Ostwald ripening is the dominant deactivation mechanism for supported metal nanocatalysts: over time, small nanoparticles dissolve into the surrounding matrix/gas phase and redeposit onto larger ones, shrinking the total catalytic surface area even though the total metal mass is conserved.
The driving force is the Gibbs–Thomson effect — a curved surface raises a particle's chemical potential relative to a flat one:
μ(r) = μ_bulk + 2γΩ / r
γ = surface energy, Ω = atomic volume, r = particle radius
Small particles (large 1/r) sit at higher chemical potential than large ones, so atoms diffuse down that gradient — away from small particles, onto large ones. In the mean-field LSW (Lifshitz–Slyozov–Wagner) approximation, each particle's radius evolves relative to the ensemble's critical radius r꜀ (its volume-weighted mean):
dr/dt = (D · γ · Ω² · Cₛ / kT) · (1/r꜀ − 1/r) / r
D(T) = D₀ exp(−Eₐ / kT) (Arrhenius diffusivity)
- Particles with r > r꜀ grow; particles with r < r꜀ shrink and eventually vanish — their volume is absorbed by the survivors.
- Temperature raises D(T) exponentially (Arrhenius), so higher-temperature calcination or operation sinters the bed far faster — the classic industrial trade-off between reaction rate and catalyst lifetime.
- Surface energy γ sets the strength of the curvature driving force; higher-γ metals (e.g. late transition metals) ripen faster than low-γ ones for the same particle size.
- Total surface area (∝ Σr²) tracks the catalyst's active-site count directly — this is why real catalysts are engineered with supports, alloying, or capping ligands to slow this exact process.
Real-world relevance: this is why fresh Pt/Pd automotive-exhaust and fuel-cell catalysts lose activity with use, and why catalyst designers chase "sinter-resistant" architectures like single-atom catalysts and oxide-anchored nanoparticles.