Atomic Layer Deposition (ALD)

• Atomic Layer Deposition (ALD) is a vapor-phase technique that uses sequential, self-limiting surface reactions to achieve atomic-scale film thickness and uniformity.
• It employs saturative chemisorption cycles with purging steps to ensure precise control of film growth, even on chemically or topographically complex substrates.
• ALD is integral to advanced applications in semiconductors, energy storage, quantum devices, and biosensors, where high conformality and ultrathin film properties are critical.

Atomic layer deposition (ALD) is a vapor-phase film growth technique predicated on sequential, self-limiting surface reactions to achieve atomic-scale control over film thickness and exceptional conformality, even on topographically or chemically complex substrates. Distinguished from conventional chemical vapor deposition (CVD) by its saturative chemisorption cycles, ALD enables sub-nanometer thickness control, pinhole-free films, and precise modulation of interfacial structure and composition—capabilities now foundational across semiconductor device manufacturing, energy storage, quantum technologies, biosensor fabrication, and nanoscale metrology.

1. Mechanistic Basis and Operational Principles

ALD is defined by cyclic exposure of a substrate to vapor-phase precursors, each of which reacts with distinct, surface-bound functionalities in a self-limiting manner. For prototypical alumina growth using trimethylaluminum (TMA) and a protic oxidant (e.g., H₂O, remote hydrogen plasma), the surface half-reactions are:

• Metal precursor adsorption (e.g., TMA):

$$*\mathrm{Si–OH} + \mathrm{Al(CH}_3)_3 \rightarrow *\mathrm{Si–O–Al(CH}_3)_2 + \mathrm{CH}_4\uparrow$$

• Oxidant exposure (e.g., H₂O or H radicals):

$$*\mathrm{Si–O–Al(CH}_3)_2 + 2\,\mathrm{H}_2\mathrm{O} \rightarrow *\mathrm{Si–O–Al(OH)}_2 + 2\,\mathrm{CH}_4\uparrow$$

or, for hydrogen-plasma-assisted cycles,

$$*\mathrm{Si–O–Al(CH}_3)_2 + 4\mathrm{H} \rightarrow *\mathrm{Si–O–AlH}_2 + 2\,\mathrm{CH}_4\uparrow$$

Each half-step saturates once all accessible surface moieties are consumed, enforcing monolayer-by-monolayer growth. Purging steps prevent precursor cross-reactions and enable uniform film formation throughout the reactor. The growth per cycle (GPC) is substrate-, temperature-, and precursor-dependent, with values for Al₂O₃ typically 0.99–1.27 Å/cycle at 100–210 °C, supporting atomic-level thickness control (Robinson et al., 1 Jul 2025, Elliot et al., 2013).

Distinct nucleation regimes arise during the early cycles: on hydroxyl-deficient metals (e.g., Al), initial H₂O pulses oxidize the metal and generate OH-terminated sites, enabling TMA chemisorption; subsequent cycles become self-limited after surface saturation. For plasma-enhanced ALD (PEALD), the remote plasma supplies reactive radicals or ions that can both oxidize and reduce surfaces, allowing for lower growth temperatures and distinct surface terminations.

2. Conformality, Penetration, and Thickness Control

ALD’s self-limiting nature yields outstanding conformality in high aspect-ratio (AR) features, nanopores, colloidal particles, and on 3D microstructures. Theoretical models (e.g., Gordon et al.'s step-coverage model) relate step coverage (SC) to precursor diffusivity, exposure time, and feature geometry:

$$SC \;\approx\; 1 - \exp\left(-\frac{D_{\text{eff}}\tau}{L^2}\right)$$

where $D_{\text{eff}}$ is the effective precursor diffusivity, $\tau$ the exposure duration, and $L$ the trench or pore depth. Full conformality is achieved only if the precursor pulse time significantly exceeds the saturation time throughout the structure (Wang et al., 2013).

Table: Typical ALD Growth Rates for Selected Oxide Systems

Material	Precursors	Growth Temp.	GPC (Å/cycle)
Al₂O₃	TMA + H₂O / H₂ plasma	100–250 °C	1.0–1.27
TiO₂	TiCl₄ + H₂O; TDMAT + H₂O	150–250 °C	~0.5–0.7
ZnO	DEZ + H₂O	100–200 °C	~1.5

Nucleation on non-hydroxylated surfaces may require service activation (e.g., plasma, ozone) or thin “wetting” layers for robust initial ALD chemisorption. Substrate temperature window is constrained both by the self-limiting chemistry (defining the “ALD window”) and by precursor thermal stability.

3. Interfacial Layer (IL) Formation and Control

The formation of an interfacial layer (IL) between the ALD dielectric and metallic substrates, particularly during the initial H₂O or oxidant exposures, is well-documented. For Al₂O₃ on Al, the first H₂O pulse at 200 °C induces thermal oxidation of Al, generating an IL of 0.5–2 nm, depending on Al wetting-layer thickness and oxidation conditions (Elliot et al., 2014, Elliot et al., 2013, Elliot et al., 2014). Linear fits of total oxide thickness ($t$) vs. cycle number ($N$) often yield:

$t(N) = \text{GPC} \times N + t_{\text{IL}}$

where $t_{\text{IL}}$ reflects the saturated IL. Minimizing $t_{\text{IL}}$ (by reducing Al thickness or using inert metal wetting layers) is critical for atomic-scale tunnel junctions and qubit applications, as the IL hosts point defects implicated in two-level fluctuators (TLFs) that degrade coherence.

For silicon substrates, hydrogen-plasma-assisted ALD using TMA enables the conversion and consumption of surface SiO₂, progressing through exhaustion of surface –OH/Si–O–Si and subsequent diffusion-limited oxygen transfer, yielding sub-nanometer AlOₓ caps with atomically sharp Si/AlOₓ interfaces (Henning et al., 2021).

4. Advanced ALD Chemistries and Monolayer Precision

Recent developments demonstrate monolayer-limited, pinhole-free ALD using oxidant-free, hydrogen-plasma-assisted cycles. On GaN, a single TMA/H∗ plasma cycle yields a continuous 2.8 ± 0.1 Å AlOₓ monolayer via site exhaustion kinetics, with layer-by-layer growth enforced by chemical deactivation of the surface—precluding island formation and suppressing subsequent nucleation (Henning et al., 2021). Such ultrathin films modulate work function (ΔΦ = –0.38 eV), support packing-limited phosphonic-acid self-assembled monolayers (density $n=4.5 \pm 0.3\;\mathrm{nm}^{-2}$), and enable chemically tailored functionalization at the atomic limit.

For 2D semiconductors, multi-step ALD processes enable the formation of wafer-scale TMDCs with independent control of layer count (via ALD cycles), stoichiometry (via chalcogenization), and crystallinity (via post-growth annealing), yielding device-grade MoS₂ with mobilities up to 55 cm²/V·s and On/Off ratios of 10⁷ (Aspiotis et al., 2022).

5. Device and Materials Applications

ALD’s combination of uniformity, thickness precision, and gentle processing conditions underpins diverse device and metrology applications:

• Tunnel barriers in Josephson and quantum devices: Nb/Al/ALD–Al₂O₃/Nb junctions with ∼1–2 nm barriers display uniformly high $R_{\mathrm{N}}A$, low subgap leakage, and low TLF density. In situ ALD–UHV integration eliminates uncontrolled native oxide formation, maintaining engineered IL thickness (Elliot et al., 2013, Elliot et al., 2014).
• Field-effect passivation and electrostatic engineering: Hydrogen–plasma ALD of AlOₓ/Si creates fixed negative charge ($|Q_f|\simeq 9 \times 10^{12}\;\mathrm{cm}^{-2}$), modulating band-bending by +340 meV (n-type) and producing 0.45 V surface potential steps (Henning et al., 2021).
• Strain management in hybrid quantum structures: 50 nm ALD Al₂O₃ reduces cryogenic strain in heterostructures, minimizing ESR linewidths and preserving $T_2$ (≈23 ms vs. 20 ms in control) (Kennedy et al., 2021).
• Hydrogen permeation barriers: 10 nm ALD Al₂O₃ on copper achieves >20× flux reduction for deuterium at 275–350 °C, transitioning transport from bulk-limited to surface/pore-mediated, with PRF = 22–34; film growth at 100–210 °C yields GPC = 1.1 Å/cycle, RMS roughness = 3.6–4.1 Å (Robinson et al., 1 Jul 2025).
• Nanoporous and high-AR templated structures: ALD enables sub-nanometer pore-size tuning determined by precursor size (e.g., TDMAT, 7 Å), with in situ GISAXS/XRF metrology confirming accessible surface area and minimum pore diameter (Dendooven et al., 2015, Wang et al., 2013). Area capacitances up to 100 μF/cm² are achieved in MIM nanocapacitor arrays (Wang et al., 2013).
• Particle and powder coating: For energy and catalytic applications, ALD enables conformal shells on nanoparticles in both static planar (lab-scale: ≤1 g/day, ≤45 nm coatings) and fluidized-bed (kg-scale) reactors. Plug-flow, high-Da reactors exhibit nearly 100% precursor utilization and self-extinguishing behavior at the monolayer limit (Yanguas-Gil et al., 23 Aug 2024, Badman et al., 2016).
• Quantitative SRM standards fabrication: Low uncertainty (σ/μ ~ 1–2%) 2D/3D ALD standards surpass NIST SRM homogeneity, with conformality extending to complex 3D-printed microstructures and direct quantization of areal density per cycle (Becker et al., 2017).

6. Limitations, Challenges, and Optimization Strategies

While ALD affords unmatched precision, several process-specific and material challenges remain:

• Incubation and nucleation: Noble metal substrates often exhibit extended nucleation delays (30–50 cycles) due to limited surface –OH; mitigation requires surface activation (plasma, ozone) or insertion of reactive wetting layers.
• Interfacial-layer (IL) management: For ultrathin tunnel barriers, even minimal IL thickness ($t_{\rm IL}$) can dominate device properties; using ultrathin wetting layers and in situ transfer/protection are critical (Elliot et al., 2013, Elliot et al., 2014).
• Throughput and scale-up: Scaling particle ALD demands reactor designs that achieve high Damköhler numbers; batch well-mixed systems require long cycles, whereas plug-flow and fluidized bed configurations enter the transport-limited regime and maximize precursor utilization (Yanguas-Gil et al., 23 Aug 2024).
• Precursor kinetics and pore closure: In mesoporous materials, the minimal pore size attainable is set by precursor steric dimensions; kinetic exclusion rather than thermodynamic wetting limits governs film closure within ultrafine pores (Dendooven et al., 2015).
• Film composition and electronic structure: For high-performance applications (e.g., batteries), ALD film stoichiometry, density, and reorganization energy (for electron tunneling) dictate passivation efficacy; tailored multi-cycle nucleation and post-treatment are often necessary (Leung et al., 2012).
• Ambient vs. in situ transfer: Sequential or ex situ processing may introduce native oxides or contamination; integration of ALD within UHV PVD platforms or gloveboxes is vital for atomic-level interface control (Elliot et al., 2014).

7. Quantitative Models, Theory, and In Situ Metrology

Mathematical modeling underpins both the understanding and optimization of ALD:

• Surface-coverage kinetics: For single-site models,

$d\Theta/dt = s_0 \beta_0 [1-\Theta] J$

where $\Theta$ is fractional coverage, $s_0$ site area, $\beta_0$ empty-site sticking probability, and $J$ incident precursor flux.

• Plug-flow and batch reactors: Growth and utilization are characterized by dimensionless time ($\tau = t/t_0$) and Damköhler number (Da). Self-extinguishing plug-flow reactors reach full coverage at $\tau_s=1$, with precursor breakthrough providing a direct process-control metric (Yanguas-Gil et al., 23 Aug 2024).
• Electron tunneling models: For electron transfer through ALD thin films in batteries, Marcus theory for non-adiabatic tunneling yields

$$k_{et} = (2\pi/\hbar) |V|^2 [1/\sqrt{4\pi\lambda k_BT}] \exp[-(\lambda + \Delta G)^2/(4\lambda k_BT)]$$

with key parameters ($|V|$, $\lambda$, $\Delta G$) extracted from constrained DFT and AIMD simulations (Leung et al., 2012).

• In situ diagnostics: Techniques such as spectroscopic ellipsometry, GISAXS, XRF, QCM, RBS, and XRR provide cycle-by-cycle calibration of thickness, composition, density, and surface area evolution (Dendooven et al., 2015, Becker et al., 2017).

Conclusion

Atomic layer deposition is a mature platform technology with unique capabilities for nanometer- and atomic-scale film growth, interfacial engineering, and chemical precision across a broad spectrum of advanced materials applications. Its continued evolution—spanning novel chemistries (plasma, oxidant-free), integration with lithography and UHV processing, and sophisticated scale-up methodologies—is enabling the systematic design of devices and interfaces at previously inaccessible lengthscales. The interplay between experimental process development, first-principles modeling, and in situ metrology remains central to addressing challenges in nucleation, interfacial layer control, and throughput, ensuring ALD’s relevance in next-generation electronics, quantum systems, energy devices, porous media, and beyond.