Third Law of Thermodynamics

The Second Law told us entropy exists, but it left a loose thread: it defined only differences of entropy, through $\mathrm{d}S = \delta Q_{\text{rev}}/T$ , leaving the additive constant undetermined. You could add any number to every entropy and nothing measurable would change. The Third Law ties that thread off. It asserts that as the absolute temperature falls to zero, the entropy of a system approaches a constant independent of thermodynamic parameters; for a perfect crystal with a non-degenerate ground state, this constant is zero. Entropy becomes absolute, not just relative. Walther Nernst introduced the law in 1906 in the weaker "heat theorem" form — that entropy differences between configurations vanish at absolute zero — motivated by low-temperature measurements of chemical affinities. Max Planck strengthened it in 1911: not just the differences but the entropy of each pure substance itself goes to zero. The modern reading is unavoidably quantum. As $T \to 0$ a system settles into its ground state, and the leftover entropy counts only the ground-state degeneracy $g_0$ through $S(0) = k_{\mathrm B}\ln g_0$ — negligible per particle unless that degeneracy grows with system size. The law has sharp, testable teeth. Because $S(T) = S(0) + \int_0^T (C/T')\,\mathrm{d}T'$ must converge, every heat capacity in thermal equilibrium is forced to vanish as $T \to 0$ under the usual conditions of the Third Law — which outlaws the classical Dulong–Petit constant and the equipartition value, and demands the Debye $T^3$ law of insulators and the linear $\gamma T$ term of metals. Thermal expansion coefficients vanish too, by a Maxwell relation. Most strikingly, the classical Sackur–Tetrode entropy of an ideal gas diverges to $-\infty$ as $T \to 0$ — a flat violation that quantum statistics must repair, which is exactly why the Third Law is a signpost pointing at the necessity of quantum mechanics. Its most quoted operational corollary is unattainability: no finite sequence of cooling steps can reach absolute zero, because the entropy curves of all configurations merge there and each step buys less and less.

Key takeaways

The Third Law (Nernst–Planck) states that the entropy of a system approaches a universal constant — conventionally zero — as $T \to 0$ , independent of all other thermodynamic parameters for a perfect crystal with a non-degenerate ground state, for which $S(0) = 0$ .
Equivalently, the isothermal entropy change of any parameter vanishes at absolute zero: $\lim_{T\to0}(\partial S/\partial X)_T = 0$ .
A direct consequence is the unattainability of absolute zero: no finite sequence of reversible steps can cool a system to $T = 0$ .
Heat capacities, thermal-expansion coefficients, and many response functions must vanish as $T \to 0$ , curing the divergence of the classical Sackur–Tetrode entropy.
Statistically the law follows from a (nearly) non-degenerate ground state: $S(0) = k_{\mathrm B}\ln g_0$ , negligible per particle unless $\ln g_0$ scales with $N$ .
Key idea: as $T \to 0$ a system collapses into its quantum ground state, so entropy approaches the parameter-independent value $S(0) = k_{\mathrm B}\ln g_0$ (zero for a non-degenerate ground state); this fixes the constant the Second Law left free and forces every heat capacity to vanish at absolute zero.

The core insight

As $T \to 0$ a system collapses into its quantum ground state, so the entropy approaches the parameter-independent value $S(0) = k_{\mathrm B}\ln g_0$ , taken as zero for a non-degenerate ground state. This fixes the additive constant the Second Law left undetermined and forces all heat capacities and entropy-derivative response functions to vanish at absolute zero.

Where this lives today

The Third Law is the working foundation of low-temperature physics and of the entire enterprise of tabulating absolute entropies. Chemists obtain standard molar entropies by integrating measured heat capacities upward from a few kelvin — $S(T) = \int_0^T (C_p/T')\,\mathrm{d}T' + \sum \Delta H_{\text{trans}}/T_{\text{trans}}$ — using $S(0) = 0$ as the lower anchor. Every entropy value in a thermodynamic data table rests on the law being true. When the calorimetric entropy disagrees with the spectroscopic (statistical) value, the gap directly measures frozen-in disorder: $R\ln 2$ for CO, $R\ln(3/2)$ for water ice — an experimental readout of ground-state degeneracy. At the cutting edge, unattainability sets the rules for the race to absolute zero. Adiabatic demagnetization of paramagnetic salts reaches the millikelvin range; nuclear demagnetization pushes to microkelvin; laser cooling and evaporative cooling of dilute atomic gases reach nanokelvin, where Bose–Einstein condensates and degenerate Fermi gases are made. Every technique runs into the same wall the Third Law describes — the entropy curves merging, each cooling step buying less — and the art is in stacking stages to claw closer to a floor that can never be touched. The law that Nernst extracted from calorimetry a century ago still bounds the coldest experiments on Earth.

The physical picture

Read entropy as the logarithm of the number of accessible microstates. As you drain energy from a system, fewer and fewer states are within reach; at $T = 0$ only the ground manifold survives. That manifold is fixed by quantum mechanics and — for a clean system — does not grow with macroscopic knobs like volume or magnetic field. So all the entropy curves $S(T; X)$ , one for each value of a parameter $X$ , approach the same limiting value as $T \to 0$ . That convergence is the whole law in one image.

Figure

All entropy curves approach the same limiting value S₀ as T→0, and the classical gas entropy diverges away

Now picture cooling as a repeated maneuver: bring the system to a low-entropy configuration, isolate it, let it relax. Each cycle pumps out some entropy. But because every configuration's $S(T)$ curve approaches the same value at $T = 0$ , the entropy you can remove per cycle shrinks toward zero as you near the bottom. You can get arbitrarily close, but never arrive. That is unattainability made visual — and it is why reaching microkelvin temperatures takes ever more elaborate machinery.

Figure

Cooling by adiabatic steps: each step's temperature drop shrinks as the curves merge

The law is fundamentally quantum, and its sharpest tell is the failure of the classical gas. The Sackur–Tetrode entropy behaves like $\ln(T^{3/2}V)$ , which plunges to $-\infty$ as $T \to 0$ and depends on volume — both flatly forbidden by Nernst. Only the discreteness and gap structure of quantum energy levels deliver a finite, parameter-independent ground-state entropy. The Third Law is where thermodynamic consistency demands quantum mechanics.

The core equations

The law's primary statement is the Nernst–Planck limit: entropy approaches a single constant, independent of every other state variable, as temperature vanishes:

\lim_{T\to0} S(T, X) = S_0 \quad (\text{independent of } X) \qquad (1)

where: $S$ — entropy; $T$ — absolute temperature; $X$ — any other state variable (volume, pressure, field, composition); $S_0$ — the universal limiting entropy, conventionally $0$ . The strength of Planck's version is the word independent: the ceiling is the same no matter how the other knobs are set. An immediate corollary is that isothermal entropy changes with any parameter die at absolute zero:

\lim_{T\to0}\left(\frac{\partial S}{\partial X}\right)_T = 0 \qquad (2)

where: $(\partial S/\partial X)_T$ — the isothermal change of entropy with the parameter $X$ . If $S(0,X)$ is the same constant for every $X$ , its $X$ -derivative there must vanish. The equation with the sharpest experimental consequences relates entropy to heat capacity by integrating $\mathrm{d}S = \delta Q_{\text{rev}}/T = (C/T)\,\mathrm{d}T$ up from absolute zero:

S(T, X) = S_0 + \int_0^T \frac{C_X(T')}{T'}\,\mathrm{d}T' \qquad (3)

where: $C_X$ — heat capacity at fixed $X$ ; the lower limit is $T'=0$ . For this integral to converge — for $S(T)$ to be finite — the integrand $C_X/T'$ must be integrable at zero, which forces $C_X(T) \to 0$ as $T \to 0$ . A constant heat capacity would make the integral diverge logarithmically, contradicting a finite entropy. This single convergence requirement is the law's most powerful lever. The microscopic content is the statistical form, valid when the ground state has degeneracy $g_0$ :

S(0) = k_{\mathrm B}\ln g_0 \qquad (4)

where: $k_{\mathrm B} = 1.380649\times10^{-23}\ \mathrm{J/K}$ ; $g_0$ — the degeneracy of the quantum ground state. For a non-degenerate ground state $g_0 = 1$ and $S(0) = 0$ ; a modest degeneracy contributes a per-particle entropy $\to 0$ in the thermodynamic limit unless $\ln g_0$ itself scales with $N$ . Through a Maxwell relation the vanishing of entropy derivatives forces thermal-expansion coefficients to zero as well:

\left(\frac{\partial S}{\partial p}\right)_T = -\left(\frac{\partial V}{\partial T}\right)_p \ \Rightarrow\ \beta = \frac{1}{V}\left(\frac{\partial V}{\partial T}\right)_p \xrightarrow{T\to0} 0 \qquad (5)

where: $\beta$ — volumetric thermal-expansion coefficient; $p$ — pressure; $V$ — volume. Since $(\partial S/\partial p)_T \to 0$ by the Third Law, the Maxwell relation drags $(\partial V/\partial T)_p$ — and hence $\beta$ — to zero at absolute zero, as observed in every material. The classical failure that makes the law necessary is the Sackur–Tetrode entropy, which violates (1) outright:

S_{\text{ST}} = N k_{\mathrm B}\left[\ln\!\left(\frac{V}{N}\left(\frac{2\pi m k_{\mathrm B}T}{h^2}\right)^{3/2}\right) + \frac{5}{2}\right] \xrightarrow{T\to0} -\infty \qquad (6)

where: $m$ — particle mass; $h$ — Planck constant. As $T \to 0$ the logarithm plunges to $-\infty$ and the result depends on $V$ — both forbidden. Quantum statistics remove the classical low-temperature divergence and, for a non-degenerate ground state, yield an entropy that approaches zero, which is the Third Law rescuing thermodynamics from its own classical limit.

Derivation AVanishing heat capacity: forced by the convergence of ∫ C/T dT

This is the article's owned structural result — the Third Law's most quantitative consequence, turning a statement about entropy into a hard constraint on heat capacity, and exposing the classical ideal gas as a violator. The goal is to prove that $C(T) \to 0$ as $T \to 0$ is forced by the requirement that entropy be finite at every temperature, then to show explicitly that a classical constant heat capacity ( $C =$ const, as in Dulong–Petit or the ideal gas $C_V = \frac{3}{2}Nk_{\mathrm B}$ ) makes the entropy diverge to $-\infty$ , while a Debye solid ( $C \propto T^3$ ) keeps it finite. The strategy is to examine the convergence of $\int_0^T (C/T')\,\mathrm{d}T'$ at its lower limit.

Step 1. Write the entropy as the integral $S(T) = S_0 + \int_0^T (C(T')/T')\,\mathrm{d}T'$ . For $S(T)$ to be finite for every $T > 0$ , the integrand $C(T')/T'$ must be integrable near $T' = 0$ . Why: a non-integrable singularity at the lower limit would make $S(T)$ infinite, which is unphysical — entropy is a finite thermodynamic quantity.

Step 2. Suppose $C \to C_0 \ne 0$ (a nonzero constant) as $T' \to 0$ . Then near zero the integrand behaves like $C_0/T'$ , and

\int_0^T \frac{C_0}{T'}\,\mathrm{d}T' = C_0\big[\ln T'\big]_0^T = C_0(\ln T - \ln 0^+) \to +\infty

so $S(T) = S_0 + \infty$ diverges. Why: $1/T'$ is exactly the borderline non-integrable singularity; a constant heat capacity therefore contradicts a finite entropy — the classical constant $C$ is ruled out.

Step 3. Apply it to the classical ideal gas, $C_V = \frac{3}{2}Nk_{\mathrm B} =$ const. The entropy integral gives $\frac{3}{2}Nk_{\mathrm B}\ln T +$ const $\to -\infty$ as $T \to 0$ — matching the Sackur–Tetrode divergence. The classical gas violates the Third Law. Why: this is the same $-\infty$ seen in Eq. (6), now traced directly to a heat capacity that fails to vanish; the cure is quantum statistics, where $C_V \to 0$ .

Step 4. Contrast a Debye insulator, $C = a T^3$ . Then $\int_0^T (aT'^3/T')\,\mathrm{d}T' = \int_0^T aT'^2\,\mathrm{d}T' = \frac{a}{3}T^3$ , which is finite and vanishes as $T \to 0$ . Likewise the metallic $C = \gamma T$ gives $\int_0^T \gamma\,\mathrm{d}T' = \gamma T$ , finite. Why: any $C \propto T^n$ with $n > 0$ converges, so the Third Law is compatible with — and in fact demands — the observed low- $T$ power laws; only $n = 0$ (a constant) fails.

Go deeper · Derivation B · 2 minUnattainability of absolute zero from the merging of entropy curves

Derivation BUnattainability of absolute zero from the merging of entropy curves

The goal is to show that the Nernst statement — all entropy curves $S(T, X)$ converge to a common value as $T \to 0$ — implies that no finite sequence of isothermal + adiabatic steps can reach $T = 0$ . The strategy follows the standard cooling cycle used in adiabatic demagnetization: alternate an isothermal change of a parameter $X$ (say magnetic field, which dumps entropy to a bath) with an adiabatic (isentropic) change (which lowers the temperature), and track how far each cycle descends as the two entropy curves squeeze together near absolute zero.

Step 1. One cooling cycle. Start at $(T_i, X_1)$ . Isothermally change $X_1 \to X_2$ , dumping entropy to a bath and moving to $(T_i, X_2)$ with lower $S$ . Then adiabatically change $X_2 \to X_1$ at constant $S$ , which drops the temperature to some $T_f < T_i$ . Why: the isothermal leg lowers entropy at fixed temperature; the adiabatic leg trades that lower entropy for a lower temperature by sliding along a constant- $S$ line to the other curve.

Step 2. Entropy bookkeeping. The adiabatic step conserves entropy, so $S(T_f, X_1) = S(T_i, X_2)$ — it maps the lower curve's value at $T_i$ onto the upper curve at the new, lower $T_f$ . Why: isentropic means the entropy at the end of the adiabatic leg equals the entropy at its start.

Step 3. Invoke Nernst. If the two curves met at a finite separation at $T = 0$ and stayed apart at all $T > 0$ , a single adiabatic step could in principle land on $T = 0$ . The Third Law forbids exactly this by forcing $S(T, X_1) - S(T, X_2) \to 0$ as $T \to 0$ : the curves merge. So the temperature drop achievable per cycle shrinks as $T$ decreases. Why: as the curves squeeze together, the horizontal (constant- $S$ ) jump between them gets shorter, so each adiabatic step buys less cooling.

Step 4. Infinite regress. The decrement satisfies $\Delta T_n \to 0$ as $T_n \to 0$ , so an infinite number of cycles would be required to reach exactly $T = 0$ . Why: a sum of ever-shrinking positive steps converges to a positive limiting temperature in any finite number of cycles — absolute zero is approached but never attained.

Limiting cases

The Third Law constrains the low-temperature asymptotics of every response function:

Case	Result	Meaning
Debye insulator	$C_V = \frac{12\pi^4}{5}Nk_{\mathrm B}(T/\Theta_D)^3$	Phonon heat capacity $\propto T^3$ vanishes at $T=0$ , so $\int_0^T C_V/T'\,\mathrm{d}T'$ converges — consistent with finite $S(0)$ .
Free-electron metal	$C_{\text{el}} = \gamma T,\ \gamma = \frac{\pi^2}{3}g(\varepsilon_F)k_{\mathrm B}^2$	Electronic heat capacity is linear in $T$ , vanishing at $T=0$ ; measuring $C/T$ vs $T^2$ separates it from the phonon $T^3$ term.
Thermal expansion	$\beta = -\frac{1}{V}(\partial S/\partial p)_T \xrightarrow{T\to0} 0$	A Maxwell relation plus vanishing $(\partial S/\partial p)_T$ forces the expansion coefficient to zero.
Classical ideal gas	$S_{\text{ST}} \sim Nk_{\mathrm B}\ln(VT^{3/2}) \to -\infty$	Constant $C_V = \frac{3}{2}Nk_{\mathrm B}$ makes $S$ diverge — a genuine violation, cured only by quantum statistics.
Residual entropy (apparent violation)	$S(0) = k_{\mathrm B}\ln g_0$ ; ice $\approx R\ln(3/2)$	Pauling's proton-disorder estimate for water ice; a frozen non-equilibrium state, not a true equilibrium ground-state degeneracy scaling with $N$ .

Common trap“The Third Law says entropy is zero at low temperature”

Not quite — and the distinction matters. The Third Law says the limit of entropy as $T \to 0$ is a single universal constant for all parameter values, conventionally set to zero for a non-degenerate ground state. It does not say $S = 0$ at some low but finite temperature, nor does it forbid a nonzero $S(0)$ . Real systems can carry residual (frozen-in) entropy — CO crystals keep $\approx R\ln 2$ from random C–O orientation, water ice keeps $\approx R\ln(3/2)$ from proton disorder, spin glasses freeze among near-degenerate states. These are kinetic failures to reach true equilibrium, not violations of the law. And beware the opposite trap: the classical ideal gas genuinely violates the Third Law ( $S \to -\infty$ ) — that is not a residual-entropy subtlety, it signals the breakdown of the classical ideal-gas description and the need for a quantum treatment at sufficiently low temperatures.

Worked examples

1 · Absolute entropy of a Debye solid from its heat capacity (confidence builder)

A crystalline solid obeys the Debye law $C_V = a T^3$ with $a = 9.0\times10^{-4}\ \mathrm{J\,mol^{-1}K^{-4}}$ at low temperature. Using the Third Law ( $S_0 = 0$ ) find its molar entropy at $T = 10\ \mathrm{K}$ , and verify the integral converges at $T = 0$ .

Step 1. Set up the entropy integral with $S_0 = 0$ : $S(T) = \int_0^T (C_V/T')\,\mathrm{d}T' = \int_0^T aT'^2\,\mathrm{d}T'$ .

Step 2. Integrate: $S(T) = \tfrac{a}{3}T^3$ . The lower limit contributes zero — the integral converges at $T = 0$ , exactly as the Third Law requires.

Step 3. Insert numbers at $T = 10\ \mathrm{K}$ : $S(10) = \tfrac{1}{3}(9.0\times10^{-4})(10)^3 = \tfrac{1}{3}(9.0\times10^{-4})(1000) = \mathbf{0.30\ J\,mol^{-1}K^{-1}}$ .

Step 4. Note the structure: $S(T) = \tfrac{1}{3}C_V(T)$ for a Debye solid, so at $10\ \mathrm{K}$ the entropy is one third of the heat capacity there ( $C_V(10) = a(10)^3 = 0.90\ \mathrm{J\,mol^{-1}K^{-1}}$ , and $0.90/3 = 0.30$ ✓). A finite, absolute entropy — anchored at zero by the Third Law — is exactly what the Debye $T^3$ law makes possible.

2 · Why the classical gas fails, quantitatively (exam level)

Show explicitly that a classical monatomic ideal gas description breaks down at low temperature by computing the entropy change from cooling $1\ \mathrm{mol}$ from $T_1 = 300\ \mathrm{K}$ toward $T \to 0$ at constant volume, and compare with a Debye solid over the same drop.

Step 1. Classical gas at constant $V$ : $C_V = \tfrac{3}{2}R = \tfrac{3}{2}(8.314) = 12.47\ \mathrm{J\,mol^{-1}K^{-1}}$ , a constant. Entropy change on cooling from $T_1$ to $T$ : $\Delta S = \int_{T_1}^{T} (C_V/T')\,\mathrm{d}T' = \tfrac{3}{2}R\ln(T/T_1)$ .

Step 2. Take the limit $T \to 0$ : $\Delta S = \tfrac{3}{2}R\ln(T/300) \to \mathbf{-\infty}$ . The gas would have to shed infinite entropy to reach absolute zero — impossible, and a clear signal of the breakdown of the classical ideal-gas approximation in the face of the finite, parameter-independent $S(0)$ the Third Law demands.

Step 3. Now the Debye solid over a finite drop, $300 \to 4\ \mathrm{K}$ , with $C_V = aT^3$ : $\Delta S = \tfrac{a}{3}(4^3 - 300^3)$ . This is finite, and as the lower temperature $\to 0$ the entropy tends smoothly to its (finite) $T=0$ value. No divergence.

Step 4. Diagnose the difference. The gas's $C_V$ stays at $12.47\ \mathrm{J\,mol^{-1}K^{-1}}$ all the way down — it never vanishes — so $\int C_V/T'$ picks up the fatal $\ln T \to -\infty$ . The solid's $C_V = aT^3 \to 0$ , killing the singularity. Under the usual regularity conditions, the Third Law implies $C \to 0$ as $T \to 0$ ; the classical gas fails precisely because equipartition keeps $C_V$ constant, and only quantum freezing-out of degrees of freedom drives $C_V \to 0$ .

Origins · 1 minWalther Nernst and how this was discovered

Walther Nernst

1864–1941 · Berlin

Nernst formulated the heat theorem — what became the Third Law — in 1906, deducing from painstaking low-temperature calorimetry that the entropy change of chemical reactions tends to zero as $T \to 0$ . His motivation was intensely practical: he wanted to predict chemical equilibria from thermal measurements alone, and the vanishing of $\Delta S$ at absolute zero let him do it. The work won him the 1920 Nobel Prize in Chemistry. Max Planck sharpened the theorem in 1911 into the stronger statement that the entropy of each pure substance itself tends to zero. Nernst also grasped the operational corollary that made the law famous: absolute zero is unattainable, a claim later refined into the statement that no finite sequence of processes can reach $T = 0$ . A gifted and combative experimentalist, he was among the last great figures who bridged physical chemistry and the new quantum physics that his own law demanded.

History

1906

— Nernst states his heat theorem: entropy differences of reactions vanish as T → 0.

1911

— Planck strengthens it: the entropy of each pure substance itself tends to zero.

1912

— Nernst argues the unattainability of absolute zero as an operational consequence.

1920

— Nernst is awarded the Nobel Prize in Chemistry for the heat theorem.

1935

— Giauque and Pauling quantify residual entropy in CO and water ice, confirming S(0) = k ln g₀.

Going further

References. Pathria & Beale, Statistical Mechanics, 3rd ed., §1.6 and §3.x · Kardar, Statistical Physics of Particles, Ch. 1 · Reif, Fundamentals of Statistical and Thermal Physics, §3.10–3.11 · Sethna, Statistical Mechanics: Entropy, Order Parameters, and Complexity, Ch. 5 · Landau & Lifshitz, Statistical Physics Part 1, §23. Key values used throughout: the Maxwell relation $(\partial S/\partial p)_T = -(\partial V/\partial T)_p$ follows from $\mathrm{d}G = -S\,\mathrm{d}T + V\,\mathrm{d}p$ ; $\int_0^T (C/T')\,\mathrm{d}T'$ diverges for nonzero constant $C$ and converges for $C \propto T^3$ and $C \propto T$ ; the Sackur–Tetrode entropy diverges as $T\to0$ ; the Debye entropy satisfies $S = \tfrac{1}{3}C_V$ , giving $0.30\ \mathrm{J\,mol^{-1}K^{-1}}$ at $10\ \mathrm{K}$ ; Pauling's ice residual entropy is $R\ln(3/2)$ ; $k_{\mathrm B} = 1.380649\times10^{-23}\ \mathrm{J/K}$ and $R = 8.314\ \mathrm{J\,mol^{-1}K^{-1}}$ (CODATA).

Builds on: Second Law of Thermodynamics and Entropy · Temperature, Pressure, and Chemical Potential as Entropy Derivatives · Maxwell Relations · Carnot Cycle and Thermodynamic Temperature. Leads to: Thermodynamic Potentials and the Fundamental Relation · Helmholtz Free Energy, Enthalpy, and Gibbs Free Energy.

Is the Third Law an independent law, or does it follow from the first two?

It is genuinely independent — the first two laws cannot produce it. The First Law is energy conservation; the Second Law defines entropy up to an additive constant through $\mathrm{d}S = \delta Q_{\text{rev}}/T$ . Nothing in either fixes that constant, and nothing forces heat capacities to vanish at low temperature. The Third Law supplies exactly the missing piece: it pins the constant (making entropy absolute) and, through the convergence of $\int_0^T C/T'\,\mathrm{d}T'$ , forces $C \to 0$ . Its deepest justification is not thermodynamic at all but quantum-statistical: a system with a discrete spectrum and a (nearly) unique ground state has $S(0) = k_{\mathrm B}\ln g_0 \approx 0$ . So the Third Law is where classical thermodynamics reaches its limit and quantum mechanics becomes indispensable.

If we can get arbitrarily close to absolute zero, why does it matter that we can't reach it exactly?

Practically, "arbitrarily close" is all any experiment needs — nanokelvin temperatures are routine in cold-atom labs. But the impossibility of reaching exactly zero is a real physical statement with teeth, not a technicality. It says the cooling power of every method fades as you descend, because the entropy curves of all configurations merge at $T = 0$ , so each cooling step removes less entropy and buys less temperature drop. That is why reaching each new decade of coldness requires a fundamentally new technique (dilution refrigeration, then nuclear demagnetization, then laser plus evaporative cooling) rather than just running the old one longer. Unattainability is the reason the approach to zero is a staircase of ever-harder steps, not a smooth glide.

What exactly is "residual entropy," and why isn't it a violation of the law?

Residual entropy is entropy a system keeps as $T \to 0$ because it gets kinetically stuck before reaching true equilibrium — not because its genuine ground state is degenerate. Water ice is the classic case: each oxygen has two nearby and two distant hydrogens, and the "ice rules" leave many nearly-equal-energy proton arrangements. As the ice cools, the protons freeze into one random arrangement long before they could find the unique lowest-energy ordering, trapping $\approx R\ln(3/2)$ of configurational entropy (Pauling's estimate). This is a failure to equilibrate, governed by kinetics, not a property of the true thermodynamic ground state. The Third Law concerns the equilibrium limit; residual entropy is the fingerprint of falling out of equilibrium, and comparing calorimetric with spectroscopic entropy is how we measure it.

Why does a nonzero heat capacity at T = 0 make the entropy infinite? I don't see the divergence.

It comes from the $1/T'$ weighting in the entropy integral. Entropy is built by $S(T) = S_0 + \int_0^T (C/T')\,\mathrm{d}T'$ , and near the lower limit that integrand is $C(T')/T'$ . If $C$ approaches a nonzero constant $C_0$ as $T' \to 0$ , the integrand behaves like $C_0/T'$ , whose integral is $C_0\ln T'$ — and $\ln T' \to -\infty$ as $T' \to 0^+$ . So $\int_0^T$ picks up a $-\ln 0^+ = +\infty$ at the bottom, and $S$ diverges. The function $1/T'$ sits exactly at the borderline of integrability: any positive power $C \propto T'^n$ with $n > 0$ tames it (the integrand becomes $T'^{n-1}$ , integrable), but $n = 0$ (constant $C$ ) fails. That is precisely why the Third Law is equivalent to $C \to 0$ .

Does the Third Law require quantum mechanics, or is it a purely thermodynamic statement?

Thermodynamically the Third Law can be stated without quantum mechanics — Nernst did, from calorimetry. But it cannot be explained classically, and worse, classical statistical mechanics actively violates it: the Sackur–Tetrode entropy of a classical ideal gas goes to $-\infty$ as $T \to 0$ and depends on volume. A finite, parameter-independent $S(0)$ requires that the system have a discrete energy spectrum with a well-defined ground state, so that as energy drains away only the ground manifold remains and $S \to k_{\mathrm B}\ln g_0$ . That discreteness is quantum mechanical. So the Third Law is a thermodynamic law that only quantum mechanics can satisfy — historically it was one of the strong hints, alongside blackbody radiation and specific heats, that classical physics was incomplete at low temperatures.

How do chemists get "absolute" entropies from this, and why can't they do it without the Third Law?

They measure the heat capacity $C_p(T)$ of a substance from as close to absolute zero as they can reach up to the temperature of interest, then integrate: $S(T) = S(0) + \int_0^T (C_p/T')\,\mathrm{d}T' + \sum \Delta H_{\text{trans}}/T_{\text{trans}}$ , adding a term at each phase transition. The catch is $S(0)$ : without a value for it, this only gives entropy differences, not an absolute number. The Third Law supplies $S(0) = 0$ for a pure, perfectly ordered crystal, which fixes the lower limit and makes the resulting $S(T)$ genuinely absolute. That is why tabulated standard molar entropies are called "absolute entropies" — they are meaningful on their own, not just as differences — and it is only the Third Law's $S(0) = 0$ that licenses calling them that.