Nickel Cobalt Separation Solvent Extraction: Flowsheet

Last updated: 2026-10-28

Nickel cobalt separation solvent extraction is the decisive step in every modern battery-metal refinery: a 4–8 stage circuit that lifts a mixed nickel–cobalt liquor into battery-grade sulfate products by swinging cobalt (with zinc and manganese) into an organic phase while nickel stays behind. Every feed — MHP, MSP, HPAL liquor, or battery leachate — funnels into the same flowsheet spine, and the project’s economics hinge on how cleanly that spine is engineered. This guide walks that spine stage by stage, and it sits inside our solvent extraction resources library alongside the wider liquid-liquid separation equipment range.

Key Takeaways

• Nickel cobalt separation solvent extraction splits Ni from Co in 4–8 mixer-settler stages using phosphonic acid extractants.
• The flowsheet is sequential: feed prep, bulk impurity removal, the Ni/Co split, cobalt polishing, crystallisation.
• D2EHPA removes manganese, zinc and calcium ahead of the split; Cyanex 272 then performs the nickel–cobalt cut.
• Feed chemistry, not reagent choice, dictates stage count: Mg, Ca, Cl and organics add circuits and cost.
• Battery-grade specs are set by the offtaker; design to their impurity schedule, not to generic limits.

What Is Nickel Cobalt Separation Solvent Extraction?

Nickel cobalt separation solvent extraction is a hydrometallurgical refining step that selectively transfers cobalt from a mixed nickel–cobalt sulfate liquor into an organic phase while leaving nickel in the aqueous raffinate, typically across 4–8 counter-current contact stages. It sits near the end of the refining chain, after impurity removal and before crystallisation, and it is the point where an intermediate product stops being a concentrate and starts becoming a cathode precursor input.

To talk about the flowsheet precisely, four terms need working definitions.

Separation factor is a ratio that expresses how much better one metal distributes into the organic phase than another under the same conditions, typically running 50–500 for the cobalt-over-nickel cut with Cyanex 272 in sulfate media at pH 5.0–5.5. A higher number means fewer stages for the same purity; a low separation factor is paid for in extra mixer-settler boxes, more organic inventory, and more entrainment risk. Our McCabe-Thiele walkthrough shows how the isotherm and this number convert into a physical stage count.

Mixed hydroxide precipitate (MHP) is an intermediate product that captures nickel and cobalt from acidic leach liquors as hydroxide solids, typically carrying 30–45% nickel plus cobalt, with manganese, magnesium, calcium and iron as the dominant gangue elements. MHP has become the dominant traded form of Indonesian HPAL output precisely because it is cheap to ship and easy to redissolve at a coastal refinery — but redissolution loads the downstream circuit with every impurity the precipitation step carried along.

Battery-grade sulfate is a product specification class that defines the impurity ceiling for nickel sulfate hexahydrate or cobalt sulfate used in cathode precursor manufacture, typically demanding cobalt below the tens-of-ppm range in nickel sulfate and nickel below the tens-of-ppm range in cobalt sulfate, with magnesium, calcium, copper, zinc and iron each individually capped. There is no single universal standard: every precursor producer publishes its own schedule, and the offtaker specification always overrides any indicative number in this article.

And nickel cobalt separation solvent extraction, as a whole circuit, is therefore the combination of these: an impurity-removal front end, a high-selectivity cobalt extraction train, a stripping and polishing system, and the crystallisation step that turns clean liquor into saleable product. The circuit is only as good as its weakest stage, which is why the flowsheet — not any single reagent — is the right unit of analysis. A deeper reagent-level treatment of D2EHPA versus Cyanex 272 lives in our comparison page; this article stays on flowsheet logic.

The Nickel Cobalt Separation Solvent Extraction Flowsheet

Every credible cobalt refining flowsheet for battery feed follows the same sequence, whatever the upstream route. The order is not arbitrary: each stage protects the next by removing the species that would otherwise poison it. The numbered sequence below is the spine; the table that follows condenses it for scoping.

1. Feed preparation and redissolution. MHP or MSP is repulped and dissolved in sulfuric acid, often with sparging air or peroxide to oxidise iron(II) to iron(III). The purpose is a clear sulfate liquor with nickel and cobalt fully in solution. The key control parameters are terminal pH and oxidation-reduction potential — under-dissolved cobalt hydroxide is metal stranded in the residue, while over-hot acid pulls extra manganese and magnesium into solution. For MSP, a prior leach or chloride-wash decision may sit here; for battery leachate, this stage is instead filtration and organics removal.

2. Iron and aluminium removal. Iron(III) and aluminium are precipitated as hydroxides at pH 3.0–4.5. Purpose: remove metals that would otherwise co-extract with cobalt and contaminate strip liquor. Key control is pH — go too high and nickel or cobalt losses to the residue begin. This stage is often combined with stage 1 in tank-based plants.

3. Bulk impurity extraction with D2EHPA. Manganese removal D2EHPA duty is the workhorse of the front end: D2EHPA extracts manganese, zinc, copper and residual calcium ahead of the Ni/Co split, typically in 2–4 extraction stages at pH 3.5–4.5. Purpose: clear the species that would otherwise follow cobalt into Cyanex 272 and contaminate the cobalt product. Key controls are stage pH and the awareness that D2EHPA co-extracts cobalt over nickel — the loaded cobalt is usually recovered by a secondary strip circuit rather than lost. This dual behaviour is why D2EHPA’s role must be understood as part of the flowsheet, not in isolation.

4. The nickel cobalt split. The heart of the circuit: Cyanex 272 (or PC-88A) extraction in 4–8 stages at pH 5.0–5.5, taking cobalt into the organic while nickel stays in the raffinate. Purpose: produce a nickel raffinate clean enough for crystallisation and a loaded organic carrying the cobalt. Key controls are per-stage pH (the separation factor is acutely pH-sensitive), O/A ratio, and temperature. Stage count is set by the isotherm: a separation factor of a few hundred closes in five stages; a degraded reagent or a cold circuit can need eight.

5. Cobalt stripping and polishing. Loaded organic is stripped with dilute sulfuric acid in 2–3 stages, then the strip liquor is polished — often with Versatic acid or naphthenic acid to squeeze out residual calcium and magnesium, sometimes a small second D2EHPA pass for zinc. Purpose: bring the cobalt liquor inside the battery-grade impurity schedule. Key controls are strip acid strength (over-stripping builds a tenor that fights crystallisation) and polisher pH.

6. Nickel raffinate finishing. The nickel-bearing raffinate gets its own polishing: residual cobalt is scavenged, calcium and magnesium are trimmed where the specification demands it, and entrained organics are removed by carbon or flotation. Purpose: deliver battery-grade nickel sulfate feed liquor. Key controls are residual cobalt tenor and organic-in-aqueous entrainment, since entrained organic fouls the crystalliser.

7. Crystallisation to battery-grade sulfate. Both product liquors are concentrated by evaporation and crystallised as nickel sulfate hexahydrate or cobalt sulfate heptahydrate. Purpose: convert clean liquor into shippable product. Key controls are mother-liquor tenor, crystal habit, and impurity partitioning between crystal and mother liquor — a clean liquor entering the crystalliser is the whole point of stages 1–6, because crystallisation concentrates whatever is left.

Hydraulic and mechanical detail — settler sizing, dispersion band, phase continuity — is common to every SX circuit and is covered in our mixer settler design guide and in the stage efficiency article; here we stay on the chemical sequencing.

Feed Types and How They Change the Flowsheet

The spine above is universal; the impurity load is not. The feed table below is the fastest way to see why two refineries running identical Ni/Co split chemistry can differ by half in stage count and organic inventory.

Leaching

Interpreting the table, four patterns matter for a project manager.

MHP is the reference case, and magnesium is the silent tax. Indonesian MHP arrives with substantial magnesium and calcium from the acid and the ore; redissolution delivers them straight to the circuit. Neither Mg nor Ca extracts with Cyanex 272 to any troublesome degree, so the Ni/Co split itself is unaffected — but both accumulate in the nickel raffinate and will crystallise into the product unless polished. That is why the Versatic or naphthenic acid polishing stage exists, and why MHP-fed refineries carry more polishing capacity than the flowsheet diagram suggests. Bulk manganese removal D2EHPA duty is likewise non-negotiable: MHP manganese tenors are high enough that skipping the front end would push manganese straight into the cobalt strip.

MSP shifts the problem upstream. Mixed sulfide precipitate carries the smelter-adjacent impurity set — copper, iron, sometimes arsenic — and, if the intermediates were produced with chloride leaching, chloride itself. Chloride changes the flowsheet in two ways: it attacks stainless equipment, and at sufficient concentration it degrades the selectivity of the nickel–cobalt separation step by shifting cobalt speciation. The responses are chloride washing of the precipitate, or a dedicated wash stage inside the redissolution train, before any organic ever sees the liquor.

HPAL liquor skips redissolution but inherits everything else. Refineries integrated with an HPAL operation feed pressure leach liquor directly into bulk extraction, saving the MHP redissolution step. What they inherit is the full laterite impurity slate — manganese, magnesium, aluminium — plus the acidity and scale-management duties of pressure leach liquor. Our HPAL application page covers the upstream; the point here is that an integrated plant trades one flowsheet stage for tighter feed discipline, because there is no MHP precipitation step left to buffer ore variance from the SX circuit.

Battery leachate is the odd one out: low impurity tonnage, exotic species. Black mass leachate has a much lower Ni:Co ratio and brings copper, aluminium, fluoride, and organic contaminants (electrolyte residues, graphite fines) that no other feed carries. Fluoride is the flowsheet driver — it attacks extractants and linings alike — so a fluoride-scavenging step typically precedes the SX train, and organic cleanup protects the organic phase from degradation. Our battery recycling application page covers the upstream flowsheet; the refinery-side consequence is a bulk extraction stage configured for copper and manganese simultaneously rather than manganese alone.

The general rule for any feed: every impurity either costs an extra stage at the front end, or costs product quality at the crystalliser. There is no third option.

Key Operating Parameters and Typical Ranges

The table below collects the operating parameters a project manager will see in every vendor proposal and pilot report. All values are typical industry ranges for sulfate-media circuits and must be confirmed by testwork on your actual liquor before they harden into a design basis — treat them as the vocabulary of the vendor conversation, not the specification itself.

Three parameters deserve more than a table row.

Per-stage pH control in the split is the highest-leverage instrumentation in the circuit. The Cyanex 272 cobalt-over-nickel separation factor climbs steeply with pH, but so does the risk of nickel co-extraction once you push toward the hydrolysis window. Plants that control each extraction stage independently — rather than dosing one lime point upstream — routinely achieve the battery-grade cobalt ceiling in nickel sulfate with one fewer extraction stage than their neighbours. Over a 4–8 stage train, that is real capital.

Temperature is a compromise, not a maximise. Raising the circuit to the high end of the range improves disengagement and selectivity, but phosphonic acid extractants degrade measurably faster when hot, and every degradation product is a future entrainment and crud problem. The optimum is usually set by the organic consumption number, not by the isotherm.

Organic consumption is where OPEX hides. Quoted reagent prices look benign per litre; the circuit consumes organic through solubility, entrainment, degradation, and crud removal. A consumption figure of a few kilograms per tonne of contained metal sounds small until you multiply it across a 50,000 t/a nickel refinery. Ask every vendor for their basis — gross inventory, consumption rate, and whether degradation losses are included — before comparing numbers.

Where Do Battery-Grade Impurity Limits Bite?

Battery-grade expectations, stated here as indicative commercial norms that must be confirmed against the offtaker specification, generally run along these lines: cobalt below roughly 50 ppm in battery-grade nickel sulfate; nickel below roughly 50–100 ppm in cobalt sulfate; and magnesium, calcium, copper, zinc, iron and manganese each capped in the single-digit-to-tens-of-ppm band, with copper, zinc and iron frequently the tightest. The exact schedule is offtake-specific and version-specific — precursor makers tighten their schedules as their cathode customers tighten theirs. The question for flowsheet design is not what the numbers are, but which impurity forces which extra stage.

ASME VIII Div 1 vs Div 2: selection rules

Manganese forces the bulk extraction stage — always. Manganese follows cobalt into every cobalt-selective extractant, and it is abundant in every laterite-derived feed. Without a dedicated D2EHPA front end, manganese rides into the cobalt product at hundreds or thousands of ppm and cannot be polished out downstream at any reasonable cost. Manganese removal D2EHPA duty is therefore the first stage in every credible cobalt refining flowsheet for these feeds.

Zinc and copper force scavenging. Both extract strongly with D2EHPA, so the bulk stage removes them mostly as a passenger benefit — but zinc’s affinity is so high that trace zinc often survives to the split, and copper is a tenacious contaminant in cobalt sulfate. Where feed tenors are high (battery leachate in particular), a dedicated copper removal step — cementation or a separate extraction pass — precedes the bulk stage.

Magnesium and calcium are the expensive ones, precisely because they are chemically well-behaved. They do not co-extract dangerously at the split; they simply accumulate, stage after stage, in the nickel raffinate and the strip liquors, and they crystallise into the product or scale the equipment. The counters are the Versatic or naphthenic acid polishing stage on each product line, careful wash-water management on the loaded organic, and bleeding calcium from the circuit before it scales the crystalliser. Magnesium polishing is the single most under-scoped item we see in junior flowsheet reviews: the reagent system is modest, but the extra mixer-settler boxes, reagent inventory and operating attention are permanent.

Chloride is a flowsheet constraint, not an impurity limit. Chloride attacks austenitic stainless equipment, complicates materials selection toward more expensive alloys, and at elevated concentration interferes with the cobalt selectivity of phosphonic extractants. The flowsheet answer is upstream: chloride washing of MSP and any chloride-contaminated intermediate, plus rinse discipline on recycled acid, so the SX circuit itself never has to fight chloride. If your feed is chloride-bearing, the materials budget moves before the reagent budget does.

Fluoride and organics are battery-leachate-specific bites. Fluoride degrades organophosphorus extractants and attacks glass and linings; electrolyte-derived organics stabilise crud and poison phase disengagement. Both are handled upstream of SX — precipitative fluoride scavenging and carbon/polishing filtration — because neither is economical to manage inside the SX train.

The pattern to carry away: the Ni/Co split is rarely the bottleneck. The impurity schedule is won or lost at the front end and the polishing stages, which is exactly why the flowsheet, not the extractant, is the design unit.

What Does the Circuit Cost and How Long Does It Take?

Every figure in this section is indicative, scoping-grade, and stated without a guarantee: actual costs depend on feed tenor, impurity load, site labour rates, and the offtake specification. The purpose is to give a project manager a shape for the cost structure, not a budget number.

Capital structure. For a solvent extraction refining section processing tens of thousands of tonnes of contained nickel per year, CAPEX is dominated by mixer-settler boxes and their linings, the crystallisation train, solution handling (tanks, pumps, heat exchange), and reagent storage and dosing. On a per-tonne-of-annual-contained-nickel basis, refining-section capital for an SX-based battery metals refinery is conventionally discussed in the range of a few thousand US dollars per annual tonne — but this number moves by a factor of two with impurity load and chloride-driven materials selection, and it excludes upstream leaching and redissolution. Treat it as a magnitude, not a quote.

Operating cost structure. OPEX per tonne of contained nickel is built from acid and alkali consumption (the largest single item, driven by pH control across 10–20 stages), steam or power for evaporation and crystallisation, organic consumption, labour, and residue disposal. Acid consumption deserves its own line item because every extraction stage consumes one equivalent of acid per mole of metal extracted, and the strip gives it back — net acid use is set by the wash and neutralisation strategy, not by the extraction itself. Reagent cost is therefore dominated by acid and base, with extractant consumption a visible but smaller share.

Organic inventory is a capital-plus-operating hybrid. The initial organic fill — extractant plus diluent across all circuits — is working capital that scales with total stage volume, and it is charged, conditioned, and commissioned before first production. Inventory also defines the exposure if degradation or crud losses run high, which is why the monitoring discussed in our extractant degradation resource belongs in the operating budget from day one.

Schedule and commissioning. Indicative lead times for the major equipment packages run in the range of 30–60 weeks for mixer-settler fabrication (longer for exotic materials of construction) and comparable for crystallisers, with both packages usually on the critical path in parallel. Construction and mechanical completion after equipment arrival typically runs several months to a year depending on scale and site readiness. Commissioning follows a sequence that should be planned, not improvised: water batching and hydraulics first; then organic charging and conditioning; then synthetic liquor circuit runs; then feed introduction at reduced rate with the crystalliser last, because it is the least forgiving of off-spec liquor. A realistic commissioning allowance — months, not weeks — protects the offtake start date far better than optimistic scheduling.

The honest summary: costs scale with impurity load and specification tightness far more than with plant size. Two refineries of identical nickel capacity can differ materially in both capital and operating cost purely because one feeds clean HPAL liquor and the other feeds high-magnesium MHP.

FAQ

How many stages does nickel cobalt separation solvent extraction actually need?
The Ni/Co split itself typically runs 4–8 extraction stages depending on the separation factor, which is set by pH, temperature, and reagent condition. Add the bulk impurity extraction (2–4 stages) and strip and polish circuits on both product lines, and a full battery-grade refinery typically carries 15–25 mixer-settler boxes in total. The isotherm, not a rule of thumb, should fix the number.

Why is Cyanex 272 preferred over D2EHPA for the nickel–cobalt split?
D2EHPA extracts nickel and cobalt close together, so the split needs many more stages and risks nickel contamination. Cyanex 272’s steric structure gives a much larger cobalt-over-nickel separation factor in sulfate media, which is why it (or PC-88A, a close relative) is the standard split reagent. D2EHPA is still used — upstream, for manganese, zinc and calcium removal, where its behaviour is an advantage.

What is mixed hydroxide precipitate and why does it dominate feed discussions?
MHP is the nickel-cobalt hydroxide intermediate produced by HPAL operations — cheap to ship, easy to redissolve in sulfuric acid at a coastal refinery. Its dominance in Indonesian supply has made it the default design feed for new refineries. Its weakness is that redissolution brings manganese, magnesium and calcium along, which is why the impurity-removal front end, not the Ni/Co split, is often the design bottleneck for MHP projects.

Can the flowsheet skip the bulk impurity extraction stage?
Only if the feed is exceptionally clean, which almost never happens with laterite or battery feeds. Manganese follows cobalt into every cobalt-selective extractant, so without a D2EHPA front end the manganese reports to the cobalt product at concentrations no offtaker will accept. Removing manganese downstream is far more expensive than removing it upstream, so the bulk stage is effectively mandatory for MHP, HPAL liquor, and battery leachate alike.

What does “battery-grade” actually mean for nickel and cobalt sulfate?
It means the liquor meets the offtaker’s impurity schedule — cobalt below the tens-of-ppm range in nickel sulfate, nickel similarly capped in cobalt sulfate, and Mg, Ca, Cu, Zn, Fe, Mn each individually limited. There is no universal standard number; each precursor producer publishes its own schedule and tightens it over time. Design to a stated offtake specification and keep margin, because the specification will get stricter during the plant’s life.

How is magnesium removed from nickel sulfate liquor?
Typically by a Versatic acid or naphthenic acid polishing circuit: at controlled pH, magnesium (and calcium) extract into the organic while nickel stays behind, producing a low-Mg raffinate fit for crystallisation. Because Mg does not extract with the cobalt split chemistry, it accumulates in the nickel line and must be handled there. The polishing stage is small but permanent — one of the most commonly under-scoped items in early flowsheet estimates.

How long does a refinery SX section take to commission?
After mechanical completion: water batching and hydraulic checks, organic charging and conditioning, synthetic-liquor circulation, then feed introduction at reduced rates with the crystalliser brought on last. Realistically this spans several weeks to a few months, depending on readiness and feed discipline. Budget a proper commissioning period because the crystalliser is unforgiving of off-spec liquor, and re-crystallising a contaminated first crop is a costly way to learn pH control.

Conclusion

Nickel cobalt separation solvent extraction is best understood as a sequence, not a reagent choice: prepare and clean the feed, remove manganese and the other hitchhikers with D2EHPA, make the nickel–cobalt cut with Cyanex 272 across 4–8 well-instrumented stages, polish both product lines against the offtake’s impurity schedule, and only then crystallise. Projects that respect that order — and that scope the magnesium, calcium and chloride problems as seriously as the split itself — reach battery-grade specification on schedule. Projects that treat the split as the whole design usually buy their missing stages later, at retrofit prices. If you are scoping a refinery based on MHP, HPAL liquor, or battery leachate, talk to our process team about your flowsheet — an early review of the feed table is the cheapest engineering you will ever buy.

References

• Thornton, J. D., Science and Practice of Liquid-Liquid Extraction, Oxford University Press.
• Lo, T. C., Baird, M. H. I., and Hanson, C., Handbook of Solvent Extraction, Wiley.
• Ritcey, G. M., Solvent Extraction: Principles and Applications to Process Metallurgy, G.M. Ritcey & Associates.
• Crundwell, F. K., Moats, M. S., Ramachandran, V., Robinson, T. G., and Davenport, W. G., Extractive Metallurgy of Nickel, Cobalt and Platinum Group Metals, Elsevier.
• Wills, B. A., and Finch, J. A., Wills’ Mineral Processing Technology, 8th ed., Butterworth-Heinemann.
• USGS, Mineral Commodity Summaries — Nickel; Cobalt. USGS Mineral Commodity Summaries
• Journals consulted for this topic: Hydrometallurgy; Minerals Engineering; Solvent Extraction and Ion Exchange.

Changelog

• 2026-10-28 — first published.