What is a nanobubble, in engineering terms

First, the terminology, because it is standardised

ISO 20480-1:2017 fixes the vocabulary, and getting it wrong is the quickest way to be marked down by a technical buyer:

Below 1 µm the physics changes rather than degrades. A 100 nm bubble is roughly 1/200th to 1/700th the diameter of a human hair, the same order of magnitude as a virus or a large protein, and one to two orders of magnitude smaller than a bacterium.

Why size dominates the outcome

Stokes' law makes rise velocity proportional to the square of diameter. Reduce a 1 mm bubble to 1 µm and it rises a million times more slowly; at 100 nm, Brownian motion and the surrounding fluid's own movement dominate buoyancy entirely. In a 10 m tank, a millimetre bubble surfaces in seconds and a nanobubble follows the water instead.

Two consequences matter commercially. First, oxygen can be delivered to the sediment or the bottom of a deep basin rather than to the top two metres. Second — and this is the one most brochures skip — the same property means the bubble stays inside the water you are trying to treat, so the gas is not lost to the atmosphere before it has done any work.

The surface is negative, and that is the useful part

Bubbles in water carry a negative zeta potential, typically −20 to −50 mV. That charge does two things: it keeps bubbles from coalescing (electrostatic double-layer repulsion plus hydrodynamic repulsion, as Seddon et al. summarised in ChemPhysChem in 2012), and it makes them adsorb onto suspended solids, oil droplets and membrane surfaces. This is the mechanism behind "small bubbles stick, big bubbles lift" — Calgaroto et al. (Minerals Engineering, 2016) showed the two populations divide the work rather than compete.

How they are made, in ISO's own categories

ISO 20480-3:2021 lists the generation families. Ours sits in the porous-membrane category, with a shear stage added:

• Pressurised dissolution — saturate water with gas under pressure, then release. The classic DAF route; bubble size depends on the release nozzle.
• Shear / swirl — cut a gas stream mechanically. Cheap, robust, but the size floor is set by the energy you are willing to spend.
• Porous membrane diffusion — force gas through a membrane whose pore size becomes the bubble size ceiling. Uniform pores give a narrow distribution; this is the route that makes 20 nm pores meaningful.
• Electrolysis — generate gas at an electrode. Very fine bubbles, but the gas is limited to what you can electrolyse and the water chemistry changes.
• Ultrasonic cavitation — common in laboratory work, rare in industrial flow for obvious reasons.

How they are counted, and what the count cannot tell you

Nanoparticle tracking analysis (NTA), dynamic light scattering (DLS), laser diffraction, resonant mass measurement and image analysis each cover a different window. NTA gives you a number concentration in a size band; it cannot tell you whether the objects it counted are gas bubbles or something else that scatters light. Scattering intensity scales with the sixth power of diameter, so a few large particles dominate the signal while thousands of small ones hide behind them. Above roughly 10⁹ particles/mL the sample has to be diluted before measurement, which changes what you are measuring.

Eight things newcomers get wrong

• "Milky white means nanobubbles." Milky means the bubbles are big enough to scatter visible light — that is a microbubble cloud. Real nanobubble water looks almost clear, with a faint opalescence.
• "Smaller is always better." For lifting floc you want lift. Flotation needs a size distribution, not a minimum.
• "Concentration alone is a specification." 10⁸ per mL at 80 nm and 10⁸ per mL at 800 nm are different products.
• "It dissolves gas, so it is just a fancy aerator." Dissolution is part of it; the residence time and the surface chemistry are what make the results different.
• "Nanobubbles pass through cell walls." They do not. The size does not fit the transport channels; see the limits page.
• "Any water works the same." Hardness, salinity, surfactants and dissolved organics all change bubble lifetime. Seawater behaves nothing like tap water.
• "A ZetaView report is a performance guarantee." It is an instrument reading on a sample. It is not an acceptance test for your plant.
• "No moving parts means no energy." Pumping is energy. A passive device with 0.3 bar of available head still consumes power somewhere upstream.

Where the literature stands

The persistence question is genuinely open, and we treat it as open: classical Epstein–Plesset dissolution theory says a 100 nm bubble should disappear in microseconds to milliseconds, while observation says hours to months. Several mechanisms have been proposed — charge stabilisation, dynamic equilibrium, high internal density, a contaminant shell — and none has won. If a supplier tells you the mechanism is settled, they are selling, not reporting.

Key sources: ISO 20480-1:2017 (terminology), ISO 20480-2:2018 (attribute classification), ISO 20480-3:2021 (generation methods) · Seddon et al., ChemPhysChem 13 (2012) on bubble coalescence suppression · Calgaroto et al., Minerals Engineering 92 (2016) on nanobubble-assisted flotation · Nanomaterials 12(13):2175 on stability models and their counterexamples. Full citations, including the ones that undercut our own claims, are on the limits page.