Tunable Coherence: Control & Applications

• Tunable coherence is a design paradigm that uses external parameters to actively control and modify the coherence properties of light, quantum states, and many-body systems.
• It leverages techniques such as pump bandwidth tuning, phase modulation, and modal conditioning to switch between enhanced and suppressed coherence in diverse platforms.
• Applications span from improving imaging resolution in nonlinear OCT to stabilizing qubit operations in superconducting circuits and many-body spin systems.

Searching arXiv for the cited topic cluster and specific IDs to ground the response in current arXiv records. Tunable coherence denotes the deliberate control of coherence properties by external parameters rather than treating coherence as a fixed attribute of a source, detector, or material system. In the literature, the term spans several non-equivalent but related operations: tuning first-order temporal or spectral coherence of optical fields, conditioning the coherence of a single photon through measurement on an entangled partner, regulating the coherence sensitivity of a POVM, suppressing unwanted interferometric coherence while preserving intended interference, and protecting or generating coherence in spin, excitonic, superconducting, and matter-wave platforms (Quesada et al., 2018, Svozilík et al., 2014, Voigt et al., 20 Jan 2025, Davis et al., 2020, Chang et al., 2022).

1. Operational scope and definitions

The central unifying feature is external control over an operationally defined coherence observable. In optics this observable is often a first-order correlation function or an interferometric visibility; in quantum measurement it is the dependence of outcome statistics on off-diagonal density-matrix elements; in many-body systems it is the survival or generation of off-diagonal density-matrix elements in a specified basis; and in qubit hardware it is a coherence time such as $T_2$, $T_{2E}$, or a logical-subspace echo time (Liu et al., 10 May 2026, Xu et al., 2019, Purkayastha et al., 2019, Levine et al., 2023).

Domain	Control parameter	Coherence quantity
Broadband optical states	Pump bandwidth, AOM phase statistics, gain	$g^{(1)}$, $G^{(2)}(\tau)$, coherence time
Heralded single photons	TSW modal acceptance $M$	$G_A^{(1)}(x,x')$, $\gamma$
Quantum detectors	LO intensity, mode overlap $\mathcal{M}$	POVM off-diagonals, $M_{\diamond}$, $\tilde{M}_{\diamond}$
Interferometers	PRN frequency, sequence length	Remaining coherence length, stray-light suppression
Spin and exciton systems	Field angle, magnetic field, bath engineering	$T_{2E}$0, valley coherence time, $T_{2E}$1
Superconducting and tunneling devices	Flux bias, gate voltage, exchange coupling	$T_{2E}$2, $T_{2E}$3, LZSM-extracted $T_{2E}$4

The formalizations differ accordingly. For a heralded photon, the reduced state is specified by the kernel of the first-order correlation function,

$T_{2E}$5

and coherence can be summarized by the incoherence parameter

$T_{2E}$6

with $T_{2E}$7 for a separable, fully first-order coherent single photon (Svozilík et al., 2014). For broadband pseudothermal light, spectral and temporal coherence are encoded in $T_{2E}$8 and $T_{2E}$9, respectively, with coherence controlled by the Schmidt-mode structure of the JSA (Quesada et al., 2018). For tunable detectors, coherence sensitivity is cast resource-theoretically through the dephasing map

$g^{(1)}$0

and the condition for detection-incoherent operations,

$g^{(1)}$1

with quantifiers $g^{(1)}$2 and $g^{(1)}$3 defined on the measurement channel $g^{(1)}$4 (Xu et al., 2019).

2. Optical-field coherence as a tunable source property

A large class of work treats coherence as a source-side degree of freedom. In broadband pseudothermal states generated from one arm of a broadband twin beam, the reduced state factorizes into thermal states in spectral Schmidt modes, and the pump spectral width $g^{(1)}$5 tunes the coherence continuously from perfect first-order coherence to complete spectral incoherence. In the pulsed regime, a nearly separable JSA yields essentially one Schmidt mode and $g^{(1)}$6; in the cw limit, energy conservation enforces

$g^{(1)}$7

eliminating cross-frequency coherences (Quesada et al., 2018).

A classical route to tunable temporal coherence uses a Mach–Zehnder interferometer with AOMs driven by programmable rf signals. Random phase jumps and controllable dwell-time statistics make the coherence time depend on the imposed phase-noise distribution rather than on the diode laser itself. For phase jumps occurring only in one arm at regular interval $g^{(1)}$8, the measured output obeys

$g^{(1)}$9

so $G^{(2)}(\tau)$0 and $G^{(2)}(\tau)$1 directly sets the coherence time. The same platform independently tailors photon-number statistics, including thermal and classical non-Gaussian states (Pandey et al., 2012).

Gain-driven tuning appears in the thulium-doped NLP laser source. There, a ring-cavity Tm-doped fiber laser with NPR mode locking and an intracavity AOTF operates across the full $G^{(2)}(\tau)$2 nm gain band, while optical gain drives a transition from coherent single-pulse dynamics through drifting, splitting, and transient behavior into chaotic NLP emission. Simulations show that the overall coherence drops from near unity at low gain to about $G^{(2)}(\tau)$3 at high gain, and the high-gain NLP regime exhibits a coherence time of about $G^{(2)}(\tau)$4 fs with an average pulse-burst duration of about $G^{(2)}(\tau)$5 ps (Liu et al., 10 May 2026). The central correlation objects are the mutual coherence function and cross-spectral density,

$G^{(2)}(\tau)$6

with normalized degrees of coherence obtained by division by the corresponding intensities and spectra.

Wave-packet-tunable photons generated by SRS in a cold $G^{(2)}(\tau)$7 ensemble provide a further optical instance in which the waveform duration and coherence are experimentally distinct. An unequal-arm fiber interferometer extracts coherence from visibility decay according to

$G^{(2)}(\tau)$8

The reported coherence times are $G^{(2)}(\tau)$9 for the write laser and $M$0 for the Stokes photons, with corresponding bandwidths $M$1 and $M$2 (Li et al., 2023).

3. Measurement-conditioned coherence and detector coherence

In another usage, tunable coherence is not a property of the source field alone but of the reduced state inferred after conditioning on a partner system. The waveguide-array proposal for Anderson localization of partially coherent light makes this explicit: photon A enters a disordered one-dimensional WGA, while photon B is projected through a TSW that supports a finite number $M$3 of guided modes. The measurement operator

$M$4

changes the conditioned state of A to

$M$5

Projecting B into one mode makes A effectively first-order coherent; projecting into more modes makes A partially coherent; and increasing $M$6 restores the source value. In the localization problem, lower coherence leads to broader localized output and therefore weaker localization, even though Anderson localization persists (Svozilík et al., 2014).

A related but distinct notion is detector coherence: the ability of a measurement to respond to coherence rather than only populations. For the WFHD, the POVM elements are reconstructed by QDT under different LO intensities $M$7 and mode overlaps $M$8. Off-diagonal entries such as $M$9 diagnose coherence sensitivity qualitatively, while the resource-theoretic measures

$G_A^{(1)}(x,x')$0

quantify it. Higher $G_A^{(1)}(x,x')$1 increases coherence detection; at nearly perfect overlap the coherence-detection measure rises monotonically with LO intensity; and for imperfect overlap the LO acts both as phase reference and incoherent background, so the dependence becomes non-monotonic (Xu et al., 2019).

These examples correct a common oversimplification: tunable coherence need not be implemented at generation. The source may remain unchanged while the observed coherence is reconfigured by modal acceptance, conditioning, or POVM geometry (Svozilík et al., 2014, Xu et al., 2019).

4. Interferometric coherence engineering in imaging and precision metrology

In nonlinear OCT, tunable coherence is used to transfer IR sample information into visible-photon interference. The Michelson-type nonlinear interferometer places a PPLN crystal inside the interferometer, generates frequency-nondegenerate SPDC pairs, sends the signal and pump to a reference mirror and the idler to the sample, and relies on induced coherence without induced emission. The detected visible intensity obeys

$G_A^{(1)}(x,x')$2

where $G_A^{(1)}(x,x')$3 is the idler amplitude reflection coefficient and $G_A^{(1)}(x,x')$4 is the normalized first-order correlation function of the SPDC biphoton field. For aligned arms, the visibility is

$G_A^{(1)}(x,x')$5

and the coherence function is

$G_A^{(1)}(x,x')$6

The axial resolution is set by the SPDC bandwidth through $G_A^{(1)}(x,x')$7. By changing the PPLN poling period and temperature, the experiment demonstrates probe/detection pairs at 1543/812 nm, 2140/708 nm, 2504/606 nm, and 3011/582 nm, enabling visible readout of IR OCT and materials characterization without IR detectors (Paterova et al., 2017).

A different interferometric agenda deliberately suppresses coherence for parasitic paths. In tunable-coherence laser interferometry, PRN phase modulation imposes a binary $G_A^{(1)}(x,x')$8-shift sequence on the laser so that desired, delay-matched interference is preserved while ghost beams, stray light, and scattered light decorrelate. The basic scales are

$G_A^{(1)}(x,x')$9

with the remaining coherence set by the chip duration and recoherence at the full sequence length. A Michelson demonstration with a 255-chip sequence achieved about $\gamma$0 dB suppression of the scattered-light peak and an artificial coherence length below $\gamma$1 cm, while cavity tests showed that the classic resonator response is recovered when the cavity round-trip length satisfies $\gamma$2 (Voigt et al., 20 Jan 2025).

Subsequent PRN work pushed modulation to $\gamma$3 GHz, reducing the remaining coherence to below $\gamma$4 cm in an interferometer and producing cavity resonance narrowing to $\gamma$5 at the largest tested $\gamma$6. In the Michelson testbed, the maximum observed suppression was just above $\gamma$7 dB; in a power-recycled Michelson interferometer the injected stray-light signal power was reduced by about $\gamma$8 dB and the signal-to-noise ratio improved by about $\gamma$9 dB (Voigt et al., 1 Aug 2025). In a tabletop Sagnac interferometer, the same controlled reduction of coherence yielded a maximum scattered-light suppression of $\mathcal{M}$0 dB and was further analyzed as a route to suppress coherent backscatter in ring resonators by preventing PRN-history matching of counter-propagating beams at scattering surfaces (Eggers et al., 11 Feb 2025).

These metrological examples show that tunable coherence can either enhance information transfer between otherwise inaccessible spectral bands, as in IR OCT, or veto unwanted interference by engineered decorrelation, as in PRN-modulated interferometry (Paterova et al., 2017, Voigt et al., 20 Jan 2025).

5. Coherence protection, generation, and regulation in many-body and solid-state systems

In cavity-mediated spin physics, tunable coherence arises from controllable anisotropy of collective interactions. The nonlocal XXZ Hamiltonian

$\mathcal{M}$1

is tuned continuously by tilting the magnetic field, with

$\mathcal{M}$2

Hamiltonian tomography and susceptibility measurements show collective-spin behavior and a symmetry $\mathcal{M}$3. More importantly for coherence, ferromagnetic XY exchange opens a many-body gap and suppresses dephasing from inhomogeneous longitudinal fields. The global contrast

$\mathcal{M}$4

reaches $\mathcal{M}$5, close to the initial coherence limit $\mathcal{M}$6, whereas the non-interacting and Ising cases dephase strongly (Davis et al., 2020).

In monolayer WS$\mathcal{M}$7, valley coherence is magnetically tunable because a vertical magnetic field lifts the $\mathcal{M}$8 degeneracy and suppresses exchange-induced pure dephasing. The suppression factor is

$\mathcal{M}$9

and the pure-dephasing rate is modeled as

$M_{\diamond}$0

The model reproduces bright-exciton coherence decay faster than population relaxation and predicts bright-exciton coherence times on the $M_{\diamond}$1 ps scale at low temperature under magnetic tuning, with dark-exciton coherence times $M_{\diamond}$2 ps even without an initially coherent state (Lan et al., 2023).

In delta-doped PECVD-grown diamond, tunable coherence is achieved indirectly through engineering of a 2D dipolar spin bath. The P1 layer thickness can be as small as $M_{\diamond}$3 nm, P1 density is measured as $M_{\diamond}$4 ppm·nm in one TEM spot by DEER, NV density is tuned between $M_{\diamond}$5 and $M_{\diamond}$6 ppm·nm by electron irradiation dose, and the maximum NV/P1 ratio reaches $M_{\diamond}$7. Single NVs outside irradiated spots retain Hahn echo times $M_{\diamond}$8, while the dense ensemble in TEM spot I shows an average $M_{\diamond}$9, with decoherence predominantly limited by dipolar interactions with the engineered P1 and NV baths (Hughes et al., 2022).

A conceptually different result is the DQD charge qubit coupled to substrate phonons. There, phonons do not merely induce decoherence; because the coupling in the energy basis has both $\tilde{M}_{\diamond}$0 and $\tilde{M}_{\diamond}$1 components,

$\tilde{M}_{\diamond}$2

the equilibrium steady state acquires nonzero coherence in the energy eigenbasis. The steady-state coherence is measured by $\tilde{M}_{\diamond}$3, its magnitude and sign are controlled by $\tilde{M}_{\diamond}$4 and $\tilde{M}_{\diamond}$5, and it is robust to equilibrium fermionic leads within the single-particle regime (Purkayastha et al., 2019). This directly contradicts the common expectation that the steady state of a phonon-coupled charge qubit must be diagonal in its own energy basis.

6. Superconducting, tunneling, and matter-wave implementations

Tunable superconducting flux qubits implement coherence-preserving frequency control by replacing one junction with an asymmetric SQUID. The effective SQUID Josephson energy is

$\tilde{M}_{\diamond}$6

which provides a tunability of typically $\tilde{M}_{\diamond}$7 GHz around the central gap frequency. The best reported coherence metrics are $\tilde{M}_{\diamond}$8 and $\tilde{M}_{\diamond}$9, with dielectric loss dominating relaxation and flux noise dominating echo dephasing even at optimal points (Chang et al., 2022).

In the dual-rail erasure qubit formed by two resonantly coupled tunable transmons, coherence is protected by encoding into the single-excitation manifold,

$T_{2E}$00

so that transmon $T_{2E}$01 events leave the logical manifold and become detectable erasures. The dual-rail splitting is set mainly by the exchange coupling $T_{2E}$02, yielding passive suppression of low-frequency detuning noise. Postselected CPMG measurements give $T_{2E}$03 from $T_{2E}$04 to $T_{2E}$05, while the erasure lifetime remains about $T_{2E}$06; the qubit also preserves $T_{2E}$07 across a broad $T_{2E}$08 MHz tuning band from $T_{2E}$09 to $T_{2E}$10 GHz (Levine et al., 2023).

In nanowire-based tunable Josephson junctions, coherence of distinct charge-transfer processes is extracted by LZSM interferometry. The tunneling object can be a single charge, multiple charges transferred through MARs, or a Cooper pair, and coherence is defined operationally as the survival of phase information across repeated LZ passages. Fourier-space line cuts of the 2D transform of the interference pattern decay as $T_{2E}$11, yielding $T_{2E}$12. Reported values include $T_{2E}$13 ns for Cooper pairs, $T_{2E}$14 ns for 2-charge first-order MAR, and $T_{2E}$15 ns for single charges at one gate setting, with magnetic field reducing coherence from about $T_{2E}$16 ns at $T_{2E}$17 to about $T_{2E}$18 ns at $T_{2E}$19 T (He et al., 2023).

A matter-wave implementation shows that tunability need not compromise coherence. In the SAT electron source with a two-counter-electrode acceleration/deceleration geometry, beam intensity and final velocity are tuned independently. The detector count rate increases by about a factor of $T_{2E}$20 at fixed final energy, while transverse coherence, fringe spacing, and contrast remain unchanged; longitudinal coherence lengths remain $T_{2E}$21, $T_{2E}$22, and $T_{2E}$23 nm for three first-electrode settings, corresponding to an energy width of $T_{2E}$24 meV for the SAT emitter (Pooch et al., 2017).

7. Applications, distinctions, and limits

The application space is correspondingly broad. Tunable optical coherence supports visible-readout IR OCT and material characterization in spectral regions where scattering and absorption contrast differ from the visible (Paterova et al., 2017). Low-temporal-coherence NLP fiber lasers target high-contrast, speckle-free full-field imaging by combining wide tunability, high average power, and low inter-shot phase correlation (Liu et al., 10 May 2026). Wave-packet-tunable photons are motivated by hybrid quantum networks and long-baseline quantum interferometric telescopes, where coherence time limits entanglement distribution distance (Li et al., 2023). Cavity-mediated spin exchange is proposed as a route to more robust spin squeezing, while dense 2D diamond spin systems are aimed at enhanced quantum sensing and quantum simulation (Davis et al., 2020, Hughes et al., 2022). Dual-rail erasure qubits use tunable coherence suppression of in-subspace dephasing to create a large erasure noise bias for hardware-efficient quantum error correction (Levine et al., 2023). PRN-based tunable coherence is directed at stray-light mitigation in ground-based gravitational-wave detectors and related precision interferometers (Voigt et al., 20 Jan 2025).

Several conceptual distinctions recur. First, tunable coherence is not synonymous with coherence suppression: it may mean reduction of mutual coherence for unwanted interferometric paths, enhancement of first-order coherence by modal projection, protection of many-body coherence by exchange gaps, or generation of steady-state coherence by a phonon bath (Voigt et al., 1 Aug 2025, Svozilík et al., 2014, Davis et al., 2020, Purkayastha et al., 2019). Second, coherence is basis-dependent. A detector may be coherence-blind in a chosen incoherent basis even while being phase sensitive in another, and a DQD may exhibit coherence in its energy eigenbasis precisely because the system-bath coupling is not diagonal there (Xu et al., 2019, Purkayastha et al., 2019). Third, tunability generally comes with residual constraints: finite EOM bandwidth and modulation-depth errors limit PRN suppression; flux noise limits tunable flux qubits; exciton-phonon scattering limits valley coherence at higher temperature; and imperfect LO overlap degrades detector coherence sensitivity (Eggers et al., 11 Feb 2025, Chang et al., 2022, Lan et al., 2023, Xu et al., 2019).

A plausible synthesis is that tunable coherence has become less a single technique than a design principle: coherence is treated as an actively engineered resource or nuisance, shaped in the domain—optical, measurement-theoretic, many-body, or circuit—the application requires.