2.2 Heterogeneous nucleation and the Harvey crevice model

The homogeneous nucleation theory of the previous lesson predicts that pure water should hold $\sim -1000$ atm of tension. Real samples tear at $\sim -0.1$ atm — ten thousand times less. The resolution is heterogeneous nucleation: bubbles do not appear de novo by thermal fluctuation; they appear at preexisting nucleation sites where the barrier is locally much lower than $\Delta G^* = 16 \pi \sigma^3 / 3 (\Delta p)^2$.

Three classes of preexisting site contribute, ordered roughly by typical importance:

1. Gas pockets trapped in surface crevices — the Harvey 1944 model, dominant for any water-bearing container with a wettable solid surface.
2. Free gas microbubbles — bubbles small enough to remain suspended and not rise out of the sample; stable because surface tension is balanced against gas pressure across the bubble wall. Typically removed by degassing.
3. Surface-active particulates — dust grains, organic films, ions with strong hydration shells. Less universally understood; for engineering practice often classed together as “particulate nuclei.”

This lesson develops the dominant Harvey mechanism, with passing reference to the other two.

The Harvey crevice model

Edward Newton Harvey at Princeton (1944) was the first to systematically observe that the bubbles that emerge during cavitation appear at specific places on container walls — not randomly distributed through the bulk. Microscopic examination of the surfaces of cavitation chambers showed the bubble release points sitting at the openings of small surface crevices. Harvey proposed that each crevice contained a long-lived gas pocket held in place by the geometry of the crevice and the surface chemistry of the wall, and that under sufficient tension the gas-liquid meniscus at the crevice mouth became unstable and released a bubble from the pocket into the bulk.

Why the gas pocket is stable

Consider a conical pit in a solid surface, filled with a small gas pocket of pressure $p_g$ and surface area appropriate to its geometry. The gas pocket is in mechanical equilibrium when the Young–Laplace condition is satisfied across the gas-liquid meniscus:

$p_g - p_\infty = \frac{2 \sigma}{R_m},$

where $R_m$ is the radius of curvature of the meniscus (positive when the meniscus is concave into the liquid, negative when concave into the gas).

Under positive ambient pressure ($p_\infty > 0$), the meniscus typically sits well inside the crevice with concavity into the liquid — the gas is pushed inward by atmospheric pressure, and the meniscus radius is whatever value makes Young–Laplace hold given the crevice geometry. The system is mechanically stable.

The critical question is what the system does as $p_\infty$ is reduced (the liquid is put into tension). The meniscus moves outward as the gas pocket expands. Whether the gas pocket eventually escapes — that is, whether at some critical tension the meniscus reaches the mouth of the crevice and a free bubble pops out — depends on the geometry of the crevice and the contact angle of the liquid against the wall material.

Why hydrophobic surfaces matter

The geometric condition is

$\theta + \beta > 90°,$

where $\theta$ is the contact angle of the liquid against the wall material (water on glass: $\sim 30°$; water on Teflon: $\sim 105°$; water on most metal surfaces: $\sim 70$–$90°$) and $\beta$ is the half-angle of the crevice at its mouth.

When this condition is satisfied (e.g., a hydrophobic surface with $\theta > 90°$ and any reasonable $\beta$), the gas pocket can be stabilised against floodings by surface tension: any infinitesimal flooding of the cavity creates a meniscus with the wrong curvature to do work against the tension, and surface energy drives the cavity back to its gas-pocket state. The gas pocket is then intrinsically stable on indefinite timescales, persisting even through cleaning cycles and dissolved-gas variations.

When the condition fails ($\theta + \beta < 90°$, i.e., a wettable surface), the gas pocket cannot be stabilised. Surface tension actively pulls liquid into the cavity. Any gas initially present dissolves into the bulk liquid (or is carried out by capillary action) and the cavity ends up filled with liquid. Such samples have no surface-crevice nucleation sites and can hold much higher tensions before some other (rarer) defect provides a nucleation event.

This is the fundamental engineering rule of cavitation nucleation: hydrophobic surfaces nucleate readily; hydrophilic surfaces resist. Surface preparation in cavitation experiments and in industrial systems aims to maximise wettability — cleaning, oxidising, and (sometimes) chemically functionalising surfaces to drive the contact angle below 90°. Briggs’s 1950 inclusion-free capillaries achieved their high tensile strengths in part because the carefully cleaned and oxidised glass was strongly hydrophilic.

Critical tension for crevice nucleation

The critical ambient pressure at which the gas pocket becomes unstable depends on the crevice mouth radius $r_c$, the geometry $\beta$, the contact angle $\theta$, and the internal gas pressure $p_g$. A simple analysis for a conical crevice gives

$p_{\infty, \text{crit}} = p_g - \frac{2 \sigma}{r_c} \cos(\theta + \beta - 90°),$

valid when $\theta + \beta < 180°$. For typical values ($p_g = 0.7$ atm, $r_c = 1$ μm, $\sigma = 72$ mN/m, $\theta = 105°$, $\beta = 20°$), the critical tension comes out to roughly $-0.2$ atm — precisely the order of magnitude observed in tap-water cavitation experiments.

The critical tension is therefore not a fundamental property of the liquid — it is set entirely by the geometry and surface chemistry of the available crevices. Different containers, different cleaning histories, and different dissolved-gas atmospheres will all produce different critical tensions, even for the same water.

Population of crevices

A real container surface contains a distribution of crevices with varying sizes and geometries. Surface-roughness measurements on typical machined or moulded surfaces find characteristic crevice mouth radii ranging from sub-micron to tens of microns. Each crevice has its own $p_{\infty, \text{crit}}$, and the sample’s overall cavitation threshold is set by the crevice with the smallest critical tension — the weakest link.

As the ambient pressure is dropped from atmospheric toward zero (toward boiling) and below (into tension), nucleation events occur one by one as each crevice’s critical tension is reached. The first bubble corresponds to the most-easily-released crevice; further pressure reduction releases additional bubbles from progressively more-stable crevices. This is why cavitation noise is broadband: the events are roughly Poisson-distributed in time, with rates set by the local crevice population and the rate at which ambient pressure varies.

What’s left to develop

The Harvey crevice model gives the dominant heterogeneous nucleation mechanism but leaves three questions for the next lessons:

• What sets the population of crevices and their critical tensions? This is the nucleation site distribution $N(R)$ — covered in Lesson 2.3.
• How does dissolved gas in the liquid bulk affect the stability of crevice gas pockets and free microbubbles? Dissolved gas is a separate parameter from the crevice geometry; high gas content stabilises smaller, harder-to-trigger nuclei. Covered in Lesson 2.3.
• How does nucleation in a flowing liquid differ from nucleation in a static sample? Pressure transients in flow can be much briefer than the equilibration time of a crevice gas pocket; the engineering inception number $\sigma_i$ captures this. Covered in Lesson 2.4.