Third Law of Thermodynamics

Why the third law matters

• Cryogenics. The unattainability principle is not just philosophy — it tells you cooling cycles get geometrically less efficient as T → 0. To go from 1 K to 1 mK takes more refrigeration stages than from 100 K to 1 K. Modern dilution refrigerators reach 5 mK; magnetic cooling stages push to μK; nuclear demagnetization to nK; trapped-atom cooling to pK.
• Absolute entropy tables. Standard entropies S°(298 K) tabulated for thousands of compounds use S(T = 0) = 0 as the reference. Engineers compute reaction entropies ΔS°_rxn = ΣS°_products − ΣS°_reactants without any unknown additive constant — a foundation of practical chemistry.
• Phase-transition consistency. Calorimetric and spectroscopic entropies must agree at high T because both anchor at S(0) = 0. Discrepancies (ice, CO, glassy materials) are diagnostic — they quantify residual disorder.
• Quantum ground states. For S(0) → 0, the ground state must be non-degenerate (or weakly so). Frustrated systems with macroscopically degenerate ground states (spin glasses, ice rules, certain antiferromagnets) violate this — hot research topic.
• Chemical equilibria at low T. ΔG = ΔH − TΔS; as T → 0, ΔG → ΔH. Reactions become enthalpy-driven; ΔS plays no role. Sets why low-T reactivity differs qualitatively from room-T behavior.
• Astrophysical limits. Cosmic microwave background T = 2.725 K is the coldest natural temperature. Future heat death implies the universe asymptotes to a non-zero T or empty de Sitter horizons. The third law constrains what "true" rest looks like.
• Quantum simulators. Cold atom and ion experiments routinely cool to nanoKelvin to study ground states. The third law sets how much work it takes — and how much it costs in dollars per qubit-second.

The math

• Nernst statement (1906). ΔS → 0 as T → 0 for any reversible process between condensed-phase states. Differences vanish; absolute value undetermined.
• Planck refinement (1911). S(T = 0) = 0 for a pure crystalline substance. Removes the additive ambiguity.
• Statistical form. S = k ln W; if W → 1 (unique ground state), S → 0. Generalization: S → k ln g if ground state has degeneracy g (small constant, often negligible).
• Calorimetric entropy. S(T) = S(0) + ∫₀^T C_p(T')/T' dT' + Σ L_i/T_i (latent heats at phase transitions). With S(0) = 0 the formula gives absolute entropy.
• C_V → 0. Required by S = ∫ C_V/T dT being finite. C_V ∝ T^n for n > 0 satisfies this — phonons (T³), electrons (T), 2D systems (T²).
• α → 0. From a Maxwell relation (∂S/∂p)_T = −(∂V/∂T)_p = −Vα. As T → 0, (∂S/∂p)_T → 0 (entropy at any p approaches zero), so α → 0.

Common misconceptions

• "Entropy is zero at all T = 0." Only for perfect crystals. Glasses, frozen solutions, and frustrated magnets retain residual entropy. Ice's 3.4 J/mol/K residual entropy is measured both calorimetrically and predicted from configurational counting (Pauling 1935).
• "S = 0 is fundamental." Convention. Nernst's original statement was about ΔS → 0 between states, not about absolute S = 0. Planck added the convention because, with quantum mechanics in 1911, a unique ground state with W = 1 makes ln W = 0 natural.
• "Asymptotic only." Yes — asymptotically approached but not reached. Even infinitely many cooling cycles still leave you at T > 0. Lab "millikelvin" environments have inhomogeneous T fluctuations larger than the nominal mean.
• "Heat capacity is constant at low T." No — vanishes. Dulong-Petit's 3R holds at high T; phonon C_V → 234·R(T/Θ_D)³ at low T. By 4 K diamond's C_V is 10⁻⁸ of room-T value.
• "Glasses violate the third law." Not violation — glasses are out of true equilibrium. With infinite annealing time they would crystallize and S(0) = 0 would hold. On lab timescales they appear to violate it.
• "Quantum mechanics is required." Helps, but classical Nernst (1906) didn't need it. Quantum mechanics provides the mechanism (discrete energy levels, unique ground state) but the empirical evidence came earlier from low-T calorimetry.
• "Negative temperatures violate the third law." Negative absolute temperatures (population-inverted spin systems) are hotter than positive ones, not colder. They reside on the high-T side of T = ±∞. The third law's T → 0 limit is approached only from the positive side.

History

• 1906 Nernst. Heat theorem: ΔS → 0 as T → 0. Empirical, motivated by chemical-equilibrium data and Berthelot's principle.
• 1911 Planck. Refines: S(0) = 0 for crystalline solids. Unifies with quantum mechanics — Planck's quantization implied unique ground states.
• 1912 Nernst. Unattainability formulation: T = 0 unreachable in finite steps. Linked entropy theorem to refrigeration limits.
• 1920 Nernst Nobel Prize. "In recognition of his work in thermochemistry."
• 1925 ice anomaly. Calorimetric vs spectroscopic entropy mismatch in ice; 3.4 J/mol/K unaccounted for. Pauling explains in 1935 via proton disorder.
• 1937 He-4 superfluidity. Below 2.17 K, He-4 transitions to a quantum fluid; provides a real low-T testbed for the third law's predictions on entropy and heat capacity.
• 1949 Giauque Nobel Prize. Adiabatic demagnetization to milli-Kelvin temperatures — first application of magnetic cooling enabled by third-law analysis.
• 1995 BEC. Bose-Einstein condensate at 170 nK in dilute Rb gas; visible third-law unattainability — each laser-cooling cycle gets less efficient.
• 2003 magnetic cooling record. 100 pK in Rh nuclei. As of 2026 the lab record sits at a few tens of pK in trapped quantum gases — still infinite ratio above absolute zero.

Applications

• Cryogenic refrigeration design. Each stage of a cascade refrigerator (compression, throttling, magnetic cooling, dilution mixing) removes a finite ΔS at progressively diminishing rate. Counted accurately via third-law entropies.
• Chemical thermodynamics tables. CRC Handbook, NIST JANAF tables. Standard reaction entropies ΔS°_rxn computed from third-law-anchored S° values.
• Battery and fuel-cell design. Free-energy analysis ΔG = ΔH − TΔS uses absolute entropies. Predicts cell voltages, side-reaction susceptibilities, and stability ranges.
• Mineral physics. High-pressure phases of mantle minerals; absolute entropies measured calorimetrically and compared to first-principles partition functions. Third law sets the calibration.
• Spin-ice and frustrated magnets. Macroscopic ground-state degeneracy gives residual entropy R·ln(3/2) per spin (Pauling). Detected by calorimetry; one of the few systems where the third law is "almost violated".
• Nuclear demagnetization. Cooling nuclear spin systems to picoKelvin via adiabatic reduction of magnetic field — a textbook unattainability-principle illustration.
• Quantum simulators. Trapped ion, optical lattice experiments need ground-state preparation; cooling protocols and times bounded by third-law arguments.
• Black-hole thermodynamics. Generalized to extremal black holes: their entropy goes to a finite area-related value, with surface gravity κ → 0 playing the role of T. Active research on whether the third law applies analogously.

Worked example: residual entropy of ice

• Setup. Pauling (1935): each oxygen in hexagonal ice has four hydrogen bonds; two H atoms bond closer (covalent), two farther (H-bond). For each O, choose 2 of 4 bond directions to be "near" — that's C(4,2) = 6 possibilities.
• Counting. Naïvely 6^N configurations, but each H is shared between two O atoms; consistency reduces it. Pauling's combinatorics: W = (3/2)^N per molecule.
• Residual entropy. S_residual = k·ln(3/2)^(N_A) = R·ln(3/2) = 8.314 × 0.405 = 3.37 J/mol/K.
• Calorimetric measurement. Giauque & Stout (1936) measured ice's molar entropy by integrating C_p/T from 10 K to 273 K, then comparing to spectroscopic value for water vapor (computed from partition functions). Discrepancy: 3.41 J/mol/K. Match to within experimental error.
• Why this matters. Ice violates strict third-law S(0) = 0 because protons are kinetically frozen — no way to find the unique ordered ground state on lab timescales. The proton-ordered "ice XI" phase exists below 72 K with KOH catalyst but takes decades to form spontaneously.
• Other examples. Solid CO at 0 K: S_residual ≈ 4.6 J/mol/K (random C-O orientations). Solid N₂O: S_residual ≈ 6 J/mol/K. Glasses generally: residual ≈ Cp_glass · ΔT_freezing/T_g, often a few J/mol/K.