Target preparation

A J30 PLD target was mixed stoichiometrically from binary oxide powders: MgO (Sigma-Aldrich, 342793), CoO (Sigma-Aldrich, 343153), NiO (Sigma-Aldrich, 203882), CuO (Alfa Aesar, 44663), ZnO (BeanTown Chemical, 211845), and Sc₂O₃ (Sigma-Aldrich, 307874). The mix was shaker-milled with 5 mm diameter yttrium-stabilized zirconia milling media for 2 h. Milling produced a wide particle distribution with diameters ranging from ~100 nm to several μm. The powder mix was pressed uniaxially into a 2.5 cm diameter pellet at 140 MPa (Carver Laboratory Press), then pressed isostatically at 350 MPa in a cold isostatic press (Autoclave Engineers, Inc.). The pellet was sintered in air at 1000 °C for 12 h (Lindberg tube furnace) and air-quenched, resulting in a geometric density of 3.5 g/cm³.

Thin film preparation

PLD employed a Coherent KrF 248 nm excimer laser, with a pulse rate of 6 Hz and target–substrate distance of 4 cm. A laser fluence of 1.3 J/cm² was achieved by focusing 0.17 J to an area of 0.13 cm². The beam was shaped using vertical slits and laser fluence was calculated for every deposition by measuring laser energy with a Coherent LabMax-Top energy meter and laser spot size with ZAP-IT^® thermal paper. Deposition rate was calibrated using x-ray reflectometry. The substrate heater was held at 400 °C for all experiments. O₂ and Ar gas purities were 99.6% and 99.999%, respectively. Substrate materials included (100)-MgO, polycrystalline Al₂O₃ (alumina), and amorphous SiO₂, which were cleaned using a solvent rinse sequence of acetone, isopropanol, and methanol followed by a 10-min ultraviolet/ozone treatment (Jetlight, Model 42A).

We prepared several sets of J30 thin films to probe the effects of kinetic bombardment, partial O₂ pressure, and epitaxy on the resulting phase and structure. Three sets of J30 films were grown on (100)-MgO substrates to a target thickness of 100 nm to provide enough sample volume for phase identification while avoiding potential structural relaxation as a function of thicknesses. The first set was grown in pure O₂ ranging in total pressure from 50 to 300 mTorr. The other two film sets on MgO employed a mixed atmosphere of O₂ and Ar. For the second set, partial O₂ pressure was constant at 50 mTorr and Ar pressure was added sequentially in 50 mTorr increments. For the third set, the O₂ partial pressure was held at 100 mTorr and Ar was again added in 50 mTorr increments. The second and third film sets are designed, to the best of our capability, to separate the potentially competing effects of particle mean free path and oxygen activity levels.

Additional films were grown on alumina and SiO₂ substrates to eliminate the possibility of epitaxial stabilization.^[Reference Gorbenko, Samoilenkov, Graboy and Kaul^24] A target thickness of 400 nm was chosen to increase diffracted signal intensity from x-ray diffraction (XRD). Thicker films were also annealed at 1000 °C for 1 h to investigate the thermal stability of as-deposited J30 films.

Sample characterization

XRD was performed to determine phase structure using a Panalytical Empyrean diffractometer with a Cu x-ray source, Bragg–Brentano HD incident optics, 1/8° divergence slits, and a PIXcel^3D area detector. Bragg–Brentano (symmetric, divergent beam θ–2θ) geometry maximizes the observed intensity of Bragg reflections from second phases. Rocking curves (ω scans at constant 2θ) and skew-symmetric diffraction (surface normal tilted out of the diffraction plane) employed parallel beam optics with a 2-bounce Ge monochromator incident optic and 0.18° parallel plate collimator with Xe detector. Asymmetric diffraction (surface normal tilted within the diffraction plane) for reciprocal space mapping employed the Ge monochromator and PIXcel^3D detector. Atomic force microscopy (AFM) was performed using an Asylum MFP3D instrument in tapping mode to view the surface morphology and determine root-mean-squared (RMS) roughness values. No compositional analysis was carried out.

PLD target phase structure

The diffraction pattern of the sintered target (Fig. S1) indicates two phases: RS solid solution (space group Fm-3 m) and bixbyite Sc₂O₃ (space group Ia-3). We estimate a lattice parameter of 4.25 Å for the cubic RS phase, based on pseudo-Voigt peak fitting. This closely matches  Sc-free J14 lattice parameters reported previously.^[Reference Rost, Sachet, Borman, Moballegh, Dickey, Hou, Jones, Curtarolo and Maria^1, Reference Bérardan, Franger, Dragoe, Meena and Dragoe^2, Reference Rost, Rak and Maria^25] The substantial presence of Sc₂O₃ in the XRD pattern indicates limited Sc incorporation into the RS phase for the thermal budgets applied and that the J30 transition to a single RS phase either takes place above 1000 °C, or requires chemical boundary conditions that facilitate the reduction and subsequent dissolution of Sc₂O₃.

Phase structure of films on MgO

The gas pressure and gas composition dependence of J30 thin film phase evolution on MgO substrates is presented graphically in Fig. 1: a phase map with O₂ fraction on the horizontal axis and total pressure (O₂ + Ar) on the vertical axis. Data points in the white phase field indicate films that are [001]-oriented, single-phase RS while data points in the gray phase field indicate samples containing both RS and Sc₂O₃ phases.

Figure 1.

Phase structure of J30 films as a function of total deposition pressure and O₂ gas fraction. Data points in the white phase field correspond to films exhibiting a single-phase rocksalt (RS) structure and points in the gray phase field correspond to films with both RS and Sc₂O₃ phases.

It was important to consider the formation of a spinel-type structure, rather than a RS structure, since Mg, Ni, Co, Cu, and Zn are all known to exist in $\hbox{A}^{2 +} \hbox{B}_2^{3 +} \hbox{O}_4^{2 -} $ spinels.^[Reference Bérardan, Franger, Dragoe, Meena and Dragoe²Reference Rak, Rost, Lim, Sarker, Toher, Curtarolo, Maria and Brenner^6] RS and spinel structures cannot be differentiated conclusively by <100> XRD scans alone, but the relatively large spinel unit cell produces a reflection from {111} planes at 2θ~18°, with reflections from spinel {222} planes appearing near reflections from RS {111} planes. For all films identified as RS, skew-symmetric θ–2θ scans, aligned to capture these reflections, show no features in the vicinity of 18°, providing evidence that no other structures or phases are present.

The transition from single-phase RS to RS + Sc₂O₃ is identified by the appearance of the bixbyite Sc₂O₃ 004 reflection in the diffraction pattern and a broadening of the RS peaks, resulting in an inability to resolve the Cu_kα doublet for film reflections. In general, at relatively low total pressure conditions, the film structure is (001)-oriented, single-phase RS, while at high pressures the films show reflections from both RS and Sc₂O₃ phases. In phase-separated films, both the RS and Sc₂O₃ phases are [001]-oriented, suggesting a RS[100]//Sc₂O₃[100] epitaxial relationship between the Sc₂O₃ grains and the RS grains. The cation–cation separation in Sc₂O₃ along <100> differs from J14 and MgO by >10%^[Reference Rak, Rost, Lim, Sarker, Toher, Curtarolo, Maria and Brenner^6, Reference Wyckoff^27] so structurally relaxed interfaces are expected. Changes in the RS phase out-of-plane lattice parameter were recorded (Fig. S2) but show no clear universal trend as a function of total pressure.

The relative Sc₂O₃ 004 peak intensity and RS 002 peak broadening tend to be greater at higher total pressures, regardless of gas composition. Higher O₂ percentages at a constant total pressure also appear to promote phase separation. A subset of the XRD patterns for films on MgO is presented in Fig. 2 to illustrate these two trends. Figure 2(a) shows the appearance of Sc₂O₃ as the Ar partial pressure is increased while maintaining a constant O₂ partial pressure of 100 mTorr. Figure 2(b) shows the second trend where increasing the oxygen fraction at constant total pressure (200 mTorr) results in phase separation. Long-range θ−2θ scans for all films on MgO can be found in supplemental Fig. S3. These data collectively illustrate relative roles of total pressure (thus particle kinetic energy^[Reference Amoruso, Aruta, Bruzzese, Maccariello, Maritato, Miletto Granozio, Orgiani, Scotti di Uccio and Wang^13–Reference Sambri, Cristensen, Trier, Chen, Amoruso, Pryds, Bruzzese and Wang^15]) and oxygen partial pressure (thus gas reactivity^[Reference Orsel, Groenen, Bastiaens, Koster, Rijnders and Boller^17]). Ultimately, both are instrumental in phase stabilization. Increasing kinetic energy by minimizing gas-phase scattering promotes solid solution, however, more kinetic energy is needed when oxygen activity is higher. We can understand this trend given the net-lower oxidation state of the cations, particularly Sc, when stabilized in the RS structure. This lower oxidation state is also predicted by DFT calculations.^[Reference Rak, Rost, Lim, Sarker, Toher, Curtarolo, Maria and Brenner^6]

Figure 2.

X-ray diffraction data for a subset of films on MgO: Bragg–Brentano scans of (a) films grown at constant O₂ partial pressure and varying total pressure; (b) films grown at a constant total pressure of 200 mTorr and varying partial O₂ pressure; and (c) reciprocal space map of J30 grown at 50 mTorr O₂. In (a) and (b), the Cu_kα doublet can be resolved in single-phase films and reflections from stray Cu_Kβ and W_L radiation are indicated with arrows. A Cu_Kα1 monochromator was used for (c).

Structural characterization of epitaxial J30

High-resolution XRD was carried out on J30 grown at optimized conditions (50 mTorr O₂) to assess crystal quality and epitaxial strain. Measurements of the J30 002, 022 and 111 reflections indicate heteroepitaxy and a tetragonally distorted unit cell with a = b = 4.20 Å and c = 4.38 Å. The alignment of the J30 and MgO 111 reflections in φ (Fig. S4) indicate an in-plane crystallographic relationship J30[100]//MgO[100]. The full-width-half-maximum (FWHM) of the rocking curves for the J30 and MgO 002 reflections are 0.015° and 0.0058°, respectively (Fig. S5). The narrow FWHM of single-phase J30 indicates a high degree of crystallinity and low mosaic spread. A reciprocal space map of the MgO and J30 204 reflections [Fig. 2(c)] illustrates the commensurate nature of the film. The peak location of the J30 204 is consistent with the lattice constants stated above. Two x-ray artifacts indicative of a high-quality film are observed in the out-of-plane diffraction pattern (Fig. S6): Pendellosung fringes^[Reference Kato and Lang^28] and a pronounced “hybrid” x-ray reflection.^[Reference Smith, King, Soukiassian, Ast and Schlom^29] Additionally, single-phase J30 is smooth with an RMS roughness <0.5 nm, while increasing O₂ pressure to 300 mTorr progressively increases the RMS roughness to 3 nm and produces a fine-grained microstructure upon phase separation (Fig. 3). The same roughness trend is observed in films grown in mixed O₂/Ar.

Figure 3.

AFM images of J30 grown on MgO in: (a) 50 mTorr O₂; (b) 150 mTorr O₂; (c) 200 mTorr O₂; (d) 300 mTorr O₂; and (e)–(h) respective line traces. Note the variable color scales in (a)–(d) and constant height scale for (e)–(h). Coordinates of line traces are denoted by dashed white lines in (a)–(d). The scale bar in (a) applies to (b)–(d).

To understand the origin of the observed tetragonal distortion, we first consider the predicted bulk lattice constant of cubic J30: 4.34 Å.^[Reference Rak, Rost, Lim, Sarker, Toher, Curtarolo, Maria and Brenner^6] If such a RS phase were forced commensurate to MgO, an out-of-plane elongation is expected. Assuming a Poisson's ratio for J30 of 0.18 (the literature value for MgO^[Reference Zha, Mao and Hemley^30]) and biaxial in-plane compression, one predicts an out-of-plane value of 4.4 Å, which is similar to the measured quantity of 4.38 Å. As such, it is reasonable to attribute J30 tetragonality to epitaxial strain. It is interesting to note that these films can withstand a tremendous amount of in-plane strain without relaxation, in this case, nearly 3% biaxial compression. Current studies are underway to identify if this capacity for strain is related to the high-entropy condition.

Thin films on alumina and SiO₂

J30 films were prepared on polycrystalline alumina and amorphous SiO₂ substrates to test for the possibility of epitaxial stabilization, as opposed to kinetic stabilization. Epitaxial stabilization refers to the growth of relatively unstable phases only when epitaxy is induced.^[Reference Gorbenko, Samoilenkov, Graboy and Kaul^24] XRD of films prepared at 50 mTorr O₂ on alumina and SiO₂ reveal a single, broad film reflection consistent with a RS 002 (Fig. S7). The peak position corresponds to ~4.4 Å and no intensity from the 004 or any off-axis reflections could be observed. We suspect the 004 reflection may be too broad to measure clearly due to a combination of small grains [Fig. 4(b)] and local lattice distortion from the Jahn–Teller effect,^[Reference Rak, Rost, Lim, Sarker, Toher, Curtarolo, Maria and Brenner^6, Reference Rost, Rak and Maria^25] which can broaden the 200 reflection from bulk ESOs.^[Reference Berardan, Meena, Franger, Herrero and Dragoe^4] There is no discernible peak splitting or reflections attributable to Sc₂O₃, suggesting the films are single phase.

Figure 4.

Collective data set for J30 films grown on alumina: (a) offset XRD plots and corresponding AFM images for (b) 50 mTorr O₂; (c) 50 mTorr O₂ + 200 mTorr Ar; (d) 50 mTorr O₂ post-annealing; and (e) bare Al₂O₃ substrate. Deposition atmospheres are labeled in (a). Data for a bare substrate is included in (a) and (e) for reference. Arrows in (d) denote scandium-rich grains. The scale bar in (e) applies to (b)–(d).

Figure 4(a) shows XRD data for films grown on alumina in 50 mTorr O₂ and 50 mTorr O₂ + 200 mTorr Ar—a condition that produced phase separation in the epitaxial case. The results on ceramic alumina are similar; additional gas-phase scattering results in the evolution of Sc₂O₃ and a RS component that shifts to a smaller interplanar spacing of 4.3 Å. A similar trend of increasing surface roughness at higher pressures also occurs on alumina substrates [Figs. 4(b) and 4(c)]. Similar trends in the case of epitaxial and textured growth indicate that PLD energetics are the more influential drivers for single-phase growth as compared to epitaxy.

Thermal stability of thin films

To test J30 metastability, 50 mTorr J30 films on ceramic alumina, (100)-MgO, and SiO₂ were annealed at 1000 °C for 1 h to equilibrate the structure in the same temperature and atmosphere the target was sintered. On alumina, after annealing the RS reflection sharpened substantially and shifted to a higher 2θ value corresponding to a lattice parameter of 4.2 Å and a strong reflection attributed to the Sc₂O₃ 222 emerged [Fig. 4(a)]. A bare substrate scan is included in Fig. 4(a) for reference, because extra reflections (attributed to impurity phases in the Al₂O₃ substrate) appear in the XRD pattern. After annealing, the polycrystalline J30 film developed grains with sharp, cubic facets and some dispersed nebulous grains [Fig. 4(b)]. SEM with EDS indicates the poorly defined grains are Sc-rich and likely Sc₂O₃ grains detected by XRD.

Annealing J30 on MgO also produced phase separation to RS and Sc₂O₃ [Fig. S8], but the results of annealing J30 on SiO₂ were inconclusive. Silica and olivine silicate reflections dominated the XRD pattern, indicating significant substrate crystallization and reaction with the film at the annealing temperature. In the absence of substrate reaction, the annealed phase structure of J30 appears the same as the phase structure of the sintered PLD target, implying metastable as-deposited films which phase separate to the equilibrium assembly given 1000 °C of thermal energy.