Fabrication of atomically abrupt interfaces of single-phase TiH₂ and Al₂O₃

Metal hydrides have been extensively studied as hydrogen storage materials for practical applications such as hydrogen fuel systems and fuel cells.^1,2 In addition, the recent reports in hydrides on ionic conduction of metal (Li⁺, Na⁺, Mg²⁺) cations³ and hydride (H⁻) ion^4,5 have paved the way for electrochemical applications. Furthermore, high transition-temperature superconductivity in H₃S⁶ and PdH_x⁷ has attracted enormous attention in the field of condensed-matter physics. Accordingly, metal hydrides are promising materials in a wide variety of fields ranging from energy devices to electronics.⁸ 

For the applications of metal hydrides in electronics,⁹ atomically abrupt interfaces between the substrate and film and/or between the films in multilayer-structured devices are essential. Since exotic physical properties such as high-mobility two-dimensional electron gases, magnetism, and superconductivity appear at the heterointerfaces of oxides,¹⁰ “heterointerfaces of hydrides” have the great potential to emerge properties using strain, superlattice, and field-effect. For the fabrication of metal-hydride epitaxial thin films and abrupt interfaces in a well-controlled fashion, the direct growth of single-phase epitaxial thin films is required. Until now, there have been only a few reports on the fabrication of metal-hydride epitaxial thin films.^9,11–13 However, epitaxial thin films of LiH¹¹ and TiH₂¹² directly grown using pulsed laser deposition contained impurity phases of Li₂O and metallic Ti, respectively. Accordingly, it is of significant importance to introduce another deposition technique for obtaining single-phase metal hydride epitaxial thin films. For this purpose, reactive magnetron sputtering is a promising candidate. The growth of polycrystalline ZrH₂ thin films using reactive magnetron sputtering has been reported,¹⁴ implying the potential of the sputtering technique to obtain single-phase metal hydride epitaxial thin films and atomically controlled interfaces.

Here, we demonstrate the fabrication of the atomically abrupt interface of single-phase epitaxial titanium dihydride (δ-TiH₂, fluorite-type, F_m−3m, a = 0.446 nm¹⁵) thin films and α-Al₂O₃ (001) substrates. We adopted the magnetron sputtering technique because we expected the assistance of hydrogenation for the thin films from reactive hydrogen in plasma, enabling the direct growth of metal hydride thin films without impurity phases. The (111)-oriented δ-TiH₂ epitaxial films were successfully fabricated at a substrate temperature (T_s) of 150°C and a hydrogen partial pressure (⁠$pH2$⁠) of 0.05 Pa. Scanning transmission electron microscopy measurements revealed that the δ-TiH₂ (111)/Al₂O₃ (001) interface is atomically abrupt. Furthermore, the temperature dependence of resistivity shows metallic behavior with no upturn over the temperature range of 4–300 K, showing that impurity phases are not present in the films. Our study shows that reactive magnetron sputtering has an advantage in fabricating atomically controlled interfaces of single-phase hydride epitaxial thin films and substrates.

Thin films of δ-TiH₂ were deposited on α-Al₂O₃ (001) (Shinkosha Corp.) substrates using reactive magnetron sputtering. A Ti metal plate (purity: 99.99%, 2 in. diameter, Toshima Manufacturing Co., Ltd.) was used as a sputtering target, and a mixture of pure Ar and H₂ gases was introduced into a vacuum chamber (background pressure: ∼5 × 10⁻⁵ Pa) for the sputtering. During the depositions, the Ar partial pressure was fixed at 1.0 Pa with a constant flow rate of 10 SCCM in the growth chamber, and $pH2$ was varied with different flow rates. The dc power supply at the Ti target was maintained at 50 W. The T_s values were varied from room temperature (RT) to 300°C. The typical film thickness was 90 nm with a growth time of 30 min. The thin-film deposition rate was 180 nm/h. As the H₂ partial pressure in the sputtering process was increased, the deposition rate decreased: 200 nm/h and 180 nm/h in Ar and Ar/H₂ (5%) atmosphere, respectively, suggesting that the Ti target surface was hydrogenated during the deposition. The structural properties of the thin films were characterized using x-ray diffraction (XRD, Bruker D8 DISCOVER). Annular dark field-scanning transmission electron microscopy (ADF-STEM) observations and selected area diffraction pattern (SADP) analyses were conducted with JEM-ARM200F (JEOL Co., Ltd.) operated at 200 kV. The thin foil for STEM observations was prepared with the SMF2000 (Hitachi High-Tech Science Co.) focused ion beam system. Electron transport measurements were carried out in a typical Hall-bar geometry using a Gifford-McMahon refrigerator with superconducting magnets (Niki Glass Co., Ltd., LTS205D-CM9T-TL-VT).

Figure 1(a) shows the obtained XRD results of the T_s dependence (⁠$pH2$ = 0.15 Pa) of the thin films deposited on Al₂O₃ (001) substrates. The out-of-plane XRD scan of the thin film deposited at T_s = 300°C shows a 002 peak of the metallic α-Ti (hexagonal, 2θ = 38.422°¹⁶). As T_s was decreased to 200°C, the α-Ti 002 peak diminished, and a peak appeared at 2θ ∼ 35.0°, corresponding to δ-TiH₂ 111.¹⁵ We here stress that the α-Ti 002 peak completely disappeared in the films deposited at T_s $≦$ 200°C (even at RT). Thus, Ti metal was fully hydrogenated in films deposited in a reactive magnetron sputtering process at low T_s, indicating the positive effect on the hydrogenation of films.

We next examined the $pH2$ dependence (T_s = 150°C) of the δ-TiH₂ thin films. Figure 1(b) shows the out-of-plane XRD patterns of the films deposited at $pH2$ = 0 (no additional H₂ gas), 0.01, 0.02, 0.03, 0.05, and 0.15 Pa. Under $pH2$ = 0 Pa, only the α-Ti 002 peak is clearly visible. As the $pH2$ is increased, the intensity of the α-Ti 002 peak became smaller, and a δ-TiH₂ 111 peak emerged. Further increase in the $pH2$ value to $∼$ 0.05 Pa led to the deposition of single-phase δ-TiH₂ (111) thin films. Note that the intensity of the δ-TiH₂ 111 peak in $pH2$ = 0.15 Pa was smaller than those in $pH2$ = 0.03 and 0.05 Pa. This is because polycrystalline films were obtained in extremely high $pH2$ (not shown).

To study the in-plane crystal orientation of the δ-TiH₂ films, pole figure measurements were performed for δ-TiH₂ 200 (2θ = 40.41°, χ = 54.7°) and Al₂O₃ 113 (2θ = 43.36°, χ = 61.2°). The pole figure measurements indicate that the peaks of TiH₂ 200 and Al₂O₃ 113 appear at the same φ-angles, [Figs. 1(c) and 1(d)], demonstrating the epitaxial growth of TiH₂ thin films. Note that the six diffraction spots for δ-TiH₂ 200 indicate existence of two rotational domains. The obtained δ-TiH₂ films were relaxed, possibly due to the large in-plane lattice mismatch of 14.2%: δ-TiH₂ (0.314 nm for Ti–Ti) to Al₂O₃ (0.275 nm for O–O). The in-plane epitaxial relationship of $[112¯]TiH2$ // $[110]Al2O3$⁠, which is consistent with a previous report using pulsed laser deposition.¹² 

The extent of hydrogenation and the crystallinity in the deposited films as a function of T_s and $pH2$ are investigated. To evaluate the extent of hydrogenation, we here introduce the δ-TiH₂ fraction as

where $STiH2_111$ and S_{Ti_002} are the areas of the XRD peak at δ-TiH₂ 111 and α-Ti 002, respectively. Figure 2(a) is a color gradation plot of the δ-TiH₂ fraction, as a function of T_s (horizontal) and $pH2$ (vertical). Single-phase δ-TiH₂ films (δ-TiH₂ fraction = 1) are shown in the red regions, while a metallic α-Ti appears in the blue regions, and accordingly, lower T_s and higher $pH2$ lead to a δ-TiH₂ phase. In Fig. 2(b), we also show a color map of a rocking curve full-width at half-maximum (FWHM) obtained at δ-TiH₂ 111 (2θ = 34.8°). Red regions indicate the δ-TiH₂ (111) epitaxial films with high crystallinity. Interestingly, higher $pH2$ led to the formation of polycrystalline films, indicating that extremely high $pH2$ hinders the fabrication of high-quality δ-TiH₂ epitaxial films [Fig. 1(b)]. According to these two plots, the best conditions for the δ-TiH₂ (111) epitaxial growth are determined as T_s = 150°C and $pH2$ = 0.05 Pa. The smallest rocking curve FWHM value obtained is 0.67°, which is smaller than the previously reported value (⁠$∼$1.4°).¹² 

We here demonstrate an atomically abrupt interface of δ-TiH₂ (111) epitaxial thin films and Al₂O₃ (001) substrates using cross-sectional ADF-STEM imaging. In a wide-view ADF-STEM image [Fig. 3(a)], the δ-TiH₂ epitaxial film shows the uniform contrast. We could observe an additional TiO_x layer on the top of δ-TiH₂, possibly formed during the deposition of an Os protective layer. The layer was deposited using plasma-assisted chemical vapor deposition with OsO_x precursor, prior to the specimen preparation for STEM measurements. The electron diffraction patterns obtained in the film region show sharp spots reflecting (111)-oriented fluorite structures with two rotational domains [Figs. 3(b) and 3(c)], consistent with the XRD results. In a close-up view near the TiH₂/Al₂O₃ interface [Fig. 3(d)], we clearly visualize the ordering of the Ti atoms in the film regions. In the out-of-plane direction, the ordering is (111)-oriented fcc stacking (ABCABC…) corresponding to Ti sites formed in the fluorite structures of TiH₂, distinct from the metallic phases of α-Ti (hcp) and $β$-Ti (bcc). Consequently, an atomically abrupt interface can be achieved even in a hydride film, illustrating the great potential of magnetron sputtering for the development of hydride electronics in future.

Finally, we performed electron transport measurements using a Hall-bar geometry on δ-TiH₂ (111) epitaxial films grown at T_s = 150°C and $pH2$ = 0.05 Pa. The temperature dependence of the longitudinal resistivity [ρ_xx(T)] shows metallic behavior over the entire temperature range of 4–300 K [Fig. 4(a)]. The residual resistance ratio [ρ_xx(300 K)/ρ_xx(4 K)] is 2.4. Note that the upturn at around 15 K reported in the previous study¹² did not appear, suggesting that an impurity phase is not present in our films. Figure 4(b) shows the magnetic field dependence of the transverse resistance (R_xy) obtained at 4 K. The slope in a linear response gives a Hall coefficient of −9.5 × 10⁻⁵ (cm³/C), and thus, the carrier concentration and mobility at 4 K are obtained as 6.6 × 10²² cm⁻³ and 2.3 cm² V⁻¹ s⁻¹, respectively. Given that the valence of Ti is 2+ (d² system), the obtained carrier concentration of 6.6 × 10²² cm⁻³ is consistent with the order of the Ti atom density (4.5 × 10²² cm⁻³).

In conclusion, we investigated the epitaxial growth of single-phase δ-TiH₂ (111) thin films on Al₂O₃ (001) substrates using reactive magnetron sputtering. We studied the growth conditions of δ-TiH₂ epitaxial films as a function of T_s and $pH2$⁠. As a result, we successfully fabricated δ-TiH₂ epitaxial films at T_s = 150°C and $pH2$ = 0.05 Pa, whereas high $pH2$ and low T_s resulted in polycrystalline films. Atomic-resolution STEM measurements clearly visualized fcc-like Ti stacking based on fluorite structures of TiH₂ and an atomically abrupt interface of TiH₂ and Al₂O₃. Furthermore, transport measurements indicate metallic behavior even at low temperatures, consistent with the result that no impurity phase is observed in the XRD and STEM. Our present study indicates that reactive magnetron sputtering is advantageous to fabricate high-quality metal hydride epitaxial films and abrupt interfaces for future hydride electronics.

This research was supported by the World Premier International Research Center Initiative (WPI Initiative) of the Ministry of Education, Culture, Sports, Science and Technology of Japan, JST-CREST (JPMJCR1523) and JST-ALCA. R.S. acknowledges Kakenhi (No. 17H05216). T.H. acknowledges Kakenhi (Nos. 26246022 and 16K14088). H.O. acknowledges Kakenhi (Nos. 15H05546 and 15K14155) and the TEPCO Memorial Foundation. S.O. acknowledges Kakenhi (No. 25220911). This work was performed under the Inter-university Cooperative Research Program of the Institute for Materials Research, Tohoku University (Proposal Nos. 16K0025 and 17K0060). STEM observations were carried out at the National Institute for Materials Science (NIMS) Battery Research Platform. We thank Professor Tsutomu Nojima for his help in designing our custom-made GM fridge instruments. We also thank Professor Akira Ohtomo, Dr. Kohei Yoshimatsu, and Mr. Takuto Soma for their support of electron transport measurements.