Phase Disengagement: 8 Causes of Slow Settling in SX

Last updated: 2026-11-01

Phase disengagement is the process by which the organic-aqueous dispersion created in a mixer breaks and coalesces back into two clean layers, and in a well-run mixer-settler it should be substantially complete within 60-120 seconds of quiescent residence. When it slows, entrainment rises in both directions: organic carries into the aqueous raffinate and aqueous carries into the loaded organic. This troubleshooting guide gives process engineers eight root causes with symptoms and fixes, the settling tests that pinpoint the problem, and the settler design rules that keep a solvent extraction circuit stable.

Key Takeaways

• Phase disengagement should reduce a fresh dispersion to two clean layers in 60-120 seconds.
• Measure it with dispersion band depth in the settler and a batch settling half-life test.
• Excess impeller tip speed, fine crud solids, and cold liquor are the three most common causes.
• Keep the dispersion band below 30-50% of settler depth to protect against entrainment and flooding.
• A daily band reading plus a monthly batch test catches degradation before a shutdown.

What Phase Disengagement Is and Why SX Settlers Depend on It

Phase disengagement is a physical separation process that governs how quickly the droplet dispersion leaving an impeller coalesces into continuous organic and aqueous layers, typically 60-120 seconds of quiescent residence in a healthy circuit. In every solvent extraction stage, the mixer deliberately shreds one phase into the other to create the interfacial area needed for metal transfer. That mixing step is a temporary investment: the moment the emulsion enters the settler, the plant needs the opposite behavior, meaning droplets must collide, coalesce, and release dissolved gases so that each phase leaves the settler essentially free of the other.

how D2EHPA vs Cyanex 272 is decided

Mixer-Settler Settler Area and Mixer Volume Calculator in practice

The settler is where the circuit pays for its mixing. Everything the settler produces depends on how fast the dispersion collapses. A shallow, fast-breaking dispersion band means the organic and aqueous inlets are quickly separated, entrainment is low, and each stage transfers metal at design recovery. A deep, persistent band means droplets are still suspended when liquor reaches the outlet weirs, and whatever has not coalesced by then leaves with the wrong phase.

Dispersion band depth is an operating measurement that tracks the thickness of the emulsion layer held at the organic-aqueous interface, typically 5-30 centimeters in a healthy industrial settler. The band is the plant’s real-time report card on disengagement. It absorbs short-term upsets — a pump surge, a momentary O/A excursion — without letting un-coalesced droplets escape. When the band grows beyond roughly 30-50% of the settler depth, the margin is gone and entrainment climbs sharply.

Three properties of the system determine how fast the dispersion breaks: droplet size leaving the mixer (smaller droplets settle more slowly, following a squared relationship between size and velocity), the physical properties of the two phases (viscosity, density difference, and interfacial tension), and the chemistry of the interface (extractant degradation products, fine solids, and dissolved air all act at the droplet surface and slow coalescence). Each of the eight faults in this article traces back to one of those three levers. For a broader treatment of the whole circuit, see our solvent extraction resource library.

How Do You Measure Phase Disengagement in a Settler?

Two complementary measurements cover nearly every troubleshooting need: the dispersion band depth read in the operating settler, and the batch settling half-life measured on a sample. Use the first as a daily health check and the second as a diagnostic that isolates liquor chemistry from settler hydraulics.

Reading band depth on a running unit is straightforward but demands consistency. Always read at the same location, between the inlet baffle and the first picket fence where the dispersion is most developed, and always at steady throughput. In open settlers the band is visible as a hazy, brown-to-amber layer between the bright organic and the darker aqueous; a calibrated sight glass, dip plate, or interface probe works where the surface is obscured. Record the number every shift in the same units so trends are meaningful.

The batch settling test removes settler geometry and hydraulic short-circuiting from the picture. It answers one question: given the liquor as it actually leaves the mixer, how fast would it break under ideal quiescent conditions?

1. Sample at the mixer outlet. Take the liquor from the overflow between mixer and settler, not from the settler itself, so the sample represents the dispersion exactly as it is fed forward.
2. Fill a graduated cylinder immediately. Use a 1,000 milliliter glass or clear acrylic cylinder, fill to a marked height, and start the clock the moment filling stops; delay lets coalescence begin before you are measuring.
3. Record the interface against time. Note the dispersion band boundaries at 0, 15, 30, and 60 seconds, then every minute until the band falls to a stable minimum; note which phase the droplets disperse into, because organic-continuous and aqueous-continuous dispersions break differently.
4. Calculate the half-life. The half-life is the time for the band depth to fall to half of its initial value; a coalescence rate can be derived from the slope of band depth versus time over the first minute.
5. Compare against your baseline. The number means nothing in isolation; keep a rolling baseline per stage and per season, and treat a doubling of half-life as a fault signal even if the settler still looks acceptable.

Typical half-life values to be confirmed by test work sit in the 20-90 second range for clean circuits. If the batch test breaks fast but the band in the settler is still deep, the problem is hydraulic — overloaded settler, short-circuiting, or excessive inlet turbulence — not chemistry. If the batch test itself is slow, the fault is in the liquor: mixing energy, solids, degradation products, temperature, or composition. That single distinction cuts the fault list in half before you open a manual.

What Happens When Phase Disengagement Fails?

Every unit of un-coalesced dispersion that reaches the outlet weirs leaves as entrainment, and it travels in both directions. The consequences are asymmetric and both are expensive.

Organic in aqueous is the more visible loss. Organic droplets carried out with the raffinate or the barren liquor represent direct reagent make-up cost — diluent, extractant, and modifier all leave the circuit and must be replaced. Beyond the cost, dissolved and entrained organic interferes with downstream units: in a copper circuit it can coat subsequent filters and carbon beds; in nickel and cobalt circuits it contaminates products that carry tight organic specifications. Plant audits commonly find organic-in-aqueous entrainment anywhere from tens to a few hundred parts per million when disengagement degrades, versus single digits in a clean circuit — indicative figures only, to be confirmed by sampling your own units. Our dedicated page on SX entrainment control covers measurement and recovery options in depth.

Aqueous in organic is often the more damaging direction. Entrained aqueous travels with the loaded organic into the strip stage and reports to the strip liquor, diluting the metal tenor and dragging impurities — chloride, manganese, iron, silica depending on the system — directly into the electrolyte fed to electrowinning. The classic symptoms in a copper tankhouse are off-spec cathode, crud on cathode surfaces, and erratic current efficiency, while the root cause sits back in the extraction settler. Aqueous entrainment also accelerates degradation of the strip-stage electrolyte balance and forces higher wash rates.

There is a third, slower consequence: lost turndown margin. As the band deepens, the useful clear-phase depth in the settler shrinks, so the circuit has less tolerance for rainwater dilution, ore-blend changes, or throughput increases. Many “sudden” flooding events are actually the visible end point of months of gradually worsening phase disengagement. Flooding itself is a distinct failure mode with its own diagnosis, covered in detail in our mixer-settler stage efficiency guide.

8 Causes of Slow Phase Disengagement and How to Fix Each

Work through this list in order of cost and likelihood. In most audits, faults 1, 2, and 4 explain the majority of slow-settling events.

1. Excessive mixing energy. Symptom: very fine, milk-like dispersion; batch half-life long; band deepest at high agitator speeds. Root cause: impeller tip speed above roughly 3-4 meters per second (indicative) shatters droplets into a size range that coalesces slowly, and over-dispersion wastes pumping energy. Fix: verify tip speed against design, reduce speed or fit a larger-diameter, lower-speed impeller, and confirm that mass transfer (extraction efficiency) holds after the change — a modest efficiency trade is usually far cheaper than chronic entrainment.
2. Fine solids feeding the interface. Symptom: a stable, mud-like interfacial layer that thickens over weeks; half-life creeps upward stage by stage. Root cause: fine insoluble solids — silica, clays, precipitated hydroxides — stabilize droplets by coating interfaces, the mechanism known as crud seeding. Fix: improve upstream clarification and filtration, and manage the accumulated layer before it seeds every stage; the chemistry and treatment options are covered in our crud treatment guide.
3. Extractant degradation products acting as surfactants. Symptom: gradual decline in break rate across all stages, worst after hot spots or oxidizing conditions; interfacial tension measurably down. Root cause: hydrolysis and oxidation of the extractant generate surface-active species that adsorb on droplets and slow coalescence. Fix: review reagent quality and operating temperature/oxidant exposure, refresh or treat the organic inventory; the mechanisms are summarized in our extractant degradation resource.
4. Temperature drop. Symptom: settling slows in winter, on night shift, or after a maintenance window; recovers when the circuit warms. Root cause: lower temperature raises both-phase viscosity and can lower interfacial tension behavior at the droplet surface; viscosity slows droplet rise and fall roughly in proportion. Fix: insulate settlers and piping, route liquor through heat recovery where available, and hold the circuit within the temperature band the reagent supplier recommends, typically within 3-5 degrees Celsius of design (indicative).
5. High ionic strength / salting-out changes. Symptom: settling changes after an ore-blend or feed-chemistry shift, even at constant temperature and speed. Root cause: dissolved salts raise aqueous viscosity and can shrink the aqueous-organic density differential; both effects slow droplet separation, and a smaller density difference is the more damaging of the two. Fix: track feed assay and liquor density as control parameters, adjust O/A ratio to restore the density gap, and re-baseline the batch test after any major feed change.
6. Wrong O/A ratio and mixer residence. Symptom: dispersion band grows with throughput even though liquor chemistry tests fine; the batch test from the mixer outlet is slower than from a well-mixed side sample. Root cause: an O/A ratio or residence time that over-disperses the feed — too much energy per unit volume, or a mixer running organic-continuous when it should be aqueous-continuous. Fix: re-optimize the O/A ratio and mixer residence time against the settler’s capacity; our mixer-settler design guide walks through the coupling between mixer duty and settler load.
7. Entrained air from mixer vortexing. Symptom: frothy surface layer, air bubbles visible in the dispersion band, band depth reads erratically. Root cause: a low mixer liquid level or missing vortex breaker lets the impeller draw air; bubbles attach to droplets and lower their effective density, so droplets neither rise nor fall cleanly. Fix: restore minimum submergence, fit or repair vortex breakers and baffles, and check for air ingress on the suction side of interstage pumps.
8. Settler overloaded. Symptom: band depth scales directly with plant flow rate; flooding at design throughput that was manageable at reduced rate. Root cause: total liquid flux across the settler surface exceeds what the dispersion can coalesce in the available residence time. Fix: confirm flux against design, restore depth, and consider baffle, picket-fence, or plate retrofits (next section) before accepting lost capacity. Flooding at full flux has its own troubleshooting sequence, so treat this as one outcome of poor disengagement and consult the mixer-settler stage efficiency guide for the flooding deep-dive.

Settler Design Rules That Keep Phase Disengagement Fast

If your circuit sits permanently near its limits, no amount of operational tuning will restore margin — the fix is geometry. Organic flux is a hydraulic loading parameter that defines how much total liquid passes through each square meter of settler surface per minute, typically 30-80 L/m²·min (0.7-2.0 gpm/ft²) of combined phases for conventional mixer-settlers, indicative values to be confirmed by test work for your system. Circuits running above their design flux have no residence buffer: any rise in half-life converts directly into entrainment, which is why flux is the first number to check when a band grows.

The governing design rule is dispersion band control. The band should occupy less than 30-50% of the settler depth under normal operation, leaving clear organic and clear aqueous depths on either side and headroom for transients. Sizing a settler to hold the band in that window at maximum sustained flux — including the density-differential and viscosity effects of your worst-case feed — is the single most important phase-disengagement decision made on the drawing board.

Geometry levers, in rough order of cost: internal baffles straighten the flow path and kill short-circuiting, so liquor that looks like it has “5 minutes” of residence actually gets it. Picket fences — vertical slotted baffles across the settler — redistribute the incoming dispersion over the full cross-section and progressively dewater it as it moves forward; upgrading an underperforming fence design is among the highest-return retrofits available. Coalescence plates or structured packing installed in the dispersion band give droplets a surface to collide against, cutting the band depth by 30-60% in favorable cases (indicative; performance is system-specific and should be piloted). Temperature management belongs in the design file too: an insulated settler wall and liquor routing that avoids cold slugs cost little at construction and remove the largest seasonal source of slow settling.

One compact alternative deserves mention where footprint or stage count is constrained: the 80 mm rotary drum centrifugal extractor uses centrifugal force — hundreds of times gravity — to separate phases in seconds, which sidesteps most gravity-settler disengagement limits for high-value, moderate-throughput duty.

How Do You Monitor Disengagement Before It Becomes a Shutdown?

Slow phase disengagement is almost never sudden; it is a trend that becomes a shutdown because nobody was trending it. A monitoring program with three tiers catches the drift while fixes are still cheap.

1. Read the dispersion band daily. Same location, same units, logged every shift with plant flow and temperature alongside it; a band that grows steadily for two weeks is a fault in progress even if entrainment has not yet moved.
2. Run the batch settling test monthly per stage. It separates liquor chemistry from settler hydraulics and gives you a half-life number to plot against reagent age, season, and feed changes.
3. Sample entrainment monthly. Collect timed samples of raffinate and loaded organic, and track organic-in-aqueous and aqueous-in-organic; sampling methods and recovery options are detailed in our SX entrainment guide.
4. Audit flux and temperature quarterly. Confirm actual throughput against design flux and verify the operating temperature band has not drifted; both are silent killers of margin.
5. Re-baseline after every change. New ore blend, new reagent batch, new impeller — rerun the batch test within days so the baseline reflects the circuit you actually operate.

Where you need data quickly — a new reagent, a retrofit candidate, an O/A study — a pilot-scale mixer-settler lets you generate half-life and entrainment numbers under controlled conditions before committing the production circuit.

What Does Slow Settling Cost?

The direct cost is organic. Entrained organic leaves with the aqueous and must be replaced: diluent, extractant, and modifier. As indicative magnitudes, an organic-in-aqueous entrainment of 100 ppm equals roughly 0.1 liter of organic per cubic meter of aqueous treated (about 0.13 fl oz per 35 cu ft), and a large circuit treating thousands of cubic meters per day converts that into hundreds of liters per day of lost organic — with extractant, the expensive fraction, carried proportionally. Recovery units claw some of this back, but every recovered liter carries flotation, filtration, and carbon-treatment cost, and recovery is never complete.

The indirect costs are usually larger. Aqueous entrained in the loaded organic degrades strip liquor quality: metal tenor drops, impurity levels rise, and the electrowinning tankhouse inherits the problem as rough cathode, powder deposits, and falling current efficiency. In nickel and cobalt circuits, organic contamination of intermediates can mean re-processing entire batches. Add the labor cost of interface cleanup and crud handling, and the profit cost of turndown margin you no longer have — a circuit running with a deep band often cannot take a feed increase that the mixers and tanks could otherwise absorb.

The pattern across audits is consistent: for most circuits the tankhouse-side and margin costs of poor phase disengagement exceed the visible reagent bill, which is why the fault deserves the same monitoring rigor as metal recovery itself.

Operating Data From SX Circuits

The following patterns are drawn from anonymized commissioning and troubleshooting experience across copper, nickel, and zinc SX duty. All figures are typical or indicative of the equipment class, not guarantees, and every number must be confirmed by test work on your own liquor system.

A copper extraction settler at a South American operation ran a dispersion band of 50-60 centimeters in a 1.3 meter deep settler — nearly half the depth — with a batch half-life of about 210 seconds and organic-in-aqueous entrainment near 180 ppm. The investigation found impeller tip speed 25% above the current design recommendation and a growing interfacial crud layer. Reducing tip speed and cleaning the interface, then re-optimizing the O/A ratio, brought the half-life to roughly 70 seconds and the band to 20-25 centimeters, with entrainment settling into the 40-60 ppm range within two months. Extraction efficiency moved by less than one percentage point.

A nickel laterite circuit saw settling degrade each winter: half-life doubled from about 45 seconds to about 95 seconds as liquor temperature fell 8-10 degrees Celsius below design. Insulating the settlers and recovering heat from the advance liquor held the temperature band and restored the baseline half-life year-round.

The common thread in these cases: the fix was found by measurement, not by inspection. The half-life test pointed at the mixer and liquor; the band trend ruled the settler in or out; the entrainment numbers confirmed recovery.

Frequently Asked Questions

What is a good dispersion band depth for an SX settler?
Keep the dispersion band below 30-50% of the settler depth under normal operation, which for a 1.2 meter deep settler means roughly 35-60 centimeters at the upper limit and considerably less in a healthy circuit. Below that ceiling, the clear-phase depths at each outlet remain protected and the settler can absorb transients. A band that trends upward day over day is a fault signal regardless of its absolute value.

How long should phase disengagement take?
In a clean, well-mixed circuit, the bulk of coalescence completes within 60-120 seconds of quiescent residence, with batch test half-lives typically 20-90 seconds depending on the system (values indicative, to be confirmed by test work). If your half-life has doubled against its own rolling baseline, you have a fault developing — even if the settler still produces acceptable liquor today.

Does raising temperature always improve settling?
Almost always, within limits. Higher temperature lowers viscosity in both phases and usually speeds coalescence significantly. But temperature also increases extractant degradation rates and diluent evaporation, changes metal distribution isotherms, and shifts which phase is continuous. Hold the circuit within the reagent supplier’s recommended temperature band, typically within a few degrees Celsius of design, rather than chasing settling speed with heat.

Can I fix slow settling without shutting the circuit down?
Often, yes. Tip speed reductions, O/A re-optimization, vortex breaker repairs, and temperature management are all done live. Interfacial crud can be removed from service with suction equipment or managed with interfacial treatment. What usually cannot wait is an overloaded settler: if flux is above what the dispersion can handle, only internal retrofits or added area restore real margin, and those need a planned window.

What O/A ratio should I run?
The ratio is set by the extraction isotherm and the desired raffinate tenor, but its hydraulic consequences matter for disengagement: the ratio determines which phase is continuous in the mixer, how large the dispersed droplet population is, and the density differential entering the settler. If settling is slow, test the mixer-outlet dispersion at neighboring ratios under controlled conditions and select the ratio that meets extraction duty with the fastest break, then confirm at plant scale.

When does crud need treatment rather than process tuning?
When a stable interfacial layer has accumulated — visible as a mud-like band that does not respond to energy or ratio changes — process tuning only treats the symptom. The solids must be removed and the upstream source (fines carryover, precipitation events) addressed. Crud chemistry, collection methods, and disposal or recovery routes are covered separately in our crud treatment resource; treat tuning as the first move and cleanup as the decisive one.

How do coalescence plates compare with adding settler area?
Coalescence plates retrofit into the existing dispersion band and give droplets surfaces to merge on, cutting band depth substantially in favorable cases at modest capital cost. Added settler area increases residence time for everything and removes flux limits outright, but costs footprint and major capital. Pilot the plates on your liquor first — their performance is system-specific — and reserve area expansion for circuits that are genuinely flux-limited.

Conclusion

Phase disengagement is the quiet variable that sets the ceiling on SX circuit performance: everything the mixers do is only banked when the settler coalesces the dispersion back into clean phases within its available residence time. The eight faults in this guide — mixing energy, crud seeding, extractant degradation, temperature, salting-out, O/A and residence, entrained air, and settler overload — cover nearly every slow-settling event you will encounter, and the half-life test plus the daily band reading will localize the fault before it reaches the tankhouse. Measure before you adjust, re-baseline after every change, and design new settlers with band depth and flux margin confirmed by test work. If you are sizing a new circuit, retrofitting a settler, or troubleshooting a persistent dispersion band, request a quote for mixer-settler equipment and pilot test support, or browse our full hydrometallurgy resource library for the companion guides in this series.

References

• Ritcey, G. M. and Ashbrook, A. W., Solvent Extraction: Principles and Applications to Process Metallurgy, revised editions.
• Lo, T. C., Baird, M. H. I. and Hanson, C., Handbook of Solvent Extraction, reprint editions.
• Rydberg, J., Cox, M., Musvik, C. and Choppin, G. R., Solvent Extraction Principles and Practice, second edition.
• Kordosky, G. A., Copper solvent extraction: the state of the art, journal review literature on copper SX reagents and circuit behavior.
• Flett, D. S., papers on the role of interfacial phenomena in solvent extraction for hydrometallurgy.
• Davenport, W. G. et al., Extractive Metallurgy of Copper, editions covering SX and electrowinning practice.
• Journal reviews of crud formation and interfacial solids in industrial solvent extraction circuits, Hydrometallurgy and related serials.

Changelog

• 2026-11-01: First publication.