Understanding chemical and physical mechanisms in atomic layer deposition
Summary¶
Atomic layer deposition (ALD) is often introduced with idealized A/B pulse cycles that deposit exactly one monolayer per pair of steps, but real processes frequently deviate: sub-monolayer growth per cycle, island nucleation, precursor ligand blocking, and interfacial mixing all arise from coupled chemisorption and diffusion phenomena. This Journal of Chemical Physics Perspective, authored by researchers in the Bent group tradition of surface chemistry, organizes how chemisorption and surface diffusion mechanisms cause departures from textbook layer-by-layer cartoons. The goal is to give experimentalists and modelers a vocabulary for rational parameter selection—temperature, precursor dose, co-reactant exposure, and reactor design—when non-ideal growth limits thickness control or creates unwanted interface profiles in oxide and dielectric stacks. ALD is central to semiconductor manufacturing; the perspective therefore doubles as a bridge between textbook introductions and the defect-tolerant processing reality in high-aspect-ratio structures.
Methods¶
4 — Review / perspective (primary article type). The piece is a J. Chem. Phys. Perspective that organizes literature on how chemisorption and surface diffusion drive departures from idealized A/B ALD cycles (sub-monolayer growth per cycle, nucleation, ligand blocking, intermixing at interfaces). Scope includes mass transport in high-aspect-ratio and porous geometries where reactor-level access limits saturation, as well as precursor families touched in the text (e.g. alkylamides, alkoxides, halides as categories in the review—not one unified data table here). Reproducibility means following cited primary ALD / surface science papers for each chemistry; this editorial does not give one laboratory or code protocol.
3 — Static QM / DFT (as a single study in this article). N/A — the Perspective is not a DFT benchmark; there is no one functional, basis, k-mesh, or NEB path for “this work.” Functional / hybrid / meta-GGA choices: N/A — see each cited first-principles ALD study. Dispersion (DFT-D, vdW): N/A at the perspective level. Plane-wave / PAW / localized basis: N/A for this page. k-sampling / Brillouin mesh: N/A for this page. Structures / pathways (TS, reaction coordinate): discussed only as reported in literature citations, not recomputed here. Properties (barriers, sticking, GPC): N/A to tabulate globally; the manuscript points readers to primary papers for each precursor** class.
1 — MD / 2 — Force-field training. N/A — not atomistic simulation papers; when the text mentions MD or FF in passing, treat those as literature pointers only.
Findings¶
Outcomes and mechanistic picture. ALD kinetics are not reducible to a saturating monolayer per half-cycle when steric hindrance, competing channels, condensation, and lateral or pore diffusion matter. Trench and porous features can segregate dose and create thickness gradients and interface broadening in stacks—i.e. morphology and composition profiles follow adsorption–diffusion physics, not a textbook 2D per-cycle slab model (themes in the article as summarized here).
Comparisons and design levers. The review is meant to help match temperature, precursor dose, co-reactant exposure, and reactor design to the actual surface process; peers in semiconductor manufacturing and R&D are the intended audience.
Limitations and corpus honesty. A perspective curates; barriers, GPC, and process windows must be taken from cited primary papers for each chemistry. Plasma-enhanced and area-selective ALD introduce additional physics not exhaustively covered in a single review length. Vacuum reactor designs also change precursor transport, which couples to the surface diffusion themes in the text.
Limitations¶
This page is a reader summary of the Perspective; it does not replace locators or numbers in the primary ALD literature the review cites.
Relevance to group¶
General surface-processing context for oxide and interface science adjacent to atomistic simulations in the corpus.
Citations and evidence anchors¶
- DOI: 10.1063/1.5133390