Cobalt almost never arrives at a refinery as cobalt. On the Central African Copperbelt it comes up as a by-product of copper, locked into oxide minerals such as heterogenite. In Finland, Canada or Australia it appears alongside nickel. By the time a concentrate leaves site, it might carry only a few per cent of the metal, tangled up with iron, manganese, copper, nickel and a long tail of trace impurities. Turning that into the white crystals a cathode plant orders is a sequence of separations, each one aimed at a specific nuisance element.

Here is how the chain actually runs, and where the hard parts sit.

What Arrives at the Refinery

The shape of the feedstock decides almost everything downstream. Sulphide concentrates — carrollite, cobaltite, or nickel-cobalt concentrates from laterite processing — can be roasted, pressure-oxidised or smelted. Oxide ores, and the hydroxide and carbonate precipitates that many Congolese operations sell as an intermediate, usually dissolve in acid at atmospheric pressure instead.

Two intermediates dominate trade. Crude cobalt hydroxide, often described as filter cake, is cheap to make and cheap to ship, but it is a wet, impure material carrying much of the manganese and magnesium from the ore. Cobalt sulphate made at or near the mine is closer to finished product, though it still needs work before a battery maker will take it. Either way, the refiner's first job is the same: know exactly what is in the drum. Feed variability, not chemistry, causes most of the upsets in a cobalt circuit.

Leaching: Getting Cobalt Into Solution

Leaching moves the metal from solid to liquid. For oxide feeds this normally means sulphuric acid, often with a reducing agent such as sulphur dioxide or sodium metabisulphite. The reduction step matters more than it sounds. Much of the cobalt in heterogenite sits in the trivalent state, which is stubbornly insoluble, and it has to be brought down to the divalent form before acid will take it.

Points that refiners watch closely:

• Acid strength and temperature — enough to dissolve the cobalt, not so much that you dissolve everything else and pay to neutralise it later.
• Redox potential — the single best on-line indicator of whether reduction is complete.
• Residence time and pulp density — under-leached residue is lost cobalt; over-leaching drags iron and silica into solution.
• Solid-liquid separation — counter-current decantation and filtration, because a dirty liquor poisons every step that follows.

Sulphide feeds take a different road: pressure oxidation in autoclaves, roasting followed by a leach, or bacterial leaching for lower-grade material. The objective is identical — a clean pregnant liquor.

First Clean-Up: Iron, Copper and the Rest

Before solvent extraction gets anywhere near the liquor, it gets a blunt clean-up. Copper usually goes first, either by cementation on scrap iron or by solvent extraction with a hydroxyoxime reagent. Iron and aluminium come out next, precipitated as hydroxide or goethite by raising pH with lime, limestone or sodium hydroxide. Getting this stage right is partly about cost: every kilogram of iron you carry into the organic circuit is acid and extractant you will never recover.

Manganese and zinc removal often begins here too. The residue volumes are substantial, and managing gypsum and iron tailings is a real part of a refinery's operating budget, not an afterthought.

Solvent Extraction: The Cobalt–Nickel Split

This is the heart of the plant. The aqueous liquor is contacted with an organic phase — an extractant dissolved in kerosene with a modifier to keep it behaving — in mixer-settlers arranged in long trains. Different extractants do different jobs, and the order matters.

An acidic organophosphorus reagent such as D2EHPA will pull out zinc, calcium and leftover manganese at low pH. The cobalt-nickel separation then falls to a phosphinic acid extractant, of which Cyanex 272 is the best known. It has a much stronger affinity for cobalt than for nickel, but the pH window where cobalt loads and nickel stays put is narrow, so plants use multiple extraction, scrubbing and stripping stages to hold the line.

Scrubbing is the unglamorous stage that decides final quality. A dilute cobalt or nickel solution is used to displace entrained impurities from the loaded organic before stripping, which is done with sulphuric acid. What comes off the strip stage is a concentrated, notably purer cobalt sulphate solution. Nickel left in the raffinate is recovered separately. Some flowsheets add ion exchange resins at this point to polish residual impurities below the levels solvent extraction alone can reach.

Electrowinning, or Straight to Crystals

From here the route forks. Metal markets — superalloys, hardfacing, catalysts, pigments — want cobalt cathode, produced by electrowinning from a sulphate or chloride electrolyte, with the metal plating onto stainless or titanium blanks that are later stripped and chopped.

Battery markets want the salt instead, and going straight to sulphate avoids dissolving cathodes all over again. The purified liquor is concentrated by evaporation and crystallised, usually by cooling, to give cobalt sulphate heptahydrate. The crystals are centrifuged, dried, screened, blended and packed into lined bulk bags. Blending matters commercially: a cathode maker buying ten tonnes wants every bag to behave the same way in its dissolution tank.

What "Battery Grade" Really Means

Battery grade is not a marketing word; it is a specification with teeth. Typical requirements cover:

• Cobalt content, tightly bracketed rather than simply "high".
• Parts-per-million limits on iron, copper, zinc, nickel, manganese, cadmium, lead, arsenic, silicon, calcium, magnesium and sodium.
• Chloride and moisture limits, since both affect downstream cathode chemistry.
• Particle size distribution, tapped density and flowability.
• Magnetic contamination, checked because stray metal particles cause cell failures.

Analytical work is continuous: ICP-OES for major elements, ICP-MS for traces, X-ray diffraction for phase confirmation, laser diffraction for particle size, plus a magnetic separation test on every batch. The reason for the fuss is that anything you leave in the sulphate ends up in the cathode, and from there in someone's vehicle or grid storage unit.

Practical Notes for Anyone Working Along the Chain

Feed assay first, always. Build your mass balance around real numbers, sample every stream, and treat any change of feedstock as a change of process until proven otherwise. Watch chloride and calcium in process water — they accumulate quietly and show up later as corrosion and crud. Reagent consumption, especially acid and lime, is where the money goes, so small efficiency gains beat large capital gestures.

Recycling fits neatly into the same flowsheet. Black mass from spent lithium-ion batteries is leached and purified along very similar lines, and a plant set up for primary concentrate can often handle it with modest adjustment.

Finally, resist the temptation to shortcut test work. A flowsheet that looks obvious on paper can behave very differently at pilot scale, and every step here interacts with the next. Bench and pilot campaigns, run with a qualified metallurgist and your actual feed, will tell you more than any general description — including this one.

Photo: Lehoo88 / Pixabay