Crystallizer Design: Yield & Supersaturation Control

Last updated: 2026-10-23

Crystallizer design is, before anything else, a mass-balance exercise: in the worked example below, a single cooling pass from 60 to 25 °C recovers 48.8% of the dissolved copper sulfate as CuSO4·5H2O crystals. Yield is fixed by the solubility curve and the mass balance; crystal size is fixed by supersaturation control and residence time. This article gives the calculation method, the operating windows, and the failure modes that separate a plant that meets its metal-recovery target from one that never quite does.

Yield and size compete for the same driving force. The leach-liquor side of any hydrometallurgical flowsheet ends in an evaporation and crystallization plant — our evaporation & crystallization overview maps the equipment spectrum — while this article drills into the two variables every crystallizer design must reconcile: how much salt leaves solution, and how large the crystals are when it does.

Key Takeaways

• Crystal yield is set by the mass balance and the solubility curve, not by equipment choice.
• Supersaturation, quantified as ΔC or relative supersaturation σ, is the driving force for both nucleation and growth.
• σ = 0.2–2% is the practical steady-state window for most inorganic salt systems.
• In MSMPR crystallizers, mean crystal size scales weakly with residence time; fines management is the stronger size lever.
• Most yield shortfalls trace to hydrate transitions and unmeasured kinetics — both closed out in pilot testing.

Crystallizer Design Fundamentals: Solubility, Supersaturation, and Yield

Three definitions anchor everything that follows.

Evaporator Crystallizer Systems for Mining Brines

The solubility curve is a boundary that separates undersaturated liquor from supersaturated liquor at equilibrium: for each temperature it gives the concentration C* of dissolved solute (expressed per unit of free water — the water that stays as solvent, excluding any water locked into the crystal lattice) at which the liquor is exactly saturated. For copper sulfate pentahydrate, C* rises from about 0.207 kg anhydrous CuSO4 per kg free water at 20 °C to about 0.400 at 60 °C. For sodium sulfate the curve is interrupted at 32.4 °C, where the stable solid switches from mirabilite (Na2SO4·10H2O) to thenardite (anhydrous Na2SO4) — a transition that has caught out more than one flowsheet.

Supersaturation is a thermodynamic state that exists when the actual dissolved concentration C exceeds the equilibrium value C* at the same temperature. It is the only driving force there is for both nucleation (new crystals appearing) and growth (existing crystals getting bigger). It is always transient in the sense that the system consumes it: every gram of salt that deposits on a crystal surface lowers C back toward C*. Practical crystallizer design is therefore the art of generating supersaturation at a controlled rate and destroying it through crystal surface area.

The metastable zone width (MSZW) is a concentration or temperature interval that spans from the solubility curve to the point where spontaneous (primary) nucleation begins. Below the MSZW, a clean liquor can be supersaturated indefinitely without nucleating; inside it, nucleation becomes probable and then violent. Seeded liquors can be worked at ΔC values well below the MSZW limit, which is exactly how industrial equipment is run — the metastable zone is the operating room, and the nucleation threshold is the wall at its edge.

Finally, the crystal size distribution (CSD) is a population statistic that describes how the crystal mass is spread across size classes. Yield tells you how many tonnes of salt you sell; the CSD tells you whether the product is a centrifugeable, free-flowing material or a fine sludge that blinds filter cloth and carries mother liquor into the product. Yield is set by the balance sheet; CSD is set by kinetics — and that is where supersaturation control earns its keep.

Throughout this article, concentrations are on a dry, anhydrous basis — kilograms of anhydrous solute per kilogram of free water — because hydrate crystals steal water as well as salt, and sloppy bookkeeping of crystallization water is the most common arithmetic error in this field.

Crystal Yield Calculation in Crystallizer Design: The Mass-Balance Method

The yield of a crystallization step follows from one solute balance and one water balance. Because a hydrate removes both, the two balances are coupled and must be solved together. Here is the step-by-step method, followed by a fully worked example (all values illustrative; verify solubility on your actual liquor).

Step 1 — Fix the basis. Choose a convenient quantity of feed liquor (here, 1,000 kg) and tabulate it on a dry basis: mass of free water and mass of dissolved anhydrous solute.

Step 2 — Get solubility at both end temperatures. From the solubility curve, read C1 (at the feed temperature) and C2 (at the final temperature), in kg anhydrous solute per kg free water. For the worked example: CuSO4 at 60 °C, C1 ≈ 0.400; at 25 °C, C2 ≈ 0.230 (values interpolated from standard solubility data; approximate).

Step 3 — Compute the hydrate ratio. R = M(hydrate)/M(anhydrous). For CuSO4·5H2O: M = 249.7 vs 159.6 g/mol, so R = 1.5646. One kilogram of product therefore contains 1/R = 0.6392 kg of anhydrous CuSO4 and 0.3608 kg of crystallization water.

Step 4 — Solve the coupled balance. For a cooling pass with no evaporation (V = 0), the yield of hydrate crystals is:

Y = S1·(C1 − C2) / [1/R − C2·(1 − 1/R)]

where S1 is the initial free-water mass. The denominator corrects for the water the hydrate removes from circulation, which raises the effective final concentration of what remains. If the pass also evaporates V kg of free water, replace the numerator with C1·S1 − C2·(S1 − V).

Step 5 — Convert to recovery. Recovery of dissolved salt = (Y/R)/(C1·S1), expressed as a percentage.

Step 6 — Cross-check both balances. Sum anhydrous solute and free water across product and mother liquor; each must close against the feed. If the water balance does not close, the crystallization-water term is wrong.

Worked example: copper sulfate pentahydrate by cooling

Basis: 1,000 kg of purified liquor at 60 °C containing 28.57 wt% anhydrous CuSO4 (i.e., C1 = 0.400 kg/kg free water, S1 = 714.3 kg). Cool to 25 °C, C2 = 0.230. Assumptions: no evaporation, no entrainment, pure CuSO4·5H2O product, illustrative solubility values.

Numerator: 714.3 × (0.400 − 0.230) = 714.3 × 0.170 = 121.4
Denominator: 0.6392 − 0.230 × 0.3608 = 0.6392 − 0.0830 = 0.5562
Y = 121.4 / 0.5562 = 218.3 kg of CuSO4·5H2O crystals

Recovery = (218.3 × 0.6392) / 285.7 = 139.5 / 285.7 = 48.8% of the dissolved copper sulfate.

Mass-balance check:

Both balances close (714.3 = 78.8 + 635.5; 285.7 = 139.5 + 146.2; 218.3 + 781.7 = 1,000.0), and the mother-liquor concentration reproduces C2: 146.2 / 635.5 = 0.230. Sensitivity: pushing the same liquor to 20 °C (C2 ≈ 0.207) gives Y = 714.3 × 0.193 / 0.5645 = 244.3 kg, a 54.7% recovery — about six extra percentage points for 5 K of additional cooling. The numbers a crystallizer design rests on are only as good as the solubility curve behind them, and on plant liquors (with iron, acid, and magnesium in solution) the curve shifts; that is why we treat literature solubility as a starting estimate, not a design input — a point our pilot-testing team quantifies before detailed engineering.

How Is Supersaturation Generated and Quantified?

Because yield comes from the balance, the only thing left to control in a crystallizer design is the rate and route by which supersaturation is created and consumed. Quantification first: the absolute supersaturation is ΔC = C − C*, and the relative supersaturation is σ = ΔC/C*. Both are dimensionless only in σ’s case — ΔC carries units (kg/kg free water). Worked micro-example: at 50 °C, C* for CuSO4 ≈ 0.336 kg/kg; a liquor at 0.340 kg/kg has ΔC = 0.004 kg/kg and σ = 0.004/0.336 = 1.2%. Industrial practice for inorganic salts typically holds σ in the 0.2–2% band at steady state — high enough for acceptable growth rates, low enough to stay clear of the nucleation wall. Treat the band as an order-of-magnitude guide from standard design practice (Mullin; Mersmann), not a per-system constant.

Four (plus one) generation methods, each with its own hardware and its own control character:

For the evaporative route, the hardware of choice is usually the forced-circulation evaporator-crystallizer, where the circulation loop keeps the heat-transfer surface starved of supersaturation and lets ΔC develop in the boiling zone instead. For size-sensitive products on cooling duties, the Oslo crystallizer keeps a fluidized bed of coarse crystals bathed in mildly supersaturated liquor, trading capacity for CSD quality.

The MSZW deserves one more paragraph, because it converts directly into control policy. Measured MSZW values for common sulfate systems span roughly 5–25 K of cooling at unseeded cooling rates of 5–20 K/h (system- and impurity-dependent; these are literature-order figures). Seeding can halve the usable spread. Judgement that comes only from field experience: we seed at roughly one-third of the measured MSZW, not at the thermodynamic saturation point, and we never let startup transients — not the steady-state balance — set the nucleation tone of a run. A vessel nucleated badly in the first thirty minutes spends the rest of the campaign making fines. Note also that the MSZW is a kinetic measurement, not a salt property: it narrows with faster cooling, impurities, and agitation, so measure it on the actual liquor at representative rates.

MSMPR Population Balance: Nucleation versus Growth

Most continuous industrial crystallizers are well described by the mixed suspension, mixed product removal (MSMPR) idealization. An MSMPR crystallizer is an idealized vessel that assumes the magma is uniformly mixed — crystals of every size experience the same supersaturation history — and that the product stream is drawn at the same size distribution as the vessel contents. Under these assumptions, the population balance gives the number density of crystals as n(L) = n⁰·exp(−L/Gτ), where n⁰ is the nucleation density (crystals per unit volume per unit length), G the linear growth rate (m/s), and τ the product residence time.

Two rate laws dominate the behavior. Growth follows G = kg·(σ)^g with g typically 1–2; nucleation (dominantly secondary — contact nucleation on existing crystals) follows B⁰ = kb·(σ)^b with b typically 2–4, i.e., nucleation is the steeper function of supersaturation. This exponent inequality is the central fact of supersaturation control: when σ rises, new crystals appear faster than existing ones grow, so the same salt production is spread across more, smaller crystals. Conversely, running at lower σ with adequate surface area favors size.

The classic sizing consequence: the mass-weighted mean size is L43 = 3.67·G·τ. Worked micro-example (illustrative): with G = 1.0 × 10⁻⁸ m/s (0.036 mm/h) and τ = 6 h, L43 = 3.67 × 1.0 × 10⁻⁸ × 21,600 = 0.79 mm (≈ 0.031 in). Does doubling residence time to 12 h double it? No — and the population balance explains why. At fixed slurry density, the mass balance requires B⁰·G³·τ⁴ = constant; substituting the rate laws gives σ ∝ τ^(−4/(b+3g)), and with b = 2g this yields L43 ∝ τ^0.2. Doubling τ then raises L43 only 2^0.2 = 1.15×, to ≈ 0.91 mm. This is the second rule of crystallizer design for size: bigger vessels buy surprisingly little; the strong levers are lowering nucleation (gentler σ, seeding, avoiding crystal breakage) and removing fines.

Whether to exploit these levers in a continuous or discontinuous flowsheet depends on scale, feed variation, and product size specification — the trade-offs are set out in our continuous vs batch crystallizer comparison. For crystallizer design purposes, the MSMPR framework’s practical value is proportional thinking: it tells you which knob moves the CSD and how weakly or strongly, before any pilot data is fitted.

What Are the Most Common Crystallizer Design Mistakes?

These recur across hydrometallurgical salt plants; each is stated as symptom → cause → remedy.

Sodium Sulfate Crystallization

1. Yield computed from the heat balance alone. Symptom: plant recovery misses the design figure by 5–15 percentage points. Cause: the design used pure-water solubility or ignored crystallization water and hydrate transitions. Remedy: redo the solubility curve mass balance on the actual liquor composition, dry basis, including water-of-crystallization bookkeeping.

2. Ignoring hydrate transitions. Symptom: sodium sulfate plant makes slushy fines below 32.4 °C or yields vanish in a warm season. Cause: mirabilite/thenardite switch means solubility, R, and water demand all change discontinuously at the transition temperature. Remedy: choose the operating temperature band deliberately — crystallize mirabilite cold for high yield (high R) or thenardite hot for free-flowing product — and design the heat train around the chosen phase.

3. Fines blowdown set too low. Symptom: product mean size drifts down week by week; centrifuge cloth blinds. Cause: the fines-removal loop is undersized or its setpoint was never commissioned against the real nucleation rate, so excess nuclei mature into product-size fines. Remedy: measure the CSD, then tune blowdown so the fines loop removes enough surface area to shift growth onto surviving crystals; expect setpoints to differ from the vendor’s default.

4. Scaling on heat-transfer surfaces from pushing ΔC. Symptom: heat duty decays over days; cleaning cycles shorten. Cause: supersaturation is being created at the wall (or at the flash surface) instead of in the bulk — the classic failure when ΔC targets exceed the metastable zone at the local skin temperature. Remedy: lower the driving force, raise circulation velocity, and move evaporation away from the exchanger into the boiling zone.

5. Size drift after feed changes. Symptom: CSD fine after an upstream upgrade raises feed tenor. Cause: higher production at fixed geometry raises the ΔC needed to balance, and since B⁰ responds more steeply than G, nucleation wins. Remedy: re-run the population-balance check at the new tenor; add seeding or fines destruction capacity rather than chasing the old size with temperature alone.

6. No MSZW measurement before detailed design. Symptom: the vessel nucleates violently on startup; weeks of off-spec product. Cause: the design assumed literature MSZW data, but plant liquor impurities narrowed it. Remedy: measure the MSZW on the real liquor, seeded and unseeded, at representative cooling rates — a small campaign, cheap relative to one bad startup.

Judgement from field experience again: most of these failures are not calculation errors but bookkeeping errors — the arithmetic of hydrate water, recycle streams, and impurity-shifted solubility. Every crystallizer design review should therefore start — and end — with the mass balance, which forgives nothing and forgets nothing. Reference data for cross-checking liquor properties can be found in our resource library, but the liquor itself always has the final say.

Cost and Energy Context: Evaporating versus Cooling

Energy separates the two generation routes more than capital does. Evaporation must supply the latent heat of water: about 2,260 kJ/kg at 100 °C and about 2,407 kJ/kg at 40 °C under vacuum. With assumptions stated: a single-effect evaporator consumes roughly 1.1–1.3 kg of steam per kg of water evaporated (including losses); a triple-effect train, 0.35–0.45 kg/kg; and a mechanical vapor recompression (MVR) unit, 20–45 kWh per tonne of water evaporated, assuming an effective heat-transfer ΔT of 4–8 K, boiling point elevation (BPE) of 3–8 K for saturated sulfate liquors, and compressor efficiency of 75–80%. For context, 30 kWh/tonne is only ~4–5% of the thermal energy content of the same duty fired with steam — the reason MVR dominates new evaporative designs where power is available.

MVR vs Multiple Effect Evaporator

MVR vs Multi-Effect Evaporation

Cooling crystallization is far cheaper thermodynamically. The duty is sensible heat plus the heat of crystallization; for CuSO4·5H2O the latter is ≈ 78 kJ/mol, or ≈ 0.31 MJ per kg of hydrate. In the worked example above: sensible duty ≈ 1,000 kg × 3.5 kJ/(kg·K) (assumed cp for a 28% sulfate liquor) × 35 K = 122.5 MJ, and crystallization heat ≈ 218.3 × 0.31 = 67.7 MJ, so the chiller sees ≈ 190 MJ per tonne of feed — roughly 53 kWh, comparable per tonne to an MVR duty but delivered as low-grade cooling rather than shaft work. The economics of the choice therefore hinge on the solubility slope: steep curves (copper sulfate) reward cooling; flat curves force evaporation regardless of energy price.

One caution on BPE: because it shrinks the effective ΔT available to both the heat exchanger and the MVR compressor, underestimating it on a concentrated liquor is a classic cause of undersized evaporative capacity — our evaporator energy consumption guide and MVR selection guide both treat it as a first-order variable, not a correction term. None of these figures are quotations; they are design-context ranges, and every crystallizer design should be energy-checked against the actual liquor’s BPE and the site’s steam-versus-power price spread.

What Yield Numbers Should You Expect for Copper, Nickel, and Sodium Sulfate?

Three illustrative cases, all on the same mass-balance method, all clearly labeled as illustrative (literature-order solubilities, pure liquors, no recycle).

Copper sulfate (cooling): as worked above, 1,000 kg of 28.57 wt% CuSO4 liquor cooled 60 → 25 °C gives 218.3 kg of CuSO4·5H2O and 48.8% recovery; deepening the pass to 20 °C lifts it to 54.7%. The steep solubility curve makes copper sulfate one of the few hydromet salts where cooling alone recovers half the dissolved metal value in one pass.

Nickel sulfate (evaporation): basis 1,000 kg of purified liquor at 30 wt% NiSO4 (dry basis), evaporate 250 kg of free water, finish saturated at 30 °C with an illustrative C2 = 0.50 kg/kg. With R = 262.85/154.75 = 1.699 for the hexahydrate: Y = (0.4286 × 700 − 0.50 × 450) / (0.5887 − 0.50 × 0.4112) = 75 / 0.3832 = 195.7 kg of NiSO4·6H2O, i.e., 38.4% of the dissolved nickel. The moderately flat curve is why nickel sulfate plants are evaporative.

Sodium sulfate (vacuum cooling): basis 1,000 kg of liquor at 25 wt% Na2SO4, cooled 40 → 20 °C with no net evaporation (V = 0, the conservative floor). With R = 322.20/142.04 = 2.269 for mirabilite: Y = 750 × (0.3333 − 0.195) / (0.4409 − 0.195 × 0.5591) = 103.7 / 0.3318 = 312.7 kg of Na2SO4·10H2O, a 55.1% recovery — the high hydrate ratio (more than half the crystal is water) drives yield up. Any flash evaporation in the vacuum cooler adds further recovery on top of this floor.

Notice the spread — 38–55% recovery from single passes on the same method, which is why crystallizer design is always salt-specific. Where the final few percent matters, mother-liquor recycle, a second effect, or a deeper ΔC is designed in; which of those pays depends on kinetics measured on your liquor, typically within a two-to-four-week pilot-testing program that converts these illustrative brackets into design numbers.

Frequently Asked Questions

What is the difference between absolute and relative supersaturation?

ΔC is the absolute driving force — the difference between the actual dissolved concentration and the equilibrium solubility C* at the same temperature, in kg/kg free water. Relative supersaturation σ = ΔC/C* expresses that difference as a fraction of solubility, and it is what nucleation and growth rates actually correlate with. Two liquors with identical ΔC can behave very differently if their solubilities differ. Use ΔC for bookkeeping and equipment sizing; use σ to compare systems and set control limits.

How much yield can a single cooling pass recover?

It depends on the solubility slope and the hydrate. The worked CuSO4·5H2O example recovers 48.8% cooling from 60 to 25 °C, and about 54.7% if deepened to 20 °C. Steep solubility curves and high hydrate ratios favor cooling; flat curves, typical of many nickel and magnesium salts, make evaporative crystallization the only economic route. Compute the number from the mass balance before choosing any equipment — the answer is never generic.

Does seeding increase crystal yield?

No. Yield is fixed by the mass balance and the solubility curve, so seeding cannot change how much salt leaves solution at equilibrium. What seeding changes is how the yield appears: it consumes the available supersaturation through growth on existing surfaces instead of spontaneous nucleation, producing fewer, larger crystals and a narrower CSD. Seed load, size, and timing are among the most sensitive operating variables in practice.

Why is my crystal size drifting smaller over months of operation?

Usual causes: rising feed tenor or impurity buildup (iron and organic species are strong growth inhibitors), fouled heat-transfer surfaces forcing higher ΔC, worn circulation pumps changing shear and contact nucleation, or fines blowdown setpoints never retuned since commissioning. Track σ and the product mean size weekly against the commissioning baseline. When both drift together, the cause is usually nucleation, not growth — and the fix is in the fines loop, not the temperature controller.

How does boiling point elevation affect evaporative crystallizer design?

Boiling point elevation — commonly 3–8 K for saturated sulfate liquors — means the liquor boils hotter than pure water at the same pressure, shrinking the effective temperature difference across the heat exchanger and the compression rise an MVR fan must deliver. Underestimating BPE is a classic cause of undersized evaporative capacity and disappointing steam economies. Measure it on the actual concentrated liquor; pure-water data and dilute-feed extrapolations will not do for design.

What data do you need before sizing a crystallizer?

The minimum set: the full solubility curve for the actual liquor, including hydrate-transition points; the metastable zone width at representative cooling rates, seeded and unseeded; nucleation and growth kinetics from a pilot campaign; boiling point elevation and heat of crystallization; and the expected variation in feed tenor and impurities. Desk calculations can bracket yield; only measured kinetics can fix crystal size — which is why a short pilot campaign is standard practice before detailed engineering.

Conclusion: Design the Balance Before the Vessel

The through-line of this article is that a competent crystallizer design is done in a fixed order — solubility curve, coupled solute-and-water mass balance, then supersaturation and population-balance thinking — because yield is decided by thermodynamics while size is decided by kinetics, and confusing the two produces plants that make the right tonnage of the wrong crystals. The single most valuable action available: bring us your liquor analysis and target salt, and run the solubility curve, MSZW, and kinetics through our pilot-testing program before detailed design — then browse the resource library for the supporting guides on energy, equipment selection, and flowsheet choice. Measure first; the arithmetic will then hold.

References

1. Mullin, J.W. Crystallization, 4th ed., Butterworth-Heinemann, Oxford, 2001.
2. Mersmann, A. (ed.) Crystallization Technology Handbook, 2nd ed., Marcel Dekker, New York, 2001.
3. Randolph, A.D. and Larson, M.A. Theory of Particulate Processes: Analysis and Techniques of Continuous Crystallization, 2nd ed., Academic Press, San Diego, 1988.
4. Green, D.W. and Southard, M.Z. (eds.) Perry’s Chemical Engineers’ Handbook, 9th ed., McGraw-Hill, New York, 2019 — crystallization and solubility-data sections.
5. Crystal Research and Technology (Wiley) — peer-reviewed journal covering nucleation kinetics, metastable zone width, and crystal growth of inorganic salts.
6. Hydrometallurgy (Elsevier) — peer-reviewed journal covering salt recovery and crystallization steps in hydrometallurgical flowsheets.
7. U.S. Geological Survey, Mineral Commodity Summaries 2025 — production and market context for copper and nickel sulfate products.

Changelog

2026-10-23 — First published.

Related reading

• Crystallization