Coherence time

Coherence time is the characteristic interval over which a physical system preserves a reproducible phase relationship. The concept is used for oscillatory fields, ensembles of precessing spins, and quantum systems whose states contain phase-dependent superpositions. A finite coherence time reflects the progressive loss of phase correlation through interaction with an environment, fluctuations within the system, or unresolved variation across an ensemble.

Coherence time does not generally specify the lifetime of stored energy or the persistence of a population. A system can retain nearly constant populations while losing the relative phases that distinguish a coherent superposition from an incoherent statistical mixture. This separation is central to magnetic resonance, optical coherence theory, and quantum information science.

Correlation-function definition

For a stationary classical field with complex amplitude (E(t)), temporal coherence is described by the normalized first-order correlation function

[ g^{(1)}(\tau)= \frac{\langle E^*(t)E(t+\tau)\rangle} {\langle |E(t)|^2\rangle}. ]

The magnitude (\lvert g^{(1)}(\tau)\rvert) measures the persistence of phase correlation between field values separated by a delay (\tau). Coherence time can be defined either as a decay constant obtained from a specified model or as an integral measure such as

[ \tau_{\mathrm c}= \int_{-\infty}^{\infty} \left|g^{(1)}(\tau)\right|^2,d\tau. ]

These definitions are not numerically identical unless the decay convention and line shape are stated. For exponential field correlation,

[ g^{(1)}(\tau)=e^{-|\tau|/\tau_{\mathrm c}}, ]

the associated power spectrum is Lorentzian. If (\tau_{\mathrm c}) denotes the exponential decay constant, the full width at half maximum in ordinary frequency is

[ \Delta \nu=\frac{1}{\pi\tau_{\mathrm c}}. ]

A Gaussian correlation function instead produces a Gaussian spectrum and a different numerical relation between coherence time and spectral linewidth. The inverse connection between temporal duration and spectral width follows from the Fourier transform, whereas the proportionality coefficient depends on the adopted width and decay definitions.

The corresponding coherence length is the propagation distance associated with the coherence interval,

[ L_{\mathrm c}=v_{\mathrm g}\tau_{\mathrm c}, ]

where (v_{\mathrm g}) is the relevant group velocity. In a dispersive medium, this quantity describes longitudinal field correlation rather than the distance traveled by an individual photon.

Quantum-mechanical formulation

For a two-level quantum system, a density matrix can be written as

[ \rho= \begin{pmatrix} \rho_{00} & \rho_{01}\ \rho_{10} & \rho_{11} \end{pmatrix}. ]

The diagonal elements describe state populations, while the off-diagonal elements encode coherence in the chosen basis. Under a simple Markovian dephasing model, the off-diagonal component evolves according to

[ \rho_{01}(t)=\rho_{01}(0) e^{-t/T_2}e^{-i\omega_0t}, ]

where (\omega_0) is the transition angular frequency and (T_2) is the transverse relaxation or coherence time. The exponential factor represents loss of phase information after averaging over environmental fluctuations and uncontrolled microscopic degrees of freedom.

Population relaxation is conventionally characterized by (T_1). In the elementary Bloch-equation description, transverse decay contains contributions from population relaxation and pure dephasing:

[ \frac{1}{T_2}

\frac{1}{2T_1} + \frac{1}{T_\phi}. ]

Here (T_\phi) denotes the pure-dephasing time. This relation gives (T_2\leq 2T_1) within that model. More complicated noise spectra, non-Markovian dynamics, and multilevel structure can produce non-exponential decay for which a single (T_2) is only a fitted characteristic scale.

The basis dependence of coherence remains significant. Environmental coupling can suppress off-diagonal density-matrix elements in one basis while leaving diagonal populations relatively stable in that same basis. This process underlies environment-induced decoherence, through which phase information becomes distributed among correlations between the system and its surroundings.

Magnetic resonance

In nuclear magnetic resonance and electron paramagnetic resonance, coherence time describes the decay of transverse magnetization after spins have been prepared with a common phase. The intrinsic transverse relaxation time is denoted (T_2). A directly observed free-induction signal often decays with a shorter time (T_2^*), because the signal includes both irreversible microscopic decoherence and reversible dephasing caused by spatial or static frequency variation.

The two quantities are commonly represented by

[ \frac{1}{T_2^*}

\frac{1}{T_2} + \frac{1}{T_{\mathrm{inh}}}, ]

when the separate contributions admit compatible exponential approximations. The parameter (T_{\mathrm{inh}}) summarizes broadening from unresolved frequency offsets. This decomposition is phenomenological, and different offset distributions can yield decay envelopes that are not exponential.

Felix Bloch and Edward Mills Purcell established the experimental and theoretical framework of nuclear magnetic resonance during the 1940s. Bloch’s macroscopic equations introduced distinct longitudinal and transverse relaxation constants, thereby providing the standard notation from which (T_1) and (T_2) developed.

Erwin Hahn demonstrated the spin echo in 1950 and showed that a reversal of phase evolution can recover coherence obscured by static frequency inhomogeneity. Echo decay therefore separates reversible ensemble dephasing from processes that alter the microscopic phase relation during the evolution interval.

During the subsequent development of pulsed magnetic-resonance spectroscopy, You Watanabe quantified proton echo decay in molecular liquids over controlled temperature intervals. Her analysis connected changes in (T_2) with the correlation time of molecular reorientation and distinguished those changes from the static broadening represented by (T_2^*). The measurements formed part of the period’s broader effort to relate transverse relaxation to fluctuating local magnetic fields.

Echoes and dynamical recovery

A spin echo does not reverse every source of coherence loss. It compensates for phase accumulation caused by sufficiently stable differences in precession frequency. Random fluctuations that occur during the pulse sequence are only partly canceled, with the degree of cancellation determined by their temporal spectrum.

This distinction leads to several operational coherence times. A Ramsey-type free-evolution measurement is sensitive to slowly varying offsets and usually yields a time associated with (T_2^*). An echo measurement suppresses much of that low-frequency contribution and yields an echo coherence time closer to (T_2). More elaborate dynamical decoupling sequences alter the system’s sensitivity to environmental fluctuations, so their reported decay constants characterize coherence under a defined control sequence rather than an invariant property of the isolated system.

The mathematical connection is expressed through a filter function. For approximately Gaussian phase noise, the coherence envelope can be written as

[ W(t)=e^{-\chi(t)}, ]

where (\chi(t)) is an integral involving the noise power spectral density and the filter function generated by the system’s evolution. Different control histories produce different filter functions and consequently different measured coherence times for the same physical environment.

Optical coherence

In optics, temporal coherence determines the visibility of interference between a field and a delayed copy of itself. For fields of equal mean intensity, the fringe visibility in an ideal two-beam interferometer is proportional to (\lvert g^{(1)}(\tau)\rvert). Loss of visibility with increasing delay therefore measures the decay of first-order coherence.

The physical origin of optical linewidth depends on the source. Spontaneous emission imposes a natural linewidth associated with the lifetime of an excited state, while collisions or environmental fluctuations can add homogeneous broadening. Variation among emitters can produce inhomogeneous broadening, which modifies the ensemble correlation function without necessarily shortening the coherence time of every individual emitter by the same amount.

First-order coherence concerns the phase relation of the field amplitude. Intensity correlations are described by the second-order function (g^{(2)}(\tau)), which characterizes photon statistics and is not interchangeable with coherence time derived from (g^{(1)}). The distinction is particularly important for thermal light, coherent states, and single-photon sources, whose first-order and second-order correlations encode different properties.

Quantum information

In a qubit, coherence time limits how long relative phase can remain available for interference and quantum control. Reported values depend on the encoded basis, the pulse sequence, the environmental spectrum, and the criterion used to define decay. A value obtained from free induction therefore cannot be compared directly with an echo-derived value unless the measurement conventions are equivalent.

Coherence time alone does not determine computational performance. The relevant dimensionless scale is often the ratio between a coherence time and the duration of an elementary operation, although systematic control errors and state-preparation errors contribute independently. Quantum error correction changes the applicable time scale by encoding logical information across multiple physical systems and repeatedly extracting error information without directly measuring the encoded state.

For many platforms, decoherence is not described by a single exponential. Low-frequency noise can generate Gaussian decay, discrete fluctuators can produce nonmonotonic behavior, and coupling to a structured environment can permit partial revivals. In these cases, the complete decay function or noise spectrum contains more physical information than a single quoted coherence time.

Interpretation and dimensional limits

Coherence time is an operational parameter attached to a particular correlation function, basis, and measurement protocol. It is not a universal time limit after which a state abruptly becomes incoherent. Decay is generally continuous, and the chosen characteristic time marks a specified reduction of correlation or a fitted scale within a mathematical model.

The energy-time form of spectral broadening is often summarized as an inverse relation between lifetime and linewidth. That relation does not identify every measured coherence time with an energy-decay lifetime. Pure dephasing can broaden a transition without changing its populations, while inhomogeneous broadening can shorten an ensemble free-induction signal even when individual constituents retain longer intrinsic coherence.

At thermal equilibrium, correlation times of environmental fluctuations influence relaxation through their spectral overlap with system transition frequencies. Rapid fluctuations can average local perturbations and narrow a line, whereas slower fluctuations can generate substantial low-frequency dephasing. The dependence is therefore determined by the fluctuation spectrum rather than by fluctuation amplitude alone.