Crystallization

How a crystal grows

Dissolve a solute in hot solvent until the solution is saturated. Cool slowly. As temperature drops, the solubility limit drops too — but the dissolved molecules don't crystallize the instant the limit is crossed. There's a barrier: forming a tiny new crystal costs surface energy that exceeds the bulk free-energy savings until the cluster passes a critical size. Below that, the cluster shrinks back. Above it, the cluster grows.

This barrier defines two regimes. The metastable zone sits between the solubility curve and a higher concentration where spontaneous nucleation kicks in. Inside this zone the solution is supersaturated — it could crystallize, but won't until something triggers it: a seed crystal, a scratch on the flask, dust, or just enough thermal noise.

Once a critical nucleus exists, molecules from solution diffuse to its surface, hop along the surface to find a low-energy site (a step or kink in the lattice), and lock in. Each addition is geometrically selective — only molecules of the right shape, size, and hydrogen-bonding pattern fit. That's why crystallization purifies: impurities can't dock without distorting the lattice and paying an energy penalty, so they remain in solution.

The solubility curve drives everything

concentration
   ▲
   │            spontaneous nucleation zone
   │     ┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄  super-solubility (kinetic limit)
   │
   │         ●  start (hot, dissolved)
   │       ╱╲       metastable zone
   │      ╱  ╲     (seed-mediated growth)
   │     ╱    ●  end (cool, partial crystals)
   │    ╱   solubility curve
   │   ╱
   │  ╱      ← below here = undersaturated
   │ ╱
   │╱
   └─────────────────────────────────▶ temperature

The vertical drop between starting concentration and final solubility limit determines the maximum theoretical yield. The horizontal width of the metastable zone determines how slowly you must cool to stay in the controlled-growth regime.

Worked example — cooling-curve yield

You have 25.0 g of a compound dissolved in 100 mL of water. Solubility data:

• At 80 °C: 30.0 g / 100 mL solvent
• At 5 °C: 4.2 g / 100 mL solvent

The starting solution is just below saturation at 80 °C — clear and homogeneous. Cool slowly to 5 °C in an ice bath. At that temperature only 4.2 g remains soluble; the rest crystallizes:

theoretical yield = (25.0 g − 4.2 g) / 25.0 g = 83.2%
crystal mass     = 20.8 g

In practice, expect 75–80% recovered after vacuum filtration and drying — the rest is lost to mother-liquor wetting on the filter cake, splashing, and incomplete equilibration. To push further, evaporate the mother liquor by half and re-cool: this "second crop" recovers another 1–2 g but at lower purity because impurities have been concentrating with each cycle.

If yield matters more than purity, switch to anti-solvent: add 50 mL of methanol to the mother liquor at 5 °C. The solubility in methanol/water mixed at 33% organic might be only 1.0 g per 100 mL, dropping the residual to ~1.5 g across 150 mL. Yield rises to ~94%.

Six methods of inducing crystallization

Method	How it works	Best for	Speed	Crystal size	Risk
Cooling crystallization	Lower temperature drops solubility	Compounds with steep solubility curves (e.g., KNO₃)	Hours	Medium-large	Oiling out if cooled too fast
Evaporation	Solvent loss raises concentration	Compounds insensitive to temperature (NaCl)	Hours to days	Variable	Salt-out of impurities
Anti-solvent (drowning out)	Add miscible solvent that the solute doesn't like	Pharmaceuticals, salts	Minutes	Small (fast nucleation)	Polymorph hop, agglomerates
Reactive crystallization	Form an insoluble product in situ (BaSO₄, AgCl)	Inorganic salts	Seconds	Small	Co-precipitation of impurities
Vapor diffusion	Slow solvent transfer through vapor phase	Single crystals for X-ray	Days to weeks	Single, large	Slow, fragile crystals
Pressure / cooling crystallization (melt)	Solidify a pure liquid below freezing	Naphthalene, fatty acids	Hours	Large	Co-crystallization at eutectic
Sublimation–condensation	Vaporize and re-deposit	Volatile solids (caffeine, iodine)	Hours	Small needles	Loss of volatile impurities only

For X-ray crystallography of a new molecule, vapor diffusion in a hanging-drop plate is the standard. Industrial pharmaceutical processes lean on cooling and anti-solvent because both scale to thousands of liters in a stirred tank.

Primary vs secondary nucleation

• Primary homogeneous nucleation — molecules in solution self-assemble. Requires high supersaturation; rare in real flasks.
• Primary heterogeneous nucleation — clusters form on dust, scratches, or impurities. The reason scratching the flask with a glass rod usually triggers crystallization.
• Secondary nucleation — existing crystals shed micro-fragments under stirring; each becomes a new crystal. Dominant in seeded industrial processes.

Classical nucleation theory predicts the nucleation rate J = A · exp(−ΔG*/kT), where ΔG* is the energy barrier. Because the exponent depends inversely on the cube of supersaturation, doubling the driving force can multiply the nucleation rate by orders of magnitude. This extreme sensitivity is why industrial crystallizers rely on tight temperature control — a 1 °C overshoot produces a fines disaster instead of a clean crystal slurry.

Crystal habit and shape

The same compound can crystallize as needles, plates, prisms, or chunks depending on conditions. Slow growth from clean solvent gives the equilibrium habit (often the most isotropic). Fast growth or impurities can selectively poison some faces, causing the crystal to elongate along the unblocked axis. Pharmaceutical formulators care because needles flow poorly and clog tableting presses, while plates are filterable but compact poorly. Habit modifiers — small additives that bind preferentially to certain faces — are often patented separately from the active ingredient.

Variants and adjacent techniques

• Recrystallization — repeating crystallization from pure solvent to drive purity upward; covered in detail in our recrystallization article.
• Co-crystallization — deliberately incorporating a second component (often a hydrogen-bonding partner) into the lattice to tune solubility or stability. Common in newer drug formulations.
• Spherical agglomeration — using a small amount of a third immiscible solvent to bind small crystals into free-flowing spheres without grinding.
• Continuous crystallization — pumping saturated feed through a temperature gradient; gives steady-state crystal size distribution and is replacing batch in modern pharma plants.

Common pitfalls

• Too-fast cooling. Drives the solution past the metastable zone. Result: a fine powder full of trapped solvent and impurities.
• Stirring too vigorously. Shears growing crystals and creates secondary nuclei everywhere; you get fines instead of big crystals.
• Wrong solvent. If the compound is too soluble, you can't drop concentration enough to crystallize. If too insoluble, you can't dissolve it in the first place. Aim for ~50 g/L solubility at 80 °C and ~5 g/L at 5 °C.
• Crashing out as oil. The compound's melting point (depressed by impurities) sits below its dissolution temperature. Switch solvent or seed.
• Contaminated seed. Even a few grains of a different polymorph can convert the entire batch. Pharma sites run dedicated seeding rooms.
• Filter loss. Wet filter cakes hold mother liquor. Wash with cold solvent before drying or you concentrate impurities into the crystals.

Why crystallization is everywhere

Crystallization is the single most-used purification step in chemical manufacturing. Roughly 90% of solid pharmaceutical products and most bulk chemicals — from sugar to fertilizer to detergent enzymes — pass through at least one crystallizer. It scales cheaply (a 10 m³ stirred tank purifies a tonne of API per batch), needs no expensive consumables (just solvent and cooling), and rejects impurities for free.