Seeking to explain thermodynamics based on moving and interacting atoms

Chapter 19 – Molecular transport: diffusion, viscosity, thermal conductivity

Chapter 18 left us with a good understanding of what an ideal gas is:

That final property is really what brings us to Chapter 19, and to the great question posed by Buys-Ballot: if gas molecules move so fast — over 400 meters per second — why does it take minutes for an odor to cross a room? Clausius rose to the challenge, answering it by defining the mean free path. Molecules are constantly colliding with each other, billions of times per second, and it’s this ceaseless collision that hinders their progress across any real distance. The mean free path is the average distance a molecule travels between collisions. For nitrogen and oxygen, with radii of about 1.8 × 10⁻⁸ cm, it comes to about 7 × 10⁻⁶ cm in air at room temperature and pressure-roughly 200 molecular diameters.

The root cause of every process in this chapter — diffusion, viscosity, and thermal conductivity — is the same: atoms and molecules in constant, restless motion, forever colliding and redirecting one another. Left alone, this ceaseless motion produces nothing but chaos. It’s only when a gradient exists — in composition, in bulk velocity, in temperature — that this same chaos organizes itself into a net, directional transport of matter, momentum, or energy from one region to another.

It’s worth pausing here to ask why this topic matters. Gas-phase diffusion governs how reactants find their way to a catalyst’s active sites, how isotopes are separated in a gas centrifuge, and how fuel and oxidizer find each other in a combustion flame. Gas-phase viscosity sets the pressure drop in a pipeline, the drag on an aircraft, and the losses inside a gas turbine. Gas-phase thermal conductivity determines why a vacuum flask keeps coffee hot, how a building gains or loses heat, and how a vehicle survives the heat of atmospheric re-entry. Three properties, one underlying mechanism.

So let’s begin with the simplest case: diffusion.

Diffusion

Recall from Chapters 5 and 6 that a system of particles evolves toward its most probable state: uniform in location, Maxwell-Boltzmann in energy. Once it gets there, all gradients have dissipated and the system is equilibrated — the only condition under which entropy properly applies.

Diffusion is one way a system gets there: the net transport of molecules from higher to lower concentration, arising purely from their random, ceaseless motion and collision. No bulk motion is required. There is no driving force involved. It’s pure statistics.

Now let’s look at what’s happening at the molecular level. As I was taught in grad school: picture yourself as a molecule in a gas. What governs how fast you make progress in any one direction?

First, your speed. Since kinetic energy (½mv²) is proportional to temperature, your average speed is proportional to √(T/m). Hotter means faster; heavier means slower.

Second, the obstacles in your way. As one of Buys-Ballot’s odorous molecules, you collide with other molecules billions of times per second. How far you travel between collisions — Clausius’s mean free path — depends on how crowded the gas is (the number of molecules per volume, N/V) and how large a target each molecule presents (its cross-sectional area, which scales with the square of the diameter, d²). The more crowded the gas and the larger the targets, the shorter your path.

So you move fast, but you never get far before being knocked in a new direction. Your journey is a random walk.

That’s one molecule. Diffusion, though, is a bulk phenomenon, built from vast numbers of molecules. Consider point 1 on one side of a room and point 2 on the other, with more odorous molecules at 1 than at 2. Every molecule wanders at random. But because more molecules start at 1, more will happen to wander from 1 to 2 than from 2 to 1. The result is a net flux from 1 to 2, proportional to the concentration gradient, dc/dx. Again, sheer statistics at work.

This is captured by Fick’s Law:

J = – DAB dcdx\huge\frac{dc}{dx}

J is the net flux of molecule A (the odorous molecule) through B (air), per unit area per unit time. The minus sign says the flux runs from high concentration to low. D is the diffusion coefficient, or diffusivity. It is not itself a rate; it is the proportionality constant that sets how quickly a given gradient disappears.

What sets D is the molecular picture above: how fast molecules move, and how often they’re stopped. The more crowded the gas and the larger each molecule, the more obstruction a molecule meets per unit distance travelled, and the shorter its mean free path:

D ∝molecular speedobstruction per unit distance∝T/m(N/V)d2=speed×mean free path\;\propto\; \frac{\text{molecular speed}}{\text{obstruction per unit distance}} \;\propto\; \frac{\sqrt{T/m}}{(N/V)\,d^{2}} \;=\; \text{speed} \times \text{mean free path}

Since N/V ∝\propto P/T for an ideal gas, this can also be written:

D ∝∝ T3m(P×d2)\frac {\sqrt{\frac{T^3}{m}}}{(P \times d^2)}

Note the use of the proportionality sign — additional constants have been left out to keep the variables in focus, and this treatment is meant to build intuition rather than deliver a rigorous derivation.

The full picture grows more complex for a mixture. The molecule you pictured moved at a speed set by its own mass, but when A diffuses through B, both are moving, and what matters is how quickly they close on each other. The mass term therefore becomes the reduced mass of the pair, mAmB/(mA + mB), and d becomes their average diameter. One consequence: the expression is symmetric, so in a two-component gas A diffuses through B as fast as B diffuses through A. Intermolecular forces, neglected here, also become significant as gases depart from ideal behavior.

Readers wanting to go deeper should consult Chapman & Cowling, “The Mathematical Theory of Non-Uniform Gases.”

This molecular picture of diffusion sets up what follows for thermal conductivity and viscosity. But before getting to those topics, let’s consider how diffusion plays a role in industrial processes.