science

Entropy: The Second Law's Enforcer

Entropy is a count, not a mess. Get that right and the arrow of time, life and information all fall into place.

Entropy measures how many microscopic arrangements (microstates) are consistent with what you observe of a system. Put another way, it measures how spread out a system's energy is. Boltzmann's formula writes it as S = k ln W. The second law says the entropy of an isolated system tends to increase, because there are overwhelmingly more high-entropy arrangements to wander into. "Disorder" is a popular shortcut, and often a misleading one.

Counting arrangements

In 1865, Rudolf Clausius needed a name for a quantity he had defined as heat divided by temperature. He built one from the Greek τροπή, which a standard lexicon translates as "turning" or "change", and which he rendered in German as Verwandlung, usually translated "transformation". He chose "entropy" partly because it echoed "energy", a concept he thought nearly "analogous in their physical significance" (Wikipedia).

The name stuck, and the quantity became the bookkeeper of the second law of thermodynamics. It also turned out to be stranger than Clausius knew. Entropy has several definitions, each correct, each showing a different side:

Clausius (1865): for a reversible process, dS = dQ/T. Heat divided by temperature, with no mention of atoms.
Boltzmann (1870s): S = k ln W. The logarithm of the number of microstates W, times Boltzmann's constant k.
Gibbs (1878): S = −k Σ pᵢ ln pᵢ, which weights each microstate by its probability and reduces to Boltzmann's formula when all are equally likely.
Shannon (1948): H = −Σ pᵢ log pᵢ, the entropy of information theory, which measures the average uncertainty in a message.

The dates for Gibbs and Shannon come from Wikipedia's entry on information entropy. The common core: entropy counts how many ways a system's parts could be arranged while looking the same from outside.

Why entropy increases

The second law says entropy increases, or at best stays constant, in an isolated system. Nothing pushes it upward. There is no entropy force.

Take a box of gas with a partition. All the molecules are on the left; the right side is empty. Remove the partition and the gas spreads out. The molecules don't "want" to spread. They just move and collide at random. But there are vastly more arrangements with molecules on both sides than with all of them on one side, so as the system wanders from microstate to microstate it ends up, almost certainly, in the spread-out ones.

OpenStax's textbook shows the counting with a tiny example: four particles in two boxes have 2⁴ = 16 possible microstates, and only one of them has all four particles in the left box (OpenStax Chemistry 2e, 16.2). Scale four particles up to 10²³ and "unlikely" becomes "never seen".

That is the whole mechanism: high-entropy macrostates are made of far more microstates.

Clausius's definition: heat over temperature

Clausius defined the entropy change as:

dS = dQ_rev / T

where dQ_rev is heat transferred reversibly and T is absolute temperature. Three consequences follow. Adding heat raises entropy. The same heat raises entropy more at low temperature, because you divide by a smaller T. And a reversible process leaves the total entropy of the system plus its surroundings unchanged; any irreversible process increases it (Feynman Lectures, Vol. I, ch. 44).

Why divide by temperature? A useful way to put it: a joule added to something already hot spreads its energy over a system that already has many accessible arrangements, so it changes the count relatively little. The same joule added to something cold makes a bigger relative difference. Lambert's paraphrase of the Clausius equation is that "entropy change measures energy's dispersion at a stated temperature" (Lambert, 2002).

Boltzmann's definition: counting microstates

Boltzmann's formula S = k ln W defines entropy as a count. W is the number of microscopic arrangements, positions and energies of every particle, that match the macroscopic state you observe. Boltzmann formulated the relation between 1872 and 1875; Max Planck later put it in its current form, around 1900, and it is carved on Boltzmann's gravestone in Vienna's Zentralfriedhof (Wikipedia).

The logarithm does three jobs. It makes entropy additive: combine two systems and their W values multiply, while their entropies add. It connects entropy to probability, because probabilities multiply too. And it tames enormous numbers.

How enormous? A mole of nitrogen gas at room conditions has a standard entropy of 191.6 J/(mol·K) (OpenStax, Appendix G). Divide by k = 1.38 × 10⁻²³ J/K and ln W ≈ 1.4 × 10²⁵. So W is about 10^(6 × 10²⁴): a number whose exponent alone has 25 digits. (An earlier version of this page gave W ≈ 10^(10²³), which understates it by a factor of about 60 in the exponent.)

Why "disorder" misleads

The popular shorthand is that entropy is disorder. It's an understandable shorthand, and it fails often enough to cause real confusion.

Frank Lambert, a chemistry educator, called disorder a "failure-prone crutch rather than a truly reliable, sturdy aid", and argued that any valid picture of an entropy change has to be about energy dispersal at the large scale, or about an increase in the number of accessible microstates at the molecular scale (Lambert, 2002). The physicist Daniel Styer called the common answer of "entropy as disorder" inadequate, warned that the alternative "entropy as freedom" is equally problematic, and suggested that the two, used cautiously and not too literally, together give real insight (Styer, 2000).

Where does "disorder" go wrong? Sometimes ordered-looking structures form spontaneously as entropy rises: crystals growing out of a solution, for instance, release heat that disperses into the surroundings, and Lambert uses cases like these to show why you have to count system plus surroundings rather than judge by appearance. And a shuffled deck of cards is not higher in thermodynamic entropy than a sorted one. The cards' arrangement is a fact about labels people care about, not about how many ways the molecules can hold their energy.

The same goes for the messy-room analogy. Rooms get messy because people move things and don't put them back, not because of the second law.

Entropy and information

Information theory reuses the mathematics, and the connection is more than a borrowed name. If you don't know which microstate a system is in, you lack information about it; high thermodynamic entropy means more bits would be needed to specify the exact state.

The link becomes physical in Rolf Landauer's 1961 argument that logically irreversible operations, such as erasing a bit, must dissipate heat, a minimum of about kT ln 2 per bit (Landauer, 1961). Erasing a bit takes a memory from two possible states to one. That is an entropy decrease in the memory, so to satisfy the second law an equal or greater entropy must appear in the surroundings, as heat. The details, and the experiments that have measured it, are in Landauer's Limit.

This idea also settled a 115-year-old puzzle. Maxwell imagined a tiny being that could sort fast molecules from slow ones and seemingly beat the second law. Charles Bennett argued in 1982 that the step that saves the second law "is not the making of a measurement... but rather the logically irreversible act of erasing the record of one measurement to make room for the next" (Bennett, 1982). That story is told in Maxwell's Demon.

Entropy in practice

Hot coffee cools. There are far more microstates with the thermal energy shared between coffee and room than with it concentrated in the coffee. Heat flows out; total entropy rises.

Ice melts above 0 °C. Liquid water has more accessible arrangements than ice: melting has ΔS ≈ +22.1 J/K per mole (OpenStax, 16.3). Above 0 °C the entropy gain wins over the heat cost, and below it the heat released by freezing wins. At constant temperature and pressure, Gibbs free energy does that bookkeeping for you, combining the system's entropy with the entropy its heat creates in the surroundings.

Entropy, time and life

Entropy is why time seems to have a direction. In isolated systems entropy never decreases, which means the arrow of entropy points the same way as the arrow of time (Wikipedia). Run a film backwards and you can tell: smoke gathering into a candle, a cup of coffee warming itself from the room. The second law essay follows that arrow further. Why the universe started with low entropy in the first place is a question for cosmology, and this essay won't pretend to answer it.

Living things are improbable, organized and far from equilibrium. They don't violate the second law because they aren't isolated. They take in free energy as food or sunlight and give off heat and waste, so the entropy of organism plus surroundings still rises. Erwin Schrödinger first wrote that life "feeds on negative entropy", then corrected himself in a later edition of What Is Life?: the true source is free energy (Wikipedia).

The same logic answers the old claim that the second law makes evolution impossible. Organisms are open systems, and local order can grow as long as more entropy is exported. A refrigerator does the same thing: it lowers the entropy inside by raising it more outside.

The count behind the curtain

Entropy is not a substance, not a force and not a mess. It is a count of the ways things could be, given what you can see. Rising entropy is the overwhelming statistics of that count, and nearly everything else, from cooling coffee to living cells, follows from taking it seriously.

The pebble: entropy is the number of secrets a system can keep while showing you the same face.

FAQ

Is entropy the same as disorder?

No. Disorder is a loose analogy that often misleads. Entropy measures the number of microstates consistent with a system's macroscopic state, or equivalently how widely its energy is dispersed. Some processes that look more ordered, like crystallization, still increase total entropy.

Can entropy ever decrease?

In an open system, yes, as long as more entropy is exported to the surroundings, which is what refrigerators and living cells do. In an isolated system, the second law says it doesn't decrease, though tiny, brief fluctuations can occur in systems of very few particles.

Does life violate the second law?

No. Living things are open systems. They take in free energy and release heat and waste, so the combined entropy of organism and environment increases even as the organism stays organized.

What does S = k ln W mean?

S is entropy, k is Boltzmann's constant (1.38 × 10⁻²³ J/K) and W is the number of microscopic arrangements consistent with what you observe. The logarithm makes entropy additive and turns astronomically large counts into manageable numbers.

Why does entropy give time a direction?

Because in isolated systems it doesn't decrease, so the direction of increasing entropy matches the direction we call the future. The underlying laws of motion work equally well backwards; the asymmetry comes from the statistics of microstates, together with the fact that entropy was lower in the past.