Grind an ore finer and more of the valuable mineral is liberated from the gangue. Grind it finer still and the flotation circuit stops being able to collect what the mill just liberated. Every concentrator lives somewhere on that curve, and the fraction below about 20 µm is where the two effects cross and the value starts reporting to tailings.
The published work on nanobubble-assisted flotation is about that fraction specifically. Before any of it is quoted, one thing has to be said plainly: Kairospace has no mining installations and no measurements of its own in this application. Every figure below is other people's laboratory work, reported as such.
Why does conventional flotation lose fine particles?
Flotation requires a particle to collide with a bubble and stay attached to it. Below roughly 20 µm, particles carry too little inertia to cross the streamlines around a rising bubble, so they follow the water around it and the collision probability collapses. What never collides never attaches, and what never attaches reports to tailings.
The mechanism is hydrodynamic rather than chemical, which is why reagent dosing alone does not fix it. A rising bubble pushes a flow field ahead of it. A heavy particle has enough momentum to punch through that field and strike the surface; a light one is carried around the obstacle like a leaf around a stone. Collision efficiency falls steeply with particle size, and it falls fastest exactly in the size range fine grinding produces.
Attachment is the second barrier. Even after contact, the thin water film between particle and bubble has to drain and rupture before a stable three-phase contact line forms, and the time that takes — the induction time — has to be shorter than the time the particle spends sliding across the bubble surface. Fines slide past quickly.
Detachment then removes part of what did attach. Turbulence in a mechanical cell strips particles from bubble surfaces, and while fines are held more securely than coarse particles, the froth phase and the launder are further opportunities to lose them.
The economics of this are worse than the tonnage suggests. Fine fractions are often the most liberated and therefore the highest grade in the feed, so the material lost is not average material.
A circuit optimized for the coarse fraction can be sending its best-liberated mineral to the tailings dam.
How do nanobubbles change fine-particle capture?
Nanobubbles change the surface a conventional bubble meets rather than the capture step itself. In published research they nucleate preferentially on hydrophobic mineral surfaces, bridge fine particles into larger aggregates, and act as seeds that make attachment to a conventional bubble more favorable. The particle is made easier to capture instead of the bubble being made stronger.
Selective nucleation is the starting point. Peer-reviewed work reports that bubbles formed by hydrodynamic cavitation nucleate preferentially on hydrophobic surfaces, because the energy cost of forming a gas phase there is lower than in the bulk. That selectivity is what makes the effect useful: the bubbles go where the valuable mineral is rather than coating everything equally.
Bridging follows. A nanobubble sitting between two fine particles forms a capillary bridge, and the aggregate that results behaves hydrodynamically like a larger particle — more inertia, better collision probability, and back inside the size range the cell was designed to float.
The seeding effect is the third mechanism reported. A particle whose surface carries attached gas presents a bubble-to-bubble contact to an arriving conventional bubble instead of a solid-to-bubble contact, and the intervening water film ruptures faster. Published flotation studies report sharply reduced induction times in the presence of interfacial nanobubbles.
Note what this does and does not describe. The literature is describing a preconditioning step ahead of a conventional cell, not a replacement for it. The froth, the collector, the cell hydrodynamics, and the operator all stay.
What recovery improvement does the literature report?
In peer-reviewed trials on fine and ultrafine chalcopyrite, flotation recovery rose by about 16–21% with nanobubbles present, compared with conventional flotation on the same feed. That result is laboratory scale and mineral specific. Kairospace has no mining installations, so every figure in this post is published research rather than a Kairospace result.
Laboratory scale, on deliberately narrow size cuts, not a plant result on a full feed distribution.
Source: Azevedo & Oliveira (2019)
Read the size fractions with the number. The chalcopyrite work that produced that range was run on deliberately narrow fine and ultrafine cuts, which is where the mechanism should help most and where a bench test can isolate it. A plant feed is a full distribution, and the coarse end of it was never the problem.
Comparable directional gains are reported elsewhere in the literature — in fine coal column flotation and in coarse phosphate, where detachment rather than collision is the limiting step. That breadth across different mineral systems is genuine evidence that something physical is happening rather than an artifact of one laboratory's apparatus.
It is also worth being precise about what a recovery gain is measured against. Recovery figures move with grade, and a technique that lifts recovery while degrading concentrate grade has moved value rather than created it. The published flotation work generally reports grade held or improved alongside recovery, which is the more meaningful result and the one to insist on in any test program.
What happens to reagent consumption?
In published research, collector and frother demand fell by up to 50% while metallurgical performance was maintained. The reported mechanism is that a surface already carrying attached gas is effectively more hydrophobic, so less chemical is needed to make it float. Reagent reduction is not reagent elimination, and the literature does not claim it is.
The secondary-collector framing is the useful one. A collector works by rendering a mineral surface hydrophobic enough for a bubble to hold it; attached nanobubbles raise the effective contact angle by physical means. The chemistry and the physics are doing the same job from different directions, and the dose required falls to the extent the physics carries part of the load.
The value of that is not only the reagent invoice. Collector and frother chemistry follows the water through the circuit into the tailings and the recycle stream, where it affects downstream separations, effluent treatment, and permitting. A dosage reduction is worth more than its price at the loading dock in most operations.
The honest limits are two.
- Ore- and circuit-specific. Dosage optimization is ore-specific and circuit-specific — a reduction demonstrated on one feed does not transfer to another, and the only way to establish a new dosage is a test program on the actual ore.
- A reduction, not an escape. This is a reduction in chemistry, not an escape from it. Hydrodynamic cavitation changes the physical state of the water so the reagents already in the circuit work harder; the reagents stay.
What is unresolved?
The nanobubble flotation literature leaves several questions unsettled. Most results are laboratory scale, on selected size fractions and single minerals. Nanobubble concentration and size distribution frequently go uncharacterized, so the dose is unknown. Real-ore work shows clay-bearing feeds can respond negatively. And durability at plant scale is not established.
Take the negative result first, because it is the most useful finding in the set. In published work on Chilean copper sulfide ores, nanobubbles improved recovery on the low-clay ore and reduced it on the high-clay ore, where they intensified non-selective slime coatings. The mechanism that helps a clean hydrophobic surface can hurt when there is clay in the feed, so the clay content of a specific feed decides which of those two results a given operation would get.
Then the dose problem. Many studies report that nanobubbles were present without reporting how many, at what size distribution, or measured by what method. Particle tracking counts scattering centres and cannot by itself distinguish a gas nanobubble from a nanoscale contaminant, so a bubble population has to be argued rather than asserted. Without a characterized dose, results cannot be compared between studies or reproduced at another site.
Scale is the third gap. Bench cells and plant circuits differ in residence time, turbulence, recycle water chemistry, and every one of those bears on whether interfacial bubbles survive to do their job. The literature does not establish how a preconditioning effect measured in a laboratory column behaves in a rougher bank running recirculated process water.
What follows is a test sequence, not a purchase order. A bench program on your own ore, with the size-by-size deportment measured and the clay content known, then a circuit trial with a proper control. How to run a pilot that produces data you can trust covers the controls. Our mining page lists the published studies, the calculators include a heap-leach model built to return zero where oxygen is not the limit, and the classroom section on flotation recovery carries the citations in full.