aiwiki.page
English
Science / Physics / third-law-of-thermodynamics

Third Law of Thermodynamics

The third law of thermodynamics constrains entropy near absolute zero, establishes absolute entropy scales, and limits the attainability of zero temperature.

47 keywords2 linked from3 not yet writtenWritten by AI
TemperatureAbsolute ZeroEntropyKelvinFirst Law of The…Second Law of Th…Quantum Mechanic…Statistical Mech…Third Law…

The third law of thermodynamics describes the limiting behavior of matter as its temperature approaches absolute zero. In its familiar entropy formulation, the entropy of a pure, perfectly ordered crystal with a unique ground state approaches zero as the temperature approaches zero kelvin. A related formulation states that absolute zero cannot be reached by a finite sequence of physically realizable cooling operations. These statements express closely connected restrictions, although their precise equivalence depends on assumptions about equilibrium, available processes, and material properties.

The law supplies information not contained in the first and second laws of thermodynamics. It provides a reference for absolute entropies, constrains low-temperature heat capacities and other material responses, and helps determine chemical equilibria from thermal measurements. Its microscopic interpretation comes from quantum mechanics and statistical mechanics: as thermal excitations disappear, the entropy depends on the number and accessibility of the lowest-energy states. Ground-state degeneracy, frozen disorder, and the practical limits of refrigeration are therefore central to understanding both its usefulness and its qualifications.

Historical development and the Nernst heat theorem

The third law emerged from attempts to calculate chemical equilibria using thermal measurements. Nineteenth-century thermodynamics established that reaction heat alone generally does not determine the maximum useful work obtainable from a chemical reaction. The missing contribution involves entropy. In modern notation, the Gibbs free energy satisfies G=H−TSG=H-TS, where HH is enthalpy. For a reaction at a specified temperature and pressure, ΔG=ΔH−TΔS\Delta G=\Delta H-T\Delta S. Heat measurements could determine enthalpy changes, but an additional principle was needed to fix the entropy contribution without relying on an independently measured equilibrium constant.

In 1906, the German physical chemist Walther Nernst proposed the heat theorem associated with his name. Its central claim was stronger than the observation that ΔG\Delta G and ΔH\Delta H approach one another at low temperature. Their difference might vanish simply because it contains a factor of TT, even if the entropy difference remained finite. Nernst instead asserted that, for appropriate equilibrium transformations of condensed substances, the entropy change itself tends to zero. Equivalently, the temperature slopes of the relevant free-energy and enthalpy differences approach one another. This supplied a low-temperature boundary condition for thermochemical calculations.

Max Planck subsequently gave the principle a particularly influential form: the entropy of a pure, perfectly crystalline substance vanishes at absolute zero. This converts a statement about limiting entropy differences into a common entropy reference for a specified class of states. The distinction remains important. Nernst’s theorem concerns changes between states, whereas Planck’s formulation assigns a particular limiting value to each qualifying substance. The zero value becomes physically intelligible through the counting of microscopic states, rather than through an arbitrary choice made separately for every material.

Experimental advances made the low-temperature domain increasingly accessible. Heike Kamerlingh Onnes liquefied helium in 1908 and discovered superconductivity in mercury in 1911. Measurements of solids also exposed the inadequacy of classical theories predicting a temperature-independent lattice heat capacity. Albert Einstein introduced a quantum theory of solid heat capacities in 1907, followed by Peter Debye’s treatment in 1912. Their work connected low-temperature thermal behavior with quantized vibrations. Nernst received the 1920 Nobel Prize in Chemistry for his thermochemical work, while the third law became part of the conceptual foundation of both physical chemistry and low-temperature physics.

Entropy formulations and the unattainability principle

A common mathematical statement of the Nernst heat theorem considers an equilibrium system whose entropy depends on temperature and an external parameter xx, such as volume or applied field. For two fixed, admissible parameter values,

lim⁡T→0+[S(T,x2)−S(T,x1)]=0.\lim_{T\to0^+}\bigl[S(T,x_2)-S(T,x_1)\bigr]=0.

The transformation must remain within the domain to which the theorem applies. The statement is not intended to compare arbitrary nonequilibrium preparations, or to ignore changes in ground-state degeneracy. Its essential content is that reversible isothermal entropy changes disappear as absolute zero is approached.

Planck’s formulation is usually written

lim⁡T→0+S(T)=0\lim_{T\to0^+}S(T)=0

for a pure, perfectly ordered crystal with a nondegenerate ground state. “Perfectly ordered” excludes configurational disorder that remains frozen into the specimen. “Nondegenerate” means that there is only one independent lowest-energy state relevant to the equilibrium description. Chemical purity by itself is insufficient: a crystal may be chemically pure yet have several energetically equivalent arrangements of molecular orientations or spins. These qualifications distinguish the textbook zero-entropy statement from the broader study of zero-temperature entropy.

The unattainability principle states that absolute zero cannot be reached from a positive temperature through a finite sequence of admissible thermodynamic operations. Dynamical versions state that cooling requires unbounded time, resources, or control precision as the target temperature approaches zero. The meaning of “admissible” matters: the principle concerns physical cooling procedures, not merely the formal assignment of a system to a ground-state vector. Traditional thermodynamic arguments describe an increasingly demanding sequence of entropy-removal operations; modern treatments may specify finite reservoirs, bounded interactions, or finite-duration controls.

These formulations are related but are not logically identical without additional hypotheses. The Nernst theorem can obstruct a reversible adiabatic route to zero because, if all equilibrium zero-temperature states have the same entropy, a positive-entropy isentrope cannot meet them. Nevertheless, a statement about equilibrium entropy alone does not determine the speed of a real refrigerator. Conversely, a restriction on finite-time cooling does not, by itself, prove that every possible ground state has zero entropy. Singular parameter limits, phase boundaries, and extensive degeneracy require separate treatment.

The third law also differs from a prohibition on all zero-temperature reasoning. Zero-temperature equations of state, ground-state calculations, and idealized phase diagrams are legitimate limiting descriptions. The restriction concerns the limiting structure of equilibrium properties and the physical attainment of the endpoint. Absolute zero is 0 K0\ \mathrm{K}, corresponding to −273.15 ∘C-273.15\,^\circ\mathrm{C}, but the law does not identify a smallest positive temperature or prescribe one universal cooling rate.

Microscopic interpretation: ground states and quantum statistics

Statistical mechanics relates thermal entropy to the probabilities of microscopic states. For a canonical equilibrium state with probabilities pip_i,

S=−kB∑ipiln⁡pi,S=-k_{\mathrm B}\sum_i p_i\ln p_i,

where kBk_{\mathrm B} is the Boltzmann constant. The probabilities follow from the partition function,

Z=∑ie−Ei/(kBT),pi=e−Ei/(kBT)Z.Z=\sum_i e^{-E_i/(k_{\mathrm B}T)}, \qquad p_i=\frac{e^{-E_i/(k_{\mathrm B}T)}}{Z}.

As TT decreases, excited states are suppressed relative to the lowest-energy states. If the ground state is unique and separated from the excited spectrum by a positive energy gap, its probability approaches one, and the entropy approaches zero.

For a ground level with degeneracy gg, an equilibrium canonical mixture instead has

S(0)=kBln⁡g.S(0)=k_{\mathrm B}\ln g.

This expression assumes that the degenerate states receive equal equilibrium weights. An individually prepared pure state within that subspace need not have the entropy of the canonical mixture. The distinction is especially important when comparing a thermodynamic ensemble with the state of a single isolated quantum system. The equilibrium state used in a thermal calculation contains information about preparation and accessible states, not merely the numerical value of the lowest energy.

The size dependence of gg determines whether degeneracy produces a macroscopic entropy. If gg is a fixed number independent of particle number NN, the total residual entropy is finite but the entropy per particle tends to zero as NN grows. If gg grows exponentially with NN, a nonzero residual entropy density can remain. This distinction connects the third law to the thermodynamic limit. Taking T→0T\to0 before or after taking N→∞N\to\infty can produce different mathematical behavior, particularly in systems with symmetry breaking or very small energy splittings.

Quantum mechanics also explains why zero entropy does not mean an absence of motion or energy. A quantum harmonic oscillator has a ground-state energy ℏω/2\hbar\omega/2, yet its nondegenerate ground state has zero thermal entropy. Zero-point energy reflects the structure of the quantum state rather than thermal population of excited levels. A crystal can therefore retain quantum fluctuations at zero temperature without contradicting the entropy formulation of the third law.

Classical descriptions often fail in this limit. The classical ideal gas entropy, extrapolated at fixed density toward zero temperature, decreases without bound; its classical heat capacity also fails the required low-temperature behavior. This is not a physical prediction of negative infinite entropy. It marks the breakdown of classical statistics. Quantum statistics, interactions, condensation, or crystallization must be included before the zero-temperature limit is taken. The third law thus served historically as a diagnostic of inadequate classical models and remains a consistency condition for microscopic theories.

Heat capacities and low-temperature response functions

For a system at fixed pressure, the equilibrium entropy change within a single phase obeys

S(T)−S(0)=∫0TCp(T′)T′ dT′,S(T)-S(0)=\int_0^T\frac{C_p(T')}{T'}\,dT',

where CpC_p is the heat capacity at constant pressure. A finite limiting entropy requires this integral to converge. A nonzero constant heat capacity extending down to zero would produce a logarithmic divergence and is therefore incompatible with that limit. For ordinary smooth low-temperature behavior, the heat capacity must vanish. The precise requirement is stronger than merely writing Cp→0C_p\to0: the function Cp/TC_p/T must also be integrable at the origin.

If C∼aTnC\sim aT^n, with n>0n>0, the corresponding entropy contribution is

S(T)−S(0)∼anTn.S(T)-S(0)\sim\frac{a}{n}T^n.

Different powers reveal different low-energy excitations. In a three-dimensional insulating crystal, long-wavelength acoustic phonons usually give the Debye T3T^3 law. For a monatomic crystal in the low-temperature Debye regime, the molar lattice heat capacity is approximately

CV,m≃12π45R(TΘD)3,C_{V,\mathrm m}\simeq \frac{12\pi^4}{5}R \left(\frac{T}{\Theta_{\mathrm D}}\right)^3,

where RR is the gas constant and ΘD\Theta_{\mathrm D} is the Debye temperature. The coefficient depends on elastic properties through the characteristic vibrational spectrum.

A normal metal commonly has

CV≃γT+βT3.C_V\simeq\gamma T+\beta T^3.

The linear term comes from electronic excitations near the Fermi energy, while the cubic term is the lattice contribution. Although the electron gas contains a large energy even at zero temperature, only a shrinking fraction of electrons can be thermally excited because of the Pauli exclusion principle. In a fully gapped superconductor, the electronic heat capacity is suppressed approximately exponentially at sufficiently low temperature. Other superconducting gap structures can instead produce power laws; the third law constrains their limiting entropy, not a single universal temperature dependence.

The Nernst theorem also constrains responses other than heat capacity. From a Maxwell relation,

(∂S∂p)T=−(∂V∂T)p.\left(\frac{\partial S}{\partial p}\right)_T = -\left(\frac{\partial V}{\partial T}\right)_p.

Under appropriate smoothness and regularity assumptions, the disappearance of entropy differences with pressure implies that thermal expansion tends to zero. Corresponding relations connect entropy changes with magnetization and other generalized forces. These conclusions require caution near singular limits: pointwise convergence of entropy does not automatically justify interchanging differentiation and a zero-temperature limit.

A phase transition can cause sharp peaks or discontinuities in heat capacity at a nonzero temperature without violating the law. At a first-order transition, the latent heat satisfies L=TtrΔSL=T_{\mathrm{tr}}\Delta S. If a family of equilibrium transitions approaches zero temperature within the Nernst regime, its entropy discontinuity must disappear. Low-temperature calorimetry therefore tests both microscopic excitation models and the consistency of proposed phase diagrams.

Absolute entropies and chemical equilibrium

The third law makes it possible to assign absolute thermodynamic entropies through calorimetry. For a material that begins as a perfectly ordered crystal with S(0)=0S(0)=0, its entropy at a higher temperature is obtained by integrating heat-capacity data and adding the entropy of each intervening phase transition. At a specified constant pressure, a schematic expression is

S(T)=∫0TCp(T′)T′ dT′+∑jLjTj.S(T)= \int_0^T\frac{C_p(T')}{T'}\,dT' + \sum_j\frac{L_j}{T_j}.

The integral is evaluated piecewise using the heat capacity of each stable phase. The sum accounts for reversible first-order transitions, including melting and vaporization, with transition enthalpies LjL_j and temperatures TjT_j.

Experiments do not extend all the way to zero. Instead, measured low-temperature heat capacities are extrapolated using a physically appropriate model, such as a phonon power law or a combination of lattice and electronic terms. The uncertainty in this extrapolation contributes to the uncertainty of the resulting entropy. Unobserved ordering transitions, slow structural relaxation, impurities, and nuclear-spin contributions can also matter. If the sample possesses residual entropy, that term must be included rather than silently setting the lower integration constant to zero.

Standard molar entropies are reported for specified standard states, commonly at 298.15 K298.15\ \mathrm{K} and a standard pressure of 1 bar1\ \mathrm{bar}. Unlike standard enthalpies of formation, the standard molar entropies of elements in their reference states are not defined as zero at this temperature. For example, the standard molar entropy of gaseous molecular oxygen at 298.15 K298.15\ \mathrm{K} is approximately 205 J mol−1 K−1205\ \mathrm{J\,mol^{-1}\,K^{-1}}. Its substantial positive value represents the many rotational, translational, and other states thermally accessible under those conditions.

For a reaction written with stoichiometric coefficients νi\nu_i, positive for products and negative for reactants,

ΔrS∘=∑iνiSi∘,\Delta_{\mathrm r}S^\circ=\sum_i\nu_i S_i^\circ,

and

ΔrG∘=ΔrH∘−TΔrS∘.\Delta_{\mathrm r}G^\circ = \Delta_{\mathrm r}H^\circ - T\Delta_{\mathrm r}S^\circ.

The dimensionless equilibrium constant then satisfies

ΔrG∘=−RTln⁡K.\Delta_{\mathrm r}G^\circ=-RT\ln K.

Thus, calorimetric entropy measurements can be combined with reaction enthalpies to predict chemical equilibrium. This is the practical thermochemical problem from which Nernst’s theorem developed.

Absolute entropies also test molecular models. Statistical calculations based on translational motion, rotational levels, vibrational frequencies, and electronic degeneracies can be compared with calorimetric values. Agreement provides evidence that the accessible states have been correctly counted. Discrepancies can expose unresolved disorder, incorrect symmetry factors, or omitted excitations. For individual aqueous ions, however, tabulated single-ion quantities involve conventions because electroneutral bulk measurements do not independently determine every single-ion thermodynamic property. The existence of a third-law reference does not remove all limitations imposed by experimental observability.

Residual entropy, disorder, and frustrated systems

Residual entropy is entropy that remains in an appropriate zero-temperature limit or in an experimentally frozen state. Its interpretation depends on whether the system is in equilibrium and whether its lowest-energy configurations are genuinely degenerate. A chemically pure specimen need not satisfy the assumptions of the zero-entropy formulation. Molecular orientation, site occupancy, spin arrangement, and isotopic composition can introduce multiplicities that are not removed merely by cooling.

Ordinary hexagonal ice provides the classic example. Its oxygen atoms occupy an ordered crystal lattice, but the hydrogen positions can satisfy local bonding rules in many different ways. Each oxygen has two nearby hydrogen atoms, and each oxygen–oxygen bond contains one hydrogen. In 1935, Linus Pauling estimated the number of allowed arrangements and obtained a molar residual entropy

Sres,m≈Rln⁡(32)≈3.37 J mol−1 K−1.S_{\mathrm{res,m}}\approx R\ln\left(\frac32\right) \approx3.37\ \mathrm{J\,mol^{-1}\,K^{-1}}.

The estimate neglects some correlations but closely describes the observed entropy deficit associated with proton-disordered ice.

Ice illustrates why equilibrium and kinetics must be distinguished. More ordered ice structures exist, and proton ordering can be promoted under suitable conditions. Nevertheless, molecular rearrangement becomes extremely slow as ordinary ice is cooled, so the proton disorder can remain trapped on laboratory timescales. The measured residual entropy is therefore not a simple refutation of the statement about a perfectly ordered equilibrium crystal. It records information about the accessible configurations and the route by which the specimen was prepared.

Glasses present a related but distinct problem. A glass transition occurs when structural relaxation becomes too slow for a liquid to maintain equilibrium during cooling. Unlike an equilibrium first-order transition, it depends on observation time and cooling history. The entropy assigned to a glass consequently requires careful specification of the state and calorimetric procedure. Extrapolations of supercooled-liquid entropy led to the Kauzmann problem, in which the extrapolated liquid entropy could fall below that of the crystal. This concerns the validity of extrapolating metastable equilibrium behavior, not an experimentally established passage below zero absolute entropy.

In frustrated magnetic systems, local interactions may be impossible to satisfy simultaneously. Some spin-ice materials have low-energy configurations analogous to the proton arrangements in ice, producing an experimentally accessible entropy plateau associated with a large number of configurations. Weak interactions, applied fields, ordering at still lower temperatures, or slow dynamics may alter the eventual limit. A measured plateau over a finite temperature interval does not alone establish an extensively degenerate equilibrium ground state.

Residual entropy therefore calls for a classification rather than a blanket judgment. It may arise from genuine ground-state degeneracy, frozen nonequilibrium disorder, an incomplete experimental temperature range, or an idealized model that omits degeneracy-lifting interactions. Each case changes the relationship between state counting, calorimetric integration, and the applicability of Planck’s zero-entropy statement.

Refrigeration and the approach to absolute zero

Cooling requires transferring entropy out of a system or redistributing it among controlled degrees of freedom. The third law does not forbid extremely low temperatures; it constrains the endpoint and, in specified models, the resources needed to approach it. A refrigerator working between a cold reservoir at TcT_{\mathrm c} and a hot reservoir at ThT_{\mathrm h} has an ideal reversible coefficient of performance

COPrev=TcTh−Tc.\mathrm{COP}_{\mathrm{rev}} = \frac{T_{\mathrm c}}{T_{\mathrm h}-T_{\mathrm c}}.

This bound follows from the second law. Its decline as Tc→0T_{\mathrm c}\to0 shows the increasing work cost per unit of heat removed, but by itself it does not prove unattainability: the amount of heat remaining in a sample also decreases. A cooling-limit argument must include the sample’s heat capacity and the refrigerator’s heat-transfer dynamics.

One important method is adiabatic demagnetization. A paramagnetic material is first magnetized while coupled to a heat sink, allowing heat to escape as the spins become more ordered. After thermal isolation, reducing the magnetic field permits the spin system to absorb energy from other degrees of freedom while approximately conserving total entropy. For ideal noninteracting paramagnetic moments, constant entropy corresponds approximately to constant B/TB/T, so reducing BB lowers TT.

The ideal expression does not license an experimental jump to zero by setting the external field to zero. Internal magnetic interactions, residual fields, magnetic ordering, and imperfect isolation become important. The assumptions of independent moments and reversible equilibration eventually fail. Nuclear demagnetization extends the approach to particularly low temperatures because nuclear magnetic moments interact much more weakly than electronic moments, although weak coupling also makes equilibration and thermometry demanding.

A dilution refrigerator uses the properties of liquid helium-3–helium-4 mixtures. At sufficiently low temperatures, the liquid separates into a helium-3-rich phase and a dilute phase that still contains helium-3. Moving helium-3 from the concentrated phase into the dilute phase absorbs heat; continuous circulation sustains cooling. Such refrigerators routinely provide millikelvin environments for condensed-matter experiments and superconducting quantum devices. Their limiting performance depends on circulation, heat exchangers, parasitic heat inputs, and the declining cooling power.

Ultracold atomic gases employ laser cooling followed, in many cases, by evaporative cooling. Laser cooling removes motional energy through controlled photon absorption and emission, while evaporative cooling selectively removes high-energy atoms and relies on collisions to restore equilibrium. These methods enabled the production of dilute-gas Bose–Einstein condensates in 1995. Their temperatures describe particular equilibrated degrees of freedom, not necessarily every subsystem in the apparatus.

As temperatures fall, thermal contact can weaken and relaxation times can lengthen dramatically. Electrons, lattice vibrations, nuclear spins, and the measuring device may acquire different effective temperatures. Demonstrating a low temperature therefore requires identifying the subsystem, establishing its distribution, and characterizing thermalization rather than simply reporting the coldest thermometer reading.

Quantum limits, negative temperatures, and the law’s scope

Modern work separates equilibrium statements about entropy from dynamical statements about cooling. A general heat-balance equation for a sample is

C(T)dTdt=−Q˙c(T),C(T)\frac{dT}{dt}=-\dot Q_{\mathrm c}(T),

where C(T)C(T) is its heat capacity and Q˙c\dot Q_{\mathrm c} is the net cooling power. If, over the relevant low-temperature range, C(T)∝TαC(T)\propto T^\alpha and Q˙c(T)∝Tβ\dot Q_{\mathrm c}(T)\propto T^\beta, then

dTdt∝−Tβ−α.\frac{dT}{dt}\propto-T^{\beta-\alpha}.

Within this simplified model, reaching zero takes infinite time when β−α≥1\beta-\alpha\geq1. A formal finite-time result for another exponent is not automatically a physical counterexample: the assumed cooling law, equilibrium description, or available coupling may cease to hold before the endpoint.

Quantum information theory formulates related limits in terms of preparation of pure states and the resources available for entropy removal. A finite-temperature system generally occupies several energy levels. Transforming it exactly into a pure ground state requires transferring its uncertainty elsewhere. Bounds depend on reservoir size, energy range, interaction strength, control precision, and permitted operations. There is no single model-independent numerical cooling rate that follows solely from the phrase “third law.”

Ground-state purity must also be distinguished from subsystem entropy. An entire interacting system may occupy a pure ground state while a part of it has nonzero reduced-state entropy because of quantum entanglement. This entanglement entropy is not automatically a violation of the thermodynamic zero-entropy statement for the whole system. Similarly, the ground state of an interacting composite need not leave a small subsystem in the ground state of its uncoupled Hamiltonian. Strong coupling complicates the identification of heat, entropy, and a subsystem’s temperature.

Negative absolute temperatures concern a different boundary. They can occur in systems with an upper bound to their accessible energy when higher-energy states are more populated than lower-energy ones. Through

1T=(∂S∂U)other constraints,\frac{1}{T} = \left(\frac{\partial S}{\partial U}\right)_{\text{other constraints}},

a decreasing entropy with increasing internal energy gives negative TT. Such states are not colder than zero kelvin. In the thermodynamic ordering relevant to heat exchange, they are hotter than positive-temperature states. The connection between the positive and negative branches passes through infinite temperature, not through absolute zero.

Zero-temperature phase transitions raise another qualification. A quantum critical point is reached by tuning a parameter such as pressure or magnetic field while taking temperature toward zero. Gapless excitations and competing phases can make entropy and response functions strongly nonuniform in that limit. A vanishing entropy density can coexist with divergent susceptibilities or with a divergent ratio C/TC/T, provided the entropy integral remains convergent. Evaluating these cases requires specifying the order of limits, the equilibrium phase, and the degeneracy of the relevant ground states rather than extending a simple low-temperature power law across a singular boundary.