Third Law of Thermodynamics

What the Third Law actually says

The First Law gives you energy bookkeeping. The Second Law gives you direction — entropy of an isolated system never decreases. But the Second Law only ever defines changes in entropy, ΔS; it leaves the absolute value floating, the way a thermometer with no marked zero leaves temperature floating. The Third Law of Thermodynamics nails down that missing zero:

The entropy of a perfect crystalline substance is exactly zero at absolute zero.
Formally: lim_T→0 S = 0 for a perfect crystal.

That single sentence does something the other laws cannot: it turns entropy from a relative quantity into an absolute one. Energy has no natural zero — we always measure energy differences and pick an arbitrary reference. Entropy is different. The Third Law says there is a real, physical bottom to the entropy scale, and a perfect crystal at absolute zero sits exactly on it.

Why entropy hits zero — the microstate count

The deepest way to see why this must be true is statistical. Ludwig Boltzmann's relation connects entropy to the number of microscopic arrangements (microstates) W consistent with the system's macroscopic state:

S = k_B · ln W, where k_B = 1.380649 × 10⁻²³ J/K.

Heat a crystal and its atoms can occupy many vibrational and electronic states; W is astronomically large and so is S. Now cool it. As thermal energy drains away, the atoms are forced into ever-lower energy states. A perfect crystal — every atom on its ideal lattice site, no defects, no isotopic mixing, a single non-degenerate ground state — has, at exactly 0 K, just one way to be arranged: W = 1. And ln 1 = 0. So:

S = k_B · ln 1 = 0.

The system has collapsed into perfect order: one microstate, zero disorder, zero entropy. This is the punchline the visualization above is built around — watch the lattice's wild thermal jitter freeze into a single rigid, motionless arrangement as the temperature counter falls to 0 K.

Nernst, Planck, and the heat theorem

The law arrived in two stages. In 1906, Walther Nernst proposed what he called his heat theorem: for reactions between pure, condensed (solid or liquid) phases, the entropy change ΔS of the reaction tends to zero as T → 0. Equivalently, the curves of ΔG and ΔH meet with zero slope at absolute zero. This was enormously useful — it let chemists predict chemical equilibria from thermal measurements alone — and earned Nernst the 1920 Nobel Prize in Chemistry.

Nernst's version only spoke about the difference in entropy between reactants and products. In 1911, Max Planck sharpened it into the modern statement: not just ΔS but the entropy of each individual perfect crystalline substance goes to zero. Planck's stronger form is what makes standard molar entropies possible. A subtlety remains: if a crystal's ground state is degenerate (some magnetic systems, or nuclear-spin degeneracy), the zero-point entropy is k_B·ln g where g is the degeneracy — usually tiny and conventionally ignored, but a reminder that "perfect crystal, non-degenerate ground state" is the precise condition.

The payoff: absolute entropies you can look up

Because we know S = 0 at the bottom, we can measure our way up. Integrate the experimentally determined heat capacity from 0 K to the temperature of interest, adding the entropy of each phase transition along the way:

S(T) = ∫₀^T (C_p / T) dT + Σ (ΔH_transition / T_transition)

For water you would integrate C_p of ice up to 273.15 K, add the entropy of fusion (ΔH_fus/T_m = 6010 J/mol ÷ 273.15 K = 22.0 J/mol·K), integrate liquid C_p to 373.15 K, add the entropy of vaporization (ΔH_vap/T_b = 40 660 J/mol ÷ 373.15 K = 109.0 J/mol·K), and continue with the vapor. Below the coldest measurable temperature, the Debye T³ law (C_p ∝ T³ for a non-metallic solid near 0 K) lets you extrapolate the integral cleanly to zero. The result is the standard molar entropy S° (298.15 K, 1 bar) tabulated in every data book.

Standard molar entropies S° (298.15 K, 1 bar) — note how order tracks down the list

Substance	State	S° (J/mol·K)	Why
Diamond (C)	solid	2.4	Rigid covalent lattice — minimal disorder
Graphite (C)	solid	5.7	Layered, slightly more freedom than diamond
Iron (Fe)	solid	27.3	Heavy metal, low-frequency vibrations
Water (H₂O)	liquid	69.9	Mobile molecules, H-bond network
Water (H₂O)	gas	188.8	Free translation + rotation
Hydrogen (H₂)	gas	130.7	Light, fast-moving molecules
Oxygen (O₂)	gas	205.2	More atoms, more modes than H₂

Solids sit lowest, liquids in the middle, gases highest — exactly the ordering the Third Law's "perfect order at the bottom" picture predicts. These absolute entropies feed directly into Gibbs free energy (ΔG = ΔH − TΔS) and so into every prediction of spontaneity and chemical equilibrium.

Why you can never reach absolute zero

A famous corollary, sometimes treated as an alternative statement of the law, is the unattainability principle: it is impossible to cool any system to exactly 0 K in a finite number of operations. The reason follows from the convergence of entropy.

Cooling works by alternately doing work on a system and letting it shed entropy — for example adiabatic demagnetization, where a paramagnetic salt is magnetized (aligning spins, lowering entropy), thermally isolated, then demagnetized (spins randomize, absorbing energy and dropping the temperature). Each cycle removes a slice of entropy. But the Third Law says all the accessible entropy curves — magnetized and demagnetized, or any two states — must converge to the same value (zero) as T → 0. So the entropy you can extract per step shrinks toward nothing as you approach the bottom. Closing the last gap would take infinitely many steps. In practice, laser-cooled and evaporatively-cooled atomic gases have reached the nanokelvin range and below, and bulk solids have hit roughly 100 picokelvin (10⁻¹⁰ K) — staggeringly cold, yet never exactly zero.

How cold can we get? Milestones toward 0 K

Temperature	System / method	Note
2.7 K	Cosmic microwave background	Coldest "natural" bath filling space
1.0 K	Pumped liquid ⁴He	Routine cryogenics
0.3 K	³He refrigerator	Single-shot cooling
~10 mK	Dilution refrigerator	Standard for quantum-computing hardware
~1 µK	Adiabatic demagnetization	Nernst's method, modern form
~100 pK	Bose–Einstein condensate (nuclear spins)	Among the coldest matter ever made
exactly 0 K	—	Forbidden by the Third Law

Residual entropy: when crystals cheat

The Third Law's clause "perfect crystal" is doing heavy lifting. Some substances cool fast enough that they freeze in a frozen-in disorder before they can settle into their unique ground state. They reach 0 K with W > 1, so they keep a small residual entropy S₀ = k_B·ln W > 0.

The classic case is carbon monoxide (CO). The molecule is nearly symmetric (C and O have similar sizes and a tiny dipole), so as it crystallizes each molecule can point C–O or O–C almost at random. With roughly two orientations per molecule, the predicted residual entropy is R·ln 2 = 5.76 J/mol·K; calorimetry finds about 4.6 J/mol·K (not quite the full ln 2 because the orientations aren't perfectly random). Water ice is the other textbook example: the hydrogen bonds can be arranged many ways subject to Pauling's "ice rules," giving a predicted R·ln(3/2) = 3.37 J/mol·K versus a measured ~3.4 J/mol·K — a beautiful match. N₂O and FClO₃ show similar orientational residual entropy. These substances don't break physics; they simply aren't perfect crystals, so the strict Third Law (which is about the idealized perfect crystal) doesn't apply to the real frozen sample.

Residual entropy from frozen-in disorder vs. the perfect-crystal ideal

Substance	Source of disorder	Predicted S₀	Measured S₀ (J/mol·K)
Perfect crystal (ideal)	none — W = 1	0	0
Carbon monoxide, CO	C–O / O–C orientation	R·ln 2 = 5.76	~4.6
Water ice, H₂O	H-bond (Pauling ice rules)	R·ln(3/2) = 3.37	~3.4
Nitrous oxide, N₂O	N–N–O / O–N–N orientation	R·ln 2 = 5.76	~4.8

Where the Third Law earns its keep

• Chemical thermodynamics. Every tabulated S° value — and therefore every ΔG and equilibrium constant computed from ΔG = ΔH − TΔS — rests on the Third Law's zero point.
• Cryogenics & quantum tech. Dilution refrigerators cooling superconducting qubits to ~10 mK, and the design limits of any cooling scheme, are governed by the unattainability principle.
• Materials & defects. Measuring residual entropy is a direct probe of frozen-in disorder, glassiness, and lattice imperfection.
• Low-temperature physics. The Debye T³ law and the vanishing of heat capacity as T → 0 are testable predictions of the law — a metal's electronic heat capacity going linearly to zero, an insulator's going as T³.
• Statistical foundations. The law is the macroscopic shadow of quantum mechanics: a unique, non-degenerate ground state is what S = 0 really means.

A built-in prediction: heat capacity must vanish

If S → 0 as T → 0 along every path, then S must be finite and continuous there, which forces the heat capacity to disappear at absolute zero: lim_T→0 C_p = 0. Classical equipartition wrongly predicts a constant C_v = 3R (the Dulong–Petit value) all the way down; experiment instead shows C_p dropping toward zero — falling as T³ for insulators (Debye) and as linear-in-T for the electron gas in metals. This collapse of heat capacity was one of the early triumphs of quantum theory and is exactly what the Third Law demands. It is also why the entropy integral ∫(C_p/T)dT converges at the bottom instead of diverging: C_p falls faster than T, so C_p/T stays integrable down to 0 K.