ALD2019 Session AF2-TuM: ALD Precursors II

From the very beginning of ALD, metal chlorides have been extensively used precursors because of their high thermal stability and reactivity as well as low cost. However, not all metal chlorides are volatile enough to be used in ALD. Cobalt and nickel chlorides for example have a polymeric structure and therefore low volatility. To circumvent this limitation, we have recently explored a strategy of adducting these chlorides with chelating ligands. With properly selected diamines the adduct complexes are monomeric and thermally stable enough to be volatilized and used in ALD. These compounds have been earlier been used for deposition of CoO [1] and intermetallics Co₃Sn₂ and Ni₃Sn₂ as a novel class of ALD materials [2].

Herein, we have studied ALD of nickel nitride and nickel thin films using NiCl₂(TMPDA) (TMPDA = N,N,N’,N’,-tetramethyl-1,3-propanediamine) as the metal precursor and tert-butylhydrazine (TBH) as a nitrogen containing reducing agent. The films were grown at low temperatures of 190–250 °C. This is one of the few low-temperature ALD processes that can be used to grow Ni₃N and Ni metal on both insulating and conductive substrates. X-ray diffraction showed reflections compatible with either hexagonal Ni or Ni₃N. ToF-ERDA analyses showed that the films were close to stoichiometric Ni₃N. Upon annealing the films at 150 °C in 10% forming gas their nitrogen content lowered down to 1.2 at.%, and the nonmagnetic nitride films were converted to ferromagnetic Ni metal.

[1] K. Väyrynen, K. Mizohata, J. Räisänen, D. Peeters, A. Devi, M. Ritala, M. Leskelä, Chem. Mater.2017, 29, 6502.

[2] K. Väyrynen, T. Hatanpää, M. Mattinen, K. Mizohata, K. Meinander, J. Räisänen, J. Link, R. Stern, M. Ritala, M. Leskelä, Adv. Mater. Interfaces2018, 12, 1801291.

Main group chemistry has been integral in the field of ALD since its inception, particularly concerning the extensive work on the most fundamental ALD process: the deposition of aluminum oxide from trimethylaluminum (TMA) and water. It is surprising that strategically functionalized analogues to this fundamental precursor have not been more extensively studied in order to optimize subsequent deposition processes. A simple, rationally designed precursor can regulate surface chemistry and limit decomposition pathways, permitting straightforward reactivity. Our focus is to enable the deposition of low-impurity aluminum nitride by ALD.

This presentation will describe the study of a set of simple molecules that are novel as ALD precursors, based on a substituted TMA framework (Figure 1a-c). Tris(dimethylamido)aluminum(III) is a direct analogue of TMA with methyl groups replaced by dimethylamido ligands. It demonstrates excellent ALD precursor behaviour with a 1 Torr temperature of 67 °C and no decomposition visible by TGA, likely attributed to the stabilization of the Al centre by the coordinative donation from the bridging NMe₂. This precursor deposited aluminum oxide and aluminum oxy-nitride films with carbon impurities lower than those reported using TMA.¹ From this, it was deduced that metal-nitrogen bonds in precursor design were beneficial to stabilizing the metal centre, and importantly, in lowering carbon contamination in deposited films by eliminating direct Al-C bonds.

Using the same TMA framework, AlH_x(NMe₂)_3-x (x = 1, 2) compounds were prepared due to the potential for the reducing ability of the hydride ligand and a maintained absence of metal-carbon bonds. These compounds were also found to be volatile and thermally stable, and thus suitable as potential AlN precursors.

This led to the study of further nitrogen-containing ligands: the family Al(^MeNacNac)_x(NMe₂)_3-x (x = 1, 3) similarly showed excellent volatility and thermal stability. Interestingly, an Al species with one monoanionic and one doubly anionic ^MeNacNac ligand was discovered (Figure 1d) and presented this novel reduced ligand for potential use in ALD precursors. This compound was synthesized via reduction of the ^MeNacNac ligand, and its thermal characterization and reactivity, as well as that of contrarily unsuitable ALD candidates using these ligands, will also be discussed in this presentation.

[1] Buttera, S. C.; Mandia, D. J.; Barry, S. T. Tris(dimethylamido)aluminum(III): An Overlooked Atomic Layer Deposition Precursor. J. Vac. Sci. Technol. A Vacuum, Surfaces, Film.2017, 35 (35), 1–128.

Lead(II) sulfide (PbS) is one of the oldest known semiconductors. It has a narrow band-gap of 0.4 eV that can be widened by quantum confinement effects. Owing to these properties, PbS is a recognized material in IR detection and quantum dot (QD) based devices such as light emitting diodes (LEDs) and photovoltaics. Photovoltaics are also central to our study. The aim is to develop an ALD PbS process that can be applied in halide perovskite solar cells.

ALD processes for PbS exist at least since 1990^[1], however we discovered that known ALD PbS processes are incompatible with halide perovskites such as CH₃NH₃PbI₃ due to their extremely low thermal budget^[2]_. To our knowledge the lowest temperature at which PbS can be deposited with existing ALD processes is 130 °C.^[3] CH₃NH₃PbI₃ film degrades completely at that temperature in our ALD reactors. Thus, we set out for a new PbS process that would work at even lower temperatures. From our previous studies on ALD of PbI₂ we already knew one lead precursor suitable for low temperature ALD: lead(II) bis[bis(trimethylsilyl)amide] or Pb(btsa)₂.^[4] Recently Bačić et al. reported another suitable candidate: lead(II) rac-N²,N³-di-tert-butylbutane-2,3-diamide or Pb(dbda).^[5] With these lead precursors and H₂S as a co-reactant we developed two new processes capable of depositing PbS even below 100 °C.

Pb(btsa)₂-H₂S process produces uniform, high-quality, crystalline PbS films in the 65 – 155 °C range. TOF-ERDA composition analysis revealed exceptionally high purity for such low deposition temperatures. The process exhibits two regions with temperature independent growth per cycle values (GPC) of 0.50 Å and 0.15 Å at 65 – 80 °C and 115 – 155 °C, respectively. The GPC saturates rapidly with respect to both precursor pulses and is independent of the purge duration and number of deposition cycles. CH₃NH₃PbI₃ films survive the PbS deposition at the lowest deposition temperatures and studies of perovskite stability after PbS deposition are ongoing.

With Pb(dbda) film deposition occurs in an even wider temperature range of 45 – 155 °C. However, XRD measurements show that at 95 °C and above, the films contain crystalline impurities. Nevertheless, high-quality polycrystalline PbS films can be deposited at temperatures as low as 45 – 75 °C. We are currently conducting studies on the influence of process parameters on the PbS film properties.

[1] Leskelä et al. Vac. 1990, 41, 1457–1459.

[2] Zardetto et al. Sustain. Energ. Fuels 2017, 1, 30-55

[3] Nykänen et al. J. Mater. Chem. 1994, 4, 1409.

[4] Popov et al. Chem. Mater. 2019, Article ASAP. DOI: 10.1021/acs.chemmater.8b04969

Nickel oxide (NiO), a wide band gap p-type oxide semiconductor, is of interest for applications in solar energy conversion,¹ electrocatalysis,² and as a tunnel barrier for metal/insulator/metal (MIM) diodes for infrared energy harvesting.³ ALD is an ideal technique for the highly conformal, uniform thin films needed for these applications. We develop a new process for ALD of NiO using Ni(^tBu2DAD)₂ and O₃­.

ALD growth of metallic Ni was demonstrated recently using Ni(^tBu2DAD)₂ and tert-butylamine.⁴ ALD of metallic Co and cobalt oxide have been reported using Co(^tBu2DAD)₂ with formic acid and O₃, respectively.^5,6 Taking a similar approach, we deposit NiO in a Picosun Sunale R-150 using N₂-purge-separated cycles of Ni(^tBu2DAD)₂ held at 150°C and an O₃/O₂ mixture of ~10%. NiO films were characterized using grazing-incidence x-ray diffraction (GIXRD), x-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), variable angle spectroscopic ellipsometry (VASE), and p-NiO/n-Si heterojunction diodes.

A plot of GPC vs. temperature for depositions using a 5/30/4/30 s Ni(^tBu2DAD)₂/N₂/O₃/N₂ pulse sequence shows a clear ALD "window" of constant GPC = 0.12 nm/cycle from 185-200 °C (Fig. 1) in which Ni(^tBu2DAD)₂ and O₃ pulses saturate after ~5 s and ~4 s, respectively (Fig. 2). Below the temperature window, condensation of Ni(^tBu2DAD)₂ occurs resulting in uncontrolled growth with 1.2% C and 0.7 % N impurities (XPS) at 150 °C. Above the window, from 200-230 °C, desorption of Ni(^tBu2DAD)₂ occurs, followed by decomposition of Ni(^tBu2DAD)₂ above 230 °C on the chamber walls upstream of the wafer, both contributing to decreased GPC. Inside the temperature window, for saturating films deposited at 200 °C, GIXRD shows randomly-oriented polycrystalline cubic NiO (Fig. 3), consistent with XPS elemental analysis showing a Ni:O ratio near unity, without any detectable C or N impurities. VASE reveals that refractive index (~2.4) and band-gap (~3.8 eV) are close to bulk values while AFM RMS roughness was 0.6 nm for an 18 nm thick film (Fig. 4). p-NiO/n-Si diodes show highly asymmetric (I₊/I_-,MAX ≈ 10⁴) behavior. All results point towards an as-deposited transparent p-type semiconducting NiO. Further growth studies, characterization, and electrical measurements will be discussed at the meeting.

1. S. Seo et al., Nanoscale 8, 11403 (2016).
2. K. L. Nardi et al., Adv. Energy Mater. 5 (2015).
3. N. Alimardani et al. Appl. Phys. Lett. 105, 082902 (2014).
4. M. M. Kerrigan et al., ACS Appl. Mat. & Interfaces. 10, 14200 (2018).
5. J. P. Klesko, M. M. Kerrigan, and C. H. Winter, Chem. Mater. 28, 700 (2016).
6. J. Kim et al., Chem. Mater. 29, 5796 (2017).

Lead-containing nanomaterials like PbS quantum dots and CH₃NH₃PbI₃ perovskite films have received a frenzy of attention because of their unique optoelectronic properties. Atomic layer deposition (ALD) is an especially attractive method for their fabrication as it can deposit uniform, conformal films over large areas at low temperatures with subnanometer thickness control. However, current lead-containing ALD processes have rarely been adopted due to their poor thermal budget, the low volatility of known precursors, and the small library of thermally robust lead complexes. Our interest in the deposition of lead-containing materials prompted us to study amido ligands because of their low cost, flexibility, and proven effectiveness of other group 14 diamido complexes as precursors for ALD.

In a pilot study,^[1] we identified the thermally stable (dec. >100 °C) and highly volatile (subl. 40 °C/0.01 Torr) 5-membered N-heterocyclic plumbylene rac-N²,N³-di-tert-butylbutane-2,3-diamido lead(II) (1), however it failed to cleanly react with our sulfur-source tert-butylthiol below its decomposition temperature. To explore safe (i.e., less reactive) co-reagents for ALD, we needed to push the thermal stability even higher through a mechanistic understanding of the decomposition pathways of diamido plumbylenes. With few reports on the mechanisms of their thermolysis, we decided to undertake a combined synthetic and theoretical approach to this problem. The two most promising examples in the literature, 1 and Pb[N(SiMe₃)₂], suggested both cyclic and acyclic scaffolds were viable.

We postulated two dominant thermolysis pathways for plumbylenes: formation of metallacycles by π-overlap of N-substituents and the 6p⁰ orbital of lead (A); and concerted pericyclic cycloreversion reactions (B). We found experimentally that known acyclic complexes were dominated by A, and 5-membered cyclic derivatives were found to be dominated by pathway B so we designed bespoke ligands to block thermolysis through two strategies: coordinatively saturating lead with a Lewis base to hinder A, and rigid ligands to block B. Ab initio calculations were performed on a model set of plumbylenes to help reveal the way forward to indefinitely robust and volatile plumbylenes for deposition, and provide a better understanding of the unique ligand-metal cooperation required for lead.

[1] Bačić, G., Zanders, D., Mallick, Devi, A., Barry, S.T.Designing Stability into Thermally Reactive Plumbylenes. Inorg. Chem.2018, 57 (14), pp 8218-8226.

The deposition and the properties of rare earth oxide thin films have been extensively studied for applications as protective coatings, optics and microelectronics. Sc₂O₃ thin films have recently attracted a strong interest for application as gate dielectrics for metal-oxide-semiconductor field-effect transistors (MOSFET) because they have a dielectric constant of 14, a band gap of 6.3 eV and excellent thermal stability. Despite its advantages, very few studies have investigated Sc₂O₃ on Si. We therefore investigated growth characteristics of Sc₂O₃ films deposited by atomic layer deposition (ALD) from Sc(cp)₃ and a novel liquid heteroleptic cyclopentadienyl containing Sc precursor, which are thermally stable (˃300 ˚C), volatile, with a good reactivity to H₂O. The depositions were done in a hot-wall cross-flow-type ASM Pulsar^® 3000 reactor connected to a Polygon 8300 platform.

The film growth properties, such as growth per cycle, impurity concentrations, and thickness uniformity on 300 mm Si (100) wafers were studied. A growth rate per cycle of 0.46 Å/ cycles and 0.52 Å/cycle was obtained at 225 ˚C with novel Sc based precursor and Sc(cp)₃, respectively. The Sc₂O₃ film non-uniformity was about 1 % and 5 % with novel Sc heteroleptic precursor and Sc(cp)₃, respectively. Sc₂O₃ films based on the novel Sc precursor were stoichiometric, with a low C impurity content of <1%, while films based on Sc(cp)₃ precursor exhibited ~4.5 % of residual carbon impurity level under similar processing conditions.

Electrical measurements were done on planar TiN gated MOS capacitor (MOSCAP) with high k thickness ~8 nm. Sc(cp)₃ based Sc₂O₃ exhibited a leakage current density of 1.5 x 10^-4 A/cm² and an effective work function (eWF) of 4.05 eV for an equivalent oxide thickness (EOT) of 3.43 nm at 1.0 V, which is significantly lower than TiN eWF (~4.8 eV) indicating considerable positive fixed charges incorporated in deposited Sc₂O₃ film. While novel heteroleptic Sc based Sc₂O₃ films exhibited much lower leakage current density of 1.0 x 10^-8 A/cm² and close to TiN eWF of 4.77 eV for an EOT of 3.15 nm at 1.0 V. It was shown that good quality and low defects Sc₂O₃ films can be obtained by using the novel heteroleptic Sc precursor.