ALD/ALE 2022 Session EM1-WeA: Emerging Materials

Alloying group 13-nitrides to ternary phases allows tuning of the bandgap from 6.2 eV for pure AlN down to 0.7 eV for pure InN. The bandgap of In_1-xGa_xN can theoretically span from UV to IR (3.4–0.7 eV), including the whole visible light range by varying x, making it promising material for optoelectronic applications. However, the ability to vary the composition of In_1-xGa_xN is limited by the theoretically predicted metastability of In_1-xGa_xN for 0.05 < x < 0.95, which leads to phase separation into their binary materials. The deposition of In_1-xGa_xN is also hindered by the low thermal stability of InN, which decomposes into In metal and N₂ at around 500 °C, making traditional CVD approaches ill-suited. We have recently shown that ALD is a promising technique to deposit InN thin films with excellent structural quality,¹ ALD is therefore a promising alternative to CVD for In_1-xGa_xN with x close to 0.5. In our efforts to deposit In_0.5Ga_0.5N we have explored two ALD approaches:

By using a short period superlattice, with alternating monolayers of GaN and InN, In_0.5Ga_0.5N deposition was attempted from repeated n InN and m GaN monolayers (n=m= 1, 2, 3…) using triethyl gallium (TEG), trimethyl indium (TMI) and ammonia plasma at 320 °C. This approach afforded single-crystalline In_1-xGa_xN with tunable x between 0.3 and 0.7 by varying the ratio between n and m. The crystalline quality of In_1-xGa_xN prepared by this multilayer approach ALD is remarkably better than that prepared by conventional continuous CVD and earlier reported ALD work using a multilayer approach with thicker layers of InN and GaN in the multilayer.

By mixing solid Ga(III) and In(III) triazenides in the same evaporator and co-subliming the two metal precursors In_1-xGa_xN was deposited using a single, mixed metal pulse and NH₃ plasma at 350 °C.² In_1-xGa_xN was successfully deposited using this approach and the value of x could be tuned by changing the sublimation- and deposition temperatures, and the ratio of the two metal precursors. An In_1-xGa_xN film with x = 0.5 was deposited and found to have a band gap of 1.94 eV. The In_1-xGa_xN film grew epitaxially on 4H–SiC(0001) without need for a buffer layer and without phase segregation or decomposition of the In_1-xGa_xN into the binary materials or In droplets.

Our results reveal a promising potential of ALD over conventional growth techniques to prepare ternary group 13-nitrides with tunable composition at low temperature, which provides the possibility to grow heterostructures with metastable alloys for device application.

Refs.

1. Hsu et al. Appl. Phys. Lett. 2020, 117, 093101.
2. Rouf et al. J. Mater. Chem. C2021, 9, 13077.

Complex oxides exhibiting metal-insulator transitions (MITs) are exemplar materials systems with strong correlation and emergent functional phenomena. Particularly the rare-earth nickelates (RENiO₃, with trivalent rare-earth RE = La, Pr, Nd, …, Lu) are of interest as their MITs occur concomitantly with a structural transition. Underlying their rich phase diagram and the MIT’s physical origin is a complex interplay of interactions; though it remains an unsolved puzzle in fundamental research, the exotic properties rooted in it have great potential for electronics applications.

Among the RENiO₃s, the MIT temperature of NdNiO₃ (T_MI = 200 K) is the closest to room temperature. Tuning the T_MI can be carried out using strain or by partial substitution of Nd with larger RE cations (see phase diagram). A more significant challenge, however, has been to develop a synthesis route that stabilizes Ni³⁺ and provides sufficient control under industrially relevant conditions. For instance, high temperatures and ultrahigh vacuum (UHV) typically facilitate epitaxy, but are incompatible with monolithic device integration.

In this talk we show that with ALD – since long embraced by the electronics industry – we can grow high-quality epitaxial NdNiO₃ thin films with excellent control of thickness, uniformity, and chemical composition. This is achieved at low temperatures (225 ^oC) without constraints to the substrate geometry or need for UHV. Thin films of stoichiometric composition show low resistivities at room temperature and a sharp MIT, which are desired properties of a functional electronic switch in future neuromorphic architectures. Quaternary oxide thin films of the form (RE,Nd)NiO₃ have been successfully deposited using ALD with the aim of tuning the T_MI close to 273 K. Further chemical and electrical characterizations are needed, however, to establish and control the effect of partial RE substitution on the T_MI.

Although much of the fundamental behavior of the RENiO₃s remains contested, their potential for applications is undisputed; in fact, many members are already found in various device concepts. The success in using low-temperature ALD to grow high-quality NdNiO₃ (stoichiometric and cation substituted) thin films with a sharp MIT could promote the implementation of such switching-materials in next-generation electronics. A complex oxide field-effect transistor may thus be more within reach than previously anticipated, offering a viable alternative and/or complement to Si-based circuitry. Based on fundamentally different mechanisms, this could pave the path for a greener and more sustainable integrated circuit technology in the future.

Thin films of insulating ferro- and ferrimagnetic complex oxides with high Curie temperatures, such as spinel ferrites, are essential for many emerging applications utilising room temperature spin-polarisation and magneto-optical effects, e. g. spintronics and sensors [1]. There is a need for a synthesis method for high quality magnetic oxides with large scale processing compatibility. Here, we have developed PEALD processes for two spinel ferrite materials, CoFe₂O₄ (CFO) and NiFe_2.5O₄ (NFO) using ferrocene and cobalt(III)- or nickel(II) acetylacetonate as precursors in direct plasma PEALD at 250°C. The CFO films were deposited with 1:2 Co:Fe ratio, while the NFO films were grown iron-rich to ensure that the ferrimagnetic property is not hampered by a parasitic antiferromagnetic nickel oxide component [2]. Stoichiometry of the grown ternary oxide films was confirmed with time-of-flight elastic recoil detection analysis measurements, which also showed that the low light element impurity content of the films (H < 2.0 at. -%, C < 0.3 at. -%,) originates mainly from the acetylacetonate sources. According to the X-ray diffraction measurements of 40 nm thick films, the PEALD CFO and NFO have the desired (inverse) spinel structure, and the films grown on sapphire substrates are strongly (111) oriented already as-deposited. Helium ion microscopy and atomic force microscopy both showed that the films are continuous and free of aggregations. The oriented CFO films on sapphire have a very smooth surface (r_rms < 0.3 nm) but the NFO with a same thickness has a higher surface roughness (r_rms > 1.5 nm), which is in accordance with the previous observations of the ALD-grown iron-rich NFO [3]. In addition to the growth and structural characteristics we will present the results of the magnetic property measurements of the films.

[1] Hirohata et al. IEEE Trans. Magnetics 5 (2015) 0800511

[2] Napari et al. InfoMat 2 (2020) 769

[3] Bratvold et al. J. Vac. Sci. Technol. A 37 (2019) 021502

Recently multilayered nanolaminates (NLs) of two dielectrics with conductivity contrast exhibiting giant dielectric constant owing to interface induced Maxwell–Wagner (M-W) relaxation have emerged as potential candidate for high density storage capacitors. The M-W polarization can be engineered precisely by controlling the thicknesses of sublayers and number of interfaces. We report growth of Al₂O₃/TiO₂ (ATA) NLs on Si and Au/Si substrates using atomic layer deposition, wherein M-W relaxation induced high dielectric constant was realized and engineered by tuning sublayer thicknesses. Trimethylaluminum (Al (CH₃)₃) and Titanium tetrachloride (TiCl₄) were used as source for Al and Ti respectively, while deionized water (H₂O) was used as source for oxygen. Depositions were carried out at 200 °C and the average growth per cycle for TiO₂ and Al₂O₃ was ~ 0.4 and 1.6 Å respectively. The thickness of Al₂O₃ and TiO₂ layers were kept same in a given NL and was reduced from ~ 2.4 to 0.17 nm in different NLs keeping the total stack thickness fixed at ~ 60 nm. X-ray reflectivity curves from these NLs with intense Bragg peaks and clean Kiessig fringes, as shown in Fig. 1, confirmed the multilayer structures with uniform thickness along with distinct interfaces. The dielectric properties of ATA NLs were studied in Au/ATA/Au device configuration using impedance spectroscopy in frequency range of 10–10⁶ Hz. The dielectric constant of ATA NLs at 10 Hz was found to increase from ~ 23 to 290 with decreasing sublayer thicknesses from ~ 2.4 to 0.17 nm (Fig. 2(a)), while the dielectric loss was initially found to reduce from ~ 0.8 to 0.06 with reduction in sublayer thicknesses down to ~ 0.48 nm and then increased up to ~ 0.24 with further reduction in sublayer thicknesses down to ~ 0.17 nm (Fig. 2(b)). The dielectric constant of ~ 290 obtained for the ATA NL with ~ 0.17 nm sublayer thickness is significantly larger than that of both Al₂O₃ (K ~10) and TiO₂ (K ~ 20) and is proposed to be due to M-W type dielectric relaxation caused by space charge polarization across the interfaces of Al₂O₃/TiO₂. Temperature dependent dispersion in dielectric constant and loss of ~ 0.48 nm ATA NL clearly revealed two sets of thermally activated relaxations, confirming existence of interfacial M-W relaxation (Fig. 3). The ATA NLs of sublayer thickness ~ 0.17 nm showed high capacitance density of ~ 43.1 fF/μm², low loss of ~ 0.24 at 10 Hz, low EOT of ~ 0.8 nm, high breakdown field of ~ 0.265 MV/cm, low leakage current density of ~ 8.5 x 10 ^-4 A/cm² at 1V and cut-off frequency of ~ 12KHz which are promising for development of next generation high density storage capacitors.