How Crystals Form — Nucleation, Bravais Lattices, and Diffusion-Limited Aggregation

1. Supersaturation and the Driving Force

Crystallization occurs when the chemical potential of molecules in solution (or a melt) exceeds that in the crystal phase — making the crystal the thermodynamically favored state. The key parameter is supersaturation:

Supersaturation is achieved by cooling (solubility decreases with T for most salts), evaporation (water lost → concentration rises), pH shift (for proteins and biominerals), or anti-solvent addition (adding ethanol to an aqueous salt solution).

2. Classical Nucleation Theory

Before a crystal can grow, a stable nucleus must form. This is energetically costly because new crystal-liquid interface must be created. Classical Nucleation Theory (CNT) balances the bulk free energy gain against the interfacial cost:

CNT predicts that nucleation rate is an extremely sensitive function of supersaturation: a two-fold increase in σ can raise J by 10 orders of magnitude. This explains why nucleation appears nearly instantaneous above a critical threshold.

3. Bravais Lattices and Crystal Systems

All crystals are periodic arrangements of atoms. The entire repeating structure can be described by a unit cell — the smallest repeating block — characterized by three edge lengths (a, b, c) and three angles (α, β, γ). There are exactly 7 crystal systems and 14 Bravais lattices that exhaust all possibilities for 3D periodic symmetry:

Combined with the 32 point groups (rotation and reflection symmetries), the 14 Bravais lattices generate exactly 230 space groups — the complete catalog of 3D crystal symmetry. X-ray diffraction (Bragg peaks) reveals which space group a crystal belongs to.

4. Growth Modes and Crystal Habit

After nucleation, crystal growth proceeds by attachment of new atoms or ions to existing lattice sites. The mechanism depends on supersaturation:

• Layer-by-layer (Frank-van der Merwe): at low σ, atoms attach at ledge steps (kink sites) and complete one layer before the next starts. Produces flat, well-faceted crystals.
• Spiral growth: a screw dislocation emerging at the surface creates a self-perpetuating step that spirals outward — explaining why growth occurs even at very low supersaturation.
• 3D island (Volmer-Weber): at high σ or poor wetting, nuclei form on top of incomplete layers, producing rough, rounded crystal faces.
• Dendritic growth: at very high σ, diffusion becomes the rate-limiting step. Corners and tips grow faster than flat faces → branching, fractal-like arms.

The resulting crystal habit (external shape) reflects the relative growth rates of different faces. Cube-habit NaCl grows rapidly on {100} faces; adding urea impurities selectively adsorbs on these faces, slowing {100} growth and producing octahedral crystals instead.

5. Snowflake Formation and Dendritic Growth

Ice crystallizes in the hexagonal system (space group P6₃/mmc). The six-fold symmetry of snow crystals directly reflects the hexagonal lattice. But why the infinite variety of shapes?

As an ice crystal falls through the atmosphere it passes through regions of varying temperature (−2°C to −22°C) and humidity (supersaturation 0–30%). Growth is exquisitely sensitive to these local conditions:

All six arms of a snowflake experience the same temperature and humidity history at any moment (since they are micrometers apart). This is why each arm is nearly identical — they grow synchronously according to the same physical conditions. Yet no two snowflakes are alike because each follows a unique micro-trajectory through the cloud.

6. Diffusion-Limited Aggregation (DLA)

DLA is a simple computational model that reproduces the branching morphology of dendritic crystals, electro-deposits, and mineral dendrites. Introduced by Witten and Sander (1981):

1. Place a seed particle at the center
2. Release a random walker from far away
3. Walker diffuses until it touches the cluster → sticks permanently
4. Repeat — result is a fractal aggregate

// 2-D DLA with off-lattice random walk
function runDLA(numParticles = 3000) {
  const cluster = [[0, 0]]; // seed particle at origin
  const r_max = () => Math.max(...cluster.map(([x,y]) => Math.hypot(x,y)));

  for (let i = 0; i < numParticles; i++) {
    const R = r_max() + 5; // release radius
    let angle = Math.random() * 2 * Math.PI;
    let x = R * Math.cos(angle);
    let y = R * Math.sin(angle);

    while (true) {
      x += (Math.random() - .5) * 2;
      y += (Math.random() - .5) * 2;
      if (Math.hypot(x, y) > R * 3) break; // escaped — restart

      const stuck = cluster.some(([cx, cy]) => Math.hypot(x-cx, y-cy) < 1.5);
      if (stuck) { cluster.push([x, y]); break; }
    }
  }
  return cluster;
}

The resulting cluster visually matches mineral dendrites (manganese oxide on limestone), lightning channels, and electro-deposited zinc — all examples of Laplacian growth in nature.

7. Gemstones and Mineral Crystals

Geological crystals grow over millions of years under extreme pressures and temperatures. The conditions dictate what minerals can form and their size:

• Diamond: pure carbon, cubic system. Forms at 150–200 km depth (5–6 GPa, ~1200°C) in lithospheric mantle, then carried to surface by kimberlite eruptions. Growth rate: ~1 mm per million years.
• Quartz (SiO₂): trigonal system. Grows hydrothermally from silica-rich fluids in veins and pegmatites. Major variety colors: amethyst (Fe³⁺), citrine (Fe⁴⁺), rose quartz (Ti).
• Snowflake obsidian: not a crystal — volcanic glass (amorphous). Cristobalite crystallites form on annealing, creating the "snowflake" pattern: devitrification.
• Giant selenite crystals (Cave of Crystals, Mexico): gypsum (CaSO₄·2H₂O), monoclinic. 10–11 m long, ~500,000 years old. Grew in hydrothermal brine at 58°C — just below the temperature at which anhydrite dissolves back to gypsum.