Colloid Science: Why Milk Doesn't Separate (Usually)

Written by MySimulator Team · Reviewed by MySimulator Editorial Review

Pour a glass of milk and leave it on the counter. Hours, even days later, it remains a uniform white liquid — the fat droplets suspended throughout, not pooled at the surface. This apparent defiance of gravity is the province of colloid science: the study of particles too small to settle quickly but too large to dissolve. Milk is an emulsion, a specific type of colloid in which liquid droplets are dispersed in another liquid. The reason it remains stable — and the reason it eventually does separate, or curdles when acid or salt is added — is elegantly explained by DLVO theory, named after Derjaguin, Landau, Verwey, and Overbeek. Understanding colloidal stability underpins industries ranging from pharmaceuticals and food science to ceramics, paints, and water treatment.

What Is a Colloid?

A colloid is a system in which particles of one material — ranging from roughly 1 nanometre to 1 micrometre in diameter — are dispersed throughout a continuous medium. This size window defines the "colloidal regime": large enough that gravity eventually matters, but small enough that Brownian motion (the thermal jostling by surrounding solvent molecules) keeps particles in suspension for timescales from hours to years. Particles smaller than ~1 nm form true solutions; particles larger than ~1 µm settle under gravity in reasonable timescales.

Colloids come in many types depending on the dispersed phase and the medium:

The vast surface area of colloidal particles — a 1 µm sphere of density 1 g/cm³ has a specific surface area of ~6 m²/g — means that surface forces, which are negligible for bulk objects, entirely dominate the behaviour. Colloidal stability is therefore a surface chemistry problem.

van der Waals Attraction: The Force That Pulls Particles Together

Between any two electrically neutral bodies, quantum-mechanical fluctuations of electron clouds create transient dipoles that induce correlated dipoles in neighbouring atoms. The resulting London dispersion force is attractive and universal. For individual atoms it decays as r⁻⁶ with distance; between extended macroscopic bodies (as colloidal particles are), the pairwise summation changes this dependence markedly.

Hamaker (1937) showed that the van der Waals interaction energy between two flat surfaces separated by a gap H is:

van der Waals interaction energy (flat surfaces) V_vdW = −A / (12π H²)

A = Hamaker constant (material-specific, typically 10−20–10−19 J)
H = surface-to-surface separation distance

For two spheres of radius R (Derjaguin approximation, H << R):
V_vdW ≈ −A R / (12 H)

The Hamaker constant A depends on the dielectric properties and refractive indices of both the particle and the medium through Lifshitz theory. For polystyrene in water, A ≈ 1.3×10⁻²¹ J; for gold in water, A ≈ 4×10⁻¹⁹ J — gold particles therefore aggregate far more readily at a given separation. Crucially, V_vdW is always negative (attractive) between like materials and decays slowly enough with distance that it can dominate at separations of a few nanometres.

Electrostatic Repulsion: The Electric Double Layer

Most colloidal particles acquire a surface charge in water — through dissociation of surface groups, preferential ion adsorption, or isomorphic substitution in clay minerals. A negatively charged particle attracts a diffuse cloud of counterions (positive ions) around it while repelling co-ions (negative ions). This structure — the charged surface plus the diffuse ion cloud — is the electric double layer, described quantitatively by the Gouy-Chapman-Stern model.

The potential decays roughly exponentially away from the surface. The characteristic length over which it falls to 1/e of its surface value is the Debye screening length κ⁻¹:

Debye screening length κ⁻¹ = (ε0εr kBT / (e² ∑ ni zi²))^(1/2)

ε0 = permittivity of vacuum, εr = relative permittivity of solvent (~80 for water)
kB = Boltzmann constant, T = temperature, e = elementary charge
ni = number density of ion species i, zi = valence

In water at 25°C: κ⁻¹ ≈ 0.304 / √I nm (I = ionic strength in mol/L)

Example: pure water (I = 10−7 M) ⇒ κ⁻¹ ≈ 960 nm
Example: 100 mM NaCl (I = 0.1 M) ⇒ κ⁻¹ ≈ 0.96 nm

When two similarly charged particles approach, their double layers overlap and the osmotic pressure of the counterion-rich region generates a repulsive force. The electrostatic repulsion energy between two spheres in the linearised Poisson-Boltzmann approximation is:

Electrostatic double-layer repulsion V_EDL = 64π R n0 kBT κ⁻² γ² exp(−κH)

γ = tanh(zeψ0 / 4kBT) (reduced surface potential parameter)
ψ0 = surface potential, n0 = bulk electrolyte concentration

Key point: V_EDL decays exponentially with the Debye length κ⁻¹;
adding salt ⇒ smaller κ⁻¹ ⇒ shorter-range repulsion.

DLVO Theory: Balancing Attraction and Repulsion

DLVO theory (developed independently by Derjaguin and Landau in 1941 and by Verwey and Overbeek in 1948) simply adds the two interaction energies:

DLVO total interaction energy V_DLVO(H) = V_EDL(H) + V_vdW(H)

= 64π R n0 kBT κ⁻² γ² exp(−κH) − A R / (12 H)

Typically V(H) shows:
• A deep primary minimum at H ≈ 0.1–0.5 nm (irreversible aggregation)
• A potential energy maximum (DLVO barrier) at H ≈ 1–10 nm
• A shallow secondary minimum at H ≈ 5–20 nm (reversible flocculation)

The height of the DLVO energy barrier determines stability. If the barrier is much greater than the thermal energy k_BT (say, >10–15 k_BT), particles cannot pass over it on any reasonable timescale and the colloid is stable. If the barrier is reduced — by adding salt (which compresses the double layer) or by adjusting pH (which changes the surface charge) — particles aggregate. This is precisely why adding vinegar or rennet causes milk to curdle: acid lowers the pH toward the isoelectric point of the casein proteins (~pH 4.6), reducing their surface charge, collapsing the energy barrier, and causing the casein micelles to aggregate into curds.

The Schulze-Hardy rule: The critical coagulation concentration (CCC) — the minimum electrolyte concentration needed to destabilise a colloid — decreases dramatically with counterion valence: CCC ∝ z⁻⁶. In practice, monovalent Na⁺ requires ~100 mM to coagulate a typical negative colloid, divalent Ca²⁺ requires ~1 mM, and trivalent Al³⁺ requires only ~0.1 mM. This 1:100:1000 ratio is widely exploited in water treatment.

Beyond DLVO: Steric, Depletion, and Hydrophobic Forces

DLVO theory works well for simple systems with smooth, rigid, uniformly charged surfaces, but real colloids often show behaviour it cannot predict. Several non-DLVO forces are now recognised:

Steric stabilisation arises when polymer chains or surfactant molecules are adsorbed onto a particle surface, projecting outward into the solvent. When two polymer-coated particles approach, the chains overlap, losing configurational entropy and increasing osmotic pressure — a purely entropic repulsion that is effective even at high ionic strength where electrostatic repulsion is screened. This is why non-dairy creamers, which often lack the protein layers present in real milk, use specific polymer additives to maintain stability.

Depletion forces appear in mixtures containing non-adsorbing polymer or smaller colloidal particles. When two large particles approach within one polymer radius of gyration R_g, the polymer molecules are excluded from the gap, creating an osmotic pressure imbalance that pushes the particles together. This depletion attraction is exploited in food science to create controlled texture and phase separation.

Hydrophobic forces act between non-polar surfaces in water. Water molecules near hydrophobic surfaces are entropically unfavourable, and when two hydrophobic surfaces approach, merging their hydration shells releases structured water, creating a strong attractive force that is not captured by DLVO. This drives the aggregation of oil droplets in the absence of stabilisers (emulsifiers).

Real-World Applications

Colloidal stability — and its deliberate disruption — underlies an enormous range of industrial and biological processes:

Frequently Asked Questions

What is a colloid?

A colloid is a mixture in which particles between roughly 1 nm and 1 µm in size are dispersed through a continuous medium. The particles are too large to dissolve but too small to settle rapidly under gravity. Examples include milk, blood plasma, fog, paint, and aerosol sprays.

What does DLVO theory predict?

DLVO theory predicts whether colloidal particles will aggregate or remain dispersed by summing the electrostatic double-layer repulsion and van der Waals attraction as a function of separation. A sufficiently high energy barrier (typically >10 k_BT) prevents aggregation and keeps the colloid stable.

Why does adding salt cause a colloid to coagulate?

Added salt ions compress the electric double layer around colloidal particles, reducing the Debye screening length and therefore the range of electrostatic repulsion. Once the energy barrier falls below the thermal energy, particles can approach closely enough for van der Waals attraction to pull them together irreversibly — coagulation.

What is zeta potential and why does it matter?

Zeta potential is the electric potential measured at the slipping plane (shear plane) around a colloidal particle. A magnitude greater than about 30 mV indicates sufficient electrostatic repulsion for stability. Values near zero predict rapid aggregation, as the double-layer repulsion can no longer overcome van der Waals attraction.

How does milk stay stable?

Milk is stabilised both electrostatically (casein micelles carry a negative charge, with a zeta potential around −20 mV) and sterically by protein and carbohydrate chains projecting from fat globule surfaces. Homogenisation reduces fat droplet size, increasing surface area and strengthening these stabilising mechanisms. Adding acid (vinegar or lemon juice) neutralises the charge, collapsing the barrier and causing curdling.

What is the Schulze-Hardy rule?

The Schulze-Hardy rule states that the critical coagulation concentration (CCC) of an electrolyte decreases sharply with the valence of the counterion: CCC ∝ z⁻⁶. So trivalent aluminium ions are approximately (3/1)⁶ = 729 times more effective than monovalent sodium at coagulating a negatively charged colloid, which is why alum is used in water treatment at low concentrations.

What are van der Waals forces between colloidal particles?

Between macroscopic colloidal particles, van der Waals forces arise from quantum-mechanical fluctuations of electron clouds across the entire particle volume, summed via the Hamaker approach or Lifshitz theory. They decay approximately as H⁻¹ between spheres (or H⁻² between flat surfaces) and are always attractive between like materials, with a strength set by the Hamaker constant A.

What is steric stabilisation?

Steric stabilisation uses polymer chains or surfactant molecules adsorbed to a particle surface as a physical brush layer. When two particles approach, the polymer layers overlap, reducing their configurational entropy and raising free energy — a repulsive force that operates even at high ionic strength where electrostatic repulsion is screened. Block copolymers with a particle-anchoring block and a solvated brush block are the archetypal steric stabiliser.

How is DLVO theory used in water treatment?

In water treatment, coagulants such as aluminium sulphate or ferric chloride are dosed into raw water to collapse the electric double layer on colloidal clay and organic particles. Trivalent Al³⁺ and Fe³⁺ ions act according to the Schulze-Hardy rule, destabilising the colloid at millimolar concentrations. The particles then aggregate into larger flocs that can be removed by sedimentation and filtration.

What are the limitations of DLVO theory?

DLVO theory assumes smooth, rigid, uniformly charged spheres in a medium described by continuum electrostatics. It ignores ion-specific effects (Hofmeister series), structured water layers at surfaces, steric forces from adsorbed polymers, hydrophobic attractions, and depletion forces from non-adsorbing species. Extended DLVO (XDLVO) models add these as extra terms to the interaction energy.

Try It Yourself

The best way to build intuition for colloidal stability is to watch particles interact and aggregate. Explore the related simulations: