ALD/ALE 2022 Session AF-MoP: ALD Fundamentals Poster Session

Ruthenium thin films are garnering significant interest as next-generation interconnects to replace Cu in future nanoelectronic devices. Especially in the back end of line (BEOL) and middle of line (MOL), ongoing scale down towards 2 nm and beyond has motivated alternative metallization approaches such as semi-damascene in which Ru outperforms Cu and Co notably.^[1,2]One deposition technique most suitable to provide thin films for challenging interconnect device architectures is ALD.^[3] A paramount factor in each ALD process is the choice of precursors and their chemistry that governs layer formation and material quality. A review of the current ALD processes for Ru thin films shows that a considerable number of often closely related precursors with their individual advantages and drawbacks have been employed.^[4] Hitherto, none of them could fully satisfy academic and industrial demands alike.

This presentation describes the synthesis and detailed characterization of an alternative Ru precursor class: Ru diazadienyl cymenes [Ru(DAD)(Cym)] and their potential for CVD and ALD applications. Two examples, Ru(tBuDAD)(Cym) and Ru(iPrDAD)(Cym) (Figure 1) were obtained in high yields in a one-pot synthesis and thoroughly assessed in terms of purity (NMR, EA) and structure (SC-XRD). The structures were reproduced by DFT studies and subjected to bond dissociation analysis which rendered them appreciably stable. Thermal stability upon evaporation was evaluated in more detail alongside overall evaporation behaviour in thermogravimetric analyses (TGA) (Figure 2). Vapor pressures of the two complexes were derived and compared to the most often used Ru(EtCp)₂. Especially Ru(iPrDAD)(Cym) showed appreciable volatilization and deemed competitive towards the reference compounds.

Thus, Ru(iPrDAD)(Cym) was employed as precursor in a PEALD process with O₂ plasma. Preliminary results demonstrated the growth of thin, pinhole-free, low roughness films on Si(100) at temperatures as low as 120 °C (Figure 3). A full process study investigating typical ALD growth characteristics in terms of saturation, temperature dependency of film growth and thickness scalability was carried out to assess the ALD behaviour. Complementarily, AFM, RBS/NRA and XPS as well as resistivity measurements were performed.

[1]G. Murdoch, et al., in 2020 IEEE IITC 1052020, p. 4.

[2]https://www.imec-int.com/en/press/imec-presents-alternative-metals-advanced-interconnect-and-contact-schemes-path-2nm.

[3]M.-J. Li, et al., in 2021 IEEE IITC 762021, p. 1.

ALD is a versatile and emerging technology, allowing for the precise coating of challenging substrates with a nanometer control over thickness. Due to its unique assets, a substantial growth of the ALD market is expected. In the current context of moving towards greener processes, the ALD technique offers the potential to become greener and to tackle environmental challenges. Indeed, the process itself has currently a consequent impact on the environment which should ideally be reduced as the technique is implemented in a wider range of products and applications. Based on a literature overview, our findings show that the duration of the process, the temperature used, and the precursor chemistry are key factors affecting the environmental impact of ALD. The principles of green chemistry are discussed considering the specificities of the ALD process, and different ways to reduce the impact are proposed, in particular the optimization of the processing parameters, the use of spatial ALD (SALD) and the chemical design of greener precursors are shown as efficient routes to lower the ALD environmental impact and improve its sustainability.

‘Click chemistry’ encompasses a set of powerful chemical reactions that have a high yield, are highly selective and specific, proceed under simple reaction conditions, use readily available starting materials, create stable products and harmless by-products.[Kolb, H.C., et al. (2001) Ang. Chem. Int.] The azide-alkyne cycloaddition reaction is considered the cream of the crop of the click reactions. Hence, surface functionalization by azido-moieties is often desired for applications in drug discovery, polymer chemistry, materials engineering and biosensor devices. [Lahann J. (2009) Click chemistry for biotechnology and materials science] This can be achieved by depositing self-assembled monolayers using azido-containing organosilanes which typically involves liquid-phase protocols. The use of vapour-phase chemistry might be beneficial regarding integration into a high-throughput production sequence and the deposition of high-quality and reproducible coatings on 3D microstructures. [Vos, R., et al. (2018) Langmuir] Because vapour-chemistry typically involves the use of high vacuum equipment, the stability of the azido-groups in high vacuum is of utmost importance.

In this work,[Vandenbroucke S.S.T. (2021) Langmuir] the stability of azido-containing self-assembled monolayers is monitored in real-time using in situ reflection FTIR at a temperature of 150 °C and a pressure of 1E-5 mbar for 8 h. The data in Figure 1 displays a clear decrease of the asymmetric azide stretching vibration at 2105 cm^-1 over time in high vacuum, suggesting the degradation of the azido-groups. The degradation is further investigated at three different temperatures and seven different nitrogen partial pressures using ex situ ATR-FTIR. The degradation is found to increase at higher temperatures and lower nitrogen pressures. This is in accordance with the theory that the degradation reaction involves the decomposition into molecular nitrogen and the formation of a highly reactive nitrene.

Many applications in the semiconductor industry require a perfect control over the surface chemistry down to the nanometre level to yield reproducible results. For the condition with the most degradation only 63% of azides are found to remain at the surface after 8 h in high vacuum. This would imply a significant loss in control over the exact surface chemistry. One should therefore always consider the stability of functional groups such as azides when depositing or post-processing functional coatings in high vacuum.

Nucleation is a crucial step in the formation of ALD films. It determines the formation of the first atomic layer of film. Proper nucleation is needed to deposit closed films with a homogeneous thickness. Since the nucleation deals with the formation of the first atomic layer of film, a very surface specific tool is needed to study nucleation: It should be able to distinguish between a closed monolayer and a half-closed double layer.

For common other techniques, such as XRF or XRR, it is not possibly to make this distinction due to a lack of surface specificity. Low Energy Ion Scattering (LEIS), on the other hand, is specific to the very first atomic layer, enabling it to determine the closure of the film (surface fractions of film and substrate), rather than the overall number of atoms per unit area that XRF or XRR provide.

We will demonstrate how LEIS works for the nucleation of AlOx films on ZrOx and the ZrOx films on AlOx. We show the normalized LEIS peak areas for Al and Zr as a function of cycle number for both depositions. The results show that in both cases, the films are almost closed after 20 cycles. In addition, we will present and discuss the following two features: In the deposition of ZrOx on AlOx, the peak area for Zr “overshoots”. This indicates that the initial deposition happens with a stoichiometry that Zr-rich compared to ZrO₂ (and has some similarity to that of the Al₂O₃ substrate). In the case of the deposition of AlOx on ZrOx, the peak area for Al “undershoots”, indicating that the initial deposition proceeds with an Al-poor stoichiometry compared to that of Al₂O₃ (with some similarity to that of the ZrO₂ substrate).

We investigated electrical properties and structure of conductive nitrides (TiN_x, ZrN_x, TaN_x, HfN_x) deposited by the PEALD process from amide precursors and H₂/Ar plasma. Ion bombardment introduces additional energy to the growing film promoting chemical reactions and film densification. In PEALD process involving H₂/Ar plasma, most of bombardment effects are related to heavier Ar ions which are responsible for obtaining highly crystalline nitride films. Ion energy depends on the pressure inside ALD reactor during plasma half-cycle. Low pressure is favorable for deposition of highly conductive nitrides. The deposited nitrides have different grain size, carbon contamination, and variation of the film density from theoretical value despite same process conditions. Films crystallinity was investigated by the HAADF-STEM (Fig.1). The deposited nitride films differ in film morphology. TiN_x films consist of large columnar grains elongated through the whole film thickness. HfN_x films has columnar-like grains with smaller grain size compared to TiN_x. ZrN_x and TaN_x films consist of small grains. Significant ion bombardment reached in the process results in the growth of (111)-oriented polycrystalline films with low O contamination (Table1). The films are polycrystalline (Fig.2) despite the significant carbon content. Crystallization is stimulated by the ion bombardment effect achieved by the lowing plasma pressure. Carbon contamination correlates with the level of C-content in the precursor molecule.

A number of models have been developed for simulating the conformality of atomic layer deposition processes.¹ Simulation models can be used, for example, for the optimization of process parameters towards improved conformality, and for the extraction of kinetic-related information through fitting the model parameters to experimental data.^2,3

Three fundamentally different simulation approaches have been described in the literature to simulate the conformality of ALD processes: diffusion-reaction, ballistic, and Monte Carlo models.^1,5,6 In diffusion-reaction models, transport inside a feature is described by a diffusion equation that includes an adsorption loss term.^1,6 Ballistic models describe particle transport in the molecular flow regime and are based on a balance of particles to compute fluxes at different locations inside a feature.⁵ In Monte Carlo models, the path of each particle at a time is simulated.¹

In this work, we compared the conformality of ALD growth inside lateral channels predicted by two fundamentally different models: a diffusion-reaction model (Model A)^6,7 and a ballistic model (Model B).^5,8 The effect of different operating conditions on the conformality of ALD is studied in the free molecular flow regime.

For all the parameters studied, the main trends of the obtained saturation profiles were similar for both models (Fig. 1 of the supp. info). However, penetration depth at half-coverage predicted by the ballistic model was greater than that predicted by the diffusion-reaction model while the predicted slope at half-coverage was greater with the diffusion-reaction model (Fig. 2 of the supp. info). The ballistic model predicted a sudden increment of coverage when the profile reaches the end of the channel. The reasons and consequences of the differences will be discussed.

This work was supported by the Academy of Finland (ALDI consortium, grant No. 331082). An earlier version of this work was presented at the 21st Int. Conf. on Atomic Layer Deposition, ALD 2021.

References

1. V. Cremers et al., Appl. Phys. Rev., 2019, 6, 021302.

2. K. Arts et al., J. Vac. Sci. Technol. A, 2019, 37, 030908.

3. J.R. van Ommen, A. Goulas, and R.L. Puurunen. Atomic Layer Deposition. In Kirk-Othmer Encycl. Chem. Technol., John Wiley & Sons, Inc., 2022.

4. A. Yanguas-Gil and J.W. Elam, Chem. Vap. Deposition, 2012, 18, 46.

5. A. Yanguas-Gil and J.W. Elam, Theor. Chem. Acc., 2014, 133, 1465.

6. J. Yim and E. Verkama et al., submitted. Preprint DOI:10.33774/chemrxiv-2021-2j4n1

7. M. Ylilammi, O. Ylivaara, and R.L. Puurunen, J. Appl. Phys., 2018, 123, 205301.

8. Machball software, available at: https://github.com/aldsim/machball (accessed: Jan. 25, 2022).

Volatile metal halides in general are good ALD precursors as they are simple, cheap, readily available, thermally stable, reactive with many non-metal precursors and small in size allowing good growth rates. Unfortunately, halides of many metals are not volatile or they possess only low volatility. Many of such metal halides can, however, be rendered more volatile by adding proper adduct ligands to the metal coordination sphere.

Earlier we have introduced diamine adducts of cobalt and nickel to ALD. While pure halides of Co and Ni have polymeric solid-state structures, high melting point and low volatility, the diamine adducts have monomeric structure and volatility that is among average for known Co and Ni precursors. Under reduced pressure these compounds sublime intact and have sufficiently large temperature window between the sublimation and decomposition temperatures. These compounds have also been shown to perform well in ALD.[1]

For metal films low deposition temperatures are desired to minimize agglomeration and thereby obtain smoother films that are continuous at lower thickness. High volatility of the metal precursor is therefore needed. Here more adducts of the transition metal halides, especially phosphine adducts, were synthesized and studied.

While diphosphine adducts of Co and Ni are thermally exceptionally stable, their volatility is low. In contrast, monophosphine adducts of the same are highly volatile but the thermal stabilities are low. For example NiCl₂(PEt₃)₂ sublimes already at 80 °C under vacuum, but signs of thermal decomposition are also seen already around 130 °C. The temperature window is still large enough to allow ALD usage as demonstrated by Ni metal film deposition using this precursor (another ALD2022 presentation).[2] Monophosphine adducts of Co have lower volatility than the Ni counterparts so that the temperature window between efficient evaporation and thermal decomposition is narrower.

In this presentation the thermal properties of cobalt and nickel halides with different neutral ligands including amines and phosphines are presented and compared. Among other things, it has been found that the thermal stability mainly increases from chlorides to iodides and the trend seems to be more pronounced with cobalt. The increasing molecular weight from chlorides to iodides causes only small decrease in volatility.

[1] K. Väyrynen (2019), Academic Dissertation, University of Helsinki, http://ethesis.helsinki.fi

[2] A. Vihervaara, T. Hatanpää, M. Ritala, ALD2022 abstract (to be published).

Dependencies of the growth rate per cycle (GPC) as a function of the temperature (T), and the so-called precursor saturation curves, are standardly observed for atomic layer deposition (ALD) and reflect its self-limiting nature. Such curves are considered as an indication of the ALD window and required to demonstrate the occurrence of ALD at various circumstances. Plotting GPC versus T is expected to be system independent, meaning that the fundamental features of the GPC curve should be similar between one reactor system and another. It looks feasible to underestimate the limits of the ALD window if there is no universal method to determine the GPC. Based on our experimental findings, we propose an underestimation of the ALD temperature window to occur in the literature study on the (self-limiting) chemisorption of Mo(CO)₆ while depositing molybdenum oxide (MoO_x) and molybdenum oxide compounds.

According to the literature, the use of Mo(CO)₆ as precursor and O₃ or O₂-plasma as co-reactant results in the self-limiting chemisorption of Mo(CO)₆ roughly between 150 and 175 ^oC. The GPC is mentioned to quickly decay for temperatures below 150 ^oC, with hardly any deposition at T<130 ^oC [1,2]. According to the experimental observations in this work, the deposition of MoO_x on a silicon substrate can indeed hardly start at T<130 ^oC, due to the practically inappropriate long incubation time. However, the deposition of MoO_x, once enabled at T=130 ^oC or higher, naturally continues at 80 ^oC with a reasonable GPC. The growth per cycle decreases only from 0.028 nm/min at T= 175 ^oC to 0.020 nm/min at T= 80 ^oC.

In our presentation, an extended look into these experimental findings including film characterization will be given. We further would like to draw attention of the ALD community to a need for verifying the methodology of ALD window determination for scientific experiments.

[1] J. Mater. Chem., 2011, 21, 705-710

[2] Appl. Mech. Mater., 2014, 492, 375-379

ZnO_1-xS_x alloys are wide-bandgap materials with high transmittance in the short-wavelength region. This makes them promising candidates as the n-type conducting buffer layer for different absorber types in thin-film photovoltaic devices. In this work, we studied the growth behaviour and composition of ternary ZnO_1-xS_x on Si wafer substrates at a fixed deposition temperature of 150°C using a thermal ALD process with diethylzinc (DEZ), H₂O and H₂S as precursors, and additionally a mixed thermal/plasma-enhanced ALD process with DEZ, H₂S and O₂ plasma. The growth rates and the composition of the films were analysed by spectral ellipsometry and EDX (energy dispersive x-ray spectrometry).

With thermal ALD we processed layers over a wide range of different H₂O:H₂S pulse ratios. The cycle ratio for H₂S is defined here as CR_H2S=k*100%/(k+l) with k being the number of H₂S pulses and l being the number of H₂O pulses in a cycle, respectively. In contrast to former works on ZnO_1-xS_x by ALD, not only the smallest possible bilayer period P=1*(k+l) for a certain CR_H2S was applied, but also larger bilayer periods P=a*(k+l) were examined (e.g. a=1,2,…,7 for CR_H2S=50% (k:l = 1:1) à P=2,4, …,14).EDX measurements showed a significant influence of the bilayer period on the sulphur content in the resulting films. Furthermore, we observed a linear relationship between the resulting S/(S+O) ratio in the film and the number of H₂S pulses for a fixed number of H₂O pulses, and vice versa. Therefore, it is possible to estimate the S/(S+O) ratio for a given pulse ratio by extrapolation. We discuss a few exceptions to the linear behaviour, e.g. for H₂O:H₂S ratio 1:1, 1:l and k:1.

ZnO_1-xS_x films grown in a mixed thermal-/plasma-enhanced ALD process showed a very different layer growth in comparison to the thermal deposition. With the purely thermal process we deposited layers with S/(S+O) ratios in a wide range from about 10 to 90 % applying moderate H₂S/(H₂O+H₂S) pulse ratios, even a higher S-content in the layers was achieved compared to the theoretically expected one. In contrast, by application of oxygen plasma (O₂Pl) as reactant for the ZnO cycle, a very high H₂S pulse fraction was necessary to reach S/(S+O) ratios >30%. With an O₂Pl:H₂S ratio of 1:24 (CR_H2S=96%) a S/(S+O) ratio of 60 % was measured.

Abstract:

Chromium oxide coatings are technologically important due to their good mechanical, chemical, magnetic, catalytic and optical properties. It has been shown that the properties of Cr₂O₃ coatings can be considerably changed if it is doped with an appropriate element. Chromium containing ternary oxide thin films have previously been grown by chemical vapor deposition or magnetron sputtering methods resulting in hard (hardness 31 GPa) [1]orsuper-hard (40 GPa) coatings [2] or showing good photocatalytic properties [3]. In this work chromium containing ternary compound (Cr-Ti-O) thin films were grown on Si (100) substrates using two different precursor combinations: CrO₂Cl₂ -CH₃OH:TiCl₄-H₂O, and Cr(thd)₃-O₃:TiCl₄-O₃by atomic layer deposition offering precise control over the concentration of dopant elements in the composition. Film density, roughness, phase composition, refractive index, hardness and Young’s modulus were studied in variation of Ti concentration in the thin films. Thin films with average thickness of 115 nm deposited using CrO₂Cl₂-CH₃OH and TiCl₄-H₂Oprecursors exhibited crystalline α-Cr₂O₃ (eskolaite) phase with density from 4.8 to 5.2 g/cm³. By increasing the number of TiCl₄ -H₂O cycles compared to CrO₂Cl₂-CH₃OH cycles, the total intensity of eskolaite reflections decreased. The films with average thickness of 45 nm deposited using Cr(thd)₃-O₃ and TiCl₄-O₃showed crystalline TiO₂ anatase phases with density from 4.0 to 4.7 g/cm³ (Fig. 1). As the growth rate of the Cr₂O₃ deposited using Cr(thd)₃-O₃ (50 pm/cycle) was remarkable lower than that for CrO₂Cl₂-CH₃OH process (70 pm/cycle), the change of the Cr(thd)₃-O₃:TiCl₄-O₃ cycle ratio from 1:30 to 1:1 only decreased the amount of the anatase phase in the film.

References:

[1] Bahrami, Amin, et al. "Structure, mechanical properties and corrosion resistance of amorphous Ti-Cr-O coatings." Surface and Coatings Technology, 374 (2019) 690-699.

[2] Mohammadtaheri, M., et al. "An investigation on synthesis and characterization of superhard Cr-Zr-O coatings." Surface and Coatings Technology, 375 (2019) 694-700.

[3] Chen, Yang, et al. "Synthesis of core-shell nanostructured Cr₂O₃/C@ TiO₂ for photocatalytic hydrogen production." Chinese Journal of Catalysis 42 (2021) 225-234.

Lead-containing precursors are an active field of study for the deposition of PbS (DOI: 10.1021/acs.chemmater.0c01887), PbO (DOI: 10.1149/1.2789286), and other lead-containing films.During our efforts to find new Pb(II) ALD precursors, we found that the acyclic diamido plumbylene lead(II) bis(N-tert-butyltrimethylsilylamide) (0) undergoes facile thermal decomposition into the homoleptic cyclic (alkyl)(amido) plumbylene (caaPb) bis(N-tert-butyl-2-aza-3,3-dimethyl-1-plumba-3-silacyclobutane) (1). Our interest in compound 1 pivoted toward the novel nature of the ligand and the potential to use it to synthesize new ALD precursors. We hypothesised that salt-metathesis between 1 and metal chlorides would produce similar heterocycles (caaM, M=metal). Lead is a good candidate for salt metathesis due to the thermodynamic driving force of the formation and precipitation of PbCl₂, driving the reaction to completion. Additionally, the high yield of the caaPb, cheap starting materials, and recyclability of the PbCl₂ by-product makes this synthetic methodology appealing, economical, and green. We tested the metathesis with ZnCl₂ because, if the salt metathesis is possible with zinc, then less electropositive metals should also undergo in the ligand exchange. During our test, we isolated PbCl₂ indicating that the metathesis was successful. Quantum chemical studies suggests the analogous reaction with other metal chlorides should also proceed.

There are early examples of titanium, zirconium, and hafnium complexes with a similar heterocyclic system (CH₂Si(Me₂)NSiMe₃), and the heterocyclic Cp₂TiCH₂Si(Me₂)NSiMe_3­ was even used as a titanium-ceramic CVD precursor.^1,2 However, the synthesis of these is unreliable.² We can now prepare similar metal complexes using ligand exchange from 1 with higher yields. Additionally, we expect to see a change in thermal properties after exchanging the trimethylsilyl group for a t-butyl group appended to the heterocycle, since the trimethylsilyl moiety is thermally active. A variety of complexes will be vetted to determine which work best as vapour-phase precursors with respect to volatility and thermal stability.

1. Planalp, R. P.; Anderson, R. A.; Zalkin, A. Dialkyl Bis[bist(trimethylsilyl)amido] Group 4A Metal Complexes. Preparation of Bridging Carbene Complexes by γ Elimination of Alkane. Crystal Structure of {ZrCHSi(Me)₂NSiMe₃[S(SiMe₃)₂]}₂. Organometallics1983 2, 16-20.

2. Simpson, S. K.; Anderson, R. A. Reaction of the Metallocene Dichlorides of Titanium(IV) and Zirconium(IV) with Lithium Bis(trimethylsilyl)amide. Inorg. Chem. 1981, 20, 3627-3629.

Currently there are few processes for the atomic layer deposition of iron metal thin films, and none without significant limitations—either by CVD or relying on very specific substrates with growth terminating after the initial substrate is covered. With an appropriate precursor, iron films could provide the basis for ferromagnetic coatings and to enable a range of iron alloy films – like stainless steel thin films – with desirable properties.

Geminal diaminosilane (gDAS) ligands are N,N’ κ² chelates with silicon bridging the coordinating nitrogen centres (Figure 1). We have previously reported a range of first row transition metal complexes with the gem-diaminosilane (^tBuNH)SiMe₂NMe₂ [2]. Transition metal gDAS precursors possess better volatility and thermal stability than their amidinate analogues while still avoiding problematic metal-oxygen bonds (compared to alkoxides) and with added stability from the chelate effect.

Here we report our initial study of new iron complexes which improve on the previously reported acetamidinate and α-imino alkoxide precursors [1]. We have synthesized, characterized, and evaluated the thermal characteristics of several Fe(II) and Fe(III) complexes containing monoanionic and dianionic gDAS ligands which show promise as precursors for the ALD of iron. Preliminary deposition studies using the previously reported Fe(gDAS)₂ monitored by quartz-crystal microbalance, as well as microscopy and initial compositional analysis will also be discussed.

[1] Kalutarage, L. C.; Martin, P. D.; Heeg, M. J.; Winter, C. H. Volatile and Thermally Stable Mid to Late Transition Metal Complexes Containing α-Imino Alkoxide Ligands, a New Strongly Reducing Coreagent, and Thermal Atomic Layer Deposition of Ni, Co, Fe, and Cr Metal Films. J. Am. Chem. Soc.2013, 135 (34), 12588–12591. https://doi.org/10.1021/ja407014w.

As high-k gate dielectrics continue scaling in the sub 1nm EOT range, the low-k interfacial layer that is pre-formed or formed during the dielectric deposition process is increasingly problematic since it reduces the effective k-value of the dielectric layer(1) . For Hafnium Oxide grown on silicon, the EOT is a combination of the silicon dioxide interface and the quality of the hafnium oxide. Consiglio (1) reported reduced thickness of the silicon dioxide interface layer with hydrogen peroxide gas when compared to ozone. In this study hafnium oxide was grown on HF cleaned silicon with either water vapor or H2O2 gas. The Keffective was improved by 27% H2O2 grown HfO2. The film grew 50% faster with H2O2 than water. Vt was 0.18V for H2O2 and -1.3V for water. Additional data will be presented to clarify if the Keffective was due to a thinner SiO2 interface layer or improved HfO2 film quality.

As the width of metallization wire in semiconductor device decreases, Cu interconnect and W word line does not scale down as fast as linewidth due to its high electron mean free path (EMFP). Meanwhile, molybdenum (Mo) and molybdenum carbides (MoC_x) are considered as promising materials for next-generation interconnect owing to its small EMFP andless resistivity size effect.In order to apply the Mo and MoC_xthin films to the 3-dimensional semiconductor devices, atomic layer deposition (ALD) of Mo and MoC_x has been required. However, Mo and MoC_x ALD using fluorine (F)-containing Mo precursor resulted in various issues such as oxide damage and deterioration of device property due to F contamination and HF by-product. Therefore, it is essential to developMo and MoC_x ALD processusing F-free Mo precursors.

In this study, δ-MoC and β-Mo₂C thin films were depositedby ALD using two different F-free precursors. Rapid thermal annealing (RTA) process was adopted to reduce MoC_x into Mo metal and to increase crystallinity of the film. β-Mo₂C showedimproved crystallinity after RTA process, whereas δ-MoC was reduced into metallic Mo (~46.5 μΩ·cm) after post-reduction annealing. The crystallinity, chemical state, impurity, and electrical characteristics of molybdenum carbide thin films will be compared before and after annealing process and reaction mechanism of the post-reduction annealing will be discussed.

Atomic Layer Deposition (ALD) is increasingly contributing to the energy field and more specifically to the engineering of solar cells. Its conformity enables deposition on nanostructured substrates and its low growth temperature allows the deposition on temperature-sensitive substrates such as perovskite.

Niobium oxide, Nb₂O₅, is a wide bandgap semiconductor that has been grown by different methods and has recently been used in solar cells. Its optical and electrical properties depend strongly of the technique used for its growth, opening access to a wide range of application, such as electron transport layer (ETL)or passivation layer [1,2]. It is also used for the doping of titanium oxide (TiO₂), a well-known ETL, to reach a better stability of the complete solar cell.

In this work, we first developed ALD-Nb₂O₅ from tris(diethylamido)(tert-butylimido)niobium (TBTDENb) and water. The growth was studied from 100°C to 200°C, on several substrates (Si, glass, FTO, ITO). The ALD process was first optimized using QCM (Quartz Crystal Microbalance). Annealing studies in different conditions (air/inert atmospheres, up to 600°C) were conducted to understand the evolution of Nb₂O₅ at temperatures which are relevant within the fabrication steps of the solar cell.

Then, niobium-doped titanium oxide was developed using tris(isopropoxide)titanium (TTIP) as titanium precursor, and Nb-doping was applied by a supercycle strategy and tuned by several methods (supercycle ratio, precursors sequences). For both materials, structural, chemical, electrical and optical properties were characterized by XRR (X-Ray Reflectivity), GIXRD (Grazing Incidence X-Ray Diffraction), ellipsometry, SEM (Scanning Electron Microscopy), XPS (X-Ray Photoelectron Spectrometry), XRF (X-Ray Fluorescence), spectrophotometry, 4-points probe.

Finally, those ALD-materials were implemented in perovskite solar cell architectures.

[1] Subbiah, et al (2019). Energy Technology, 8(4), 1900878. https://doi.org/10.1002/ente.201900878

[2] Macco et al, (2018). Solar Energy Materials and Solar Cells, 184, 98-104. https://doi.org/j.solmat.2018.04.037

Phase change memory (PCM) is attracting attention as the need for high-performance memory semiconductors that can process large amounts of data quickly increases with the development of the 4th industry, such as big data and artificial intelligence. At the same time, research on 3dimensions cross-point (3D X-point) memory using memory cells and selectors in the area where word lines and bit lines intersect in 3D to create a highly integrated PCM is also being conducted. However, the most important factor for high efficiency and integration of these PCM devices is to reduce the current required for device operation. Recently, studies have been actively conducted to increase the joule heating efficiency by increasing the specific resistance of the electrode. In this respect, carbon, which exhibits various resistivities (conductor (graphite, sp² bond) – insulator (diamond, sp³ bond)) ^{[1], [2]} depending on the state of atomic bonding, is an attractive material as a PCM electrode. So far, PVD (Physical Vapor Deposition) and CVD (Chemical Vapor Deposition) processes have been used to deposit carbon thin films. However, it is difficult to deposit a uniform carbon thin film on a three-dimensional structure with the conventional method, and it is difficult to control the physical properties of carbon. Therefore, in this study, a carbon process using ALD (Atomic Layer Deposition) was developed. Furthermore, the analysis of the carbon thin film properties according to the process conditions (reactant, temperature, pressure, time, etc...) was analyzed, and the corresponding mechanism was studied.

As the dynamic random access memory (DRAM) has been continuous downscaling, zirconium oxide, a high-k dielectric material, is widely used as a substitute for conventional SiO2. Because it has a relative high dielectric constant (15~22), a high breakdown field (15–20 MV/cm), a large band gap (5–7 eV), and a good thermodynamic stability up to 800 ℃ when in contact with the silicon substrate.¹

Among the various methods of deposition, ALD has been recognized as the leading candidate to process the high-k dielectrics owing to characteristics such as self-limiting reaction, excellent conformality, easy controlled thickness, and large area uniformity.²

The choice of precursor is very important because the film properties are totally different depending on which kinds of precursor is used. For instance, in case of halide ligand precursor, it has been widely used since there is no carbon incorporation into the films. However there is limitation like low volatility, particle issue and harmful byproduct. Also alkoxide ligand having a strong metal-oxygen bonding is required a relatively high temperature and it is easily decomposed by heat and contains high concentration of carbon impurity in the thin film. ³ Among various types of precursors, ZrO2 thin film using Cp ligand precursor have been widely used recently due to their good thermally stable allowing deposition temperature above 300℃ and excellent crystallinity. However it has disadvantage in that the growth rate is decreased due to its bulky size.⁴

Like this, several kinds of precursors have been studied for ALD of ZrO2 thin film but it is still needed for research on precursors. In this paper, new linked-Cp zirconium precursor ((linked CpZr(N(CH3)2)3) ) was synthesized for ALD of ZrO2 thin film and compared with the commonly used (CpZr(N(CH3)2)3) precursor in terms of growth characteristic and film property. For ALD ZrO2 process, oxygen plasma was used as the oxidant.

Various experimental methods, including spectroscopic ellipsometry, X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), X-ray Reflectivity (XRR) and density functional theory calculations (DFT)) were used for analyzing growth characteristic and film property. Additionally for the analysis electrical property, capacitance-voltage (C-V) and current-voltage (I-V) measurements were conducted.

¹ M. Balog, M. Schieber, Thin Solid Films 1977, 47, 109

² H. J. Kim, H. -B. -R. Lee, Thin Solid Films 517, 2563–2580 (2009)

³ J. Niinistö, K. Kukli, Advanced Engineering Materials vol. 11 223–234 (2009)

⁴ J. –S. Jung, S. –K. Lee, Thin Solid Films. Thin Solid Films.Volume 589 831-837 (2015)

Indium oxide (In₂O₃) has received much attention for a wide range of applications, including in optoelectronics as a transparent conducting material. We report ALD of In2O3 using a recently reported indium(III) triazenideprecursor¹ together with water. The deposition process was studied between 150 and 520 °C using a homebuilt crossflow ALD reactor at 50 hPa with indium(III) triazenide pulsed for 4s, and water pulsed for 3s as precursors and purging with nitrogen for 10s in between. The deposition process is self-limiting at ~1.0 Å/cycle and temperature window between 270 and 385 °C. XRD analysis shows that the films are polycrystalline with a preferred (222) orientation. The films are substoichiometric with have low levels of C impurities. Optical transmittance is high, >70% in visible light, and the resistivity was found to be low signifying high conductivity. These results are on par with the current state-of-the-art reported for thermal ALD of In₂O₃ from a formamidinate precursor.² We make a direct comparison between the indium triazenide precursor and the indium formamidinate precursor in the same reactor and show that they render very similar temperature windows and film quality optical transparency, and conductivity of the deposited films.

Refs.:

1. O’Brien et al. Chem. Mater., 2020, 32, 4481.
2. Kim et al. Chem. Eur. J.,2018, 24, 9525.

Ta₂O₅-Al₂O₃ nanolaminates are potentially applicable as high dielectric strength insulators [1], resistive switching media [2] and corrosion resistant coatings [3]. For most applications the mechanical properties of the thin films have a significance on the reliability of the devices. Double- and triple-layer amorphous Al₂O₃-Ta₂O₅ laminates with an overall thickness of about 70 nm were atomic layer deposited while changing the sequence of the layers from surface to substrate. Hardness and elastic modulus of the laminates were measured with nanoindentation and influence from the layer thickness and sequence on the mechanical properties were analysed.

Figure 1 depicts the change of hardness along the depth of the films for double-layered laminates. Layered structure caused an uneven rise in hardness with depth dependent on the sequence of Al₂O₃ and Ta₂O₅. Figure 2 describes the hardness of triple-layered laminates, showing an even incline for the laminate with a middle Ta₂O₅ layer surrounded with Al₂O₃ whereas the laminate with middle Al₂O₃ between Ta₂O₅ layers resembled the double-layered laminate behaviour. An additional quadruple-layered laminate was measured and showed a steady increase of hardness with depth. The elastic modulus for all the laminates was similar, steady along the depth and fell between 145 – 155 GPa.

It can be concluded that the hardness of layered films is affected by both the thickness of the layers and their sequence. Lowering a single constituent layer thickness can smoothen the difference in hardness along the film depth which could reduce internal stresses, defects, and delamination. Adding an extra layer to a thin film could increase its mechanical resilience in a cost-effective way.

References:

[1] B.V.T. Hanby, B.W. Stuart, M. Gimeno-Fabra, J. Moffat, C. Gerada, D.M. Grant, Appl. Surf. Sci. 492, 328 (2019) https://doi.org/10.1016/j.apsusc.2019.06.202.

[2]W. Song, W. Wang, H.K. Lee, M. Li, V.Y.-Q. Zhuo, Z. Chen, K.J. Chui, J.-C. Liu, I.-T. Wang, Y. Zhu, N. SinghAppl. Phys. Lett. 115, 133501 (2019) https://doi.org/10.1063/1.5100075

[3] B. Díaz, E. Härkönen, J. Swiatowska, A., V. Maurice, M. Ritala, P. Marcus, Corr. Sci 82, 208 (2014) http://dx.doi.org/10.1016/j.corsci.2014.01.024

Boron nitride (BN) has been considered a promising dielectric material for 2D material-based electronics. The atomic layer deposition (ALD) technique is a good method to grow conformal and ultrathin materials at relatively low temperatures. However, the growth mechanism is still not clear because the surface events with self-termination resulting from chemical reactions and physical dynamics are complicated. We aim to understand the growth mechanisms of the BN-ALD process from BCl₃ and NH₃ by experiments and simulations using the reactive force-field molecular dynamics (ReaxFF MD).

First, we investigated temperature profiles using a thermal ALD system described in Fig. (a). BN films were grown on the Si(100) surface in a hot walled horizontal reactor through the general ALD cycles as follows; BCl₃ exposure, Ar purge, NH₃ exposure, and Ar purge. Fig. (b) shows the thickness of the grown BN as a function of growth temperature measured on thechamber wall. We obtained a relatively small dependency of growth per cycle on the temperature in the range of 700-900 °C, suggesting ALD growth. The thin film property was characterized using IR spectroscopy as shown in Fig. (c). The remarkable peak at 1367 cm⁻¹ is originated from sp² BN associated with the in-plane stretching.These experimental results mean proper BN growth on the Si(100) surface by ALD. Second, we started to develop a new force field for the ReaxFF MD, which can simulate the surface events including complicated chemical reactions and physical dynamics. The initial force field is based on the two types of force field; boron nitride nanostructure formation and dynamical crack propagation in silicon, by Lele et al. and Buehler et al., respectively. The initial force field is mainly trained for boron chloride species and the reaction of BCl₃ on the OH-terminated Si(100) surface. We simulated one ALD cycle after thermal annealing to relax the initial system shown in Fig. (d). The Ar purge steps are simply replaced as the removals of gas molecules with thermal annealing. The simulations revealed some growth mechanisms: the -BCl₂ chemisorbed on the OH-terminated surface in the feed step, -NH₂ chemisorbed on the Cl-terminated surface in the reaction step, and HCl was generated with the chemisorption in these two steps. Our simulations are still one ALD cycle, however, these growth mechanisms are chemically straightforward.

We will show you thin film properties such as crystallinity and composition and compare them with the experimental results. The experimental and theoretical study of the ALD can be applied not only to the BN system but also to critical materials such as TiN and GaN in the future.

In thermoelectric materials, phase boundaries are crucial for carrier/phonon transport. Manipulation of carrier and phonon scatterings by introducing continuous interface modification has been shown to improve thermoelectric performance. In this work, a strategy of interface modification based on powder atomic layer deposition (PALD) is introduced to accurately control and modify the phase boundary of pure bismuth. Ultrathin layers of Al₂O₃, TiO₂, ZnO and Sb₂O_x are deposited on Bi powder by typically 1–20 cycles. All of the oxide layers significantly alter the microstructure and suppressed grain growth. These hierarchical interface modifications aid in the formation of an energy barrier by the oxide layer, resulting in a substantial increase in the Seebeck coefficient that is superior to that of most pure polycrystalline metals. Conversely, taking advantage of the strong electron and phonon scattering, an exceptionally large decrease in thermal conductivity is obtained. It’s worth noting that a substantial decrease of κ_tot from 7.8 to 5.7 W⸱m^-1⸱K^-1 was obtained with just 5 cycles of Sb₂O_x layers and a 16% reduction of κ_lat. Finally, a maximum figure of merit, zT, of 0.15 at 393 K and an average zT of 0.14 at 300–453 K were achieved after 5 cycles of Al₂O₃-coated Bi. The ALD-based approach, as a practical interfacial modification technique, can be easily applied to other thermoelectric materials to enhance their performance.

Scaling down of Si-based metal-oxide-semiconductor (MOS) has been main issue for semiconductor industry. With the demand for faster and smaller devices, channel length and thickness of gate dielectric have rapidly shortened. To prevent high leakage current for thin SiO₂, scaling down requires materials with higher dielectric constant. Among many materials, HfO₂ has been widely used for its superior properties. It has appropriate band offset with Si (~1.4 eV) and superb dielectric properties, such as high dielectric constant of about 25. For deposition of gate dielectric HfO₂, atomic layer deposition (ALD) technique has been widely used due to its superior characteristics, including excellent conformality, easily controlled thickness at atomic scale, atomic level composition control, large area uniformity, low impurity, and low growth temperature.

Many kinds of precursors, such as halides, alkoxides, alkylamides, and cyclopentadienyls have been studied for HfO₂ ALD process. Among these various type of precursors, alkylamide precursors are attracting attention due to their superior characteristics. They have relatively weak metal-N bonds, and weaker bond attributes to high reactivity in low temperature. Furthermore, the ligands of alkylamide precursors effectively prevent neighboring precursors from bonding to the metal center, which attributes to low melting point and high volatility. However, their weak metal-N bonds lead to decomposition of the precursors at high temperature. It limits ALD window and results in incorporation of impurities in film at high process temperature. In addition, the upper process temperature limit also prevents the formation of films with higher density and higher crystallinity.

To overcome this low thermal stability issue, heteroleptic precursors including cyclopentadienyl (Cp) ligand were introduced. Cp ligand has been widely employed for enhancing thermal stability and volatility of precursor. To take step further in terms of stability, we studied precursor with linking between Cp ligand and alkylamide ligand. By comparing the two Cp-based heteroleptic precursors with the linking between Cp ligand and alkylamide ligand and the one without, we investigated the impact of linked ligand structure on growth characteristics, chemical compositions, crystallinity, and electrical properties. Furthermore, density functional theory (DFT) calculations were introduced for revealing the reaction energy and pathways.

In the early stage of dynamic random access memory (DRAM) development, the thickness of the dielectric SiO₂ has been reduced in the Si based two dimensional structure to obtain high integration density and capacitance. However, as the DRAM devices have been continuously scaled down, thickness of SiO₂ reached a fatal limit of reaching the physical thickness at which leakage current due to tunneling effect occurs. To break through this, SiO₂ was replaced by high dielectric constant (k) materials[1]. Among the various high-k materials, ZrO₂ is one of the promising materials since it has good thermal stability, high dielectric constant (k~30) and wide bandgap (5.16-7.8 eV)[2,3]. However, when high-k thin film is deposited on TiN, which is widely used as an electrode, the electrical properties of the thin film are deteriorated such as increased leakage current density due to the interlayer formed between high-k film and TiN[4,5].

The interlayer formed between the substrate and dielectric film occurs due to the oxidation of the electrode during the film deposition process, which is affected by the potential barrier height of the oxidation reaction[6]. Since the potential barrier height depends on the work function of the metal electrode, proper selection of the electrode is required. Studies have been reported on replacing the top electrode with Pt, Au, Ag, etc.[6], but these have the disadvantages of being expensive, which makes difficult to easily apply to real industry. Recently, Ru is recognized as a promising material because of its good thermal stability, low resistance, high work function, and relatively inexpensive among noble metals[7]. In this study, we used sputtered Ru as a top electrode and compared effects of bottom electrode using Ru and TiN on ZrO₂ film properties. In addition, to find the proper oxidant to reduce the oxidation of the substrate, the film properties using the two oxidants (oxygen plasma and hydroperoxide) were also compared and analyzed. For ultrathin high quality ZrO₂ film, atomic layer deposition (ALD) has been used, which can secure atomic layer controlled ZrO₂ film with high conformality and high uniformity.

*Corresponding author: hyungjun@yonsei.ac.kr

[1] Chemical Reviews 99, 1823-1848 (1999)

[2] Reports on Progress in Physics 69, 327-396 (2006)

[3] Journal of Materials Science 53, 15237-15245 (2018)

[4] Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structure 27, 378 (2009)

[5] Solide-State Electronics 115, 133-139 (2016)

[6] Microelectronic Engineering 87, 98-103 (2010)

[7] Microelectronic Engineering 85, 39-44 (2008)

Boron carbide (B_xC) films enriched in ¹⁰B is a promising neutron converter material for the next generation solid-state neutron detectors. Upon neutron irradiation, ¹⁰B produce detectable particles by the nuclear reactions ¹⁰B + n⁰ → ⁷Li (0.84 MeV) + ⁴He (1.47 MeV) + ϒ (0.48 MeV) and ¹⁰B + n⁰ → ⁷Li (1.02 MeV) + ⁴He (1.78 MeV) that have 94% and 6% probability, respectively. Since the world has a ³He shortage, ¹⁰B solid-state detectors can potentially be a replacement to ³He detectors in large scale neutron facilities. For the next generation high resolution ¹⁰B detectors, films enriched in the ¹⁰B isotope must be deposited on pixelated sensor-chip substrates with high aspect-ratio morphologies. For such geometries, the currently employed magnetron sputtering technology is limited which highlight the need for alternative deposition techniques such as chemical vapor deposition (CVD) and atomic layer deposition (ALD). In addition, a low temperature process is required since the detector requires ohmic contacts which needs to be coated before converter layer deposition.

For coating high aspect-ratios, ALD would be the obvious choice, but the lack of ALD carbon precursors as well as carbide processes makes conformal continuous CVD processes a promising synthesis route. From our investigation, we report moderate temperature CVD of B_xC thin films on silicon substrates with 8:1 aspect-ratio morphologies, using triethylboron (TEB, ^natB(C₂H₅)₃) as single source CVD precursor. Step coverage (SC) calculated from the cross-sectional scanning electron microscopy measurements shows that films deposited at ≤450 °C were perfectly conformal (SC = 1). We attribute this to the low reaction probability at low substrate temperatures enabling more gas phase diffusion into the features. The quantitative analysis using time of flight elastic recoil detection analysis (ToF-ERDA) and X-ray photoelectron spectroscopy (XPS) reveals that the as deposited films at 450 °C deposition temperature are B rich carbide material i.e., around 80% Boron, 15% carbon and 5% impurities, which is a promising B/C ratio for neutron detector application if ¹⁰B isotope enriched TEB is used. The density of the material, measured by X-ray reflectometry measurements, varies from 1.9 to 2.28 g/cm³ for the deposition temperatures varying from 425 to 550 ⁰C in steps of 25 °C.

Since its discovery in 1959, vanadium dioxide (VO₂) is well-known for its Insulator-to-Metal Transition (IMT) around 70 °C just above room temperature (RT). Over the last decade, it has gained interest because of its potential applications for resistive switching systems, energy storage in lithium-ion batteries, supercapacitors, infrared detectors, or thermochromic materials¹. Due to its low-temperature process, Atomic Layer Deposition (ALD) offers a real industrial advantage in the growth of VO₂, allowing to run deposition with low energy consumption and therefore low cost, and on substrates that can be temperature sensitive.

In this context, vanadium oxide films were deposited on both silicon (100) and glass substrates at 240 °C from vanadium tri-isopropoxide (VTIP) and water as an oxidizing agent. Films were grown with different thicknesses, ranging from 30 nm to 120 nm, and then annealed for one hour under forming gas (95% Ar, 5% H₂) at a temperature ranging from 400 °C to 550 °C by 50 °C steps.

According to a structural analysis held by X-ray diffraction (XRD) and transmission electron microscopy (TEM) as-deposited vanadium oxide films were amorphous and change morphology to polycrystalline with an admixture of VO₂ and V₂O₅ phases having crystallites size reaching 300 nm after annealing. The elemental analysis was performed by RBS and XPS studies will also be presented for a deeper understanding of their composition and stoichiometry.

VO₂ films were also characterized electrically and optically using resistivity measurements, and spectroscopic techniques such as Raman, FTIR, ellipsometry. All experimental results show a reversible and reproducible IMT around 70 °C for films annealed at 500 °C. For electrical measurements, the resistivity decreases down to 10^-1 Ω.cm upon IMT temperature. The optical ones show a particularly interesting result by FTIR spectroscopy on VO₂ films on silicon substrate, with an optical absorbance of 0.1 OD at RT which increases up to 0.9 OD above IMT temperature on a wavenumber range extending from 4000 to 1000 cm^-1. The phase transition is also observable in the UV-visible range on VO₂ films on glass substrate and is correlated with the appearance or disappearance of low-temperature VO₂ peaks in Raman spectroscopy.

In conclusion, vanadium oxide films deposited by ALD were analyzed between RT and 100 °C and present promising properties tunable with temperature, especially in the IR range, and paves the way for future applications in thermochromic materials.

1. Atomic layer deposition of vanadium oxides: process and application review. Mater. Today Chem. 12, 396–423 (2019).

Yttrium oxide (Y₂O₃) thin films have been a subject of intensive research, particularly as an alternative high-k dielectric, due to its high relative permittivity (e_r ~17-20) and large bandgap of 5.5 eV. Furthermore, its high chemical resistivity and mechanical strength facilitate its application as a passivation layer in many fields.

Atomic layer deposition (ALD) has been established as one of the most promising techniques for the growth of high-quality layers for the above mentioned applications. As the precursor selection plays a pivotal role in an ALD process, the development of compounds with an optimal combination of volatility, reactivity and thermal stability is needed.

Besides the established Y precursors, such as β-diketonates or cyclopentadienyls, the all-N coordinated class of the amidinates¹ and guanidinates² has emerged as a promising class for the fabrication of yttrium-based materials. Recently, the structurally related formamidinate (famd) ligand class (N,N’-dialkyl-formamidinato) was explored, exhibiting a high volatility, reactivity, and stability.³

In this study, we focused on the systematic molecular engineering of Y formamidinates to fine–tune the physicochemical properties through a variation of the alkyl side chains. Among the four evaluated precursors, the tertbutyl-substituted [Y(^tBu₂–famd)₃] showed an increased thermal stability and high reactivity towards H₂O, as revealed by thermal analysis and density functional theory (DFT) studies, respectively.

Subsequently, a thermal ALD process for Y₂O₃ using H₂O as co-reactant was developed, yielding dense fcc-Y₂O₃ films on Si substrates with smooth topography. Owing to the appealing structural, compositional and morphological quality of the layers, the process was used to deposit Y₂O₃ as a dielectric component in metal insulator semiconductor (MIS) capacitor structures.⁴ The promising electric properties set a strong platform for in-depth studies to understand the interplay between precursor chemistry, ALD process development and integration in capacitor structures.

Literature:

¹ de Rouffignac, P., Park, J.-S., Gordon, R. G., Chem. Mater., 2005, 17, 19, 4808-4814.

² Mai, L., Boysen, N., Subasi, E., de los Arcos, T., Rogalla, D., Grundmeier, G., Bock, C., Lu, H.-L, Devi, A., RSC Adv., 2018, 8, 4987.

³ Boysen, N., Zanders. D., Berning, T., Beer, S. M. J., Rogalla D., Bock, C., Devi, A., RSC Adv., 2021, 11, 2565-2574.

Titanium dioxide (TiO₂) is a wide bandgap (3.0-3.20 eV) material that presents high transparency in the visible range; besides, it has a high refractive index. TiO₂ doping is widely investigated to induce structural and electronic modifications that may improve electronic properties and photocatalytic activities. Due to their additional electron compared to Ti and their atomic radius slightly above and below the one of Ti, Nb and V are the most promising candidates among the potential dopants of TiO₂. In this context, this work aims to synthesize Nb/V-doped TiO₂ films by atomic layer deposition (ALD) with various dopant concentrations and growing conditions.

Nb/V-doped TiO₂ films were deposited on Si wafer and glass substrates with water as oxidizing agent using a shared ALD window for precursors niobium (V) ethoxide or vanadium (V) oxytriisopropoxide and titanium isopropoxide, respectively. The dopant content was adjusted by the dopant ratio (R_Nb or R_V : range from 0 to 1), which is the number of dopant cycles over the total number of ALD cycles. Films were then annealed in N₂ or forming gas (95% N₂ and 5% H₂) at different temperatures of 400, 500, and 600 °C for 1 hour. To investigate the characteristics of films, various techniques, such as spectroscopic ellipsometry, spectrophotometry, GI-XRD, Raman, FTIR, HRTEM, XRR, and four-probe resistivity were used for the as-deposited and annealed Nb/V-doped TiO₂ films.

It was observed that when introducing Nb or V into the TiO₂ matrix up to a doping ratio of 0.025, the thickness increased slightly compared to the undoped TiO₂ film, and then gradually decreased as the dopant ratio increases further. For both dopants, the refractive index and the electronic density of films were found to evolve similarly and in the opposite way to their thickness. In the as-deposited Nb-doped TiO₂ films with R_Nb = 0-0.025, the presence of the crystalline anatase phase was identified, but the peak intensity decreased and progressively shifted as R_Nb increased. In the case of the as-deposited V-doped TiO₂ film, only an amorphous phase was obtained which transform into crystalline phase with annealing.

After annealing, the optical transmittance and electrical resistivity of anatase phase crystallized films were measured for Nb/V-doped TiO₂ films. With increasing R_Nb, the films showed transmittance ranging from 60 to 80% in the visible range, which increase with the conductivity (∼10² S.cm^-1). The optical and electrical properties of V-doped TiO₂ film are also performed and will be further discussed.

This work highlights the significant role of Nb and V dopants in tuning the structural, optical, and electrical property of TiO₂.

Since the reaction of atomic layer deposition (ALD) strongly depends on surface property, understanding of surface reaction mechanism between substrates and precursors is essential to predict and interpret thin film deposition in ALD. Recently, the many researchers have studied chemical reactions of ALD using density functional theory (DFT) and physical reaction using molecular dynamic (MD). However, DFT is suit for calculating a few of molecule adsorption but not simultaneous multiple adsorptions, and MD is not proper for a large scale simulation due to huge computing resource and long calculation time. In addition, although the steric hindrance effect between the molecules is an important physical factor for simulation of ALD, but it was not considered as a main variable for simulation. In this study, by adopting several assumptions and approximations, we developed a simple simulation method to understand physical steric hindrance effects of ALD precursors by using Monte Carlo (MC) without huge computing resources and applied the method to study surface reaction mechanism of ALD and area selective ALD (AS-ALD). We calculated the areal coverage of precursor on a specific surface used in ALD and AS-ALD using the MC simulation with a random adsorption model. The simulation results show high consistency agreement with experiment data. Based on the 2D model developed first, we extended the MC simulation to 3D system, and obtained reliable results in bulky precursor systems. The simulation method developed in this study could be applied to many of ALD precursors and AS-ALD inhibitor systems just by using a laptop computer.

With its high permittivity and large band gap of E_g = 5.7 eV, HfO₂ is of significant interest for high-κ dielectric layers and excellent resistance ratios as well as fast switching speeds are reported for HfO₂ based memristor devices. To realize these microelectronic components, the deposition of pinhole-free thin films with an excellent uniformity and conformality is required. Owing to its saturative growth characteristics, atomic layer deposition (ALD) intrinsically fulfills these requirements. Plasma-enhanced ALD (PEALD) furthermore allows thin film depositions at low temperatures, as required for future flexible electronics, and is thus the method of choice. To enable the beneficial features of ALD, the physico-chemical properties of the precursor need to be carefully fine-tuned in order to optimize its thermal stability, volatility as well as reactivity. Looking for Hf precursors that fulfill these prerequisites, we investigated new heteroleptic Hf complexes. Starting from the parental Hf dialkylamide, we introduced a chelating formamidinate ligand that stabilizes and shields the Hf center, thereby increasing the thermal stability of the resulting complexe while retaining adequate reactivity and volatility. The resulting bis-(dialkylamido)-bis-(formamidinato) Hf(IV) precursor was synthesized on a multigram scale, structurally characterized and evaluated by thermogravimetric analysis and subsequently employed as a precursor for the deposition of HfO₂ in a PEALD process. Using an oxygen plasma, the typical ALD characteristics of precursor saturation, linearity and ALD temperature window were demonstrated with constant growth of 0.7 Å per cycle from 125 °C to 200 °C on Si(100) substrates. The resulting HfO₂ films were further characterized by RBS/NRA, XPS and AFM, revealing the formation of pure and smooth HfO₂ layers. Compared to our previous work [1] with a closely related guanidinate precursor, shorter plasma pulses were sufficient to achieve ALD growth, preventing the formation of an interfacial SiO₂ layer, as revealed by transmission electron microscopy (TEM). By coating polyimide (PI) foils at temperatures as low as 150 °C, there is a potential of implementing the presented low-temperature HfO₂ PEALD process into the development of flexible electronic devices in the future.

[1] D. Zanders et al., ACS Appl. Mater. Interfaces, 106 (2019) 28407

PE-ALD of SiO₂ has been well studied in the past years, and has found a lot of applications as e.g. for advanced lithograpy, dielectric material in microelectronic devices, photovoltaics and optical applications. For these purposes where a high film quality is required in combination with a limited throughput, the use of more expensive precursors is justified. However, in recent years PE-ALD of SiO₂ found its way towards high surface applications as e.g. the encapsulation of OLEDS or deposition of anti-reflective coatings where precursor cost is a much bigger issue.

In this work, the PE-ALD growth characteristics of SiO₂ was investigated using four precursors in a different price setting: Bis(diethylamino)silane (BDEAS), (3-Aminopropyl)triethoxysilane (APTES),Tetraethyl orthosilicate (TEOS) and Hexamethyldisilazane (HMDS) (in order of high to low precursor cost price), in combination with oxygen plasma as a reactant.

Although it was possible to deposit SiO₂ with all four ALD processes, a significant difference in growth rate was observed. The growth rate varied from 1 Å/cycle for BDEAS to 0.2 Å/cycle for the TEOS-based process (Figure 1). Further the conformality of the four processes was investigated using macroscopic lateral structures with an equivalent aspect ratio (EAR) of 22. While BDEAS and APTES showed an excellent conformality, TEOS and HMDS showed a coated EAR of only 1 and 2.5 respectively. Here the coated EAR is defined as the EAR where the deposited film thickness equals 50% of the original film thickness.
In order to enhance the growth rate and conformality of these last two processes, TiO₂ subcycles were added to the TEOS and HMDS- based processes using titanium(IV)isopropoxide (TTIP) as a low-cost precursor. Both Ti_xSi(1-x)O₂ films showed a strong improvement in growth rate, from 0.2 to 0.5 Å/cycle for the TEOS-based process and from 3.5 to 6.2 Å/cycle for the HMDS-based process, using a 1 TTIP :9 TEOS/HMDS subcycle ratio. Besides the growth rate, also a strong improvement in conformality was observed as is shown in Figure 2. The coated EAR for the TEOS-based process increased from 2.5 to 17.5 and for the HMDS based process from 1 to 12 (in both cases for a 1:9 (TiO₂:SiO₂ subcycle ratio)).
These results could be promising when (no-pure) SiO₂ films are required for high surface area applications where a low precursor cost is relevant.

Cobalt(II) fluoride, among some other first-row transition metal fluorides, is an excellent cathode candidate for lithium-ion and sodium-ion batteries. These metal fluorides have high theoretical potentials and high energy densities compared to the current oxide-based cathodes. However, although ALD is recognized in lithium-ion and sodium-ion battery research in general, the number of ALD processes for fluoride-based battery materials has remained small. Particularly, these first-row transition metal fluorides have been lacking ALD processes. In this work, we present the first ALD process for cobalt(II) fluoride.

CoF₂ was deposited using CoCl₂TMEDA* and NH₄F as precursors. The films were characterized with XRD, EDS, XPS, ToF-ERDA, SEM, and AFM. The deposition temperature range studied was 180–300 °C, and all the deposition temperatures resulted in tetragonal CoF₂, as measured by XRD. Also XPS and ToF-ERDA confirm the films to consist of CoF₂. The impurity content measured with ToF-ERDA is low. Most importantly, the amounts of Cl and N, which are constituents of the precursors, are low, for example 0.53 and 0.08 at-% for a film deposited at 250 °C. The combination of a chloride-based precursor and NH₄F thus seems to work well in this case. The saturation of the growth per cycle with respect to pulse and purge lengths was confirmed at 250 °C, and the growth per cycle saturates to ~1.1 Å. In addition, the film thickness is linearly dependent on the number of applied cycles. Like many ALD metal fluorides, these films are rough, as seen in SEM and AFM. At a deposition temperature of 250 °C, for example, a ~60 nm film has a roughness of 12.6 nm.

*TMEDA: N,N,N′,N′-Tetramethylethylenediamine

The wurtzite form from Indium Nitride has semiconducting behavior that, combined with advantageous electron transport properties, has offered potential applicability of this material in the field of electronics and light-emitting diodes.¹ The InN thin film is preferentially obtained using atomic layer deposition (ALD) techniques, with lower temperatures that are beneficial for the crystal stability and enable the utilization of ammonia precursor at such conditions.² In the process, trimethylindium is a well-known In precursor that might undergo partial decomposition in the gas phase,³ resulting in the CH₃ radical release that is expected to affect the initial steps of the reaction mechanism. In addition, this precursor usually leads to high level of carbon impurities that is inconvenient for large scale production. Therefore, this project aims at fully understanding the mechanistic aspects of the adsorption and reaction-related processes leading to the In-rich layer formation for InN crystal growth by using a multiscale approach that comprises density functional theory (DFT) and Kinetic Monte Carlo (KMC) computational techniques.

The atomic-scale periodic calculations are carried out within the Perdew–Burke–Ernzerhof (PBE) level of theory in VASP⁴. Initially, the thermal effects are neglected to enable a more extensive investigation of the relevant reaction pathways. Such effects are then added to approximate the model to the actual experimental conditions (T= 593 K, 1 bar). The outcome indicates that the initial decomposition steps whether they occur in the gas phase or at the surface both lead to the final product, i.e.methylindium (MI) chemisorbed at the hcp site and ethane. However, the N- rich layer leads to an activation of this process that is found to facilitate the precursor dissociation at the surface environment, with an activation enthalpy Δ^‡H <20 kJ/mol for TMI/DMI displacement towards other stable adsorption sites. In a second step, the hydrogen atoms are subsequentially removed through the involvement of two additional precursor molecules to produce low-weight hydrocarbons. The results also suggest the origin of the carbon impurities to be the CH₃ radical released during the process that in turn can form a strong chemical bond with the N-rich layer. From this point, all data necessary for the KMC simulation at the mesoscale level are acquired, which shall also be presented.

References:

1-Bhuiyan, A. et al. J. Applied Phys. 94, 2779 (2003).

2-Deminskyi, P. et al. J. Vac. Sci. Technol. A 37, 020926 (2019).

3-Hwang, J. et al. J. Electrochem. Soc., 155, 2 (2008).

4-Gresse, G. et al. Comput. Mat. Sci. 15, (1996).

Phosphides are interesting materials among wide variety of scientific fields. InP and GaP are the most profound semiconductors with application in photovoltaics or electronics. Several ALD depositions of metal phosphides were already presented. Apart from one reported reaction of P(NMe₂)₃ with GaMe₃ affording GaP,^[1] PH₃ or its alkylated analogue tBuPH₂^[2]are usually used as a source of P^-III ion. Nevertheless, except high toxicity of PH₃, depositions using this precursor are often accompanied with lower reactivity supplemented by plasma activation^[3] or laser irradiation.^[4] PH₃ may also decompose during the deposition causing high content of phosphorus resulting in non-linear growth.

Trialkylsilyl ligand were utilized in ALD of As,^[5] Sb^[6] Se^[7,8] and Te^[7] several times. Its electropositive nature generates negative charge on the deposited atom ensuring high reactivity, while keeping good volatility and thermal stability. For example, tris(trimethylsilyl)phosphide is a favorite precursor for metal phosphide quantum dots. It is fairly volatile and can be distilled even at laboratory pressure. Interestingly, no ALD deposition using this class of precursor has been reported for so far. Therefore, this work aims to investigate tris(trialkylsilyl)phosphides as a potential ALD precursors. Preparation of these molecules will be discussed along with structure/thermal properties relationships and selected phosphides will be tested for thin film deposition in ALD.

[1] E. Graugnard, V. Chawla, D. Lorang, C. J. Summers, Appl. Phys. Lett.2006, 89, 211102.

[2] N. Otsuka, J. Nishizawa, H. Kikuchi, Y. Oyama, J. Vac. Sci. Technol. A Vacuum, Surfaces, Film.1999, 17, 3008.

[3] A. V. Uvarov, A. S. Gudovskikh, V. N. Nevedomskiy, A. I. Baranov, D. A. Kudryashov, I. A. Morozov, J. P. Kleider, J. Phys. D Appl. Phys.2020, 53, 345105.

[4] M. Yoshimoto, A. Kajimoto, H. Matsunami, Thin Solid Films1993, 225, 70–73.

[5] T. Sarnet, T. Hatanpää, M. Laitinen, T. Sajavaara, K. Mizohata, M. Ritala, M. Leskelä, J. Mater. Chem. C2016, 4, 449.

[6] V. Pore, K. Knapas, T. Hatanpää, T. Sarnet, M. Kemell, M. Ritala, M. Leskelä, K. Mizohata, Chem. Mater.2011, 23, 247–254.

[7] T. Hantapää, V. Pore, M. Ritala, M. Leskelä, Electrochem. Soc.2009, 25, 609–616.

[8] R. Zazpe, J. Charvot, R. Krumpolec, L. Hromádko, D. Pavliňák, F. Dvorak, P. Knotek, J. Michalicka, J. Přikryl, S. Ng, V. Jelínková, F. Bureš, J. M. Macak, FlatChem2020, 21, 100166.

1 Ke Sun. etal. Energy Environ. Sci. 2012, 5 (7), 7872–7877

Atomic Layer Deposition (ALD) is a booming technology to deposit thin films and has been applied in several fields. This technique is based on surface-chemical reactions, and relies on the gas phase transport of metal containing molecules into a reaction chamber. However, not any molecule is suitable to be used as precursor, as they must be thermally robust while being sufficiently volatile and chemically labile to react with the surface functional groups. Organometallic chemistry offers an infinite set of options to design new efficient precursors, though predicting their volatility and reactivity in the ALD chamber remains tricky.^[1] Establishing a new method to assess the physical and chemical properties of complexes would grant access to new ALD precursors and a better understanding of surface reactions.

This communication focuses on the development of new efficient gallium precursors to be used in the ALD of oxygen-free gallium-containing sulfide thin films.^[2] A series of gallium complexes with chelating nitrogen based ligands (guanidinate, amidinate and triazenides) were synthesized and characterized (NMR, XRD) thanks to modular procedures.^[3,4] To assess the thermal stability of the reagents and shed light on their transport in ALD, thermal analysis (TGA, DSC) were realized under N₂ and vacuum to mimic transport conditions. Finally, reactivity studies in solution of established ALD precursors and synthesized complexes provide an insight of surface reactions which might take place in an ALD chamber.

References:

[1] S. E. Koponen, P. G. Gordon, S. T. Barry, Polyhedron2016, 108, 59–66.

[2] N. Schneider, M. Frégnaux, M. Bouttemy, F. Donsanti, A. Etcheberry, D. Lincot, Materials Today Chemistry2018, 10, 142–152.

[3] A. P. Kenney, G. P. A. Yap, D. S. Richeson, S. T. Barry, Inorganic Chemistry2005, 44, 2926–2933.

[4] S. Dagorne, R. F. Jordan, V. G. Young, Organometallics1999, 18, 4619–4623.

The detailed characterization of Bis(tri-isopropylcyclopentadienyl)strontium(Sr(iPr₃Cp)₂) precursor was conducted to understand growth mechanism in atomic layer deposition of SrTiO₃ thin films. First, the adsorption behavior was studied using an in-situ attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR). The band of the Sr(iPr₃Cp)₂ spectrum on the Ge crystal surface was identical to that of the spectrum measured in the gas phase, but peak intensity was different. In addition, the absorption characteristics studies were carried out over the Ge crystal temperature in the range of 40∼100^oC. Upon increasing the temperature, a reduction of absorption was observed. Second, to investigate the volatility of Sr(iPr₃Cp)₂, vapor pressure curve was determined using thermogravymetric analysis. This method can reduce both precursor amount and time required for the vapor pressure measurement. Furthermore, molecular simulation was applied to explain the interrelationship between those properties evaluated in this study and molecular structure. Our study to understand the detailed behavior of precursor can be provided as useful information for optimization of ALD process and new precursor design.

The first syntheses of transition metal nitrides are derived from metallurgical processes and consisted in nitriding a powder of the metal or one of its oxides (Oyama, 1992; Toth, 1971). These nitrides were synthesized under severe conditions (T > 1200 °Cz and had low specific surfaces (Marchand et al., 1996). Subsequently, the development of catalytic applications requiring nitride powders with large specific surfaces made it necessary to use processes with more moderate temperatures (700 °C to 900 °C) (Marchand et al., 1996, 1999). These “softer” synthetic routes have developed and been applied to the formation of nitrides from transition metal sulfides.

Among the transition metal disulfides, molybdenum disulfide (MoS₂₎is one of the most widely studied materials in recent years to synthesis molybdenum nitride due to its availability (E. R. Braitwaite & J. Haber, 1994). MoS₂ has a natural two-dimensional structure with the sandwich-like S-Mo-S layers serving as building blocks, in which the atoms in the layer are bonded with strong covalent bonding, while the layers are packed together with weak interlayer forces (Jariwala et al., 2014; Li & Zhu, 2015). In recent years, the emergence of 2D materials and the increase in the demand of metal nitrides nanosheets due to their remarkable physical and chemical properties such as high electrical conductivities, catalytic properties, energy storage, and conversion efficiency aroused a particular interest (Khazaei et al., 2013; Wang & Ding, 2018; Xiao et al., 2017, 2018; Zhong et al., 2016). Thereby, some research group(Sun et al., 2018)^,(Cao et al., 2020) have demonstrated the transformation of MoS₂ nanosheets exfoliated from bulk material into molybdenum nitride nanosheets with ammonia and urea reactive heat treatments, respectively.

In this work, we proposed a method to transform a well-controlled uniform MoS₂ thin film deposited by Atomic Layer Deposition into molybdenum nitride (MoN_x) nanosheets via an ammonia reactive heat treatment at 700 °C supported by in-situ reflectance measurements and ex-situ Raman and X-Ray Photoelectron spectroscopy characterizations.

This work reports on the application of Self Plasma Optical Emission Spectroscopy (SPOES) to perform Process Gas Analysis (PGA) to monitor and optimize thin film atomic layer deposition (ALD) process, carried out using metalorganic precursors and water vapor or oxygen plasma as oxidizers. Depositions were carried out at 150℃ using N₂ or Ar as carrier and purging gas. The ALD cycle consisted of four steps: (a) metalorganic precursor pulse, (b) purge, (c) oxidant pulse, and (d) purge. Purge times varied in the range of 2 – 120 seconds to find the optimal value based on the PGA results. We performed SPOES PGA using Broadband Plasma Emission Monitoring (2B-PEM) of an inverted magnetron-based plasma reactor attached to the ALD process exhaust. The miniature plasma reactor can operate at pressures 7.5e−7 Torr to 7.5 Torr. The sensor signals derived from SPOES data react instantly to composition changes revealing trace amounts of constituents of the process material. We demonstrate how real-time process diagnostics, pump-down monitoring, process condition recognition and end-point detection, all taking place in parallel, facilitate ALD process yield maximisation and reaction by product residue in thin films prevention. Furthermore, the gas analysis technology demonstrated does not require additional (differential) pumping systems to perform analyses.

Boron nitride (BN) is a binary compound of boron and nitrogen alternatively linked which can exist in various crystalline forms, all of them analogous to the carbon allotropes. In each of its forms, BN presents interesting properties that make it a useful material in different applications, especially remarkable in nanotechnology.

The aim of this work is to develop and characterize a new Atomic Layer Deposition (ALD) process for the growth of boron nitride thin films. So far, this has been achieved by using ammonia as the nitrogen providing precursor in combination with a boron halide at temperatures above 400^oC. The interest is to obtain boron nitride thin films at reaction temperatures below 275^oC. In this work, ammonium carbamate (NH₄[H₂NCO₂]) and boron tribromide (BBr₃) are used as precursors. NH₄[H₂NCO₂] is a solid at room temperature which easily decomposes giving a CO₂/NH₃ mixture that can substitute ammonia, making the laboratory work safer and simpler. This work focuses on the analysis of the dependence of the growth rate of the process on its different parameters, aiming to optimize the deposition and predict the thickness of the grown films.

The thin films deposited using this process have been analyzed by means of X-Ray reflectivity, X-Ray photoelectron spectroscopy and electron energy loss spectroscopy among other techniques to conclude that the material deposited is amorphous boron nitride with a 1:1 stoichiometry.

The major challenge in metal ALD is the reduction process to yield the metallic target film from metal source that usually comprise of metal cation surrounded by anionic ligands. Existing strategies involve using reducing agent or, counterintuitively, oxidizing agent as second co-reagent. Using reducer as co-reagent, e.g. H₂, can lead to an abbreviated cycle, and reduced rate of deposition, when stable metal hydrides are not available. In process using oxidizing agent, e.g. O₂, transient metal oxide surface may be generated and that can greatly facilitate noble metal ALD [1]. In this scenario the reduction of metallic center is a result of precursor decomposition at this catalytic surface. The higher growth rate is thus achieved, because the metallic film forms also in the processes of combustion of ligands by the oxidizing agent, however with the danger of surface poisoning and oxide deposition.

Thus, the second co-reactant role is crucial. It is used to eliminate the surface bound species of the metal pulse and, at saturation, to produce a reactive overlayer – the catalytic oxide surface, which is characteristic for a particular noble metal. We therefore first investigate the thermodynamics to understand the self-limiting surface chemistry of the oxidizing co-reagent. We use density functional theory (DFT) to establish order of reactivity as a function of temperature and pressure of noble metals (Ru, Rh, Pd, Ag, Os, Ir, Pt, Au) to form oxides. Next, we examine the thermodynamics of ALD process that includes the transient generation of noble metal oxide.

Finally, we investigate reaction steps involved in the metal nucleation on the example of Pt ALD from MeCpPtMe₃ and O₂. We evaluate whether the nuclei of the catalytic surface can be formed during the O₂ co-reactant pulse, i.e. when oxidizing agent is introduced into the ALD chamber to combust hydrocarbon-based ligands into the volatile by-products (e.g. CO₂, H₂O). We discuss the possibility of production of transient surface bound OH groups predicted in previous study [2] and other by-products, e.g. CH₄ identified in the experiment [3]. The factors that facilitate nucleation are examined. This will allow to propose appropriate reagents and chemical processes to control and improve efficiency of the atomic layer deposition of noble metals.

1. The Journal of Chemical Physics, 2017, 146, 052822.
2. Langmuir, 2010, 26, 9179-9182.
3. Physical Chemistry Chemical Physics, 2018, 20, 25343-25356.

The thermal ALD of TiN is a well-documented process. The common precursors combination is TDMAT/NH₃, for thermal as well as for plasma-enhanced ALD (PE-ALD). However, PE-ALD offers a larger variety of co-reactant (N-sources): N₂, H₂ and N₂/H₂. Building the ALD window of TiN using the recipes from our manufacturer, the PE-ALD processes exhibit a longer cycle duration associated to a slower GPC regarding to the thermal ALD one. This is unexpected since the plasma should enhance the production of reactive species and then promote the deposition process. The GPC should therefore be higher. The aim of this work is to optimize the PE-ALD recipes using two different N-sources: NH₃ and N₂. In addition to shorten the deposition duration, the effects of those N-sources as well as their dilution in Ar and the plasma power on the final properties of the films are also studied. The influence of those parameters has been monitored by in situ characterizations (ellipsometry and optical emission spectroscopy, SE and OES, respectively) and by ex situ characterizations (morphology, composition, crystalline structure and electric properties).

The recipe parameters are adjusted to limit the recombinations of the reactive species generated between the remote plasma source to the substrate. For instance, a large Ar dilution of both N₂ and NH₃ limits the film growth (lower GPC). A gas ratio of 1:1 for N-source and Ar flow is set to the optimal values. This is correlated with the OES spectra presenting that in diluted condition, the intensity of the Ar pics is predominant compared to the N-sources one. This suggests that high Ar dilution hinders the generation of N-reactive species leading to a reduction of the number of reactive species involved in the deposit growth and then, to lower GPC, mainly for N₂ plasma. The applied power (50 to 300 W) has no significant effect on the GPC with NH₃ plasma while, for N₂-based plasma, the GPC is maximum at the highest power. This is consistent with the expected low reactivity of N₂ (inert without plasma activation) as compared to highly reactive NH₃. Note that using NH₃, a thermal contribution cannot be discarded. Those results indicate that producing less active species facilitate their transport by limiting recombinations.

The films grown from both N-sources have a similar roughness, composition and morphology. However, the conductivity, conformality on high aspect ratio substrates (1:25) and the growth rate are better using NH₃-based plasma. The N₂ plasma process exhibits an acceptable film quality and it should be considered as well since it uses a non-harmful gas.

To satisfy the requirements for applications of silicon nitride (SiN_x) at the highly scaled logic and memory devices, high quality films (i.e., low contamination, low roughness, etc.) with high conformality at low temperatures are demanded. To meet the stringent requirements for SiN_x applications, plasma enhanced atomic layer deposition (PEALD) is being extensively investigated as the deposition technique. However, the minimization of plasma damage and lowering the process temperature still remain as issues for SiN_x PEALD processes. In this work, the properties of PEALD SiN_xfilms deposited at a low process temperatureof100^oCwithdi(isopropylamino)silane(DIPAS)andN₂plasmaexcitedbyveryhighfrequency(VHF,162MHz)capactivelycoupledplasma (CCP) sources with a floatingmulti-tile type electrode and a conventional diode type electrode are investigated and compared in addition to the plasma characteristicsofbothplasmasources. ThePEALDSiN_xfilmdepositedusing thefloating multi-tile electrodeexhibitedhighergrowthrate(~0.6Å/cycle),morestoichiometricfilm (N/Si~0.98), andhigherconformalityinatrenchcomparedtothosedepositedbytheconventional VHF CCP.Inaddition,theimprovedelectricalcharacteristicscouldbeobtainedwiththefloatingmulti-tile electrode.The improved properties are believed to be related to the higher plasma density, higher radical density, and lower ion energy bombarding the substrate observed for the multi-electrode type through the enhanced power coupling efficiency between the pairs of multi-electrodes in the plasma source.

Atomic layer deposition (ALD) is often chosen over other techniques for thin film growth or surface modification because of its conformality, which originates from the self-terminating nature of the reactions used [1]. It is of paramount importance to understand how the conformality in high-aspect-ratio (HAR) surface features evolves with time and depends on process parameters and chemistry. Many simulation frameworks are available to model ALD growth in HAR features [2,4]: diffusion–reaction models, Monte Carlo models and ballistic models. Most simulation frameworks work in the free molecular flow conditions, where kinetic information of the reactions can be extracted from an experimental thickness profile by a simple slope method [3,4].

As seen in Figure 1, the thickness profile characteristics such as the half-thickness penetration depth x_50% and the slope at this half-thickness penetration depth, strongly depend on the Knudsen number in other diffusion conditions than free molecular flow. While x_50% can be taken as a simplistic measure for “conformality”, the slope carries information of reaction kinetics, specifically of the sticking coefficient. To make interpretations on kinetics from experimental thickness profiles, understanding the flow conditions is of central importance. Specifically, assuming Knudsen flow when it is in reality not valid, would lead to incorrect (too high) interpretation of the sticking coefficient.

Recently [4], we showed that the way the process parameters affect the simulated thickness profile in HAR structures, depends on the diffusion regime: free molecular flow (Knudsen number Kn >> 1) has partly different trends than transition flow (Kn ~1). In this work, we extend the simulations to continuum conditions (Kn << 1). In addition to the previously used 1d diffusion–reaction model [4,5], in this work we also use computational fluid dynamics (CFD) calculations to investigate the processes in 3d.

[2] V. Cremers et al., Applied Physics Reviews, 6(2), 021302, (2019).

[3] K. Arts et al., J. Vac. Sci. Technol. A, 37, 030908, (2019).

[4] J. Yim and E. Verkama et al., Phys. Chem. Chem. Phys., in press,

https://doi.org/10.1039/D1CP04758B

[5] M. Ylilammi, O. Ylivaara, and R.L. Puurunen, J. Appl. Phys., 123, 205301, (2018).

The application of photocatalytic materials in air treatment and air-borne pollutant remediation has been well established. Photocatalytically active materials used in urban areas on a commercial base can include paints, tiles, and concretes which mainly utilize TiO₂ as the photocatalyst material. Feasibility studies demonstrated the potential of those materials for the use of mineralizing organic compounds into carbon dioxide, water and corresponding mineral acids. Of special interest is the decomposition of nitrogen oxides with a main focus on NO and NO₂. A review of the photocatalytic effectiveness for outdoor applications is not always possible due to the simultaneous variation of other parameters, such as traffic density and weather conditions. Therefore, for the simulation of air pollutants numerical models for the release calculation in the atmosphere, the transport of pollutants in the gas phase, and the interaction with solid surfaces are used.

The determination of the deposition rate of available photocatalytic materials is currently limited to examining those embedded in matrices and formulations, e.g. for concrete surfaces, roof tiles or plaster, since these currently have the largest commercial proportion of photocatalytically active products. An evaluation of other, especially vacuum-based, coating processes for the deposition of TiO₂ layers has not yet taken place. To create a basic understanding of the essential process parameters influencing photocatalytic NO oxidation thermal atomic layer deposition is used.

In this work suitable process windows for the deposition of photocatalytic TiO₂ are identified and evaluated with a main focus on the precursor combinations TiCl₄/H₂O, TiCl₄/O₃, TTIP/H₂O, TTIP/O₃ and TiCl₄/TTIP. Process parameters affecting the crystallinity of the TiO₂ layers and thus the photocatalytic effectiveness and the process-related layer properties with different process temperatures and layer thicknesses are determined.

For a cross-method and unified comparison of the TiO₂ layer properties the photocatalytic oxidation of methylene blue in aqueous solution and the degradation of nitrogen monoxide in a photoreactor are compared with each other as well as the kinetic modelling which was observed and simulated during NO degradation.

In high-vacuum chemical vapour deposition (HV-CVD), heated substrates are exposed to continuous precursor fluxes from orifices in the precursor delivery system. Owing to the HV conditions, the precursor trajectories are ballistic, therefore the fluxes can be evaluated analytically. By blocking of individual effusion orifices, a range of different combinations of precursor fluxes can be explored in a single synthesis, which vastly accelerates the process optimization for the desired film properties. This approach is referred to as combinatorial deposition.

Moreover, due to HV conditions the probability of gas-phase collisions between precursor molecules is negligible, and thus, the chemical reactions occur strictly on the substrate surface. This characteristic allows for ALD-like synthesis of the films at a higher growth rate than ALD, provided that the substrate temperature is below the pyrolysis threshold of the precursors. In one of the previous works of our group (Reinke et al., J. Phys. Chem. C, 2015, 119, 50, 27965–27971), we presented HV-CVD of TiO₂ using ALD precursors TTIP and H₂O at typical ALD temperatures 175-225°C, achieving growth rates of up to 2nm/min. The growth was demonstrated to follow ALD chemistry and kinetics. The combinatorial HV-CVD allowed to extract a range of kinetic parameters of the precursor system under study, thus proving the HV-CVD highly valuable as a tool for the fundamental study of ALD processes. Moreover, it shows that ALD processes can be adapted in the HV-CVD system for higher growth rates.

Owing to the negligible gas-phase reactions, in the CVD mode of HV-CVD, highly crystalline films are attainable. The HV-CVD-grown films rival the quality of molecular beam epitaxy results, achieved at lower temperatures, and providing much-improved process scalability and cost-efficiency. Our recent efforts have been focused on the epitaxial growth of BaTiO₃ on SrTiO₃-buffered substrates for integrated electro-optical devices using Ba(iPr₃Cp)₂, TTIP and O₂ as precursors (Borzì et al, Materialia, 2020, 14, 100953). Employing the combinatorial mode of growth, we have established the optimal precursor fluxes for the correct stoichiometry, validated by Rutherford backscattering and Elastic recoil detection elemental analysis. We also demonstrated the films to be epitaxial by XRD as well as <0.5nm root mean square roughness, as shown by AFM, both of which being crucial for the application.

Gallium oxide (Ga₂O₃) is an emerging ultrawide-bandgap (UWBG) semiconductor attracting significant interest for high-power and high-frequency electronics that features comparable electrical properties (larger bandgap ~4.9 eV, high dielectric constant 6-8 MV/cm) to wide-bandgap GaN and SiC. However, growing device-level high-quality (Ga₂O₃) films have been mainly possible at high substrate temperatures (700 – 1000 °C) using complex reactor settings, which substantially increases the production cost and limits the application space.