014
).
35.
Y.
Yao
and
D. D.
Klug
,
Phys. Rev. B Condens. Matter Mater. Phys.
88
,
1
(
2013
).
36.
D. F.
Urban
,
O.
Ambacher
, and
C.
Elsässer
,
Phys. Rev. B
103
,
1
(
2021
). arXiv:2002.08143.
37.
U. P.
Verma
and
P. S.
Bisht
,
Solid State Sci.
12
,
665
(
2010
).
38.
C. E.
Ekuma
,
D.
Bagayoko
,
M.
Jarrell
, and
J.
Moreno
,
AIP Adv.
2
,
032163
(
2012
).
39.
M. A.
Moram
,
T. B.
Joyce
,
P. R.
Chalker
,
Z. H.
Barber
, and
C. J.
Humphreys
,
Appl. Surf. Sci.
252
,
8385
(
2006
).
40.
T.
Ohgaki
,
I.
Sakaguchi
, and
N.
Ohashi
,
Materials
11
,
22
(
2018
).
41.
M. T.
Hardy
,
B. P.
Downey
,
N.
Nepal
,
D. F.
Storm
,
D. S.
Katzer
, and
D. J.
Meyer
,
Appl. Phys. Lett.
110
,
162104
(
2017
).
42.
S.
Fichtner
,
N.
Wolff
,
G.
Krishnamurthy
,
A.
Petraru
,
S.
Bohse
,
F.
Lofink
,
S.
Chemnitz
,
H.
Kohlstedt
,
L.
Kienle
, and
B.
Wagner
,
J. Appl. Phys.
122
,
035301
(
2017
).