Colloids and Their Classification

Milk looks like a uniform white liquid, but it is not a solution at all. It is a swarm of fat droplets and protein clusters suspended in water, each too small to see but far larger than a single molecule. This intermediate world, between true solutions and coarse mixtures like sand in water, is the domain of colloids. Colloid science matters because most real materials are colloidal: paint, blood, smoke, fog, ink, gels, and countless pharmaceutical formulations. Understanding how particle size and phase combination govern behavior lets chemists predict stability, control texture, and design new materials. This article defines colloidal dispersions and fixes their size range. It classifies them by dispersed and continuous phase. It closes with worked examples, common misconceptions, and a preview of the surface chemistry that governs their stability.

Key takeaways

A colloid is a dispersion of particles roughly 1 nm to 1000 nm across, suspended in a continuous medium.
This size range sits between true solutions (sub-nm) and coarse suspensions (above 1000 nm, which settle quickly).
Colloids are classified by combining the physical state, solid, liquid, or gas, of the dispersed phase and the continuous phase.
Named systems include sols, emulsions, foams, aerosols, and gels, each a specific phase pairing.
Gas-in-gas mixtures are never colloidal; gases always mix at the molecular level.
Key idea: Colloidal behavior arises purely from particle size, not chemical identity, so the same substance can be a true solution or a colloid depending on how finely it is divided.

The core insight

A colloidal dispersion is defined not by what it is made of but by how big its dispersed units are: particles, droplets, or bubbles with at least one dimension between about 1 nm and 1000 nm, small enough to stay suspended against gravity through thermal (Brownian) motion, yet large enough to scatter light and present a huge internal surface area that dominates the system's physical behavior.

The physical picture

Ordinary matter can be divided into three regimes based on the size of the particles dispersed in a medium. In a true solution, the solute exists as individual ions or molecules, typically under 1 nm across; sodium chloride dissolved in water is a familiar example. In a coarse (mechanical) suspension, particles exceed roughly 1000 nm (1 μm) and settle out under gravity within seconds to hours, like sand stirred into water. Between these two extremes lies the colloidal regime. A colloidal particle is large enough that it does not diffuse away like a small molecule, but small enough that random thermal collisions with solvent molecules, Brownian motion, keep it suspended indefinitely rather than sedimenting. This single size criterion, not chemical composition, is what makes a system colloidal. Starch dissolved gently in cold water is a true solution of small molecules; the same starch, gelatinized, forms a colloidal paste. Gold atoms in solution are not colloidal; gold nanoparticles of 20 nm are the classic colloidal sol first studied systematically by Michael Faraday. Because colloidal particles are so small, their surface-area-to-volume ratio is enormous. A single 1 cm cube of material has a surface area of 6 cm². Divided into 1 nm cubes, the same volume of material presents a total surface area on the order of 6 × 10⁷ cm², a ten-millionfold increase. This is why colloid chemistry is inseparable from surface chemistry: interfacial energy, not bulk energy, controls stability, aggregation, and the counterintuitive properties of colloidal systems (foreshadowing later discussions of lyophilic versus lyophobic colloids and the electrical double layer). Classification of colloids rests on a second, independent variable: which phase, solid, liquid, or gas, is dispersed, and which phase forms the continuous surrounding medium. Because there are three possible states for each of the two roles, nine combinations are conceivable. One combination, gas dispersed in gas, is never observed as a colloid because gases mix completely at the molecular level and cannot form a stable, phase-separated dispersion of one gas in another.

Figure

The three size regimes of dispersed matter, spanning true solution to coarse suspension

Figure

Classification grid of colloids by dispersed phase and continuous phase

The core relations

1\ \text{nm} \lesssim d_{\text{particle}} \lesssim 1000\ \text{nm}

where: $d_{\text{particle}}$ is the characteristic dimension of the dispersed unit defining the colloidal size window.

\frac{A}{V} = \frac{6}{d}

where: $A/V$ is the surface-area-to-volume ratio of a spherical (or cubic) particle of diameter $d$ , showing how surface area explodes as particles shrink.

\langle x^2 \rangle = 2Dt

where: $\langle x^2 \rangle$ is the mean-square displacement of a colloidal particle undergoing Brownian motion in time $t$ , and $D$ is its diffusion coefficient (Einstein relation, the microscopic reason colloids resist settling).

D = \frac{k_{\mathrm{B}}T}{6\pi \eta r}

where: $k_{\mathrm{B}}$ is Boltzmann's constant, $T$ the temperature, $\eta$ the solvent viscosity, and $r$ the particle radius (Stokes–Einstein relation).

Derivation AWhy the colloidal size window has a lower and an upper bound

Goal. Show, using elementary kinetic and gravitational arguments, why colloidal particles are bounded below by roughly 1 nm and above by roughly 1000 nm.

Step 1: The lower bound comes from distinguishability from true solutes. A species is a "true" solute, not a colloidal particle, when its size approaches that of solvent molecules themselves, roughly 0.3 to 1 nm for water and small organics. Below about 1 nm there is no meaningful separate "particle" with its own interface; the species behaves as an individually solvated ion or molecule governed by ordinary solution thermodynamics (colligative properties, Raoult's law), not by interfacial phenomena. The colloidal regime therefore cannot begin below the scale of molecular dimensions.

Step 2: The upper bound comes from balancing gravitational settling against Brownian agitation. Consider a spherical particle of radius $r$ and density $\rho_p$ in a fluid of density $\rho_f$ and viscosity $\eta$ . Its downward gravitational force, corrected for buoyancy, is

F_{\text{grav}} = \frac{4}{3}\pi r^3 (\rho_p - \rho_f) g

Under Stokes drag, this force produces a terminal settling velocity

v_{\text{sed}} = \frac{2r^2(\rho_p - \rho_f)g}{9\eta}

Step 3: Compare settling to diffusive (Brownian) motion over the same time. The characteristic distance a particle diffuses in time $t$ is of order

x_{\text{diff}} \sim \sqrt{2Dt}, \qquad D = \frac{k_{\mathrm{B}}T}{6\pi \eta r}

A particle remains effectively "suspended" (colloidal) as long as diffusive displacement dominates over sedimentation displacement on laboratory timescales, i.e., $x_{\text{diff}} \gtrsim x_{\text{sed}} = v_{\text{sed}}\,t$ . Substituting typical values for a mineral particle in water at 298 K ( $\rho_p - \rho_f \approx 1500\ \text{kg m}^{-3}$ , $\eta \approx 1.0\times10^{-3}\ \text{Pa s}$ ) and solving for the radius at which the two displacements become comparable over an hour-long observation gives $r$ on the order of several hundred nanometers.

Step 4: Interpret the crossover as the practical upper size limit. Above roughly $r \sim 500\ \text{nm}$ (diameter ~1000 nm), gravitational settling wins decisively and the system behaves as a coarse, non-colloidal suspension that visibly separates on standing. Below this, Brownian motion keeps particles kinetically suspended indefinitely, even though they remain thermodynamically distinct from the medium (unlike true solutions).

\boxed{1\ \text{nm} \lesssim d \lesssim 1000\ \text{nm} \ \text{defines the colloidal domain, bounded below by molecular scale and above by the onset of gravitational sedimentation}}

Worked examples

Example 1 — Surface area gained by subdividing a particle into colloidal fragments (foundational). A 1.00 cm cube of silica (density irrelevant here, only geometry matters) is broken into cubic colloidal particles of edge length 10.0 nm. Step 1: original surface area. A cube of edge $a = 1.00\ \text{cm}$ has surface area

A_0 = 6a^2 = 6(1.00\ \text{cm})^2 = 6.00\ \text{cm}^2

Step 2: number of small cubes obtained. Volume is conserved:

V = a^3 = (1.00\ \text{cm})^3 = 1.00\ \text{cm}^3 = 1.00\times10^{21}\ \text{nm}^3

Each small cube has edge $a' = 10.0\ \text{nm}$ , volume $a'^3 = 1.00\times10^{3}\ \text{nm}^3$ . The number of small cubes is

N = \frac{1.00\times10^{21}\ \text{nm}^3}{1.00\times10^{3}\ \text{nm}^3} = 1.00\times10^{18}

Step 3: total surface area of the small cubes. Each small cube has surface area $6(10.0\ \text{nm})^2 = 600\ \text{nm}^2 = 6.00\times10^{-14}\ \text{cm}^2$ . Total area:

A = N \times 6.00\times10^{-14}\ \text{cm}^2 = (1.00\times10^{18})(6.00\times10^{-14}\ \text{cm}^2) = 6.00\times10^{4}\ \text{cm}^2

Step 4: ratio.

\frac{A}{A_0} = \frac{6.00\times10^{4}}{6.00} = 1.00\times10^{4}

Subdividing down to 10 nm colloidal particles increases surface area ten-thousandfold, from 6 cm² to 6 m². This directly explains why colloidal sols and gels expose enormous interfacial area, and why surface phenomena (adsorption, surface charge) dominate their chemistry. Sanity check: halving the edge length always doubles surface area for fixed volume, and here the edge shrank by a factor of $10^6$ (cm to nm scale), giving area scaling by $10^6$ ... but volume-conservation redistributes into $N=10^{18}$ pieces, correctly yielding the $10^4$ factor found, consistent with the general result $A/A_0 = a/a'$ for cubes of the same total volume.

Example 2 — Does a 200 nm polystyrene particle in water stay suspended, or settle, over one hour? (applied, exam-level). A polystyrene latex sphere of radius $r = 100\ \text{nm} = 1.00\times10^{-7}\ \text{m}$ (diameter 200 nm, well within the colloidal window) is dispersed in water at $T = 298\ \text{K}$ . Polystyrene density $\rho_p = 1050\ \text{kg m}^{-3}$ ; water density $\rho_f = 997\ \text{kg m}^{-3}$ ; water viscosity $\eta = 8.9\times10^{-4}\ \text{Pa s}$ . Compare the sedimentation distance to the diffusion distance over $t = 3600\ \text{s}$ . Step 1: sedimentation velocity (Stokes' law with buoyancy).

v_{\text{sed}} = \frac{2r^2(\rho_p-\rho_f)g}{9\eta}

v_{\text{sed}} = \frac{2(1.00\times10^{-7}\ \text{m})^2(1050-997\ \text{kg m}^{-3})(9.81\ \text{m s}^{-2})}{9(8.9\times10^{-4}\ \text{Pa s})}

Numerator: $2 \times (1.00\times10^{-14}\ \text{m}^2) \times 53\ \text{kg m}^{-3} \times 9.81\ \text{m s}^{-2} = 1.04\times10^{-11}\ \text{kg m}^{-1}\text{s}^{-2}$ . Denominator: $9 \times 8.9\times10^{-4} = 8.01\times10^{-3}\ \text{Pa s}$ .

v_{\text{sed}} = \frac{1.04\times10^{-11}}{8.01\times10^{-3}} = 1.30\times10^{-9}\ \text{m s}^{-1}

Step 2: sedimentation distance in 1 hour.

x_{\text{sed}} = v_{\text{sed}} \times t = (1.30\times10^{-9}\ \text{m s}^{-1})(3600\ \text{s}) = 4.68\times10^{-6}\ \text{m} = 4.68\ \mu\text{m}

Step 3: diffusion coefficient (Stokes–Einstein).

D = \frac{k_{\mathrm{B}}T}{6\pi\eta r} = \frac{(1.381\times10^{-23}\ \text{J K}^{-1})(298\ \text{K})}{6\pi(8.9\times10^{-4}\ \text{Pa s})(1.00\times10^{-7}\ \text{m})}

Numerator: $4.12\times10^{-21}\ \text{J}$ . Denominator: $6\pi \times 8.9\times10^{-4} \times 1.00\times10^{-7} = 1.678\times10^{-9}$ .

D = \frac{4.12\times10^{-21}}{1.678\times10^{-9}} = 2.45\times10^{-12}\ \text{m}^2\text{s}^{-1}

Step 4: root-mean-square diffusive displacement in 1 hour (one dimension).

\sqrt{\langle x^2\rangle} = \sqrt{2Dt} = \sqrt{2(2.45\times10^{-12}\ \text{m}^2\text{s}^{-1})(3600\ \text{s})}

= \sqrt{1.76\times10^{-8}\ \text{m}^2} = 1.33\times10^{-4}\ \text{m} = 133\ \mu\text{m}

Step 5: compare. Diffusive displacement (133 μm) exceeds sedimentation displacement (4.68 μm) by a factor of about 28. Brownian motion dominates comfortably, so this 200 nm particle behaves as a genuine, kinetically stable colloid over an hour, consistent with its diameter lying well inside the 1–1000 nm window. Sanity check: had we used $r = 5\ \mu\text{m}$ instead (a coarse suspension particle), $v_{\text{sed}}$ scales as $r^2$ (2500× larger) while $D$ scales as $1/r$ (50× smaller), flipping the comparison decisively toward sedimentation, as expected for non-colloidal particles.

Limiting cases & common traps

Situation	Correct view
Particle size below ~1 nm	True (molecular) solution, not colloidal; governed by solution thermodynamics
Particle size above ~1000 nm	Coarse suspension; settles under gravity within a practical timescale
Gas dispersed in gas	Never colloidal; gases always mix homogeneously at the molecular level
Solid dispersed in solid	Still colloidal if domain size is in range (e.g., ruby glass, some alloys)
"Colloid = liquid mixture"	False; colloids can be solid-in-solid, solid-in-gas (smoke), or gas-in-liquid (foam)
Same chemical substance	Can be a true solution OR a colloid depending purely on particle size, not identity

The chemist behind it

Thomas Graham, working in the 1850s–1860s, coined the term "colloid" (from Greek kolla, glue) after noticing that substances like starch, gelatin, and albumin diffused through parchment membranes far more slowly than salts and sugars, and could not be crystallized easily. He classified substances as "crystalloids" (fast-diffusing, crystallizable) versus "colloids" (slow-diffusing, glue-like). Although his mechanistic explanation, that colloids were a distinct class of chemical substance, was later superseded by the recognition that colloidal behavior is a matter of particle size rather than chemical identity, his diffusion experiments and terminology founded the field. Michael Faraday's mid-19th-century work on stable gold sols, showing that finely divided gold particles gave distinctive ruby-red colors and remained suspended for years, provided the classic physical demonstration that colloid science later explained quantitatively.

Origins · 1 minThomas Graham and how this was discovered

Thomas Graham

1805–1869 · Scottish chemist who coined "colloid" and pioneered dialysis and diffusion studies distinguishing colloids from true solutions

Ask the tutor: How does the electrical double layer explain why some colloidal sols stay stable for years while others coagulate within minutes?

Is there a sharp, universally agreed cutoff between a colloid and a true solution?

No. The 1 nm to 1000 nm range is a practical convention, not a fundamental law. The lower bound reflects the size of small molecules and ions; the upper bound reflects where gravitational sedimentation begins to dominate Brownian motion on laboratory timescales (minutes to days). Some texts use 1 nm to 500 nm; others extend to 1 micrometer. The physically meaningful criterion is always the competition between diffusion and settling, illustrated in the second worked example, not a rigid numerical wall.

Why can the same chemical substance be either a true solution or a colloid?

Because colloidal behavior depends on particle size, not chemical composition. Dissolved starch molecules in dilute cold water behave as a true solution because individual polymer chains are small and well solvated. Gelatinized or aggregated starch forms clusters in the colloidal size range, producing a paste with light-scattering (Tyndall effect) and settling resistance typical of colloids. The chemistry is unchanged; only the degree of aggregation differs.

How does the classification grid help predict a colloid's real-world behavior?

Knowing which phase is dispersed and which is continuous tells you immediately what stabilizing forces matter. In a sol (solid in liquid) or emulsion (liquid in liquid), surface charge and steric stabilization at the solid-liquid or liquid-liquid interface control aggregation, foreshadowing the electrical double layer. In a foam (gas in liquid), thin liquid film drainage and surfactant packing control lifetime. The phase pairing is the first diagnostic step before deeper interfacial analysis.

Why is gas-in-gas never a colloid, while gas-in-liquid (foam) is?

Gases lack a stable interface between two gas phases because gas molecules intermix completely and rapidly at the molecular level; there is no way to maintain discrete gas "particles" within another gas, so no interface, and hence no colloid, can persist. In a foam, the continuous liquid phase provides a stable film and surface tension that can trap and maintain distinct gas bubbles as a genuine dispersed phase with a real interface.

Are colloids thermodynamically stable, like true solutions, or just kinetically stable?

This depends on the type of colloid, a distinction developed further when comparing lyophilic and lyophobic colloids. Many colloids, especially those stabilized by surface charge or added surfactant, are only kinetically stable: they resist aggregation on observable timescales due to energy barriers (foreshadowing the electrical double layer), but are not at the global thermodynamic minimum. Others, like certain polymer solutions, approach true thermodynamic stability.