Nonequilibrium Phase Transitions

Updated 27 February 2026

Nonequilibrium phase transitions are abrupt shifts in macroscopic order observed in systems far from equilibrium, driven by mechanisms like long-range interactions, external fields, and kinetic constraints.
They exhibit discontinuous jumps or smooth continuous changes in order parameters across classical, quantum, and field-theoretic models, exemplified by magnetization in the HMF model and conductivity in driven quantum systems.
Theoretical frameworks such as the Vlasov equation, path-action methods, and entropy production diagnostics are crucial for identifying and classifying unique nonequilibrium universality classes and critical behaviors.

A nonequilibrium phase transition is a nonthermal abrupt change of macroscopic order that occurs in a long-lived, non-Boltzmannian regime of a system driven by mechanisms such as long-range interactions, external driving currents, kinetic constraints, dissipation, or intrinsic breaking of detailed balance. Unlike equilibrium phase transitions, which are characterized by non-analyticities of a thermodynamic potential in thermal equilibrium, nonequilibrium phase transitions involve systems far from equilibrium, often in steady states or in long-lived quasi-stationary states, and frequently display dynamical phenomena inaccessible to equilibrium statistical mechanics. These transitions manifest in classical, quantum, and field-theoretic settings, spanning driven spin models, long-range Hamiltonian systems, open quantum systems, and fluctuating environments.

1. Fundamental Mechanisms and Definitions

Two prototypical mechanisms underlie nonequilibrium phase transitions: (i) dynamical self-organization into ordered, symmetry-broken, or oscillatory macrostates through kinetic, transport, or long-range mean-field effects; and (ii) abrupt changes in transport, structural, or activity observables in response to nonconservative external parameters.

A canonical example is the transition from magnetized ( $M>0$ ) to non-magnetized ( $M\approx0$ ) quasi-stationary states (QSSs) in the Hamiltonian Mean Field (HMF) model as the energy per particle $\varepsilon$ is increased. Here, the order parameter is the magnetization $M = \sqrt{M_x^2 + M_y^2}$ , and the system is driven far from equilibrium by long-range mean-field interactions, with dynamics governed by the Vlasov equation in the thermodynamic limit. Violent relaxation—a rapid, collisionless evolution—brings the system to QSSs where sharp jumps of $M$ signal nonequilibrium phase transitions (Filho et al., 2012).

Nonequilibrium transitions also manifest in quantum systems such as current-driven supersymmetric Yang-Mills theory studied via AdS/CFT methods. Under strong electric fields, these systems exhibit first-order phase transitions in conductivity $\sigma(J)$ with discontinuous jumps and critical endpoints inaccessible in linear response, as well as nonequilibrium critical exponents distinct from equilibrium universality classes (Nakamura, 2012, Endo et al., 2023).

2. Theoretical Frameworks and Model Systems

2.1 Long-Range Hamiltonian Systems

The HMF and its generalizations (gHMF-XY with both ferromagnetic and nematic couplings) provide paradigmatic settings for the study of nonequilibrium transitions in long-range interacting systems. In these models, the dynamics do not equilibrate via two-body collisions but are dominated by Vlasov mean-field evolution and incomplete mixing, so standard Boltzmann-Gibbs statistical mechanics (BG) is inadequate.

The Vlasov equation captures the evolution of the one-particle distribution $f(\theta,p,t)$ .
Nonequilibrium phase boundaries are identified by kinetic criteria such as linear instabilities $\lambda_n^2$ of marginal modes and by direct observation of discontinuous jumps in order parameters (magnetization $m_1$ , nematic order $m_2$ ) in molecular dynamics.
Equilibrium (BG) ensembles often predict qualitatively incorrect phase diagrams and critical behaviors compared to numerical dynamics, especially in the presence of long-range interactions and kinetic trapping (Filho et al., 2012, Teles et al., 2012).

2.2 Driven-Dissipative Quantum and Classical Systems

Nonequilibrium transitions are found in externally driven or dissipative quantum and classical models:

In AdS/CFT models (D3/D7 brane), one constructs a steady-state analog of a thermodynamic potential from gravity duals, identifying first-order jumps in observables such as conductivity as nonequilibrium phase transitions, with nonequilibrium critical points and universal critical exponents consistent with mean-field values: $\beta \approx 0.5$ , $\delta \approx 3$ , $\gamma \approx 1$ (Nakamura, 2012, Endo et al., 2023).
Light-driven superconductors (Floquet BCS models) exhibit multi-photon-induced first-order transitions and nested hysteresis behaviors. Here, phase transitions are induced by the interplay of Floquet sidebands, population inversion, and photon-assisted tunneling, with phase diagrams fragmented by drive amplitude and frequency (Zhang et al., 13 Mar 2025).
Driven spin models (Potts, Ising variants with friction or double-bath driving) demonstrate novel sequences of continuous–tricritical–first-order transitions, splitting of the conventional critical point, and mean-field character at interfaces, tied to energy flow and symmetry-breaking induced by the drive (Igloi et al., 2010, Forão et al., 2024).

2.3 Absorbing-State and Constrained Kinetic Models

In kinetically constrained or absorbing-state models, nonequilibrium transitions can be induced by irreversibility, breaking of detailed balance, or temporally fluctuating environments:

The constrained adsorption model with irreversibility parameter $\alpha$ exhibits a continuous nonequilibrium transition with order-parameter exponent $\beta=1$ , defining a universality class absent in equilibrium (Sellitto, 2019).
Temporal disorder in growth/spreading processes leads to “infinite-noise” criticality, with diverging effective noise amplitudes and infinitely broad order-parameter distributions at criticality—markedly different from equilibrium scaling (Vojta et al., 2015).

2.4 Oscillatory and Replica-Like Nonequilibrium Transitions

Mean-field spin models can display nonequilibrium transitions to time-periodic (oscillating) macrostates, characterized not by static order parameters but by the amplitude of deterministic phase-space trajectories (e.g., the effective Hamiltonian $H(m,\dot m)$ ). The oscillating phase supports a nontrivial overlap distribution $P(q)$ mimicking replica symmetry breaking, despite the absence of quenched disorder (Guislain et al., 2022).

3. Methodologies for Characterization

Nonequilibrium phase transitions are diagnosed using diverse order parameters and methods:

Measurement of steady-state or quasi-stationary order parameters (magnetization, nematicity, conductivity, photon population, etc.) as control parameters (energy, current, field strength) are varied.
Detection of discontinuities (first-order), cusps/tricriticality, and continuous (second-order) vanishing of order parameters.
Evaluation of higher-order moments and occupation statistics (e.g., moments $\mu_4,\mu_6$ in HMF QSSs) corroborates the order of the transition (Filho et al., 2012).
Nonequilibrium analogs of “thermodynamic potentials” (e.g., Legendre-transformed D7-brane Hamiltonians in AdS/CFT) are constructed for stability and Maxwell-type criteria (Nakamura, 2012, Endo et al., 2023).
Entropy production $\Pi$ serves as a universal classifier: it vanishes only in equilibrium and shows distinct signatures—a continuous profile with singular derivative (continuous transition) or a jump/hysteresis (first-order), as verified in the majority-vote model and $Z_2$ -symmetric systems (Noa et al., 2018).
Dynamic path-based actions and quasi-potentials (minimum action methods) generalize the Arrhenius law and identify transition rates and barriers even in non-gradient or non-Gaussian systems, often yielding distinct forward/backward paths and broken time-reversal symmetry (Zakine et al., 2022).

4. Classification, Universalities, and Critical Behavior

Unlike equilibrium, universality classes in nonequilibrium phase transitions are determined not only by symmetry and dimensionality but also by the mechanism of irreversibility, the range of interactions, and the type of external driving or constraint.

Model Class	Transition Type	Control Parameter	Key Exponent(s)/Feature	Reference
HMF (long-range)	First-order/reentrant	$\varepsilon$ (energy)	Magnetization jump, reentrant cascades, velocity moments	(Filho et al., 2012)
AdS/CFT current- or field-driven QFT	First-/Second-order	$J$ , $E$ (current/field)	Conductivity jump, nonequilibrium critical point, $\beta\approx0.5$	(Nakamura, 2012, Endo et al., 2023)
Driven Potts models with friction	Continuous/tricritical/1st-order	$K_f$ (interface coupling)	Mean-field critical exponents (even in 2D), hysteresis	(Igloi et al., 2010)
Constrained adsorption (1D)	Continuous	$\alpha$ (irreversibility)	$\beta=1$ (linear vanishing of order parameter)	(Sellitto, 2019)
Temporal disorder (infinite-noise)	Infinite-noise critical	disorder strength	Infinitely broad order-parameter PDF, non-power-law scaling	(Vojta et al., 2015)
Driven Ising models (double-bath)	Split transitions	$\epsilon$ (coupling)	Distinct critical exponents, split transition, engine operation	(Forão et al., 2024)
Light-driven superconductors	First-order, hysteretic	Drive amplitude $A$	Nested hysteresis, multi-photon-induced transitions	(Zhang et al., 13 Mar 2025)

Scaling laws in these systems can be mean-field-like, as in large- $N$ models or driven interface transitions, or highly nontrivial, as in absorbing or infinite-noise critical points, where broad distributions and activated scaling replace standard power laws (Vojta et al., 2015).

5. Uniquely Nonequilibrium Effects

Nonequilibrium phase transitions display phenomena with no equilibrium analogs:

Cascades of reentrant ordered/disordered phases (e.g., multiple on-off magnetization transitions in HMF QSSs as energy is swept) (Filho et al., 2012).
Splitting of traditional order-disorder critical points into separate transitions for symmetry-related ordered phases, with distinct critical exponents (as in driven-bath Ising models) (Forão et al., 2024).
Mean-field criticality at two-dimensional driven interfaces, independent of spatial dimension, due to driving-induced long-range couplings (Igloi et al., 2010).
Existence of nonequilibrium transitions (or persistence of criticality) in systems where equilibrium arguments (e.g., random-field suppression of symmetry breaking) would forbid ordering—such as the one-dimensional contact process with random fields (Barghathi et al., 2012).
Time-periodic (limit-cycle) phases where the order parameter is the amplitude of oscillation, and dynamical overlap distributions realize continuous replica-like symmetry breaking without disorder (Guislain et al., 2022).
Infinite-noise universality classes induced by temporal disorder, with noise-amplitude divergence leading to novel aging and fluctuation statistics (Vojta et al., 2015).
Non-reciprocal, trajectory-dependent instantons in path-action landscapes, with nonequilibrium transition rates that differ forward and backward due to lack of detailed balance (Zakine et al., 2022).

6. Broader Implications, Applications, and Open Directions

Nonequilibrium phase transitions are now established as generic in driven, open, or long-range-interacting many-body systems. They provide the foundation for understanding far-from-equilibrium order/disorder phenomena in fields ranging from quantum transport and driven condensed matter (e.g., Floquet systems, Kerr resonators), quantum optics, and ultrafast superconductor dynamics to biological switching, climate models, and active matter pattern formation. Open challenges include:

Systematic classification of nonequilibrium universality classes, especially in the presence of complex drives, dissipation, and spatial/temporal disorder.
Extension of macroscopic fluctuation theory and path-action methods to high-dimensional, interacting, or non-Markovian systems.
Development of universal entropy production–based diagnostics for experimental identification and quantitative analysis of transitions in real systems (Noa et al., 2018).
Machine learning of nonconservative force-fields for dynamical nonequilibrium phase switching in correlated quantum materials (Fan et al., 12 Jan 2026).
Exploration of the interplay between inertia, transport, and nonequilibrium ordering (e.g., tuning between active and equilibrium-like crystalline phases via inertia) (Omar et al., 2021).
Understanding the operational optimization of nonequilibrium thermodynamic machines around phase transitions in driven materials (Forão et al., 2024).

These advances demonstrate that nonequilibrium phase transitions are not merely extensions of equilibrium theory but require new conceptual, analytical, and computational tools to capture the emergent phenomena intrinsic to irreversible, driven, and dynamically constrained many-body systems.