Atomic layer deposition builds a nanocoating one self-limiting half-reaction at a time, alternating two gas-phase precursors separated by inert purges:
θ_A(dose) = 1 − exp(−k·D) (Langmuir-type saturation)
Δh/cycle = GPC₀ · θ_A · f(T) (ideal, self-limiting)
f(T) = 1 inside the "ALD window"
f(T) < 1 below window (slow/condensing)
parasitic growth > 0 above window (precursor decomposes, CVD-like)
- Pulse A — the first precursor chemisorbs onto reactive surface sites. Coverage θ_A rises with dose D but saturates near 1 once every site has reacted — pushing the dose higher cannot add a second layer in the same pulse. That self-limiting saturation is what makes ALD depth-independent and atomically conformal.
- Purge A — excess, unreacted precursor is flushed away; only the chemisorbed monolayer of ligands remains.
- Pulse B — the second precursor reacts with those surface ligands, completing one atomic layer and releasing a volatile byproduct (e.g. CH₄ for TMA/H₂O).
- Purge B — byproducts are flushed out and the surface is reset to identical reactive sites, ready for the next cycle.
- Substrate temperature — every material has an "ALD window". Below it, ligand exchange is incomplete and growth-per-cycle (GPC) drops. Above it, the precursor thermally decomposes and deposits continuously like CVD — growth stops being self-limiting, thickness runs ahead of the ideal Å/cycle line, and the film roughens (visible as uneven column heights).
- Reactivity slider — models the intrinsic sticking coefficient of the chosen precursor chemistry on this surface; it scales how fast θ_A approaches saturation for a given dose.
Real-world relevance: this four-step ABAB cycle is exactly how semiconductor fabs deposit high-κ gate oxides, and how the anticorrosion, antireflective and barrier nanocoatings described on this page's article are actually grown — nanometre by nanometre, with sub-ångström thickness control.