Colloids & Emulsions

The state between dissolved and settled

Stir sugar into water and the crystals vanish: the molecules break apart into individual species smaller than a nanometre, surrounded by water, never to be seen again. Stir sand into water and it sinks: the grains are tens of micrometres across, heavy enough that gravity drags them to the bottom in seconds. Between those two extremes lies a third regime — particles big enough to stay clumped as little packets of many molecules, yet small enough that the random buffeting of solvent molecules keeps them permanently afloat. That is a colloid.

The Scottish chemist Thomas Graham coined the word in 1861 (from the Greek kólla, "glue") after noticing that substances like gelatin and albumin diffused through a membrane far more slowly than salts, and could not be crystallized. He divided matter into crystalloids (which pass freely and crystallize) and colloids (which do not). We now know there is no sharp chemical divide — the same substance can be dissolved, colloidal, or a coarse suspension depending only on how finely it is divided. The defining variable is particle size, conventionally 1 nm to 1 µm (1000 nm).

Every colloid has two parts: the dispersed phase (the particles or droplets) and the continuous phase or dispersion medium (the stuff they float in). Milk is fat droplets (dispersed) in water (continuous). Fog is water droplets in air. Paint is solid pigment in a liquid vehicle. Because each phase can independently be solid, liquid, or gas, there are eight named colloid types — covered in the table below.

Why they never settle: Brownian motion vs gravity

The reason a colloid is permanent while a suspension is not is a contest between two forces. Gravity pulls a particle down at a sedimentation speed given by Stokes' law:

v_sed = 2 r² (ρ_p − ρ_f) g / (9 η)

where r is the particle radius, ρ the densities of particle and fluid, g = 9.81 m/s², and η the viscosity. The crucial term is r²: halve the radius and the settling speed drops fourfold. Against this, every particle is being kicked at random by collisions with solvent molecules — Brownian motion — with a mean-square displacement set by the Einstein–Smoluchowski relation:

⟨x²⟩ = 2 D t,    D = k_B T / (6 π η r)

Here k_B = 1.38 × 10⁻²³ J/K and D is the diffusion coefficient. Plug in numbers: a 100 nm latex sphere in water (η ≈ 1 × 10⁻³ Pa·s, ρ_p ≈ 1050 kg/m³, so it is only slightly denser than water) settles at just ~3 × 10⁻¹⁰ m/s — barely 0.02 mm per day — while Brownian motion shuffles it a few micrometres every second in a random walk. The jitter utterly overwhelms the sinking, so the particle never reaches the bottom; the dispersion is kinetically stable. For a 10 µm sand grain, by contrast, settling jumps to ~5 × 10⁻⁵ m/s (centimetres per minute) and Brownian motion becomes negligible — it sinks. The 1 µm boundary of the colloidal range is exactly where these two rates cross over.

The Tyndall effect: seeing the particles with light

Colloidal particles are comparable in size to the wavelength of visible light (400–700 nm), so they scatter it strongly. Shine a beam through a colloid and the beam's path lights up sideways — the Tyndall effect, named for John Tyndall who explained the blue of the sky and the visibility of sunbeams the same way. A true solution shows nothing, because its sub-nanometre particles are far too small to scatter. This is the single quickest bench test for a colloid: a laser pointer makes a visible track through milk diluted 1:1000, fog, or a starch sol, but not through salt water.

The scattering intensity for particles much smaller than the wavelength follows Rayleigh's law, I ∝ 1/λ⁴, which is why scattered light is bluish (short wavelengths scatter more) and transmitted light reddens — the same physics that paints the sky blue and sunsets red. Modern instruments turn this into measurement: dynamic light scattering (DLS) reads the time-fluctuation of scattered laser light to recover the diffusion coefficient D, then inverts the Einstein relation to report particle size, routinely down to a few nanometres.

The eight colloid types — and everyday examples

Dispersed → in Continuous	Name	Everyday example
Liquid in gas	Liquid aerosol	Fog, clouds, hairspray mist
Solid in gas	Solid aerosol / smoke	Smoke, dust, soot
Gas in liquid	Foam	Whipped cream, beer head, shaving foam
Liquid in liquid	Emulsion	Milk, mayonnaise, butter (water-in-oil)
Solid in liquid	Sol	Paint, blood, ink, gold sol
Gas in solid	Solid foam	Pumice, styrofoam, aerogel, marshmallow
Liquid in solid	Gel	Jelly, cheese, opal, agar
Solid in solid	Solid sol	Ruby glass (gold in glass), pigmented plastics

The only combination missing is gas-in-gas: all gases mix at the molecular level, so that is always a true solution, never a colloid. Note that emulsion is specifically the liquid-in-liquid case — milk, mayonnaise, vinaigrette — while sol is solid-in-liquid. The two words are often blurred in casual use, but the size rule (1–1000 nm) is the same for all eight.

Surfactants: lowering the cost of making interface

Dispersing oil in water creates an enormous amount of interface. One millilitre of oil broken into 1 µm droplets has a total surface area of about 6 m². Every square metre of oil–water interface stores energy equal to the interfacial tension γ — roughly 50 mN/m for clean oil–water. That 6 m² therefore costs ≈ 0.3 J, and the system relentlessly tries to shed it by letting droplets coalesce back into one blob with minimal area. This is why a fresh oil-and-vinegar mix breaks within minutes.

A surfactant (surface-active agent) fixes this. It is an amphiphile — a molecule with a hydrophilic head and a hydrophobic tail, like sodium dodecyl sulfate (SDS, CH₃(CH₂)₁₁OSO₃⁻Na⁺) or a phospholipid. At the interface it sits with its tail in the oil and head in the water, so neither end is forced into a hostile phase. This adsorption drops the interfacial tension from ~50 mN/m to a few mN/m, slashing the energy cost of small droplets and letting them form and persist. The amount of surfactant that prefers the interface, and the curvature it favours (which decides oil-in-water vs water-in-oil), is captured by the HLB scale (hydrophilic–lipophilic balance, 0–20): HLB 3–6 favours water-in-oil emulsions, HLB 8–18 favours oil-in-water.

Above a threshold called the critical micelle concentration (CMC ≈ 8 mM for SDS in water), surplus surfactant in the bulk self-assembles into micelles — tiny spheres with tails tucked inside, away from water. Micelles are the engines of detergency: their oily cores solubilize grease that water alone cannot touch, which is the entire chemistry of soap.

DLVO theory: the energy barrier that keeps particles apart

Even with low interfacial tension, two colloidal particles will still merge if they touch, because van der Waals forces always pull them together at close range. What keeps a colloid stable is a repulsion that erects an energy barrier before they can touch. The quantitative account is DLVO theory (Derjaguin, Landau, Verwey, Overbeek, 1940s), which sums two interaction energies as a function of separation:

V_total(h) = V_attraction(h) + V_repulsion(h)

V_attraction = − A·r / (12 h)          (van der Waals, always pulls in)
V_repulsion  ∝ + exp(−κ h)             (electric double-layer, pushes out)

κ⁻¹ = Debye length (thickness of the ionic atmosphere)

Particles in water usually carry a surface charge (from ionized surface groups or adsorbed ions), surrounded by a diffuse cloud of counter-ions — the electric double layer. When two like-charged particles approach, their double layers overlap and repel. The net V_total(h) has a primary maximum — an energy barrier — that incoming particles must surmount before van der Waals attraction can snap them together. If that barrier is more than about 15–25 k_BT, thermal collisions almost never clear it and the colloid stays dispersed for years.

The measurable proxy for the surface charge is the zeta potential ζ, the electric potential at the slipping plane around a moving particle. As a rule of thumb, |ζ| > 30 mV (positive or negative) means strong enough repulsion for a stable dispersion; |ζ| near 0 mV — the isoelectric point — means no barrier and rapid aggregation. Tuning pH, salt, or surfactant to move ζ across that line is how formulators stabilize a paint or, deliberately, break a colloid.

Breaking a colloid: coagulation, the Schulze–Hardy rule, and deltas

To destroy a colloid you collapse the repulsive barrier. Adding electrolyte compresses the double layer (shrinks the Debye length κ⁻¹), lowering the DLVO barrier until particles aggregate — coagulation. The dramatic part is the charge dependence: the Schulze–Hardy rule states that the critical coagulation concentration scales as the inverse sixth power of the counter-ion charge, so the coagulating power of Na⁺ : Ca²⁺ : Al³⁺ runs roughly 1 : 64 : 729. A trivalent ion is hundreds of times more effective than a monovalent one at the same concentration.

This explains river deltas. River water carries negatively charged colloidal clay. Where it meets the sea, the Na⁺, Mg²⁺ and Ca²⁺ of saltwater compress the double layer, the clay coagulates, and it drops out of suspension — building the Mississippi, Nile and Ganges deltas grain by grain. Water-treatment plants do the same on purpose, dosing alum (aluminium sulfate, Al₂(SO₄)₃) so the Al³⁺ coagulates suspended colloids into flocs that settle and can be filtered. The reverse trick — peptization, adding a stabilizing ion to re-charge and re-disperse a coagulate — is how a freshly washed precipitate annoyingly slips through filter paper.

For emulsions, "breaking" can also proceed by: creaming (droplets rise or sink without merging, like cream on raw milk — reversible by shaking); flocculation (droplets stick in loose clumps); coalescence (droplets actually merge, irreversible); and Ostwald ripening (large droplets grow at the expense of small ones because smaller droplets have higher internal pressure, set by the Laplace relation ΔP = 2γ/r). Homogenizing milk to sub-micron droplets and coating them with protein blocks all four, which is why a carton lasts weeks without separating.

Where colloids run the world

• Food. Milk (oil-in-water emulsion stabilized by casein), butter (water-in-oil), mayonnaise (egg-lecithin-stabilized, up to 80% oil yet still pourable), whipped cream (foam), cheese and jelly (gels). Emulsifiers like lecithin (E322) and mono/diglycerides (E471) are among the most-used food additives on Earth.
• Biology. Blood is a colloidal sol of proteins and a suspension of cells; the cytoplasm is a crowded colloidal gel; bile salts emulsify dietary fat into micelles so lipase and the gut can absorb it. Without emulsification you cannot digest fat.
• Industry. Paints and inks are pigment sols engineered to a high zeta potential so they don't settle in the can; latex, asphalt emulsions for road surfacing, and pharmaceutical creams and lotions are all emulsions; photographic film was a silver-halide colloid in gelatin.
• Environment & medicine. Clouds and fog are aerosols; smog and diesel soot are pollutant aerosols; intravenous lipid nutrition (Intralipid) is a sterile soybean-oil emulsion; the lipid nanoparticles that deliver mRNA vaccines are engineered colloidal carriers ~80–100 nm across.

Common misconceptions & pitfalls

• "A colloid is just a weak solution." No — the dispersed particles are intact clusters of thousands to millions of molecules, not individual solvated species. That is why colloids scatter light (Tyndall) and solutions don't, and why colloids are filtered by fine membranes (dialysis, ultrafiltration) that pass true solutes.
• "Colloids are thermodynamically stable." Almost never. They are kinetically stable — trapped behind a DLVO energy barrier in a metastable state. Given enough time, energy, or salt, every colloid wants to coalesce and minimize interface. The exceptions are association colloids like micelles, which are genuinely at equilibrium.
• "Emulsion type follows whichever liquid you add first." It is set mainly by the surfactant, not the mixing order. Bancroft's rule: the phase in which the surfactant is more soluble becomes the continuous phase. A water-soluble surfactant (high HLB) gives oil-in-water; an oil-soluble one (low HLB) gives water-in-oil — which is why the same oil and water can form either emulsion depending on the emulsifier.
• "More surfactant always means a more stable emulsion." Only up to a point. Past the CMC, extra surfactant just forms bulk micelles and can even destabilize via depletion flocculation, where crowding micelles push droplets together. Stability is about interfacial coverage and the resulting barrier, not raw surfactant dose.
• "Filtering removes colloidal particles." Ordinary filter paper has pores of tens of micrometres and lets colloids straight through — that's the whole point of peptization being a nuisance. Removing them needs coagulation first, or a membrane with sub-micron pores (ultrafiltration / dialysis).