Characterization of epitaxial titanium nitride mediated single-crystal nickel oxide grown on MgO-(100) and Si-(100)

Nickel oxide (NiO) has been used in the optoelectronic applications of group-III nitride semiconductors to manufacture devices with high wall-plug efficiency in recent decades because it has the properties of p-type semiconductors with a wide bandgap. In the context of the physical properties of NiO, Newman and Chrenko researched its absorption using the halide decomposition method, where the results showed a rapidly increasing absorption coefficient from 3.5 eV to 4 eV.¹ This revealed that the energy bandgap of NiO is 3.5 eV, which is suitable for visible-light applications. Sato et al. acquired a thin film of p-NiO by using radio-frequency (RF) magnetron sputtering and measured a hole concentration of 1.3 × 10¹⁹ cm⁻³and transmittance of 80% for wavelengths above 350 nm.² Albert and Lee found that NiO is an antiferromagnetic material.³ Mironova-Ulmane et al. synthesized NiO nanoparticles and observed that their particle size is responsible for the phase conversion from antiferromagnetic to paramagnetic. The sizes of the nanoparticles have also been found to affect the spectrum of Raman scattering.⁴ 

The novel applications of NiO to electrochemistry,⁵ flexible supercapacitors,⁶ and in the case of NiO-based compounds, superconductivity continue to attract the interest of the scientific community.⁷ The existence of a NiO layer is beneficial for photoelectric devices, including those composed of organic and inorganic materials. Widjonarko et al. used a thin, sputtered film of NiO as a hole transport layer (HTL) to improve the performance of bulk-heterojunction organic solar cells, and examined the stoichiometry of the film as well as its influence on the work function of the HTL.⁸ Chan et al. also used NiO as an HTL to improve the performance of a tris(8-hydroxyquinoline) aluminum(III) (Alq₃)-based organic light-emitting diode (OLED),⁹ whereas Liu et al. combined NiO and poly(9-vinylcarbazole) (PVK) to form a new hole transport layer that increases the quantum yield of photoluminescence.¹⁰ For inorganic InGaN blue LEDs, Horng et al. proved that hole-rich NiO works as a carrier bridge to connect p-GaN and ITO.¹¹ In addition to the use as HTL and bridge layers, the NiO thin film is expected to find new integration effort in group-III oxide and heterogeneous group-III oxide/nitride semiconductors. Li et al. combined lithium-doped NiO with n-Ga₂O₃ to generate a new p–n junction for the application of a deep ultra-violet photodetector using type-I energy band alignment, where the device showed a responsivity of 415 mA/W at −7 V under a 260-nm incident light.¹² 

A thin film of NiO is typically deposited using a sputtering tool at a low temperature in industrial applications to form polycrystalline NiO. The quality of NiO influences its semiconductor properties and device performance. If polycrystalline NiO, which exhibits inferior electrical transport,¹³ is used in optoelectronic devices, the resulting leakage current through grain boundaries will affect the performance of the resulting device where NiO is used as a p-layer of a p–n or p–i–n junction. A single crystal NiO is therefore desired. However, epitaxial-quality NiO requires considerably more refined process conditions, and single-crystal NiO requires preparing a preorienting insertion layer. Budde et al. studied growing NiO under different conditions through plasma-assisted molecular beam epitaxy (PA-MBE) on an MgO substrate,¹⁴ both materials of which are similar in terms of the crystal structure and the lattice constant. Yamauchi et al. grew NiO on sapphire (0001) substrates because they are commonly used in the industry. Owing to structural mismatch, a stepped sapphire was used to characterize epitaxy by high-resolution transmission electron microscopy (HRTEM).¹⁵ However, MgO and sapphire are both insulators, and a highly conductive seed layer is desirable for the subsequent growth. Like MgO, TiN has a cubic lattice structure with a similar lattice constant to that of NiO.¹⁶ It is also a degenerated n-type material that can be grown as a contact layer for device implementation.¹⁷ Hence, it is suitable as a platform growing NiO.

While single crystalline TiN can be obtained by using MBE¹⁸ and magnetron sputtering,¹⁹ little research explored its use as a seed or preorienting layer for the subsequent growth. Therefore, the authors here develop epitaxial NiO of the single-crystalline quality using an epitaxial-TiN interlayer on a lattice-matched MgO (100) substrate as well as lattice-mismatched silicon (100) substrate for the large-scale preparation of a NiO binary compound. The insertion of TiN is motivated by the need to form highly conductive bottom contact for the implementation of devices developed in the future, and to mitigate the effect of the lattice mismatch resulting from growing NiO directly on Si (100). Systematic structural, morphological, and optical characterizations of the growth of single-crystalline epitaxial NiO are detailed in this investigation. The template developed here can assist in subsequent research on the efficiency of the injection or extraction of highly charged carriers in group-III oxide and group-III nitride semiconductor devices, and for the development of such electrochemical applications as solar hydrogen generation, p–n or p–i–n junction photodetectors and nitrogen reduction reactions.

Substrates of magnesium oxide (MgO) and Si with purity higher than 99.95% were arranged in the (100) orientation. Titanium nitride (TiN) was first deposited using radio-frequency (RF) magnetron co-sputtering of two Ti targets at a power of 180 W each, in a reactive atmosphere of 5 mbar consisting of 92.5% Ar and 7.5% N₂. The TiN thin films were deposited at a substrate temperature of 800 °C and −120 V substrate bias for 120 min. Before deposition, the Si and MgO substrates were annealed for 30 min at 800 °C in an Ar RF plasma of 50 W. The TiN/MgO and TiN/Si samples were then transported under ambient atmospheric condition for further processing. Before the deposition of NiO, annealing was required for a single-crystal outcome. This was carried out at 450 °C with 50 mTorr of oxygen for 1 h in a pulsed laser deposition (PLD) chamber. NiO was then grown on top of the annealed TiN layer at a temperature of 550 °C using a 254-nm laser operating at a pulse frequency of 5 Hz and an energy of 300 mJ. Oxygen as an ambient gas was supplied at 50 mTorr during NiO deposition. A total of 30 000 laser pulses were used at a deposition rate of NiO of 85 mÅ/pulse, which yielded a thickness of 255 nm.

The surface morphology was examined using the SPM tool Dimension Icon in tapping mode using a RTESPA-300 AFM tip composed of antimony (n)-doped Si at 0.01–0.025 Ω cm. The scan area was 5 × 5 μm², and the measurements were conducted at room temperature. The crystallinities of the NiO/TiN/MgO and NiO/TiN/Si structures were examined using a Bruker D8 Advance X-Ray Diffractometer with Cu Kα (λ = 1.54 Å) radiation, while the φ-scan and rocking curve were measured using a Bruker D8 Discover X-Ray Diffractometer with Cu Kα (λ = 1.54 Å) radiation.

For the section requiring high-resolution transmission electron microscopy (HRTEM), the bright-field cross-sectional images and the information on atom stacking were collected by an FEI Titan at an operating voltage of 300 kV. The TEM specimens were prepared in an FEI Helios SEM using a focus ion beam (FIB) and an Omniprobe nanomanipulator, while the elemental distributions of the as-grown NiO/TiN/MgO and NiO/TiN/Si samples were checked by scanning transmission electron microscopy in conjunction with an energy-dispersed x-ray spectroscopy (STEM-EDX) in the FEI Titan at an acceleration voltage of 300 kV.

The out-of-plane orientations of as-grown NiO/TiN/MgO were analyzed by x-ray diffraction (XRD) 2θ-scan, as shown in Fig. 1(a). Because NiO, TiN, and MgO have the same crystal structure and similar lattice constants, their corresponding (200) and (400) peaks were relatively close. To distinguish among them, we used zoom-in XRD plots of the NiO/TiN/MgO samples, as shown in Figs. 1(b) and 1(c). Figure 1(b) shows that the (200) peaks of NiO, MgO, and TiN occurred at 43.60°, 43.12°, and 42.57°, respectively, in the reverse order of the first appearance. The higher Miller index groups of the (400) peaks are presented in Fig. 1(c), and appear separated due to the Bragg diffraction condition of 2d sin θ = nλ, where d is the atomic spacing in a certain orientation, θ is the incident angle of the x ray with respect to the surface of the planes of the crystal, n is the order of diffraction, and λ is the wavelength of the incident x ray. The peaks for NiO, MgO, and TiN occurred at 95.56°, 94.1°, and 92.66°, respectively. The results indicate the presence of the out-of-plane tensile strain in the NiO layer, while the TiN layer underwent compressive strain with respect to the MgO substrate. Based on these XRD results, the out-of-plane orientation relationship was NiO (200) ∥ TiN (200) ∥ MgO (200). The lattice constants were subsequently accurately determined using atomic line profiles derived from the TEM characterization.

The measured rocking curves (RC) of NiO and TiN are shown in Fig. 2. Owing to the small differences in the lattice constants of NiO, TiN, and MgO, the values of the full-width at half-maximum (FWHM) were small—0.30° for NiO and 0.21° for TiN, as shown in Figs. 2(a) and 2(b), respectively. For the in-plane RC scanning, the FWHMs of NiO (220) and TiN (220) were 0.40° and 0.30°, as shown in Figs. 2(c) and 2(d), respectively. Moreover, the sharp peaks located beside the peaks of the in-plane (220) rocking curves of NiO and TiN originated from the interplanar diffraction of MgO. Peaks of MgO, marked by black arrows, occurred owing to the small differences in lattice constants.

In the measurements of the φ-scan for the (220) in-plane orientation, i.e., the in-plane azimuthal rotation scan, the four peaks were related to the fourfold symmetry of the cubic crystal structure, as shown in Fig. 3. Moreover, the peaks of NiO, TiN, and MgO were aligned, which confirmed that the as-grown lattices were cube-on-cube stacked and the in-plane orientation relationship was NiO (020) ∥ TiN (020) ∥ MgO (020).

To prepare a large wafer-scale template of the NiO substrate, we used a silicon substrate for the potential uptake at an industrial scale. Because the substrate used was Si with a (100) orientation, a large lattice mismatch was expected from the overgrown, cubic NiO and TiN. This was evident from the out-of-plane orientations, as measured using XRD, showing peak positions at 43.55° for NiO (200) and 42.66° for TiN (200) in Fig. 4.

Rocking curves of the out-of-plane orientation obtained using XRD, shown in Fig. 5, reveal that the as-grown NiO layer was grown with a large tensile strain on the TiN/Si template substrate. This resulted in degradation in the quality of the crystal and, therefore, a larger FWHMs, 1.02° for the NiO layer and 1.01° for the TiN layer, as shown in Figs. 5(a) and 5(b), respectively. For the in-plane rocking curves obtained using XRD, the FWHMs were substantially larger: 1.42° for NiO and 1.35° for TiN, as shown in Figs. 5(c) and 5(d), respectively.

Even though there was a large tensile strain in the structure, the fourfold peaks of NiO, TiN, and Si in the (220) phi (φ)-scan show that the atoms were stacked in a cube-on-cube orientation without the rotation of the crystal, and the in-plane orientation relationship was NiO(020) ∥ TiN(020) ∥ Si(020), as shown in Fig. 6.

The surface analysis was conducted using an atomic force microscope. Owing to the similar lattice constants and the identical crystal structure (rock salt cubic) of TiN and MgO, the atomic force microscopy (AFM) scan on the surface of the TiN/MgO sample exhibited a relatively flat surface with square patterns, as shown in Fig. 7(a), inherited from the bare surface of MgO. The rms of roughness was as small as 0.423 nm. For the TiN/Si sample, a needle-like morphology was observed on the surface. This might have occurred due to the large lattice mismatch between the Si substrate and the TiN layer, which resulted in a rougher surface than that of the TiN/MgO sample, as shown in Fig. 7(b). The rms of roughness was higher, 0.981 nm.

The surface morphology changed when NiO was deposited on top of the TiN layer. Instead of a flat morphology, NiO deposition led to columns gathering on the surface of TiN/MgO, as shown in Fig. 8(a). These columns made the surface rougher than that of TiN/MgO with an rms of roughness of 22.2 nm. For NiO deposited on TiN/Si, no column formation was observed on the surface such that its morphology was more uniform and flatter than that of NiO/TiN/MgO, as shown in Fig. 8(b). It had an rms roughness of 18.6 nm.

Figure 9 shows the cross section of the HRTEM of NiO/TiN/MgO and NiO/TiN/Si. It shows that the NiO/TiN/MgO samples were grown layer by layer on the MgO substrate. The thick layer on the top of MgO was the TiN layer, 200 nm, while the NiO layer was 50 nm. As the atomic numbers of Ni (28) and Ti (22) are higher than that of Si (14), the Si layer was lighter in color, and thus the NiO/TiN and TiN/Si interfaces were distinguishable. To determine the crystallinity of the layers, we examined the interfaces between pairs of layers by using the fast Fourier transform (FFT), inverse FFT, and line profiles, as shown by the red and yellow boxes in Figs. 9(a) and 9(b).

Because NiO, TiN, and MgO have the same crystal structure and similar lattice constants, the zoom-in images of the interface among NiO, TiN, and MgO of NiO/TiN/MgO showed an aligned atomic arrangement, which indicates coherently strained growth. The FFT images revealed distinct spots at different orientations, which show that the as-deposited layers were single crystalline, as shown in Figs. 10(a) and 10(b). Further simulations were performed on VESTA using the HRTEM images of NiO, TiN, and MgO, and the schematic diagrams of their atomic alignments were obtained to clarify their structures in the (020) orientation, as shown in Fig. 10(c). These diagrams show that the nitrogen and oxygen atoms overlapped with the metal atoms in this orientation. Therefore, these atoms were undetectable, and only the metal atoms were observable. The diagrams also show that the structures of NiO, TiN, and MgO were cube-on-cube stacked, which fit the results of the φ-scan data shown in Fig. 3. The lattice constants and mismatch between different layers were then calculated from inverse FFT and (020) atomic line profiles of NiO, TiN, and MgO, as shown in Fig. 11.

The (200) orientation was selected in FFT images of NiO, TiN, and MgO to determine the atomic distances because the surface of the cross section was in the (200) phase, as shown in Fig. 11. Because it was a classic rock salt cubic structure, the lattice parameter was given by

where d is the interplanar spacing, a is the lattice constant, and (h, k, l) are the Miller indices representing the orientation of the crystal. Based on Eq. (1), the out-of-plane lattice constants were calculated to be 4.192 Å for NiO, 4.228 Å for TiN, and 4.214 Å for MgO in the (020) orientation. The lattice mismatch was calculated by

Based on Eq. (2), a 0.85% mismatch was found at the interface of NiO and TiN, and −0.33% at the interface of TiN and MgO. The small mismatch values indicate similar lattice constants, which is correlated with the XRD results. Because MgO has a smaller lattice constant than TiN, the lattice mismatch between them has a negative value.

To further confirm the deposited materials, we performed high-angle annular dark-field based on scanning transmission electron microscopy (STEM-HAADF) and STEM-EDX measurements on the TEM image of Fig. 12(a). In the STEM-EDX mapping, the blue, green, and red regions reveal the elements of Mg, Ti, and Ni accordingly, which represent the MgO substrate, TiN layer, and NiO deposition, respectively [see Fig. 12(b)]. In the STEM-EDX spectrum, Mg K_α, Ti K_α1 and K_β1, and Ni K_α1 and K_β1 were detected, which confirms Mg, Ti, and Ni as constituents, as shown in Fig. 12(c).

In Figs. 13(a) and 13(b), the atomic stacking of TiN and NiO on Si (001) was starkly different from that on MgO (001) in Fig. 10. The corresponding effect was revealed by the HRTEM images and the FFT images, as shown in Fig. 13. The atoms of the NiO and TiN layers in the zoom-in images in Figs. 13(a) and 13(b) are not seen as sharp as those of NiO/TiN layers grown on the MgO substrate [Figs. 10(a) and 10(b)], indicating the reduced crystalline quality of the NiO/TiN structure grown on Si as compared to that grown on MgO. This qualitative result correlated well with the conclusions derived from the analysis of FWHM values of the corresponding XRD rocking curves [Figs. 2(c) vs 5(c) and Figs. 2(d) vs 5(d)]. Same conclusion results from the FFT images; the FFT spots in Fig. 13 are dimmer and more diffused in comparison with those in Fig. 10. Because the Si substrate had a diamond crystal structure while TiN and NiO had a rock salt crystal structure, they showed different atomic arrangements in the (022) plane. For the TiN and NiO layers, the Ti and Ni atoms were interleaved with a nitrogen atom and an oxygen atom, respectively. In the Si substrate, the atoms were paired and had a hexagonal interval arrangement, as shown in Fig. 13(c). Note that owing to their smaller atomic weights, atoms of O and N were only barely detected by TEM.

We further calculated the atomic distances of the three layers and the lattice mismatch at their interfaces. Because the cross section was in the (022) orientation, we performed (022)-focused FFTs and examined the profiles of the atomic lines, as shown in Fig. 14. Using Eq. (1), the lattice constants under the (022) orientation were calculated as 5.952 Å for NiO, 5.992 Å for TiN, and 7.716 Å for Si. The lattice mismatches calculated using Eq. (2) were 0.67% for the NiO/TiN interface and 22.34% for the TiN/Si interface.

The STEM-HAADF image and the STEM-EDX mapping in Figs. 15(a) and 15(b) show the elemental distribution of the structure. The red, blue, and green areas show the corresponding elements of Si, Ti, and Ni, which represent the Si substrate, the TiN layer, and NiO deposition, respectively [see Fig. 15(b)]. The STEM-EDX spectrum further analyzed the signals of Ni K_α1, Si K_α, and Ti K_α1, as shown in Fig. 15(c). By combining the results of STEM-EDX mapping, the STEM-EDX spectrum, and STEM-HAADF, we completed the elemental characterization of the NiO/TiN/Si structure.

Finally, the (perpendicular) lattice strain (σ), which is the lattice mismatch, is redefined in Eq. (3) to yield the perpendicular lattice strain of NiO on TiN/MgO of 0.0085, and that of TiN on MgO of −0.0033. The lattice misfit (f) is further defined in Eq. (4) with reference to bulk NiO (a_{bulk_NiO}) to obtain the Poisson ratio (ν) for NiO on the TiN/MgO template substrate. The lattice constant of TiN (a_TiN) is expected to vary according to the strain of substrates (MgO or Si), and hence we adopted the approach of using the measured a_TiN in obtaining ν. The critical thickness (h_c) of NiO on TiN/MgO is then calculated using Eq. (6),²² 

where b is the magnitude of Burgers vector for NiO (b = $12$a√2, a is lattice constant), λ is 45° (rock salt cubic system), β is the angle between the Burgers vector and the dislocation line (β = 90°). For NiO on TiN/MgO, we obtain b = 2.964 Å, cos λ = 0.707, ν = $0.0085140.008514+2×0.0121$ = 0.26, cos β = 0, and f = $4.228−4.17694.228$ = 0.0121. The critical thickness of NiO on TiN/MgO is therefore ∼45 Å. The value is close to the calculated critical thickness (75 Å) obtained for the NiO/MgO system reported without the TiN interlayer.²² 

By the same calculation, the parallel lattice strain of NiO on TiN/Si is 0.0067, and that of TiN on Si is 0.2234. For NiO on TiN/Si, we have b = 2.964 Å, cos λ = 0.707, ν = $0.006670.00667−2×0.3029$ = 0.0108, cos β = 0, and f = $5.992−4.17695.992$ = 0.3029. However, the calculation of the critical thickness of NiO on TiN/Si did not yield practical values, which indicated a significantly smaller critical thickness of NiO on TiN/Si or breakdown of the model.

Although the adopted critical thickness model predicted a relatively small critical thickness of ∼45 Å for NiO on TiN/MgO, the actual lattice relaxation process is known to be hindered by various kinetic barriers. The determination of the actual critical thicknesses will therefore require growth experiments on the respective template substrates having multiple samples grown at varying thicknesses for subsequent XRD and TEM characterization, which is beyond the scope of the current investigation.

In this work, we grew single-crystalline NiO epitaxy on conductive-TiN-sputtered substrates and confirmed the properties of the NiO and TiN layers by multiple tests. TiN had a closer lattice constant to that of NiO, and therefore, is useful as an intermediate layer for a non-conductive substrate, such as MgO, and for mitigating the large lattice mismatch in NiO grown on Si (100). The STEM-HAADF and STEM-EDX confirmed the nature of the deposited materials and the elemental cross section distribution. The XRD out-of-plane 2θ and in-plane φ scans revealed the cubic structures of the NiO and TiN layers and the cube-on-cube epitaxial growth of the NiO/TiN/MgO-(100) and NiO/TiN/Si-(100) structures. The XRD rocking curve measurements and the cross section HRTEM imaging and the corresponding FFT analysis demonstrated the higher crystalline quality of the NiO/TiN structure grown on the MgO-(100) substrate than on the Si-(100) substrate. The lattice mismatch between NiO/TiN and TiN/MgO was lower than 1%, while that at the TiN/Si interface was above 20% based on the atomic line profile extracted from the inverse FFT images. The epitaxial single-crystal NiO-based template substrate can thus be used to grow group-III-oxide- and nitride-based semiconductors and related devices.

J.-W.L. and K.-H.L. contributed equally to this work.

The data that support the findings of this study are available within the article.

This research was supported by the King Abdullah University of Science and Technology (KAUST) baseline funding, Grant No. BAS/1/1614-01-01, and by the Romanian Ministry of Education and Scientific Research through the National Core Program, Grant No. 18N/08.02.2019. The authors further acknowledge the access of the Nanofabrication Core Lab as well as the Imaging and Characterization Core Lab facilities at KAUST.