Fick's Laws of Diffusion: Understanding How Particles Move

In 1855, a 26-year-old physiologist named Adolf Fick published a paper that would outlast him by nearly two centuries. Fick was trying to understand how gases exchange across the lung's membrane, and in the process, he wrote down two equations so simple and so general that they became the foundation of diffusion science. Every model of diffusion since — from the spread of pollutants in the atmosphere to the doping of semiconductors to, remarkably, the mathematics of AI image generation — traces its lineage to Fick's two laws.

What makes Fick's achievement remarkable is that he derived these laws by analogy with Fourier's heat equation, not from first principles. He didn't know about atoms or molecules — the atomic theory was still controversial in 1855. He simply observed that the flux of a substance should be proportional to its concentration gradient, just as the flux of heat is proportional to the temperature gradient. The guess turned out to be exactly right, and it was only decades later, with Einstein's analysis of Brownian motion, that the molecular basis for Fick's equations was fully understood.

Fick's First Law: The Flux Equation

Fick's first law describes the rate at which particles move. It states that the diffusion flux — the amount of substance passing through a unit area per unit time — is proportional to the negative of the concentration gradient:

This equation captures something fundamental: diffusion is driven by concentration differences, not by any external force. If the concentration is uniform everywhere (the gradient is zero), there is no net flux, even though individual molecules are still moving. Net flow appears only when there's a difference in concentration, and the steeper that difference, the faster the flow.

The diffusion coefficient D is where the physics lives. It depends on the temperature, the viscosity of the medium, and the size of the diffusing particle. For a small molecule like oxygen in water at body temperature, D is about 2 × 10⁻⁹ m²/s — meaning it takes about 0.5 seconds for oxygen to diffuse across a distance of 30 micrometers (the thickness of a capillary wall). For larger molecules like proteins, D is much smaller, and diffusion over biological distances becomes impractically slow.

The Intuition Behind the Gradient

Why does diffusion depend on the gradient rather than the concentration itself? The answer lies in the random walk. Each molecule moves randomly — it has no preferred direction. But in a region of high concentration, there are more molecules, so more of them will randomly wander into the adjacent low-concentration region than will wander the other way. The net flow is proportional to the difference in concentration across the boundary, which is exactly the gradient.

This is why a flat concentration profile (same everywhere) produces no net flow, while a steep profile (rapidly changing) produces fast flow. It's not that the molecules move faster in a steep gradient — they move at the same thermal speed. It's that the statistical imbalance is larger, so more molecules cross in one direction than the other.

Fick's Second Law: The Evolution Equation

Fick's first law tells you the flux at a point. But often what you want to know is how the concentration at a point changes over time. For that, you need the second law, which combines the first law with the principle of conservation of mass:

This is the diffusion equation, and it is one of the most important equations in mathematical physics. It says that concentration doesn't change in proportion to the gradient (that would be the first law) but in proportion to how the gradient is changing. A linear concentration gradient — constant slope — produces no change in concentration over time, because the flux is the same everywhere. But a curved profile — where the gradient is itself changing — means that more particles flow in from one side than flow out the other, so the local concentration changes.

The physical interpretation is beautiful: diffusion smooths out concentration. Any peak (where concentration is higher than its surroundings) will be eroded as particles flow away in all directions. Any valley will be filled as particles flow in. The process continues until the concentration is uniform everywhere — equilibrium.

Solutions and Behavior

The diffusion equation has well-known solutions. For a point source — an instantaneous release of particles at a single location — the concentration spreads as a Gaussian (bell curve) that widens over time. The width of the Gaussian grows as the square root of time: to spread twice as far, you need four times as long. This is the √t scaling that appears in Einstein's relation for Brownian motion, and it is a universal feature of diffusive processes.

This square-root scaling has profound implications. It means diffusion is fast over short distances and catastrophically slow over long distances. A signaling molecule might cross a cell (10 micrometers) in about 25 milliseconds, but to cross a centimeter by diffusion alone would take about 70 hours. This is why larger organisms need circulatory systems: diffusion alone cannot deliver oxygen across centimeter-scale distances fast enough to sustain life.

The Diffusion Coefficient: What Controls D

The diffusion coefficient D is the single parameter that captures the physics of a specific diffusing substance in a specific medium. It depends on three things:

• Temperature: Higher temperature means faster molecular motion and faster diffusion. D is proportional to absolute temperature.
• Viscosity: A thicker medium resists motion. D is inversely proportional to viscosity. This is why diffusion in honey is far slower than in water.
• Particle size: Larger particles diffuse more slowly. The Stokes-Einstein relation gives D = kT / (6πηr), where r is the particle radius, η is viscosity, k is Boltzmann's constant, and T is temperature.

The Stokes-Einstein equation is remarkable because it connects the macroscopic diffusion coefficient D to the microscopic size r of the diffusing particle. This is the equation Einstein used in 1905 to connect Brownian motion to Avogadro's number — measuring how far particles spread in a given time allowed him to calculate the number of molecules per mole, providing evidence that atoms were real.

Where Fick's Laws Apply

Fick's laws apply wherever diffusion is the dominant transport mechanism — that is, wherever there are no significant convective flows (bulk movement of the medium) or external forces. In practice, this covers a remarkable range of phenomena:

• Gas exchange in lungs: Oxygen and carbon dioxide diffuse across the alveolar membrane, the original problem Fick was studying.
• Semiconductor doping: Impurity atoms diffuse into silicon wafers at high temperature to create the p-n junctions that form transistors.
• Drug delivery: Controlled-release medications rely on diffusion of the drug through a polymer matrix.
• Food processing: Salting, curing, and brining are all governed by Fick's laws, as described in our article on diffusion in everyday life.
• Environmental science: Pollutants spreading through groundwater follow Fick's laws (alongside advection).

The Connection to the Fokker-Planck Equation

Fick's second law is, in fact, a special case of the Fokker-Planck equation — the case with no drift and constant diffusion coefficient. The Fokker-Planck equation generalizes Fick's law to include external forces (drift) and spatially varying diffusion coefficients. Fick described diffusion with no forces and uniform medium. The Fokker-Planck equation describes diffusion with forces and non-uniform conditions — the general case.

This lineage matters because the Fokker-Planck equation is also the mathematical foundation of AI diffusion models. The same equation that describes ink spreading through water — Fick's second law, the simplest diffusion equation — is the seed from which the entire theory of generative diffusion models grew. The connection runs from Fick in 1855 to the Fokker-Planck equation in the early 20th century to the score-based generative models of today. We trace this lineage in our article on the connection between physical and AI diffusion.

A Law That Endured

Adolf Fick derived his laws to understand the lung. He could not have anticipated that 170 years later, the same equations would describe how noise is added to images during AI training, or that the diffusion coefficient would become a tunable parameter in generative models. But the mathematics doesn't care about the context. Diffusion is diffusion, whether it's oxygen crossing a membrane or pixels turning to noise. Fick's genius was to capture the universal principle in two simple equations — and those equations have been spreading into new domains ever since.