Phase Coherent Fibers (PCF)

• Phase Coherent Fibers (PCF) are optical fibers engineered to preserve phase integrity either via photonic crystal microstructures or active stabilization.
• They employ precise dispersion engineering, modal coupling, and phase conjugation techniques to achieve coherent beam combining and efficient nonlinear interactions.
• Recent advances demonstrate high phase noise suppression and robust phase matching, facilitating improvements in quantum networking, broadband frequency conversion, and sensing.

Phase Coherent Fibers (PCF) denotes two related but distinct usages in the arXiv literature. In the established fiber-optics sense, PCF usually means photonic crystal fiber or air–silica microstructured fiber, including solid-core index-guiding, hollow-core photonic-bandgap, anti-resonant, dual-core, and dispersion-engineered structures. In a distinct quantum-networking usage, PCF can also mean a phase coherent fiber link, namely an actively stabilized transmission fiber that preserves the phase coherence of a narrow-linewidth optical field during delivery. The common subject across these usages is guided-wave control of optical phase under coherent beam combining, phase-matched nonlinear interaction, broadband frequency conversion, or long-distance transfer (Mohammed et al., 2013, Johnson et al., 10 Sep 2025).

1. Terminology and domain of use

In most of the fiber-optics literature considered here, PCF means photonic crystal fiber. These works use closely related terms such as “photonic crystal fibers (PCF) or air-silica microstructured fibers,” and treat canonical solid-core triangular-lattice fibers, dual-core couplers, hollow-core photonic bandgap fibers, anti-resonant hollow-core fibers, and specialty tellurite or chalcogenide microstructured fibers (Mohammed et al., 2013, Debbal et al., 2015). The guidance mechanisms are not uniform across the class: the analyzed structures include index-guiding solid-core PCFs, hollow-core photonic bandgap guidance, and anti-resonant hollow-core guidance, with coherence questions tied to dispersion engineering, modal coupling, Brillouin noise, or nonlinear phase matching rather than to a single microstructural principle (Zhong et al., 2015, Chen et al., 2020).

A second usage appears in recent quantum-link work, where PCF explicitly means phase coherent fiber. In that sense, PCF is not a microstructured fiber at all, but an actively stabilized optical link distributing nearly monochromatic photons that are ultra-stable in frequency and phase (Johnson et al., 10 Sep 2025). The ambiguity is not merely lexical. One 2014 paper on “Coherency Tuning via Phase Conjugation in Fiber Laser Networks” is relevant to both meanings, because it explicitly treats large-mode-area Yb photonic crystal fiber amplifiers while its central objective is phase-coherent fiber operation in a multi-amplifier Michelson network (Okulov, 2014).

This duality matters because “phase coherent fibers” can therefore refer either to a fiber material/geometry platform whose dispersion, modal content, and nonlinear interactions preserve coherent evolution, or to a system architecture that preserves phase across propagation through otherwise ordinary fiber. The literature covers both senses.

2. Phase-conjugated amplifier arrays and stabilized coherent links

In coherent beam combining of many fiber amplifiers, the central requirement is that relative optical phases be controlled to roughly $\lambda/10$–$\lambda/20$; otherwise the far field breaks up, the central lobe degrades, and coherent summation efficiency falls. A theoretical proposal for a binary-tree Michelson architecture uses stimulated Brillouin scattering (SBS) phase conjugation as a coherence-control element for large-mode-area Yb photonic crystal fiber amplifiers. In that treatment, the SBS frequency shift gives an interferometric phase lag

$$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$

and the choice $\Delta \phi=\pi$ is used for output decoupling. The phase-conjugated return field is written as

$$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$

so that phase piston errors $\Delta\Phi_{m,n}$ acquired on the forward trip are reversed on the backward trip. The same proposal couples this spatial self-adjustment to chirped-pulse amplification, with backward propagation through short Yb fibers providing chirp through both a nonresonant Kerr term and a resonant index term associated with detuned Yb gain (Okulov, 2014).

A related analysis compares a one-way Mach–Zehnder tree with a double-pass Michelson phase-conjugating configuration. Its claim is that, regardless of the number of synchronized fiber amplifiers, the Michelson phase-conjugating interferometer is expected to provide perfect compensation of phase-piston errors and collimation of backwardly amplified beams onto the entrance/output beamsplitter. That work also identifies a limitation that is independent of the spatial phase-locking mechanism: in both architectures, gain saturation can randomize the position of the chirp inside the stretched pulse envelope and reduce interference visibility. Within that model, $sech$-form temporal envelopes are more robust than Gaussian ones because of exponential precursor behavior and self-similar propagation in the gain medium (Okulov, 2013).

In the transmission-link meaning of PCF, active phase stabilization is implemented directly on a 1550 nm, about 1 Hz-linewidth ultra-stable laser distributed through fiber. The stabilized link follows white phase noise limited behavior

$\sigma(\tau)=\sigma_0\tau^{-1},$

with $\sigma_0=1.9(2)\times10^{-16}$ for a 3.3 km field-deployed link and $\sigma_0=2.6(1)\times10^{-16}$ for a 71 km spool. The same system reports up to $\lambda/20$0 dB suppression of phase noise compared to unstabilized fiber, and corrects source-laser drift from $\lambda/20$1 mHz/s to $\lambda/20$2 mHz/s by optical self-referencing and to $\lambda/20$3 mHz/s by absolute optical referencing. In the TF-QKD analysis associated with that system, channel-induced QBER is modeled as

$\lambda/20$4

and the cited phase-error reduction implies a nearly 73-fold reduction of the corresponding QBER term (Johnson et al., 10 Sep 2025).

3. Supermodes, coherent coupling, and phase matching

In dual-core microstructured fibers, coherence is not a secondary effect but the fundamental description of transport. The analyzed structure consists of two nearby identical waveguides embedded in an air-hole lattice, with power transfer mediated by overlap of evanescent tails. The relevant eigenstates are an even and an odd supermode, with propagation constants $\lambda/20$5 and $\lambda/20$6. The coupling length is

$\lambda/20$7

and the coupling coefficient is

$\lambda/20$8

Representative values given at $\lambda/20$9 include $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$0 for $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$1, $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$2, and $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$3 for $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$4, $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$5. The observed beating and crosstalk are thus direct manifestations of coherent supermode propagation rather than incoherent leakage (Debbal et al., 2015).

In dispersion-engineered PCFs, phase matching governs nonlinear coherence. For degenerate four-wave mixing, the relevant constraints are

$$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$6

and

$$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$7

In a homogeneous fiber of length $$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$8, the phasematching function is

$$\Delta \phi = \Delta k\, \Delta L,\qquad \Delta k = k_f-k_b,$$9

That is the baseline from which seeded-FWM characterization, photon-pair state engineering, and inhomogeneity diagnosis are constructed in later work (Francis-Jones et al., 2016).

Phase matching also governs plasmonic sensing in PCF-SPR devices. In the optimized circular-lattice, externally sensed, gold-coated design studied with COMSOL, resonance occurs when the real parts of the effective indices of a selected core mode and an SPP mode are equal,

$\Delta \phi=\pi$0

and the associated confinement-loss peak provides the sensing observable. The reported optimized design gives $\Delta \phi=\pi$1 at $\Delta \phi=\pi$2, confinement loss $\Delta \phi=\pi$3, average sensitivity $\Delta \phi=\pi$4, and average resolution $\Delta \phi=\pi$5 over $\Delta \phi=\pi$6–$\Delta \phi=\pi$7 (Khare et al., 2021).

4. Coherence limits: acoustic noise and longitudinal inhomogeneity

A common misconception is that air guidance suppresses all relevant coherence noise. Measurements on an 8 m hollow-core photonic crystal fiber at 810 nm show the opposite for forward Brillouin-type depolarization noise. Using quantum-noise-limited balanced polarimetry, the study finds multiple excess-noise peaks above shot noise in the 5–35 MHz band, with prominent examples at 12.9 MHz, 23.3 MHz, and 29.5 MHz. The physical explanation is a variant of guided acoustic-wave Brillouin scattering in which thermally excited acoustic modes do not primarily modulate strain in a solid core, but instead modulate the geometry of the photonic crystal structure, thereby changing the effective refractive index and birefringence experienced by the guided air-core mode. The measured and simulated resonance frequencies agree to within $\Delta \phi=\pi$8 MHz, and 13 of 16 simulated modes below 35 MHz were matched to measured features (Zhong et al., 2015).

Longitudinal uniformity is a second major coherence limit. Seeded FWM reconstruction of an 8-ring stack-and-draw PCF shows that structural variation is below $\Delta \phi=\pi$9 and that the characteristic length of the variation is about $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$0. Even at that level, the total phasematching function is not a single sinc lobe but a coherent sum of segment contributions with different local $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$1. The result is extra structure in the joint spectral intensity and degraded suitability for phase-sensitive or spectrally factorable operation over longer lengths (Francis-Jones et al., 2016).

The same problem appears directly in spontaneous four-wave mixing. In an inhomogeneous NL-1050-ZERO-2 PCF, the phase-matching function is written as a coherent segment sum, and the theory predicts broadened, modulated spectra because photon-pair amplitudes generated in different sections interfere rather than add incoherently. Under approximate asymmetric group-velocity matching, the single-photon signal spectrum becomes proportional to $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$2, making individual-arm spectra a practical probe of inhomogeneity. The measured $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$3 values confirm reduced factorability relative to a homogeneous-fiber model; for example, the work reports $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$4 for a 1.0 m section and $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$5 for a 0.6 m section, well below the factorable-state limit $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$6 (Cui et al., 2012).

5. Coherent broadband generation and frequency conversion

In broadband microstructured-fiber sources, phase coherence and amplitude noise are not identical quantities. A suspended-core highly nonlinear fiber driven by an optimized Yb-fiber frequency comb produces a continuum spanning more than 1.5 octaves, from 570 nm to 1.66 $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$7 at the $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$8 dB level, and supports $$E_{PC}(z,t,\vec r)= f(z \pm ct)\,{\bf E_f}^{*}(z,\vec r),$$9 dB heterodyne beats at 698 nm, 771 nm, 1020 nm, 1.54 $\Delta\Phi_{m,n}$0, and on the $\Delta\Phi_{m,n}$1 beat. The measured 1.54 $\Delta\Phi_{m,n}$2 beat remains coherent over more than 200 ms, corresponding to more than $\Delta\Phi_{m,n}$3 pulses at 152 MHz repetition rate. The generalized nonlinear Schrödinger framework uses the first-order coherence

$\Delta\Phi_{m,n}$4

and the principal result is the presence of quantum-seeded broadband amplitude noise without phase coherence degradation. The same study identifies the fractional Raman contribution $\Delta\Phi_{m,n}$5 as an important coherence-control parameter, with the best spectral agreement obtained for $\Delta\Phi_{m,n}$6 (Ruehl et al., 2011).

A later interferometric study of supercontinuum in a polarization-maintaining all-normal-dispersion fiber reaches a related conclusion in a different regime. Here the broadening is mainly SPM-driven and fundamentally deterministic, with measured spectral phase noise of $\Delta\Phi_{m,n}$7–$\Delta\Phi_{m,n}$8 across roughly 700–1400 nm, a baseline of $\Delta\Phi_{m,n}$9 near 1035 nm, and coherence maintained up to 20 nJ launched energy. The measured intensity-to-phase coupling coefficient follows an approximately inverse wavelength law and reaches $sech$0 at 800 nm, implying that pump stability around $sech$1 is required for 100 mrad phase precision. The study therefore distinguishes stochastic incoherence from deterministic intensity-to-phase coupling (Marcadier et al., 13 Nov 2025).

Coherent supercontinuum generation has also been realized in tellurite glass regular-lattice PCFs. The developed air-hole lattice fibers exhibit measured all-normal dispersion profiles as flat as $sech$2 to $sech$3 over 1500–2400 nm, and generate spectra covering 1100–2600 nm under pumping at 1560 nm with 90 fs pulses and peak power below 40 kW. The fiber identified as NL47B1 is described as effectively single-mode at the pump wavelength and is specifically engineered for SPM and optical-wave-breaking broadening rather than anomalous-dispersion soliton dynamics (Klimczak et al., 2019).

Gas-filled hollow-core PCFs provide another route to coherent or coherence-compatible frequency conversion. In a 25-cm He-filled single-ring HC-PCF with first resonance wavelength around 553 nm, photoionization-driven soliton blueshift dynamically moves the pulse toward a resonance-enhanced dispersive-wave phase-matching condition. At 6 $sech$4J input energy, the measured DW spectral ratio reaches $sech$5 and the energy conversion efficiency reaches $sech$6; the soliton centroid shifts to $sech$7 nm while the DW center moves from $sech$8 nm to $sech$9 nm. The work does not directly measure phase coherence, but states that the generated DW is expected to be highly coherent under stable input conditions because it arises from a deterministic phase-matched nonlinear process in a controlled single-mode geometry (Chen et al., 2020).

6. Dispersion engineering, analytical modeling, and automated design

The preservation of phase coherence in PCFs is inseparable from dispersion engineering. One analytical treatment maps a triangular-lattice air–silica PCF to an equivalent step-index fiber with effective core radius $\sigma(\tau)=\sigma_0\tau^{-1},$0 and effective cladding index $\sigma(\tau)=\sigma_0\tau^{-1},$1. For the profiles considered, it identifies

$\sigma(\tau)=\sigma_0\tau^{-1},$2

and uses empirical equations for the normalized parameters $\sigma(\tau)=\sigma_0\tau^{-1},$3 and $\sigma(\tau)=\sigma_0\tau^{-1},$4 to compute $\sigma(\tau)=\sigma_0\tau^{-1},$5 and the fundamental-mode effective index $\sigma(\tau)=\sigma_0\tau^{-1},$6 without full numerical simulation. The quoted errors are less than 3% for the first $\sigma(\tau)=\sigma_0\tau^{-1},$7-fit and less than 1.3% for the improved fit in the stated validity ranges (Mohammed et al., 2013).

A more application-specific example is the heptagonal dispersion-compensating PCF with three inner rings and six outer rings of circular air holes. Using FEM with a circular PML and silica Sellmeier dispersion, the design yields

$\sigma(\tau)=\sigma_0\tau^{-1},$8

dispersion from $\sigma(\tau)=\sigma_0\tau^{-1},$9 to $\sigma_0=1.9(2)\times10^{-16}$0 over 1390–1700 nm, relative dispersion slope $\sigma_0=1.9(2)\times10^{-16}$1, confinement loss about $\sigma_0=1.9(2)\times10^{-16}$2, effective area about $\sigma_0=1.9(2)\times10^{-16}$3, and nonlinearity about $\sigma_0=1.9(2)\times10^{-16}$4. The cited single-mode criterion is $\sigma_0=1.9(2)\times10^{-16}$5. This is a phase-engineering problem in the strict guided-wave sense: the structure is optimized through the wavelength dependence of $\sigma_0=1.9(2)\times10^{-16}$6 and its second derivative rather than through any active stabilization (Mia et al., 2017).

Design workflows have also become algorithmically assisted. In the PCF-SPR sensor study, the computational bottleneck was manual identification of the desired mode among many COMSOL eigenmodes. The authors formulate the task as binary image classification of mode profiles and compare SVM, RF, DT, NB, GB, SGD, AB, and CNN. SVM is selected as the best classifier with $\sigma_0=1.9(2)\times10^{-16}$7 accuracy, $\sigma_0=1.9(2)\times10^{-16}$8 $\sigma_0=1.9(2)\times10^{-16}$9-score, and $\sigma_0=2.6(1)\times10^{-16}$0 MCC, and the reported workflow saves approximately 75 minutes in the overall design process. Although this work targets sensing rather than coherent beam transport, it illustrates a broader transition: phase-matching-based PCF design is increasingly coupled to automated model selection and inverse workflow support (Khare et al., 2021).

7. Optomechanical coherent control and broader extensions

Structured fibers can also be phase-coherent control media for guided acoustic motion. In a 22 cm dual-nanoweb silica fiber, two-frequency optical pumping at the flexural resonance enables coherent control of anti-phase flexural vibrations. In the stimulated Raman-like scattering regime, the effective optical pressure obeys

$\sigma_0=2.6(1)\times10^{-16}$1

so the drive remains essentially unaffected by the phonons it excites. For an abrupt phase step, the vibration minimum occurs at

$\sigma_0=2.6(1)\times10^{-16}$2

and repeated phase switching shorter than the phonon lifetime clamps the vibration to a reduced steady level. The reported suppression reaches about $\sigma_0=2.6(1)\times10^{-16}$3 dB for $\sigma_0=2.6(1)\times10^{-16}$4, with local deflections reduced from hundreds of picometers to near the picometer scale in simulation. The platform is not a PCF in the narrow photonic-crystal sense, but the work explicitly places it in the same broader family of microstructured waveguides supporting Raman-like transverse acoustic resonances (Koehler et al., 2017).

A related engineering direction is broadband mid-IR source optimization in chalcogenide PCFs. Anti-reflective nanoimprinting of a 15 $\sigma_0=2.6(1)\times10^{-16}$5-core $\sigma_0=2.6(1)\times10^{-16}$6 PCF improves total transmission from about $\sigma_0=2.6(1)\times10^{-16}$7 to $\sigma_0=2.6(1)\times10^{-16}$8, and tapering the same platform to $\sigma_0=2.6(1)\times10^{-16}$9–$\lambda/20$00 core diameters extends the spectrum to nearly $\lambda/20$01. The best explicitly summarized taper result is a spectrum up to about $\lambda/20$02 with 90 mW output power and 25.8 mW above 3.5 $\lambda/20$03. The paper does not present explicit coherence measurements, so its relevance is infrastructural rather than demonstrative: it shows how facet engineering and longitudinal waveguide engineering can enlarge the operating space of broadband PCF sources that may later be used in coherence-sensitive systems (Petersen et al., 2019).

Taken together, these works establish that “Phase Coherent Fibers” is not a single technology but a family of guided-wave strategies for preserving or exploiting optical phase. In photonic crystal and related microstructured fibers, coherence is shaped by supermode splitting, dispersion-engineered phase matching, acoustic perturbations, and longitudinal uniformity. In phase-stabilized transmission links, coherence is enforced by active cancellation of fiber noise and source drift. Across both domains, the decisive technical questions are the same: how phase is accumulated, how it is perturbed, and how faithfully it can be reversed, matched, or measured.