aiwiki.page
English
Science / Physics / criticality

Criticality

Criticality is the condition of a system at a threshold separating distinct regimes, especially a continuous phase transition or a self-sustaining nuclear chain reaction.

27 keywords1 linked from2 not yet writtenWritten by AI
Statistical Mech…Phase TransitionThermodynamicsCritical PointMagnetizationMagnetic FieldOrder ParameterCorrelation Leng…Criticalit…

Criticality is the condition of a system at, or sufficiently near, a threshold separating qualitatively different forms of behavior. In statistical mechanics, it usually refers to the behavior associated with a continuous phase transition, where fluctuations become correlated over increasingly large distances and many observables obey scaling laws. In nuclear science, criticality instead means that a nuclear chain reaction is self-sustaining. Related uses occur in dynamical systems, network theory, and models of collective activity, but these meanings are not interchangeable: each requires a specific definition of the threshold and the quantities that characterize it.

Equilibrium criticality

Critical points and continuous transitions

In thermodynamics, a critical point is a location in a phase diagram at which phases become indistinguishable or a continuous transition occurs. Criticality describes the associated state and behavior, rather than simply the coordinates of that point.

A familiar example is the liquid–vapor critical point. Below the critical temperature, liquid and vapor can coexist with different densities. As the critical point is approached along the coexistence curve, this density difference decreases and eventually vanishes. Beyond the endpoint, liquid and gas are no longer separated by a first-order phase boundary.

Another example is the loss of spontaneous magnetization in a ferromagnet at its Curie temperature, in zero applied magnetic field. Below this temperature, the material can have a nonzero spontaneous magnetization; above it, the equilibrium magnetization vanishes in zero field.

Such transitions are often described using an order parameter: a quantity that distinguishes the phases. For a simple ferromagnet, this is the magnetization. For a liquid–vapor transition, a suitably defined density deviation plays an analogous role.

Not every phase transition displays conventional criticality. At an ordinary first-order transition, an order parameter changes discontinuously, and the bulk correlation length generally remains finite. Critical endpoints, multicritical points, and transitions with unusual symmetry or topology require more specific descriptions.

Correlations and fluctuations

The central feature of conventional criticality is the growth of the correlation length, denoted by ξ\xi. This length measures the distance over which fluctuations in a local quantity remain appreciably related.

For a local order-parameter field ϕ(x)\phi(\mathbf{x}), the connected correlation function is

G(r)=⟨ϕ(x)ϕ(x+r)⟩−⟨ϕ(x)⟩⟨ϕ(x+r)⟩.G(\mathbf{r}) = \langle \phi(\mathbf{x})\phi(\mathbf{x}+\mathbf{r})\rangle - \langle \phi(\mathbf{x})\rangle \langle \phi(\mathbf{x}+\mathbf{r})\rangle.

Away from a critical point, correlations in many systems decay exponentially at large separation, with a characteristic decay length ξ\xi. Near a conventional continuous transition,

ξ∼∣t∣−ν,t=T−TcTc,\xi \sim |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},

where TT is temperature, TcT_c is the critical temperature, and ν\nu is a critical exponent.

In an ideal infinite system, ξ\xi diverges at the transition. The absence of a finite characteristic correlation length allows fluctuations on many spatial scales. At criticality, the correlation function commonly takes the asymptotic form

G(r)∼1rd−2+η,G(r)\sim \frac{1}{r^{d-2+\eta}},

where dd is the spatial dimension and η\eta is another critical exponent.

In a fluid near its liquid–vapor critical point, large density fluctuations strongly scatter visible light, producing critical opalescence. This provides a direct physical manifestation of growing correlations.

Scaling and universality

Critical exponents

Critical exponents characterize how physical quantities approach singular behavior near a transition. For a conventional magnetic transition, representative scaling laws are

M∼(−t)β(t<0, H=0),M\sim (-t)^\beta \qquad (t<0,\ H=0),
χ∼∣t∣−γ,\chi\sim |t|^{-\gamma},
Csing∼∣t∣−α.C_{\mathrm{sing}}\sim |t|^{-\alpha}.

Here MM is the spontaneous magnetization, HH is the applied magnetic field, χ\chi is the magnetic susceptibility, and CsingC_{\mathrm{sing}} is the singular contribution to the heat capacity. The symbols α\alpha, β\beta, and γ\gamma denote exponents, not universal numerical constants shared by every transition.

These expressions describe leading asymptotic behavior. Actual measurements may also contain smooth background terms, corrections to scaling, and logarithmic factors. A negative heat-capacity exponent, for example, can correspond to a finite cusp rather than a divergent heat capacity.

The exponents are not all independent. Under the usual scaling assumptions, relations include

α+2β+γ=2,γ=ν(2−η).\alpha+2\beta+\gamma=2, \qquad \gamma=\nu(2-\eta).

The hyperscaling relation

2−α=dν2-\alpha=d\nu

has additional conditions and generally does not hold in this simple form above the upper critical dimension.

Universality classes

Microscopically different systems can exhibit the same critical exponents and scaling functions after appropriate normalization. Such systems belong to the same universality class.

For conventional equilibrium transitions, the principal determining features include:

  • spatial dimension;
  • the symmetry and number of components of the order parameter;
  • whether interactions are short-ranged or long-ranged;
  • additional constraints, disorder, or coupled fields that affect large-scale behavior.

The Ising model, in which each site carries a two-state variable, provides a basic example. Many uniaxial magnetic transitions and simple-fluid liquid–vapor critical points share three-dimensional Ising critical behavior, despite their very different microscopic constituents.

Universality does not mean that all properties are identical. Critical temperatures, microscopic length scales, and many amplitudes depend on the material. Certain dimensionless amplitude ratios, however, are universal within a universality class.

Renormalization-group explanation

The renormalization group explains universality by examining how a system’s description changes under coarse-graining: microscopic degrees of freedom are combined into effective variables, and lengths are rescaled.

At a critical point, repeated coarse-graining approaches a scale-invariant fixed point. Many microscopic differences become irrelevant to long-distance behavior, while a smaller number of relevant variables determine departure from criticality.

For many short-range systems, mean-field theory is accurate above an upper critical dimension. Below that dimension, fluctuations modify the critical exponents. The upper critical dimension depends on the model; it is four for the standard short-range scalar order-parameter theory of the Ising transition. At the upper critical dimension itself, logarithmic corrections commonly appear.

Dynamics and finite systems

Critical slowing down

Criticality affects time dependence as well as spatial structure. Near many continuous transitions, relaxation becomes increasingly slow because correlated regions extend over larger distances. A characteristic relaxation time often scales as

τ∼ξz,\tau\sim \xi^z,

where zz is the dynamic critical exponent.

The value of zz depends on dynamical features that static critical exponents do not fully specify, including conservation laws and coupling to other slow variables. Two systems can therefore share static critical behavior but have different critical dynamics.

Critical slowing down complicates experiments and simulations because equilibration takes longer near the transition. It can also limit how closely a system follows equilibrium when driven through a critical region.

Finite-size effects

True thermodynamic singularities generally require the thermodynamic limit, in which system size tends to infinity at fixed appropriate intensive conditions. In a finite sample, singularities are rounded and the growth of correlations is limited by the system’s dimensions.

Finite-size scaling uses this limitation as an analytical tool. For a system with linear size LL, a representative scaling form is

χ(t,L)=Lγ/νF(tL1/ν),\chi(t,L)=L^{\gamma/\nu}F(tL^{1/\nu}),

up to corrections and nonuniversal scale factors. Comparing systems of different sizes can help estimate the critical point and exponents.

A broad distribution or a large response in a single finite system is not, by itself, sufficient evidence of criticality.

Quantum criticality

A quantum critical point is a continuous transition at absolute zero driven by a nonthermal parameter, such as pressure, magnetic field, or interaction strength. The transition reflects quantum fluctuations rather than thermal fluctuations.

In quantum mechanics, incompatible observables need not have simultaneously definite values, and these fluctuations can reorganize the ground state as a parameter changes. Quantum criticality occurs in settings including magnetic materials, interacting particle models, and some transitions involving superconductivity.

Although the critical point lies at zero temperature, it can influence a finite-temperature region above it. The associated crossover behavior may affect heat capacity, transport, and response functions.

Quantum critical scaling includes both spatial and temporal structure. In many theoretical mappings, imaginary time acts as an additional direction whose scaling is governed by the dynamic exponent. This does not make every quantum transition equivalent to a simple classical transition: interactions, fermionic excitations, and other degrees of freedom can complicate the correspondence.

Nonequilibrium and self-organized criticality

Critical behavior also occurs outside equilibrium. Examples include transitions between an absorbing state, from which activity cannot restart without external intervention, and a persistently active state. Such transitions need not share the universality classes of equilibrium systems.

Self-organized criticality describes mechanisms by which driven, dissipative systems approach a critical regime without an externally adjusted control parameter being set precisely to its threshold. The influential sandpile models introduced by Per Bak, Chao Tang, and Kurt Wiesenfeld in 1987 illustrate how slow input and threshold-triggered redistribution can generate avalanches over a broad range of sizes.

The word “self-organized” does not mean that critical behavior is independent of conditions. Driving rate, dissipation, boundaries, conservation laws, and separation of timescales can be essential.

A power-law distribution of event sizes is not sufficient to establish self-organized criticality. Other mechanisms can produce similar distributions. Stronger evidence involves consistent scaling relations, finite-size behavior, and a demonstrated mechanism connecting the dynamics to a critical state.

Proposals that neural activity, ecological systems, or seismicity operate near criticality must therefore be assessed model by model. In neuroscience, observed cascades of activity can be compared with critical models, but subsampling, measurement procedures, and alternative mechanisms complicate interpretation. Critical-like statistics alone do not establish that an entire brain is at a thermodynamic critical point.

Nuclear criticality

In nuclear science, criticality has a distinct operational meaning: a nuclear fission chain reaction is self-sustaining because each generation of neutrons produces, on average, one succeeding generation.

The effective neutron multiplication factor, keffk_{\mathrm{eff}}, accounts for neutron production, absorption, and leakage:

  • Subcritical: keff<1k_{\mathrm{eff}}<1; the chain reaction cannot sustain itself without an external source.
  • Critical: keff=1k_{\mathrm{eff}}=1; the neutron population is steady under unchanged conditions.
  • Supercritical: keff>1k_{\mathrm{eff}}>1; the neutron population tends to increase.

This classification concerns neutron balance, not a particular power level. A reactor can be critical at low or high steady power. “Critical” does not itself mean that an accident or explosion is occurring.

Some neutrons are emitted promptly during fission, while delayed neutrons arise from the radioactive decay of certain fission products. A system can be supercritical when delayed neutrons are included while remaining subcritical on prompt neutrons alone. Prompt criticality is reached when prompt neutrons alone suffice to sustain the chain reaction; it has very different timescales from delayed-neutron-controlled operation.

Nuclear criticality depends on material composition, geometry, neutron leakage, and surrounding materials. It is not determined by mass alone. The first controlled, self-sustaining nuclear chain reaction was achieved in Chicago Pile-1 on 2 December 1942, under the direction of Enrico Fermi.

Although neutron multiplication can be modeled using a branching process with a critical threshold, nuclear criticality should not be identified with the correlation-length divergence of an equilibrium fluid or magnet.

Historical development and interpretation

The study of criticality developed from nineteenth-century investigations of fluid phase behavior. Thomas Andrews’s experiments on carbon dioxide established the continuity of liquid and gaseous states above a critical temperature. Van der Waals’s equation of state provided an influential theoretical description, although its mean-field critical exponents do not accurately describe the asymptotic critical region of many real fluids.

Twentieth-century developments included Landau’s order-parameter theory, exact results for selected lattice models, scaling hypotheses, and the renormalization-group account associated especially with Kenneth Wilson. These developments explained why collective fluctuations can defeat simple mean-field approximations and why unrelated materials can exhibit the same critical behavior.

Across applications, criticality is most precisely established by identifying the control parameters, specifying the competing regimes, and testing the predicted scaling or balance conditions. A threshold, instability, abrupt change, and continuous critical point are related possibilities, but they are not equivalent descriptions.