Atomic layer deposition (ALD) is a technique by which surface-based reactions
between a precursor molecule (often metal-containing) and a co-reactant (e.g. H2O, O2
or H2) yield highly uniform and conformal (ultra-)thin films. The precursor and co-reactant
are each delivered in the gas phase, separated from one another by inert gas purge steps.
The self-limiting nature of these surface-based reactions allows the thickness of the film
to be controlled solely by the number of ‘precursor – purge – co-reactant – purge’ cycles.
This nano-scale control of film thickness allows for a large number of applications such
as in flat panel displays, fuel and solar cells, and microelectronic devices.
The first goal of this project was the pursuit of new low-temperature methods for
main group elemental ALD using silyl-substituted precursor molecules. The second goal
of the project was the development of alternative methods for thin film deposition of group
4 (M = Hf, Zr) oxides that would encourage effective (ie. void-free) filling of narrow (<20
nm) trenches in high-aspect-ratio (HAR) substrates. This thesis includes the development
of new precursor molecules and reaction pathways, evaluation of precursor molecular
structures, thermal stability, volatility and solution reactivity, identification of appropriate
experimental conditions for ALD, and characterization of the resulting thin films.
ALD of elemental antimony was achieved on hydrogen-terminated silicon (H-Si)
and SiO2/Si substrates using Sb(SiMe3)3 (2-1) and SbCl3 in the temperature range 23-
65 °C. The mirror-like films were confirmed to be composed of crystalline antimony by
XPS (for the film deposited at 35 °C) and XRD, with low impurity levels and strong
preferential orientation of crystal growth relative to the substrate surface. To the best ofour knowledge, this is the first example of room temperature thermal ALD (with
demonstrated self-limiting growth) of a pure element. Film growth at 35 °C exhibited a
substrate-enhanced mechanism, characterized by faster film growth for the first ~125
ALD cycles, where substantial deposition is occurring on the original substrate surface
(GPC (growth-per-cycle) = 1.3 Å on SiO2/Si, and 1.0 Å on H-Si), and slower film growth
(GPC = 0.40 Å on SiO2/Si, and 0.27 Å on H-Si) after ~125 cycles, once much of the initial
substrate surface has been covered. Films deposited using 500-2000 ALD cycles were
shown to be continuous by SEM. The use of less than 250 cycles afforded discontinuous
films. However, in this initial growth phase, when deposition is occurring primarily on the
original substrate surface, in-situ surface pre-treatment by Sb(SiMe3)3 or SbCl3 (50 x 0.4
or 0.8 s pulses), followed by the use of longer precursor pulses (0.4 or 0.8 s) during the
first 50 ALD cycles resulted in improved nucleation. For example, on H-Si, a continuous
6.7 nm thick film was produced after initial pre-treatment with 50 x 0.8 s pulses of SbCl3,
followed by 50 ALD cycles using 0.8 s pulses. The use of longer ALD pulses in the first
50 ALD cycles following surface pre-treatment is likely required in order to achieve
complete reactivity with an increased density of reactive surface sites.
Boranes featuring bulky silyl or sterically unencumbered trimethylgermyl groups,
in combination with a stabilizing dimethylamido group, were pursued as potential
precursors for ALD of elemental boron. This ALD process would employ a boron trihalide
(BX3; X = F, Cl, Br, I) co-reactant, exploiting the thermodynamically favourable formation
of tetrel-halide bonds as a driving force. This work required multistep syntheses of alkali
metal silyl reagents, {(Me3Si)3Si}Li(THF)2 (3-1) and tBu3SiNa(THF)n (3-2), and previously
un-isolated [Me3GeLi(THF)2]2 (3-3), and their reactions with B(NMe2)Cl2 (3-4). The boranes {(Me3Si)3Si}2B(NMe2) (3-8) and (tBu3Si)(Me3Ge)B(NMe2) (3-12) were
successfully synthesized, spectroscopically and crystallographically characterized, and
assessed for their suitability as precursor molecules for boron ALD. Unfortunately,
deposition attempts on SiO2/Si using 3-8 and BCl3 led to minor film growth (GPC = 0.01
Å). However, the enhanced volatility and solution-state reactivity of 3-12 in comparison to
3-8 makes it a promising precursor candidate for future ALD reactor studies. Attempts to
synthesize bis(trimethylgermyl)(dimethylamido)borane from the 2:1 reaction of 3-3 with
3-4 resulted in the formation of a lithium trigermylamidoborate,
{(Me3Ge)3B(NMe2)}Li(THF)2 (3-13).
ALD can give rise to uniquely uniform and conformal ultra-thin films, but voids often
remain after attempted filling of narrow high-aspect-ratio trenches. To achieve void-free
trench-filling, ALD (or CVD; chemical vapour deposition) methods which deposit a
flowable material are desirable, and this initially-deposited material can be converted to
the target material (e.g. a metal oxide) by post-deposition annealing, or potentially at the
deposition temperature on a longer timescale than flowable behaviour. In this work, a new
HfO2 ALD process was developed using [Hf(NMeEt)4] in combination with β-
hydroxyisovaleric acid (IVA; CMe2(OH)CH2CO2H) that introduces the potential for
flowability. Self-limiting growth was observed at 100, 250, and 300 °C, with a GPC of 1.5-
2.2 Å on planar SiO2 substrates. Films deposited at 100 °C consisted of amorphous HfO2
with significant carbon content (~22 at%) and <1 at% nitrogen. After annealing at 400 °C
in vacuo for 1 hour, the films were composed of amorphous HfO2 with low (<1 at%) carbon
content. The co-reactant in this work, β-hydroxyisovaleric acid, was chosen with the
following criteria in mind: Firstly, the carboxylic acid group may be sufficiently acidic to cleave linkages between chemisorbed hafnium species and the surface, generating
flowable non-surface-tethered hafnium carboxylate species (with low volatility, so that
they are not lost from the surface). Secondly, the hydroxyl groups of the ligands can
potentially serve as reactive sites for the hafnium precursor delivered in the next pulse.
Thirdly, fairly low-energy pathways should exist for deprotonated IVA ligands to
decompose to generate oxide or hydroxide ligands with release of volatile by-products,
such as CO2 and isobutene, or acetone and ketene. Experiments to gain insight into the
nature of reactivity between [Hf(NMeEt)4] and IVA and a structurally similar carboxylic
acid are described. These include (a) solution-state reactions between [Hf(NMeEt)4] and
IVA or pivalic acid (tBuCO2H), with formation of [H2NMeEt]2[Hf(κ2-O2CCH2CMe2OH)2(κ2-
OC(O)CH2CMe2O)2] (4-1) and [Hf5(μ3-O)4(κ2-O2CtBu)4(μ-O2CtBu)8] (4-2), (b) attempted
ALD using pivalic acid (which lacks a hydroxyl group) in place of IVA, and (c) roomtemperature
solution reactions between [Hf(NMeEt)4] and 4 equiv. of IVA to form 4-1,
followed by removal of volatiles, heating at 200 °C, and volatile/soluble product analysis
by NMR spectroscopy and GC-MS headspace analysis. Compounds 4-1 and 4-2 were
isolated and crystallographically characterized.
Heteroleptic zirconium(IV) complexes were designed, synthesized,
spectroscopically and crystallographically characterized, and assessed as potential
precursor molecules to enable flowable ZrO2 ALD. The envisaged process would operate
via the deposition of oligomeric, one-dimensional chains that, if grown untethered on a
functionalized substrate, could potentially flow to the bottoms of trenches. Reaction of
one equivalent of H2(acen), H2(cis-Cyacen) or H2(trans-Cyacen) with [Zr(CH2SiMe3)4] at
room temperature afforded [Zr(acen)(CH2SiMe3)2] (5-1), [Zr(cis-Cyacen)(CH2SiMe3)2] (5-2) or [Zr(trans-Cyacen)(CH2SiMe3)2] (5-3), respectively (acen = C2H4(NCMeCHC(O)Me)2;
Cyacen = 1,2-C6H10(NCMeCHC(O)Me)2). These alkyl compounds are trigonal prismatic
in the solid state, and whereas 5-1 and 5-3 decomposed without sublimation above
120 °C (5-10 mTorr), 5-2 sublimed in >95% yield at 85 °C (5-10 mTorr). However, heating
solid 5-2 at 88 °C under static argon for 24 hours resulted in extensive decomposition to
afford H2(cis-Cyacen) and SiMe4 as the soluble products. Compound 5-2 reacted cleanly
with two equivalents of tBuOH to afford [Zr(cis-Cyacen)(OtBu)2] (5-4), but excess tBuOH
caused both SiMe4 and H2(cis-Cyacen) elimination. The 1:1 reaction of H2(acen) with
[Zr(NMeEt)4] did not proceed cleanly, and 8-coordinate [Zr(acen)2] (5-5) was identified as
a by-product; this complex was isolated from the 2:1 reaction. A zirconium amido
complex, [Zr(acen)(NMeEt)2] (5-6) was accessed via the reaction of 1 with two equiv. or
excess HNMeEt, but decomposed readily in solution at room temperature. More sterically
hindered [Zr(acen){N(SiMe3)2}2] (5-7) was synthesized via the reaction of [Zr(acen)Cl2]
with two equivalents of Li{N(SiMe3)2}, but was also thermally unstable as a solid and in
solution at room temperature. Compounds 5-1 to 5-3, 5-5 and 5-7 were
crystallographically characterized. / Dissertation / Doctor of Science (PhD) / The focus of this work is the development of new processes to deposit ultra-thin
films of main group elements and transition metal oxides. The deposition method utilized
in this work is atomic layer deposition (ALD), which involves the use of a precursor
molecule (which contains the target element) and a co-reactant. These chemical species
must be appropriately reactive towards one another, and display adequate volatility and
thermal stability. The feasibility of a precursor/co-reactant combination can be assessed
using solution-state reactivity studies.
For main group element ALD, silyl-containing compounds (E(SiR3)3, E = Sb, B)
have been investigated as precursors in combination with EX3 (X = F, Cl, Br, I) coreactants,
due to the potential for thermodynamically favourable Si-X bond formation to
drive the required surface-based reactions. For metal oxide ALD (MO2; M = Hf, Zr), new
ALD methods have been proposed to enable gap-free filling of narrow trenches on the
surface of a silicon wafer. This work involved the design, synthesis, and evaluation of new
ALD precursor molecules and reactions, ALD reactor studies for thin film deposition, and
characterization of the resulting films.
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/28704 |
| Date | January 2023 |
| Creators | Al Hareri, Majeda |
| Contributors | Emslie, David, Chemistry |
| Source Sets | McMaster University |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0032 seconds