{
  "id": "quantum-phase-transition",
  "title": "Quantum Phase Transition",
  "lang": "en",
  "source_lang": "en",
  "summary": "A quantum phase transition is a zero-temperature change of a system’s ground-state phase driven by a nonthermal control parameter.",
  "category": "science",
  "subcategory": "physics",
  "type": "concept",
  "aliases": [
    "Quantum phase change",
    "zero-temperature phase transition"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/quantum-phase-transition/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/quantum-phase-transition/",
      "json": "https://aiwiki.page/zh/quantum-phase-transition.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "A quantum phase transition is a change between distinct phases of matter at absolute zero, produced by varying a parameter such as pressure, magnetic field, interaction strength, or particle concentration. Unlike a conventional thermal phase transition, which is driven by thermal fluctuations, it reflects a reorganization of the lowest-energy state allowed by quantum mechanics. Continuous quantum phase transitions are associated with fluctuations extending over increasingly large distances and long times; discontinuous transitions instead involve an abrupt change of ground state.\n\nAlthough absolute zero cannot be reached experimentally, quantum phase transitions can strongly influence measurable properties at finite temperature. They provide a framework for understanding quantum magnets, interacting ultracold gases, metal–insulator transitions, unconventional metals, and changes between topological phases. Their study combines condensed-matter physics, statistical mechanics, and quantum field theory, with particular emphasis on the relationship between microscopic interactions and universal large-scale behavior."
    },
    {
      "heading": "Ground states and the meaning of a quantum transition",
      "text": "An equilibrium quantum system is described by a Hamiltonian operator \\(H(g)\\), where \\(g\\) denotes a tunable parameter. Its ground state minimizes the energy for that value of \\(g\\). A quantum phase transition occurs when changing \\(g\\) produces a nonanalytic change in an appropriate ground-state quantity in the thermodynamic limit, in which particle number and volume become infinite at fixed density. This limiting procedure distinguishes a phase transition from a rapid but smooth change in a small system. Finite systems can exhibit exact level crossings protected by symmetry, but they do not generally possess the full singular behavior of a bulk transition.\n\nThe competing terms in a Hamiltonian often favor incompatible ground states. Magnetic exchange can favor aligned spins, whereas a transverse magnetic field favors alignment along a different spin direction. In a lattice gas, particle motion promotes delocalization, while repulsion can favor an integer number of particles on each site. Because the competing operators do not generally commute, their effects cannot be minimized simultaneously. Quantum fluctuations arise from this incompatibility; they are not simply thermal motion persisting at low temperature, nor do they require the ground state itself to evolve randomly in time.\n\nAt a continuous transition, the characteristic correlation length diverges and a relevant low-energy excitation scale vanishes. The ground-state energy can remain continuous, together with some of its derivatives, while response functions become singular. At a first-order transition, two competing phases exchange stability, commonly producing a discontinuity in a quantity such as magnetization or density. The closing of an excitation gap is therefore an important indicator of continuous criticality, but it is not a sufficient definition of a transition: some phases remain gapless throughout an extended parameter range.\n\nMany transitions are classified using an order parameter, a quantity that distinguishes phases through symmetry. A ferromagnet has a nonzero spontaneous magnetization, whereas a symmetric paramagnet does not. Other transitions change a topological invariant or the long-range organization of quantum entanglement without acquiring a conventional local order parameter. Ground-state phase diagrams consequently require more than a catalogue of broken symmetries. They may also distinguish gapped from gapless states, localized from conducting states, and phases with different responses to external fields."
    },
    {
      "heading": "The transverse-field Ising model",
      "text": "The transverse-field Ising model provides a particularly clear example of a quantum phase transition. For a one-dimensional chain with nearest-neighbor coupling, its Hamiltonian can be written as\n\n\\[H=-J\\sum_i \\sigma_i^z\\sigma_{i+1}^z-h\\sum_i\\sigma_i^x,\\]\n\nwhere \\(J>0\\) favors ferromagnetic alignment, \\(h\\) is the transverse-field energy scale, and \\(\\sigma_i^\\alpha\\) are Pauli matrices. The field parameter \\(h\\) includes the appropriate conversion from physical magnetic field to energy. The exchange term favors two oppositely magnetized states along the \\(z\\) axis, while the field term favors spins polarized along \\(x\\).\n\nAt \\(h=0\\), the two fully aligned configurations minimize the energy. For \\(h/J\\ll1\\), the transverse field admixes configurations containing reversed spins, but ferromagnetic order survives in the infinite chain. At \\(h/J\\gg1\\), the ground state approaches a product of \\(x\\)-polarized spins and has no spontaneous \\(z\\)-magnetization. The Hamiltonian preserves a global twofold symmetry that reverses every \\(z\\)-spin. The ordered phase breaks this symmetry spontaneously; the high-field phase preserves it. In a finite chain, the exact ground state can respect the symmetry even when the corresponding infinite system has ordered states.\n\nThe nearest-neighbor chain is exactly solvable through a transformation from spin variables to fermionic variables. In this normalization, its critical point is \\(h=J\\), and the elementary excitation gap away from criticality is\n\n\\[\\Delta=2|J-h|.\\]\n\nOn the ordered side, the spontaneous magnetization is\n\n\\[m_z=\\left[1-(h/J)^2\\right]^{1/8}.\\]\n\nThese expressions exhibit a continuous transition: the order parameter vanishes continuously, while the gap closes and the correlation length diverges. The critical exponents include \\(\\nu=1\\), describing the correlation length, and \\(z=1\\), describing the relative scaling of time and space.\n\nPierre Pfeuty’s 1970 exact treatment made the chain a central reference model for quantum criticality. Its critical behavior is related to that of the two-dimensional classical Ising model, despite the quantum chain having only one spatial dimension. This correspondence illustrates how quantum dynamics supplies an additional dimension in an imaginary-time description. It also exposes an important contrast: a one-dimensional classical Ising chain with finite-range interactions has no ferromagnetic order at nonzero temperature, yet the quantum chain has an ordered zero-temperature phase. Experiments on magnetic chains, including cobalt niobate, have explored realizations of this physics, although interchain coupling and additional interactions complicate direct comparison with the ideal model."
    },
    {
      "heading": "Scaling, universality, and the quantum–classical correspondence",
      "text": "Near a continuous quantum critical point, microscopic details can become less important than symmetry, dimensionality, interaction range, and the kinds of low-energy excitations. This is the basis of universality. For a dimensionless distance \\(r\\) from criticality, the correlation length and characteristic correlation time commonly scale as\n\n\\[\\xi\\sim |r|^{-\\nu}, \\qquad \\xi_\\tau\\sim \\xi^z.\\]\n\nHere \\(\\nu\\) is the correlation-length exponent and \\(z\\) is the dynamic critical exponent. For a transition with a single characteristic low-energy scale,\n\n\\[\\Delta\\sim \\xi^{-z}\\sim |r|^{\\nu z}.\\]\n\nThe exponent \\(z\\) need not equal one: quantum dynamics may scale differently from spatial variations.\n\nThe connection with classical critical phenomena follows from the equilibrium partition function,\n\n\\[Z=\\operatorname{Tr}e^{-\\beta H}, \\qquad \\beta=\\frac{1}{k_{\\mathrm B}T},\\]\n\nwhere \\(k_{\\mathrm B}\\) is the Boltzmann constant. Dividing the exponential into many factors introduces an imaginary-time direction of extent \\(\\hbar\\beta\\), with \\(\\hbar\\) the reduced Planck constant. At zero temperature this extent becomes infinite. A quantum model in \\(d\\) spatial dimensions can therefore resemble a classical statistical model with an additional, generally anisotropic, dimension. This is a mathematical representation of quantum equilibrium, not an assertion that laboratory time is an ordinary extra spatial coordinate.\n\nFor \\(z=1\\), many short-range quantum models map to familiar classical universality classes in \\(d+1\\) dimensions. More generally, the combination \\(d+z\\) often controls power counting and the relevance of interactions. The mapping is not unrestricted: topological terms, fermionic signs, long-range interactions, and coupling to gapless particles can prevent a simple interpretation as a conventional classical model. Integrating out metallic electrons, for example, can generate nonlocal interactions in an effective theory for an order parameter.\n\nThe renormalization group describes how a theory changes as short-distance fluctuations are removed and lengths are rescaled. A critical point corresponds to a scale-invariant fixed point, while relevant perturbations move the system away from it. When hyperscaling applies, the singular free-energy density satisfies\n\n\\[f_{\\mathrm s}(r,T) =b^{-(d+z)} f_{\\mathrm s}(rb^{1/\\nu},Tb^z),\\]\n\nwith nonuniversal conversion factors understood. This relation generates predictions for response functions and finite-temperature behavior. Above an upper critical dimension, or in the presence of a dangerously irrelevant coupling, naive hyperscaling can fail. “Irrelevant” here means that a coupling decreases under coarse graining, not that it has no observable consequences: it may still set an ordering scale or control a singular limit."
    },
    {
      "heading": "Quantum criticality at nonzero temperature",
      "text": "The influence of a zero-temperature critical point often extends into a finite-temperature region called the quantum critical region. A common schematic boundary is set by\n\n\\[k_{\\mathrm B}T\\sim A|r|^{\\nu z},\\]\n\nwhere \\(A\\) is a nonuniversal energy scale. When thermal energy exceeds the scale associated with distance from criticality, the system can probe critical fluctuations rather than the characteristic behavior of either neighboring ground-state phase. This produces the familiar fan-shaped region in many temperature–control-parameter diagrams, although real phase diagrams need not display a geometrically simple fan.\n\nTemperature limits the available extent in imaginary time. Consequently, the zero-temperature divergence of the correlation time is cut off by a scale of order \\(\\hbar/(k_{\\mathrm B}T)\\), and the corresponding thermal length behaves as \\(\\xi_T\\sim T^{-1/z}\\), up to dimensional factors. Exactly at critical tuning, observables can acquire characteristic power laws. Where hyperscaling holds and no additional low-energy sector dominates, the singular contribution to the heat capacity per unit volume follows\n\n\\[C_{\\mathrm s}\\propto T^{d/z}.\\]\n\nAn observed heat capacity may also contain lattice vibrations, ordinary electronic contributions, nuclear terms, or other backgrounds.\n\nCritical dynamics can appear in spectroscopy and scattering measurements. A response function may depend on frequency and temperature through a scaling combination such as \\(\\hbar\\omega/(k_{\\mathrm B}T)\\), accompanied by a temperature-dependent prefactor and momentum-dependent scaling variables. Such frequency-over-temperature scaling is valuable because it tests the organization of fluctuations over multiple energy scales. It is not guaranteed at every quantum critical point, and a limited experimental scaling interval does not uniquely identify a microscopic theory.\n\nA finite-temperature transition and a quantum critical crossover are different phenomena. In a three-dimensional magnet, a line of genuine thermal ordering transitions may terminate at a zero-temperature critical point. Sufficiently close to a nonzero ordering temperature, long-distance fluctuations are usually governed by an appropriate classical universality class. By contrast, some low-dimensional systems cannot sustain the corresponding order at any nonzero temperature, even though they have an ordered ground state. Their finite-temperature behavior consists of crossovers rather than a thermal transition into that state. Superconductivity or another intervening phase may also conceal the low-temperature region in which a proposed quantum critical point would otherwise be approached."
    },
    {
      "heading": "Interacting bosons and the superfluid–Mott transition",
      "text": "Interacting particles on a lattice provide an experimentally controllable setting for quantum phase transitions. The Bose–Hubbard model is\n\n\\[H=-t\\sum_{\\langle ij\\rangle} (b_i^\\dagger b_j+b_j^\\dagger b_i) +\\frac{U}{2}\\sum_i n_i(n_i-1) -\\mu\\sum_i n_i.\\]\n\nHere \\(b_i^\\dagger\\) creates a boson on site \\(i\\), \\(n_i=b_i^\\dagger b_i\\), \\(t\\) is the hopping amplitude, \\(U>0\\) is the on-site repulsion, and \\(\\mu\\) is the chemical potential. Hopping favors delocalization and phase coherence; repulsion penalizes multiple occupancy and can suppress number fluctuations.\n\nAt sufficiently large hopping, the system can exhibit superfluidity, characterized by phase stiffness and a coherent response to imposed phase twists. At integer filling and sufficiently strong repulsion, it can instead form a Mott insulator, with a gap to particle addition or removal and zero compressibility at zero temperature. The insulating behavior is interaction-driven rather than simply the consequence of filling a noninteracting electronic band. Number and phase fluctuations are complementary: a state with tightly constrained site occupation has strongly fluctuating phase, whereas a coherent superfluid permits occupation fluctuations.\n\nThe zero-temperature phase diagram contains Mott regions often drawn as lobes in the \\(\\mu/U\\)–\\(t/U\\) plane. At a generic boundary, the transition is driven by adding particles or holes, and the standard short-range model commonly has \\(z=2\\). At a lobe tip, particle and hole excitations soften together; an emergent particle–hole symmetry gives \\(z=1\\). In two spatial dimensions, the tip transition belongs to the three-dimensional classical XY universality class. One-dimensional systems require particular care: the commensurate transition at a lobe tip has Berezinskii–Kosterlitz–Thouless behavior, with an exponentially diverging correlation length rather than a simple finite-\\(\\nu\\) power law.\n\nIn 1989, Matthew Fisher, Peter Weichman, G. Grinstein, and Daniel Fisher developed an influential account of boson localization, including the role of disorder. In 2002, Markus Greiner and colleagues observed the superfluid-to-Mott-insulator transition using ultracold rubidium atoms in an optical lattice. Increasing lattice depth reduced tunneling relative to interaction energy, and changes in interference patterns revealed the loss of long-range phase coherence. Trapping potentials produce spatially varying density, so superfluid and insulating regions can coexist within one cloud. Later measurements of site occupation, compressibility, and correlations have provided more detailed tests than interference contrast alone. These systems illustrate both the power of parameter control and the need to distinguish the ideal homogeneous phase diagram from a finite, inhomogeneous experiment."
    },
    {
      "heading": "Quantum criticality in metals",
      "text": "Quantum phase transitions in metals are complicated by the presence of low-energy electronic excitations on a Fermi surface. In a conventional Fermi liquid, excitations near this surface behave as long-lived quasiparticles. A fluctuating magnetic or other order parameter can scatter these quasiparticles strongly, while the electrons modify the dynamics of the collective fluctuations. The critical theory must therefore account for both the proposed order and the conducting background, rather than treating the electrons as an inert environment.\n\nJohn Hertz’s 1976 work and Andrew Millis’s 1993 treatment of finite-temperature behavior established a widely used approach: integrate out the fermions and construct an effective theory for the order parameter. Decay into electron–hole excitations produces damping and often a dynamic exponent greater than one. In standard idealized cases, antiferromagnetic dynamics can have \\(z=2\\), while clean ferromagnetic dynamics can have \\(z=3\\). These values depend on assumptions about electronic structure and coupling. In some metallic systems, integrating out gapless fermions generates singular terms that invalidate a simple local expansion. A proposed continuous ferromagnetic transition can also be preempted by first-order behavior or by spatially modulated order.\n\nHeavy-fermion systems provide prominent examples. In compounds containing elements such as cerium or ytterbium, localized magnetic degrees of freedom interact with conduction electrons. The Kondo effect favors screening of local moments, whereas indirect exchange can favor magnetic order. Near certain magnetic phase boundaries, measured heat capacities, magnetic responses, and electrical resistivities depart from conventional Fermi-liquid expectations. Cerium copper gold alloys and ytterbium rhodium silicide have been extensively studied in this context. Their behavior motivates theories ranging from itinerant spin fluctuations to critical changes in Kondo screening and Fermi-surface structure.\n\nNo single observed anomaly proves the existence or nature of a metallic quantum critical point. Approximately linear temperature-dependent resistivity, for example, can arise in multiple settings and does not by itself determine a universality class. Disorder, phonons, multiband transport, and nearby ordered phases complicate interpretation. Moreover, electrical resistance requires a mechanism for relaxing total momentum; rapid scattering alone does not necessarily produce finite direct-current resistivity in a perfectly translation-invariant model. Superconductivity frequently appears near magnetic instabilities and may obscure the putative normal-state critical point. Critical fluctuations can contribute to electron pairing, but a nearby superconducting dome is not sufficient evidence that one particular fluctuation mechanism causes it."
    },
    {
      "heading": "Topological transitions and criticality beyond local order",
      "text": "The Landau framework classifies many phases by symmetry and local order parameters, but it does not exhaust quantum phase structure. In a topological insulator, for example, occupied electronic states possess a global property that cannot be continuously removed while preserving the relevant symmetries and maintaining a bulk gap. A continuous transition between distinct insulating phases in a noninteracting band description requires a bulk gap closing. A typical low-energy description involves a Dirac-like Hamiltonian whose mass parameter changes sign. The two sides can have identical ordinary symmetry yet different boundary states and quantized responses.\n\nThe quantum Hall effect supplies another important class of examples. Changes between Hall plateaus involve changes in quantized transverse conductance, with localization and delocalization playing central roles. Fractional quantum Hall states additionally exhibit topological order, including fractionalized excitations and ground-state structure that depends on spatial topology. Their transitions cannot generally be characterized by the onset of a simple local expectation value. In interacting systems, topological phase changes can also be first order, so a continuous gap-closing scenario should not be assumed universally.\n\nA different extension is deconfined quantum criticality, developed prominently in 2004 by T. Senthil and collaborators. Its canonical setting is a possible direct transition in a two-dimensional quantum magnet between a Néel antiferromagnet and a valence-bond solid. Néel order selects a spin direction, whereas valence-bond order selects a pattern of strong singlet bonds and breaks lattice symmetry. A conventional theory based only on these two competing order parameters often favors a first-order transition or an intervening phase. The deconfined proposal instead describes criticality through fractionalized spin-carrying fields coupled to an emergent gauge field.\n\nIn this picture, excitations that are confined into conventional objects away from criticality can become effectively deconfined at the transition. Topological defects of one order may carry quantum numbers associated with the competing order, linking structures that appear unrelated in a purely local description. Numerical lattice models have provided substantial evidence for this framework, but determining the ultimate character of particular proposed transitions remains difficult. Very weak first-order behavior can imitate continuous scaling over large system sizes. Apparent enlarged symmetries and slowly changing exponents must therefore be assessed alongside finite-size drift, competing length scales, and the possibility of an extended crossover rather than a true scale-invariant point."
    },
    {
      "heading": "Disorder, slow dynamics, and passage through a transition",
      "text": "Static disorder can alter quantum criticality more strongly than an equivalent amount of rapidly fluctuating noise. A random exchange interaction or local potential is fixed throughout the imaginary-time direction, so its influence persists over the full quantum fluctuation history. For spatially uncorrelated quenched disorder, the Harris criterion states that a clean critical point is stable against weak randomness only if\n\n\\[d\\nu>2,\\]\n\nwith equality requiring further analysis. The criterion compares fluctuations of the local tuning parameter within a correlation volume with the distance from criticality. Its spatial dimension is \\(d\\), not automatically \\(d+z\\), because the disorder is correlated in imaginary time.\n\nDisordered systems can contain rare spatial regions that locally favor the competing phase. Their slow dynamics can produce Griffiths effects, including singular responses over a range of parameters rather than only at one point. In some random quantum spin chains, the transition is controlled by an infinite-randomness fixed point. The usual relation \\(\\xi_\\tau\\sim\\xi^z\\) is then replaced by activated scaling,\n\n\\[\\ln(\\xi_\\tau/\\tau_0)\\sim \\xi^\\psi,\\]\n\nwhere \\(\\tau_0\\) is a microscopic time and \\(\\psi\\) is an exponent. Other combinations of disorder, symmetry, and dissipation can smear a transition, as sufficiently large rare regions develop static order independently.\n\nPassing through a continuous quantum transition in time raises a related dynamical problem. The closing of the excitation gap makes adiabatic evolution increasingly difficult. The Kibble–Zurek mechanism estimates the scale at which a driven system falls out of equilibrium by comparing the intrinsic relaxation time with the time remaining to criticality. For a linear ramp with timescale \\(\\tau_Q\\), conventional power-law critical scaling gives\n\n\\[\\hat{\\xi}\\sim\\tau_Q^{\\nu/(1+\\nu z)}.\\]\n\nWhere independently formed excitations are separated by this characteristic length, a corresponding density scales as \\(\\hat{\\xi}^{-d}\\). Domain walls and other extended defects require attention to their dimensionality and to subsequent coarsening.\n\nThese relations are scaling predictions, not universal formulas for every experimental sweep. The ramp protocol, finite size, initial temperature, conservation laws, environmental coupling, and post-transition dynamics all matter. An isolated system evolves unitarily, whereas a system coupled to an environment can exchange energy and relax. Quantum decoherence does not simply turn every slow quantum sweep into a classical equilibrium process. A finite-rate crossing must also be distinguished from a dynamical quantum phase transition defined through a nonanalyticity in a real-time quantity, such as a return amplitude after a quench. The latter need not correspond to crossing an equilibrium ground-state phase boundary."
    },
    {
      "heading": "Experimental identification and computational tests",
      "text": "Because experiments operate at nonzero temperature and finite resolution, identifying a quantum phase transition requires several complementary observations. Researchers first establish how a nonthermal control parameter changes the phases, then determine whether an ordering scale approaches zero. Measurements of excitation gaps, correlation lengths, order parameters, and thermodynamic derivatives help distinguish a continuous transition from a first-order boundary or a crossover. Hysteresis and phase coexistence can support a first-order interpretation, but their absence is not decisive when discontinuities are small or equilibration is slow.\n\nDifferent probes reveal different parts of the critical behavior. Neutron scattering measures momentum- and frequency-dependent magnetic correlations; nuclear magnetic resonance probes local magnetic environments and low-frequency spin dynamics. Calorimetry measures entropy accumulation through heat capacity, while thermal expansion and magnetostriction track the sensitivity of characteristic energy scales to pressure and field. Under suitable assumptions, ratios of thermal expansion to heat capacity, or magnetic analogues derived from magnetocaloric response, diverge near a quantum critical point. Transport measurements are experimentally accessible and sensitive, but their interpretation depends on how charge and heat currents relax.\n\nComputational methods provide controlled tests in selected models. Exact diagonalization yields complete spectra for small systems, while tensor-network methods exploit entanglement structure and are especially effective for many one-dimensional ground states. Quantum Monte Carlo can access large systems when statistical weights are nonnegative. For many fermionic or frustrated models, however, the Monte Carlo approach encounters a sign problem: cancellations between positive and negative contributions cause severe statistical difficulties. No single numerical method reliably covers all dimensions, interactions, temperatures, and entanglement structures.\n\nFinite-size scaling connects these calculations to thermodynamic critical behavior. At a conventional continuous transition, the lowest characteristic gap scales as\n\n\\[\\Delta_L\\sim L^{-z},\\]\n\nand dimensionless ratios can approach size-independent values at criticality. Simulations must also choose an inverse temperature large enough relative to \\(L^z\\) to access ground-state behavior. Apparent crossings can drift because of corrections to scaling, and a small system can mistake a weak first-order transition for a continuous one. Reliable identification therefore combines larger sizes, multiple observables, and explicit comparisons with competing descriptions. Cold atoms, trapped ions, superconducting circuits, and programmable atom arrays extend these tests through quantum simulation, while introducing their own finite-size, calibration, interaction-range, and coherence-time constraints."
    }
  ],
  "links": [
    "absolute-zero",
    "pressure",
    "magnetic-field",
    "phase-transition",
    "quantum-mechanics",
    "temperature",
    "condensed-matter-physics",
    "statistical-mechanics",
    "quantum-field-theory",
    "hamiltonian-operator",
    "energy",
    "density",
    "magnetization",
    "order-parameter",
    "quantum-entanglement",
    "ising-model",
    "universality-class",
    "partition-function",
    "boltzmann-constant",
    "planck-constant",
    "renormalization-group",
    "heat-capacity",
    "spectroscopy",
    "scattering",
    "boson",
    "chemical-potential",
    "superfluidity",
    "fermi-surface",
    "electrical-resistance",
    "superconductivity",
    "topological-insulator",
    "quantum-hall-effect",
    "topological-order",
    "gauge-theory",
    "quantum-decoherence",
    "neutron",
    "nuclear-magnetic-resonance"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}