Solvent Extraction in Hydrometallurgy: Process & Equipment

Last updated: 2026-10-18 · Reviewed by: Senior Metallurgical Engineer, {BRAND}

A single well-designed solvent extraction stage transfers 90–95% of the target metal out of the aqueous feed in one pass, and organic inventory typically runs 0.6–1.0 m³ of organic for every m³/h of aqueous feed treated — a capital line comparable to the contactors themselves. Both figures are typical industry ranges for copper circuits; the exact values belong to test work.

Key Takeaways

• Solvent extraction purifies and concentrates metal between leaching and electrowinning; single-stage recovery of 90–95% is normal.
• Extractant choice is chemistry-first: oximes for copper, D2EHPA and Cyanex 272 for Ni–Co–Zn, amines for U and V.
• Mixer-settlers are the default above roughly 50 m³/h; centrifuges buy residence time; pulsed columns buy confinement.
• Performance is set by O/A ratio, mixer residence, settler flux and entrainment — confirm each by test work.
• Organic inventory is a capital item; entrainment, not solubility, dominates ongoing organic losses.

What Is Solvent Extraction in Hydrometallurgy?

Solvent extraction is a liquid-liquid separation operation that transfers a dissolved metal ion from an aqueous leach liquor into an immiscible organic phase, and then back into a second aqueous phase, so that the metal leaves the step both purer and more concentrated than it entered. In a copper plant it sits between leaching and electrowinning: the organic selectively picks copper out of pregnant leach solution, rejects iron and chloride, and hands a clean, concentrated electrolyte to the tankhouse. The same architecture serves nickel, cobalt, zinc, uranium, vanadium and the rare earths, with different reagents.

Copper Solvent Extraction

Extractant Degradation

Solvent Extraction

Mixer-Settler Settler Area and Mixer Volume Calculator in practice

Three definitions carry most of the technical weight, because vendor proposals tend to use them loosely.

A mixer-settler is a compartmented contacting device that mixes the two phases for a controlled residence time and then releases them into a quiescent settler, where they disengage by gravity. One mixer-settler is one theoretical stage, minus the efficiency loss.

A diluent is a purified hydrocarbon carrier, usually a kerosene cut, that dissolves the extractant and sets organic viscosity, density and flash point. It does almost none of the chemistry but controls most of the phase-disengagement behaviour.

Crud is a stable gummy emulsion of fine solids, organic and trapped aqueous that collects at the interface and, if ignored, migrates into every downstream tank.

The economics of the step are unusual for a hydrometallurgical plant. There is almost no thermal duty — SX runs at 30–45 °C in most circuits — so operating cost is dominated by reagent losses and pumping rather than energy. Capital is dominated by settler area and by the organic charge itself. And the step is where product purity is actually made: a leach circuit can tolerate a sloppy impurity profile if the SX reagent rejects it. That is why reagent selection and contactor selection deserve separate, deliberate decisions rather than a single vendor package, and why the rest of this page treats them separately.

One boundary is worth drawing precisely, because it is asked in every feasibility review: ion exchange (IX) resins perform the same load-and-strip function, and at flows below a few tens of cubic metres per hour IX is often cheaper, because no settler area is needed and no organic is lost to entrainment. Above roughly 30–50 m³/h the comparison reverses — solvent extraction wins on reagent cost per kilogram of metal, on scale of equipment, and on tolerance to the suspended solids that blind resin beds. Most copper, nickel-cobalt and uranium flowsheets above demonstration scale are built on SX for exactly those three reasons, and the equipment sections below assume that scale.

Which Extractant Family Fits Which Metal?

Selectivity lives in the extractant molecule, not in the equipment. Three families cover most metallurgical duty, and the choice within a family is a pH-stripping trade-off.

Oximes for copper. Modified aldoxime and ketoxime reagents extract Cu²⁺ from dilute sulphuric acid with copper-to-iron selectivity high enough that EW-grade cathode is routine. They strip cleanly into spent electrolyte at 150–180 g/L H₂SO₄, which conveniently closes the loop with the tankhouse. Typical concentration is 7–12% v/v in diluent; more extractant raises copper loading but slows phase disengagement and raises losses.

Acidic organophosphorus reagents for zinc, nickel and cobalt. D2EHPA (di-2-ethylhexyl phosphoric acid) extracts zinc and trivalent impurities strongly and is the workhorse of zinc SX and bulk impurity removal. Cyanex 272 (a phosphinic acid) sacrifices some strength for a much cleaner cobalt-over-nickel split near pH 5.0–5.5. The two are constantly compared in Ni–Co projects, and the trade-off is set out in detail in D2EHPA vs Cyanex 272; as a screening rule, D2EHPA for zinc and bulk removal, Cyanex 272 when the cobalt-nickel split itself is the product.

Amines for uranium and vanadium. Tertiary amines extract the anionic sulphate complexes of U(VI) and V(V) from acid leach liquors, and strip into carbonate or chloride media. Quaternary amines extend the range to weaker-acid systems. Amine circuits demand clean feed, because amines are the most crud-prone family in industrial service.

Table 1 — Principal extractant families in metallurgical solvent extraction (typical ranges, project-specific)

Two practitioner notes. First, ferric iron is the quiet antagonist of every acidic extractant: it loads strongly, strips poorly and oxidises the reagent, so iron control is a feed-chemistry problem before it is a reagent problem. Second, extractant concentration is a disengagement variable, not only a capacity variable — raising concentration beyond the isotherm’s needs usually shows up as slower settling and higher entrainment within weeks.

Mixer-Settler, Centrifugal Extractor or Pulsed Column: Which Contactor Fits Your Duty?

Contactor selection is a residence-time, throughput and cleanliness decision, and it resolves cleanly once those three are known.

Mixer-settlers are the default for flows above roughly 50 m³/h of aqueous, for circuits with any solids or crud exposure, and for chemistry that needs minutes of residence — which describes every copper oxime duty. They are the lowest-CAPEX option per stage, they are visually inspectable, and a malfunctioning stage is obvious to an operator walking the deck. The design details that decide whether they work well — box proportions, draft tubes, baffling — are covered in mixer-settler design and in our mixer-settler equipment page.

Centrifugal extractors buy residence time and cleanliness. When extraction is fast but the system is hydrolysis-prone, or when the stream is high-value and small — rare earths, niobium, refinery bleed streams — a machine that contacts and separates in seconds inside a spinning bowl wins on footprint, entrainment and inventory. The price is rotating equipment: seals, bearings and a maintenance skill set that a tank farm does not need. Details in the centrifugal extractor equipment page.

Pulsed columns buy confinement and single-vessel throughput. With no internal moving parts, one column replaces many stages, which is why they dominate where operator exposure or criticality control matters — refined uranium duty — and where feed is clear and flows are large. They flood, they are hard to sample stage by stage, and they cannot be retrofitted into a settler-based layout cheaply.

Table 2 — Contactor comparison for liquid-liquid separation duty (indicative, to be confirmed by vendor data and test work)

Read as a decision rule: proven chemistry plus large, dirty flows → mixer-settlers; fast kinetics plus small, clean, valuable streams → centrifuges; confinement or clear-feed bulk duty → columns. Hybrid layouts are legitimate — a pulsed column ahead of mixer-settler strip stages is a common uranium arrangement.

Which Design Parameters Set Solvent Extraction Performance?

Five numbers define a SX flowsheet, and four of them are settled by bench work before a single plate is cut. They are summarised below and then discussed in the order in which a design usually fixes them.

O/A ratio. The organic-to-aqueous flow ratio is fixed by the extraction isotherm, not by preference: it is the ratio that reaches target raffinate grade at the chosen stage count. Copper extraction typically runs near 1:1; copper stripping runs at 2:1 to 5:1 O/A because loaded organic carries only 40–60 g/L of copper. An O/A far from 1:1 in extraction is a warning sign that the isotherm or the reagent concentration is wrong.

Mixer residence time. Oxime systems reach near-equilibrium in 1.5–3 minutes; slower chemistries and viscous organics can need up to 5. Residence below the kinetic requirement shows up as falling stage efficiency, which is misdiagnosed as bad reagent far more often than it is measured.

Settler flux. Design total-phase flux of 4–8 m³/(m²·h) is the typical copper-SX envelope, with well-baffled modern settlers reaching 10 under proven disengagement behaviour. Flux is not a table lookup — it is measured, and the measurement method and its traps are covered in phase disengagement testing.

Entrainment. Typical targets are below about 50 ppm aqueous-in-organic and below 10–20 ppm organic-in-aqueous ahead of electrowinning, using residence, picket fences and, where needed, flotation polishers. Entrainment is an operating cost measured in litres per hour; see the cost section.

Extractant concentration and diluent. Concentration follows the isotherm (Table 1); diluent composition — aliphatic content, aromaticity, flash point — is chosen for disengagement speed and safety, and it changes how the same extractant behaves more than most reagent suppliers admit.

A sixth number, stage efficiency, is an output rather than an input: 90–98% is normal for well-designed mixer-settlers, and every point below it buys another stage or another mixer minute.

Table 3 — Design parameter envelope for mixer-settler solvent extraction circuits (typical industry ranges)

The envelope exists to be tested against, not adopted from. A project that accepts every value in this table without measuring its own isotherm and disengagement behaviour has outsized its design risk to the one discipline where bench data is cheapest to obtain.

How Many Stages Do You Need? McCabe-Thiele Staging in Practice

Stage counting is graphical, and the diagram is worth drawing even when software exists, because it exposes the assumptions.

Table 4 — Worked staging example: copper extraction, modified oxime, typical figures

The procedure is standard: measure the batch isotherm on real liquor at design temperature and reagent concentration, draw the operating line with slope minus A/O, and step off equilibrium stages at the isotherm. Two refinements separate a working design from a textbook one. First, divide by stage efficiency — a 95% efficient stage turns two theoretical stages into a little over two real ones, while an 80% efficient amine stage turns four theoretical stages into five. Second, add margin for isotherm shift: feed tenor, acid strength and temperature all move the curve over the ore body’s life, and a design with zero margin at year-one assays will miss raffinate grade in year four. The full construction, with a worked isotherm and the common errors, is in McCabe-Thiele and extraction isotherms.

The staging logic also explains why solvent extraction circuits are so rarely redesigned after start-up: stage count, O/A and reagent concentration are coupled, so a raffinate that misses grade cannot be fixed by one variable alone. The usual plant response is to raise reagent concentration first — it shifts the isotherm without touching hardware — but that move costs disengagement speed and entrainment, which is why the margin belongs in the original design rather than in the operating manual.

Why Does Crud Form in Solvent Extraction, and How Do You Treat It?

Crud formation is a feed-quality symptom wearing a process-chemistry disguise. The initiators are fine solids reaching the interface — colloidal silica and clays from upstream settling, precipitated iron, aluminium or manganese, humic material — sometimes helped along by bacteria growing in the diluent and by excess agitation energy emulsifying everything together. Treatment follows a fixed sequence, and skipping the first step is why crud programs fail:

1. Characterise the crud. Sample it at the interface, settle or centrifuge a known volume, and measure ash content and loss on ignition. An inorganic crud points at upstream clarification; an organic, oily crud points at diluent degradation or biological growth. The treatment is different and the sampling costs a shift, not a project.
2. Remove the feed solids. Polish filtration or additional clarification to hold suspended solids below roughly 10–20 mg/L to SX feed, fix the upstream thickener, and stop blending in ore zones that shed colloidal silica. This is where 80% of crud problems are actually solved.
3. Adjust the chemistry. Rebalance extractant concentration, review diluent aromatic content, and introduce biocide in organic storage where loss-on-ignition shows biological activity. Small chemistry moves beat heroic hardware.
4. Recover the organic. Treat crud through a dedicated centrifuge or acid wash; typical recovery of 40–70% of the organic held in crud pays back the unit quickly, and the recovered organic must be assayed before it re-enters the circuit.
5. Manage the interface. Hold interfacial level with real control, keep pumping suctions away from the interface, and treat the settler’s first metres as the crud trap they are.

The failure modes behind each step, with photographs and disposition options, are in crud identification and treatment. Treat crud as a monitored plant metric — interfacial band thickness — rather than as a housekeeping chore.

What Does Solvent Extraction Cost, Including Organic Inventory?

Three cost lines behave differently from the rest of a hydromet plant, and each deserves its own line in the feasibility model.

Organic inventory is a capital item. At 0.6–1.0 m³ of organic per m³/h of aqueous throughput, a 500 m³/h copper plant carries 300–500 m³ of diluent and extractant charged at commissioning. This inventory rivals the cost of the contacting equipment itself, it is financed on day one, and it never depreciates — it is top-upped forever. Size it explicitly; do not let it hide inside “equipment supply”.

Operating losses are entrainment-dominated. At 100 m³/h of raffinate flow, every 10 ppm of organic-in-aqueous entrainment is roughly 1 L/h of organic leaving the plant — on the order of 8 m³ per year. That arithmetic is why entrainment targets belong in the equipment specification and why polishers frequently pay back within a year. Total organic make-up in a well-run copper SX plant is typically 2–8 L per tonne of cathode, site-specific and dominated by entrainment plus reagent degradation rather than by solubility.

Energy is modest and should be claimed honestly. Mixer agitation typically draws 0.4–0.8 kW per m³ of mixed liquor, and total SX plant energy for copper is typically an order of magnitude below the electrowinning duty it feeds — a few percent, not a cost driver. The exception is centrifugal service, where bowl drive power raises the figure and belongs in the comparison when contactor options are being priced.

The full context of where SX sits in the copper value chain, with the EW numbers next to it, is in copper SX-EW applications. All figures here are indicative industry ranges on a 2025 cost basis, stated for orientation; project numbers require vendor quotations and current reagent pricing.

FAQ

Q1. What causes crud in solvent extraction?
Crud is usually triggered by fine solids reaching the aqueous-organic interface — colloidal silica and clays from incomplete upstream settling, precipitated iron, aluminium or manganese, sometimes humic material or bacteria growing in the diluent. Over-agitation then emulsifies the solids with organic into a stable interfacial layer that hardens over weeks. The cure starts upstream: clarify the feed first, then manage interface level and agitation energy.

Q2. What is the difference between a mixer-settler and a centrifugal extractor?
Mixer-settlers contact and separate the phases by gravity in two large vessels per stage, giving 1.5–3 minutes of residence and high solids tolerance at the lowest capital cost. Centrifugal extractors do both inside a spinning bowl within seconds, cutting entrainment and footprint but adding rotating-equipment maintenance. Above roughly 50 m³/h of aqueous flow, mixer-settlers usually win on total cost of ownership.

Q3. How is the O/A ratio chosen?
The organic-to-aqueous ratio is set by the extraction isotherm, not by preference. It is the ratio that reaches the target raffinate grade at the chosen number of stages, verified by stepping off a McCabe-Thiele diagram. Copper extraction typically runs near 1:1, while stripping runs at 2:1 to 5:1 organic-to-aqueous, because loaded organic carries only about 40–60 g/L of copper.

Q4. How much entrainment is acceptable between the phases?
Typical design targets are below about 50 ppm aqueous-in-organic and below 10–20 ppm organic-in-aqueous ahead of electrowinning, achieved by settler residence, coalescing media and flotation polishers. Every 10 ppm of organic lost in raffinate at 100 m³/h is roughly 1 L/h of make-up, so entrainment control pays for itself quickly. Treat the targets as indicative and measure them on your own circuit.

Q5. When is Cyanex 272 chosen instead of D2EHPA?
Cyanex 272 separates cobalt from nickel far better than D2EHPA, extracting cobalt selectively near pH 5.0–5.5 while nickel largely remains in solution. D2EHPA extracts zinc, cobalt and trivalent impurities more strongly but then struggles to strip them apart cleanly. Choose Cyanex 272 for cobalt-nickel separation duty, D2EHPA for zinc or bulk impurity removal, and confirm the split with bench isotherms.

Q6. How many stages does a copper SX circuit need?
Two extraction stages and one or two strip stages are typical, delivering better than 90% overall copper extraction with modified oxime reagents. The count comes from stepping stages off the batch isotherm on a McCabe-Thiele diagram at the design O/A ratio and dividing by stage efficiency. Shallow isotherms, common in cobalt and nickel systems, can push extraction to four or five stages.

Q7. Why is organic inventory treated as a capital item?
Inventory is charged at commissioning and never leaves the plant. At 0.6–1.0 m³ of organic per m³/h of aqueous throughput, a 500 m³/h copper plant carries 300–500 m³ of diluent and extractant, a figure that can rival the cost of the contacting equipment. Ongoing make-up through entrainment and degradation adds an operating line, so inventory sizing and loss control belong in the feasibility study.

Conclusion

Solvent extraction is where a hydrometallurgical flowsheet earns its purity: the extractant family sets selectivity, the contactor sets capital and robustness, and staging, residence and settler flux set recovery. Treat organic inventory as capital, entrainment as an operating cost, and crud as a feed-quality symptom rather than a mystery. The deeper treatments behind every section above — equipment detail, design methods, test procedures — are indexed in our resources library.

If the isotherm, settler flux or crud risk on your project is still unresolved, send us the liquor analysis and target specification through our pilot testing and metallurgical test work service. A four-week bench campaign settles O/A ratio, stage count and entrainment targets before they are cast in steel.

References

1. B.A. Wills & J. Finch, Wills’ Mineral Processing Technology, 8th ed., Elsevier, 2016 — hydrometallurgy and solvent extraction chapters.
2. SME Mineral Processing & Extractive Metallurgy Handbook, Society for Mining, Metallurgy & Exploration, 2019 — liquid-liquid separation practice. smenet.org
3. USGS Mineral Commodity Summaries 2025 — copper, nickel, cobalt and zinc supply context. usgs.gov
4. G.M. Ritcey & A.W. Ashbrook, Solvent Extraction: Principles and Applications to Process Metallurgy, revised ed., 1984.
5. ISEC — International Solvent Extraction Conference proceedings, recent editions (extractant chemistry and contactor design papers).
6. ALTA Metallurgical Project Development proceedings — nickel-cobalt-copper SX sessions.
7. MetSoc Conference of Metallurgists / Hydroprocess proceedings — copper SX-EW plant practice and crud case studies.

Changelog

• 2026-10-18 — Initial pillar refresh for the liquid-liquid separation cluster. Extractant family table, contactor comparison, design parameter envelope, crud treatment sequence, inventory and cost basis.

{
“@context”: “https://schema.org”,
“@graph”: [
{
“@type”: “TechArticle”,
“headline”: “Solvent Extraction in Hydrometallurgy: Process & Equipment”,
“about”: [“solvent extraction”, “liquid-liquid separation”, “mixer-settler”, “centrifugal extractor”, “D2EHPA”, “Cyanex 272”, “crud”],
“datePublished”: “2026-10-18”,
“dateModified”: “2026-10-18”,
“author”: {“@type”: “Person”, “name”: “Process Metallurgist, {BRAND}”},
“publisher”: {“@type”: “Organization”, “name”: “{BRAND}”}
},
{
“@type”: “FAQPage”,
“mainEntity”: [
{“@type”: “Question”, “name”: “What causes crud in solvent extraction?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “Fine solids reaching the aqueous-organic interface — colloidal silica, clays, precipitated iron, aluminium, manganese, humics or bacteria — combined with over-agitation. Clarify the feed first, then manage interface level and agitation energy.”}},
{“@type”: “Question”, “name”: “What is the difference between a mixer-settler and a centrifugal extractor?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “Mixer-settlers contact and separate by gravity in 1.5–3 minutes with high solids tolerance at lowest capital cost; centrifugal extractors do both in seconds with lower entrainment but rotating-equipment maintenance.”}},
{“@type”: “Question”, “name”: “How is the O/A ratio chosen?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “By the extraction isotherm: it is the ratio that reaches target raffinate grade at the chosen stage count. Copper extraction typically runs near 1:1; stripping runs at 2:1 to 5:1 O/A.”}},
{“@type”: “Question”, “name”: “How much entrainment is acceptable between the phases?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “Typical targets are below about 50 ppm aqueous-in-organic and 10–20 ppm organic-in-aqueous ahead of electrowinning. Every 10 ppm in raffinate at 100 m³/h is roughly 1 L/h of organic make-up.”}},
{“@type”: “Question”, “name”: “When is Cyanex 272 chosen instead of D2EHPA?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “When the cobalt-over-nickel split is the product: Cyanex 272 separates Co from Ni near pH 5.0–5.5, while D2EHPA suits zinc and bulk impurity removal. Confirm with bench isotherms.”}},
{“@type”: “Question”, “name”: “How many stages does a copper SX circuit need?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “Typically two extraction and one or two strip stages for better than 90% extraction with oxime reagents, counted from a McCabe-Thiele diagram at the design O/A ratio.”}},
{“@type”: “Question”, “name”: “Why is organic inventory treated as a capital item?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “At 0.6–1.0 m³ of organic per m³/h of aqueous throughput, inventory is charged at commissioning, never leaves the plant, and can rival the cost of the contacting equipment itself.”}}
]
},
{
“@type”: “BreadcrumbList”,
“itemListElement”: [
{“@type”: “ListItem”, “position”: 1, “name”: “Home”, “item”: “/”},
{“@type”: “ListItem”, “position”: 2, “name”: “Liquid-Liquid Separation”, “item”: “/liquid-liquid-separation/”},
{“@type”: “ListItem”, “position”: 3, “name”: “Solvent Extraction”, “item”: “/liquid-liquid-separation/”}
]
}
]
}