ALD2018 Session EM-MoA: Laminate, Multicomponent, and Nitride Materials

Nanolaminate coatings can offer significant improvements to traditional single-layer materials. An interesting class of nanolaminate materials is one based on alternatively ordered thin films of transition metal oxides with nanoscale thickness [1]. Using the TiO₂/Al₂O₃ nanolaminate concept of our previous work [1], it is possible to obtain at high temperature a ‘stable amorphous film’ which is very attractive for applications in microelectronics and optics. It is known that amorphous films of certain materials ‘crystallize’ during the various processing steps (deposition, metallization and lift-off) required for manufacturing a device. This crystallization modifies the fundamental properties of the film, which makes it suitable for the application. Herein, the thermal atomic layer deposition (ALD) and plasma enhanced atomic layer deposition (PEALD) of TiO₂/Al₂O₃ nanolaminates on silicon(100) and glass substrates were studied in order to discuss the growth mechanism and material properties of the films. We use the nanolaminate concept where each TiO₂/Al₂O₃ nanolaminate incorporates a certain number of Al₂O₃ partial-monolayers (between 10 to 90) during 2700 total reaction cycles of TiO₂ under temperature of 250ºC [1]. TiO₂/Al₂O₃ films were deposited by a TFS-200 ALD system from Beneq. TMA and TTIP were used as metallic precursors, while H₂O or O₂ plasma were used as ligand. The growth mechanisms and fundamental properties of the TiO₂/Al₂O₃ nanolaminates were inferred from measurements of the film thickness, chemical composition, microstructure and morphology. In addition, some optical and mechanical characteristics were determined and correlated with fundamental properties. Results evidenced that the model proposed for thermal ALD TiO₂/Al₂O₃ nanolaminate is valid for PEALD, however it was necessary a higher number of Al₂O₃ layers for stop the TiO₂ crystallinity. This allows obtaining a nanolaminate with improved properties in comparison with thermal ALD, as for example higher transmittance, lower refractive index (near the value of 3.2 eV), low resistivity, and higher hardness and young modulus. These properties are interesting for sensing application, such as UV detection.

[1] J. Azadmanjiri, C. C. Berndt, J. Wang, A. Kapoor, V. K. Srivastava, C. Wen, J. Mater. Chem. A 2 (2014) 3695–3708

[2] G. E. Testoni, W. Chiappim, R. S. Pessoa, M. A. Fraga, W. Miyakawa, K. K. Sakane, N. K. A. M. Galvão, L. V. Santos, H. S. Maciel. Journal of Physics D: Applied Physics, 49 (2016) 375301.

PbTi_xZr_1-xO₃ (PZT) is a perovskite ferroelectric material that is widely studied for its high dielectric constant and piezoelectric coefficients. Thin-film PZT has found several commercial uses, namely ink-jet print heads, ferroelectric random access memory (FRAM), MEMS gyroscopes, and more recently PZT-MEMS based speakers. Niche technological applications include MEMS resonators, actuators, and transducers. Thin-film PZT is currently deposited by primarily 2-dimensional (2-D) deposition techniques such as sol-gel, sputtering, and MOCVD. Thin-film PZT-based devices could benefit greatly from an atomic layer deposition (ALD) PZT process capable of transitioning traditionally 2-D architectures to 3-dimensional (3-D) architectures utilizing etched high aspect-ratio sidewalls to greatly enhance areal density especially for actuator technologies. The literature on PZT deposited by ALD is sparse and falls short of providing industry with an off-the-shelf solution. Additionally, there is little effort dedicated to controlling the texture of ALD PZT films, which is a desired component for many of the aforementioned technologies. Here we explore several techniques for controlling the texture of ALD PZT by depositing nucleation seed layers. First, we attempt to deposit 1-3nm of ALD grown TiO_x followed by 1-5nm ALD grown PbO_x and anneal in a PZT sputter chamber which has a PbO_x partial pressure. The PbO_x is expected to react with the TiO_x to form PbTiO₃ (PTO), but will be self-limited to what can be chemically incorporated because excess lead oxide will boil off the surface with the chuck maintained at 600° C or hotter. Second, we deposit 1-5nm of ALD grown TiO_x and anneal in an RTA that is expected to have a small PbO_x partial pressure due to re-deposition of lead oxide from prior processing of chemical solution deposited PZT films onto a quartz glass liner that is within 1 cm of the wafer surface. Third, as a control we deposit 40nm of PTO by sol-gel with 30% lead excess which crystallizes primarily in the 100/001 orientations following post-deposition anneal. Each of the ALD PZT films are deposited in an amorphous state and later crystallized by rapid thermal anneal in an O₂ atmosphere. To further evaluate the contribution of the nucleation layers to the texture, sol-gel PZT films are deposited in the first and second cases to be analyzed alongside the ALD PZT films. The films are characterized primarily by spectroscopic ellipsometry for film thickness, scanning electron microscopy for film thickness and microstructure, x-ray diffraction for phase identification and texture, and for select samples STEM with EDS for detailed structure and chemical analysis.

The mixed TiO₂-SiO₂ oxides have multiple possible optical applications with demonstrated use in waveguides, laser mirrors and rugate filters. They are also considered as an alternative dielectric for high-k applications and they attracted considerable attention in the area of photocatalysis because they are more active than pure TiO₂. Ti_xSi_(1-x)O₂ were deposited on Si substrate by plasma enhanced atomic layer deposition using tetrakis(dimethylamido)titanium and tris(dimethylamino)silane precursors for Ti and Si, respectively. The overall stoichiometry of the final thin film was varied by changing the relative number of TiO₂ and SiO₂ cycles as 1:1, 2:1, 1:2 and 3:1. The chemical bonding in the films was confirmed by X-ray photoelectron spectroscopy (XPS). Optical properties were determined in the wide spectral range from the ellipsometry (NIR-UV spectroscopic ellipsometer J. A. Woollam V-VASE, 0.6-6.4 eV, angles of incidence 60, 65, 70 and 75°) and reflectance (VUV spectrometer McPherson VUVAS 1000, 5.6–10.3 eV, near-normal incidence) measurements. The results on dielectric function and band gap were in good agreement with the density functional theory (DFT) predicted optical properties of amorphous Ti_xSi_(1-x)O₂ solid solutions. The electrical properties of the films were investigated in the MOS capacitor structures using a wafer probe station and the 4200 Keithley semiconductor parameter analyzer. The mixed oxide dielectric constant is deduced from the C-V curves measured at 1 MHz in accumulation regime whereas the film stability is assessed from the C-V loops recorded by accumulation (A) to inversion (I) and I to A. The dielectric layer insulating characteristics were assessed from the measured leakage current and breakdown voltage.

Composite nanoscale materials are under intense investigation, opting for merging the functionalities of the constituting materials. Fe₂O₃-TiO₂ core-shell nanoparticles (NPs) are among the intensely investigated composite NPs for their application potential in a broad variety of fields, such as, sensing, selective separation, biomedicine, or photocatalysis [1,2]. The combination of a titanium dioxide (TiO₂) shell and a magnetic iron oxide core can provide added value to the composite by tuning the biocompatibility, stability and/or photocatalytic properties through the TiO₂ shell and at the same time, the possibility to use the materials for hyperthermia, imaging or controlled positioning through magnetic fields [3].

In this work, Fe₂O₃ nanoparticles are coated with TiO₂ and concertedly turned magnetite Fe₃O₄-TiO₂ core-shell NPs through atomic layer deposition (ALD) at moderate temperatures (Figure 1). The generated coatings prevent agglomeration of the nanoparticles and re-oxidation to γ-Fe₂O₃ even at high temperatures, allowing for the use of post-process annealing and thus crystallization of the amorphous TiO₂ to anatase. The occurring reduction of Fe₂O₃ to Fe₃O₄ is a function of the applied precursor and the processing temperature. This presentation will detail on our process and the characterization of the materials by means of X-ray diffraction (XRD), magnetometry, Energy Dispersive X-Ray analysis (EDX) and electron microscopy.

[1] M. Stefan et al., “Synthesis and characterization of Fe₃O₄-TiO₂ core-shell nanoparticles,” J. Appl. Phys., vol. 116, no. 11, 2014.

[2] W. Wu, X. Xiao, S. Zhang, F. Ren, and C. Jiang, “Facile method to synthesize magnetic iron oxides/TiO₂ hybrid nanoparticles and their photodegradation application of methylene blue,” Nanoscale Res. Lett., vol. 6, pp. 1–15, 2011.

[3] S. Exhibit, C. Strobel, R. Herrmann, A. A. Torrano, and I. Hilger, “Biocompatibility of titanium dioxide nanoparticles for diagnostic and therapeutic purposes in personalized nanomedicine,” pp. 1–17, 2013.

Aluminum nitride (AlN) has a wide band gap, good thermal and chemical stability, thermal conductivity and interesting electro-acoustic properties. It is used in several applications, of which many depend on the crystalline structure of the material. Barrier, passivation and dielectric layers benefit from amorphous structures [1]. However, highly oriented crystalline films are needed for instance in optoelectronic and surface acoustic wave devices [2].

In this work the aim was to deposit AlN films with varying crystallinities to separately address the needs of different application areas. TMA and AlCl₃ were utilized as the metal precursors, while NH₃ or NH₃ plasma were used as the nitrogen source. The deposition temperatures were varied from 220 to 550°C.

PEALD AlN films were deposited at 220°C from TMA and NH₃ plasma in a Beneq TFS 200 ALD reactor. On a 200 mm wafer a uniformity of 1.7% was achieved. The films were amorphous and had a refractive index of 1.88.

Thermal ALD process from TMA and NH₃ was studied at 350–475°C with a Beneq P400 batch ALD reactor. Significant effect of the pulsing sequence on the growth characteristics and film properties was observed. With the optimized process at 350°C, a GPC of 1.37 Å/c and refractive index of 2.10 were measured. The bulk oxygen and carbon contents were 1.0 and 0.4 at.%. The films were polycrystalline with randomly oriented hexagonal structure, and the degree of crystallinity was increased with increasing deposition temperature (Figure 1).

The Beneq P400 batch ALD reactor is equipped for studying ALD processes at high temperatures. This enabled us to increase the temperature range of the thermal ALD AlCl₃–NH₃ process from the previously reported 500 up to 550°C [3,4]. Significant improvement on the film quality was observed with increasing deposition temperature. The GPC and refractive index increased from 0.67 to 0.76 Å/c and 1.99 to 2.06. Simultaneously the bulk oxygen and chloride impurity contents decreased from 3.2 to 2.5 and 1.8 to 0.7 at.%. Furthermore, these results were achieved with a uniformity of 2.5% on a batch set-up with a planar surface area of 20 000 cm². All the films were polycrystalline with highly oriented hexagonal [001] structure as the only observed peak was (002) (Figure 2).

[1] M. Bosund et al., Appl. Surf. Sci. 256 (2010) 7434-7437.

[2] G.F. Iriarte at al., Mater. Res. Bulletin 45 (2010) 1039-1045.

[3] K.-E. Elers et al, J. De Phys. IV France 5 (1995) C5-1021–C5-1027.

[4] V. Rontu, J. Vac. Sci. Technol. A 36 (2018) 021508.

In the field of III-V semiconductor materials, research is focused mostly on epitaxially grown films. Developing their polycrystalline counterparts (e.g., polycrystalline gallium nitride or poly-GaN) may lead to their more widespread adoption in application areas, such as in microelectronics and lighting. Unlike traditional epitaxy, atomic layer deposition (ALD) of poly-GaN films allows for less stringent growth requirements such as reduced temperatures, atomic-level thickness control and direct growth on a variety of substrates without the need of buffer layers.

The majority of reports on low temperature deposition of poly-GaN from trimethylgallium (TMG) and ammonia (NH₃) precursors mention an additional means of activation (e.g., plasma) to dissociate NH₃. This is done to increase the chemical activity of ammonia (by forming radicals) with the TMG-chemisorbed surface. However, in this work, we have used a different (radical-free) chemistry to deposit GaN films by a purely thermal route; i.e., without plasma or other radical-production means. Such a deposition approach can be beneficial, for example, in industrial-scale batch-type reactors where uniform radical delivery is a challenge.

We have utilized the strong adduct-forming chemistry of the Lewis-acid TMG and the Lewis-base NH₃ on the growth surface (Fig. 1). At the deposition temperature of 400^oC the TMG-NH₃ adduct dissociates into a Ga-NH₂-Ga linkage, thereby forming the first monolayer of GaN in a self-limiting fashion. Due to the reversible nature of this adduct formation, at optimal gas-pressure conditions, alternate pulses of TMG and NH₃ with in-between purges result in the efficient film growth. This pressure-dependency of adduct formation is revealed by in-situ monitoring the growth with spectroscopic ellipsometry (SE), which shows a strong dependence of the growth per cycle (GPC) with the NH₃ partial pressure (Fig. 2). Using this approach, we obtained GPC values as high as 0,045 nm/cycle.

The efficient removal of all methyl groups of TMG during GaN deposition in the adduct-assisted mechanism is revealed from the low carbon content (1 at. %) in the films, as obtained from depth profiled X-ray photoelectron spectroscopy (Fig. 3). The cross-section scanning electron microscope (SEM) image (Fig. 4) shows a 23-nm GaN film deposited on aluminium nitride (AlN). Grazing-angle X-ray diffraction (GIXRD) spectrum (Fig. 5) reveals that the GaN film is polycrystalline and has a wurtzitic structure.

In our presentation, the purely thermal radical-free deposition of poly-GaN films will be detailed: focussing on the growth mechanism, determining the ALD window, and presenting several film properties.

The group 13 nitrides (13-Ns) are essential electronic device materials for present and future technologies. Thin film deposition of AlN and GaN are well explored by CVD at relatively high temperatures (800-1000°C). The high temperature aids epitaxial growth and ligand removal, rendering high crystalline quality films with low impurity levels. As 13-N CVD almost exclusively uses the trimethyl complexes of the group 13 metals, carbon is a major impurity with concentrations in the 10¹⁷ cm^-3 range.

ALD is less explored for the 13-Ns but would open to topographically more advanced 13-N structures, higher control when depositing multilayer superlattice 13-N structures and possibly low deposition temperatures. A major challenge for ALD is the limitation in deposition temperature set, mainly by the thermal stability of the monolayers formed by the surface chemical reactions in the ALD cycle. The thermal stability of the group 13 precursors also limits the deposition temperature; AlMe₃ decomposes at about 330°C and previous studies have shown that the carbon content in the films increases rapidly when deposition temperature starts to approach the decomposition temperature [1]. The lower temperatures used in ALD compared to CVD hampers the crystalline quality and the removal of ligands leading to an increase in impurity levels, mainly carbon and oxygen, to the several atomic percentage levels.

Here we discuss the possibility to use the time-resolved precursor supply in ALD with an additional pulse between the precursor pulses, i.e. an ABC-type pulsing, to decrease impurities and increase crystalline quality. To improve the crystalline quality, we intentionally move out of the ALD temperature window and did time-resolved CVD of 13-Ns at 480°C, allowing thermal decomposition of the 13Me₃ precursors. We then introduced a pulse of H₂ or N₂ gas or H₂-, N₂- or Ar plasma as B-pulse between the 13Me₃ (A-pulse) and NH₃ (C-pulse) to study how it can aid the removal of ligands, as probed by the impurity levels in the deposited films. XPS measurements shows that the carbon content in AlN can be reduced from about 3 at. % to below 1 at. % (which is on the detection limit of XPS) by a H₂ gas or H₂ plasma as the B-pulse. H₂ gas as B-pulse also increases the GIXRD peak intensity, which is likely an effect of a higher crystalline quality, and the growth per cycle. Initial results do not show any improvement on the carbon content for N₂ plasmas as B-pulse in AlN deposition.

Financial support from the Swedish Foundation for Strategic Research (SSF RMA 15-0018) is gratefully acknowledged.

[1] H. Van Bui et al. ECS J. Sol. Stat. Sci. Technol. 2014, 3, P101-P106.