betaFeedback

WorksPhysics and Astronomy

Inline inter-core phase control in multi-core fiber with a commercial polarization controller

Abstract

Inter-core phase control is a prerequisite for interferometry in multi-core fiber (MCF), but no compact inline element for this purpose is presently available, and existing approaches require fanout to single-mode fiber or free-space optics. We show that a commercial inline polarization controller, a low-loss and widely available component, can be operated as an inline inter-core phase shifter. Two-core interferometry on a jacketed four-core patchcord gives a reversible inter-core phase of 0.95±0.020.95 \pm 0.02~rad from a single device, with an operating-point-independent lower bound of 0.380.38~rad, driven through repeated load-unload cycles at a rate set by the actuator. A diametrically compressed disk model is used to predict the operating condition: with the input polarization aligned to a principal stress axis the device imparts a relative phase between cores without rotating the polarization state of either, which we verify by polarimetry on all four cores. The response is polarization dependent, falling to less than 0.580.58~rad for the orthogonal input. The result is a practical, inline route to inter-core phase control and stabilization in all-MCF interferometers, requiring neither fanout to single-mode fiber nor free-space optics.

Introduction

Multi-core optical fibers (MCFs) have emerged as a powerful solution for the escalating demands of both classical and quantum optical communications [1]. Through space-division multiplexing, each core acts as an independent optical channel, increasing aggregate bandwidth while conserving physical footprint [2, 3, 4]. Beyond mere parallelism, it is the inter-core phase stability between core modes that gives the platform its distinctive capabilities for interferometric applications. In classical networks this inter-core phase underlies interferometric core routing and fast core-selective switching [5, 6], key building blocks for the SDM platforms developed to relieve the capacity limits of single-mode fiber [7, 8], and it is the basis of MCF interferometric sensors, in which the cores act as the arms of a compact common-path interferometer so that an external perturbation is transduced into an inter-core phase read out from the interference spectrum [9, 10, 11, 12, 13]. In the quantum domain, MCFs implement programmable multiport beam splitters and interferometers for high-dimensional quantum information processing [14, 15, 16] and support the distribution of high-dimensional quantum states and keys over deployed fiber [17, 18, 19, 20, 21]. All of these applications suffer from the fact that in fiber-based interferometers the phase drifts with temperature and mechanical stress and must be actively controlled [22]. An inline element that imparts a controllable inter-core phase, without removing light from the MCF, would therefore be a valuable tool across all of these settings.

A distinct way in which mechanical effects couple to the guided light is through stress-induced birefringence. A transverse stress makes the fiber locally birefringent and, in an MCF, imparts a phase to each core that depends on its transverse position. This is the basis of a broad class of MCF strain, load, bend, and shape sensors, in which an unknown external stress is reconstructed from the induced birefringence and inter-core phase [23, 24, 25, 26, 27, 28], and of recent modeling of the intrinsic residual stress frozen into multi-core preforms [29]. Mechanical perturbations are also known to couple the cores of an MCF opto-mechanically, and guided acoustic modes launched in one core have been used to modulate the light in a neighboring one [30]. In all of these cases the mechanical perturbation is either unknown and to be reconstructed, as in sensing, or intrinsic to the fiber. Here we invert the picture and ask whether a deliberately applied, externally controlled stress can be used to impose a chosen inter-core phase.

Ideally, the stress should be applied by a simple, inline, plug-and-play device. Inline polarization controllers (IPCs), which apply a localized mechanical squeeze to the jacketed fiber [31, 32], are a particularly good match: they impart a well-defined stress geometry and act without disconnecting the fiber or removing light from the system, requiring no fanout to single-mode fiber or free-space optics. These advantages suit deployed and integrated systems, including all-MCF interferometers, multi-core quantum networks, and inline sensing.

In this paper we test a commercial IPC as an inter-core phase controller on a jacketed fiber. The way we operate ate the device departs from its intended use. An IPC is normally adjusted by rotating the clamped fiber section about the central axis while pressure is applied by the clamp, so that the birefringence axis is steered as the fiber is twisted. Instead, we use the rotational degree of freedom to fix the load axis at a chosen angle relative to the core lattice, which is a meaningful setting only for a multi-core fiber. Then the pressure alone is varied. As a theoretical guide, we adapt a diametrically-compressed-disk stress solution to the MCF geometry and use it to describe both the intra-core birefringence and the inter-core phase imparted by an IPC to a four-core fiber (Section II). Though this simplified model does not predict the magnitude of the inter-core phase, it yields a useful operating condition: when the input polarization is aligned with a principal stress axis, the device acts as a pure inter-core phase shifter, imparting a relative phase between cores without rotating the polarization state of either. In Sec. III we present first a set of polarimetry experiments that qualitatively verify this on all four cores independently. A second experiment measures the inter-core phase directly by two-core interferometry. We obtained close to one radian of reversible, polarization-selectable phase from a single off-the-shelf component.

Inline Polarization Controller and Stress-Induced Birefringence

IPCs are commercial devices that apply mechanical stress to a section of an optical fiber to create optical birefringence. The stress is induced by a contact pad that applies external pressure to the fiber via a screw or actuator. Here we tested a commercially available device from Thorlabs [31], originally designed for single-mode fiber (SMF), applied to a four-core MCF, as illustrated in Figure 1. The IPC device used here is designed to apply pressure to the outside of a 900 μ\mum diameter tight-buffered fiber. The MCF patchcord (Fibercore SM-4C1500(8.0/125)/001) used was composed of four cores placed within a 125 μ\mum diameter cladding material, which is covered by a coating material (Dual Acrylate) reaching 245 μ\mum diameter, all within the buffer material [33].

Figure 1. a) Illustration of an inline polarization controller applied to a single-mode fiber and MCF. b) Different pressure directions applied to the multi-core fiber. Red corresponds to a horizontal force from the right (rh); blue to a diagonal from the right (rd); green to a vertical direction (v); light blue to a diagonal from the left (ld); and pink to a horizontal force from the left (lh).

The IPC was placed a few cm from the connector of the patch cord to minimize any twisting of the fiber, such that the orientation of the cores at the IPC device was very close to that of the FC/PC connector, as shown in Figure 1(b). While the usual use of the device involves applying birefringence and rotating the clamped section of the fiber around the central axis, here we focus instead on the stress-induced birefringence on each core of the MCF, so that the only degrees of freedom are the force applied and the angle of orientation with respect to a certain core geometry.

IPCs change the polarization state of light through stress-induced birefringence produced by mechanical pressure on a section of the fiber [34, 35]. The stress-optic effect produces a birefringence [36]

Δn=C(σ1−σ2).(1)\Delta n = C(\sigma_1-\sigma_2). \tag*{(1)}

Here C=C2−C1C=C_2-C_1 is the relative stress-optic coefficient, the difference of the parallel (C1C_1) and perpendicular (C2C_2) coefficients that set the index change for light polarized along and across a given stress axis. While the birefringence depends only on CC, the absolute phase accumulated by each polarization depends on C1C_1 and C2C_2 separately.

Over a length LL of stressed fiber this gives a retardance (i.e. a differential phase between the slow and fast axes) δ=kLΔn\delta=kL\Delta n, where kk is the wavenumber and LL the length of the region under stress. This retardance δ\delta is precisely the differential phase δ=ϕh−ϕv\delta=\phi_h-\phi_v that appears in the Jones matrix Eq. (12) below. The terms σ1\sigma_1 and σ2\sigma_2 are the principal stresses, corresponding to the eigenvalues of the stress matrix

σ=(σxxσxyσyxσyy).(2)\sigma= \begin{pmatrix} \sigma_{xx} & \sigma_{xy} \\ \sigma_{yx} & \sigma_{yy} \end{pmatrix}. \tag*{(2)}

The eigenvectors of matrix (2) determine the direction of the fast and slow optical axes. Typically, point pressure on a cylinder results in a compressive stress in the direction of the pressure and a tensile stress in the perpendicular direction. In principle, at each transverse point (x,y)(x,y) in the fiber, a corresponding set of principal axes and retardance Δn(x,y)\Delta n(x,y) is obtained.

In search of the simplest model for the stress within the MCF, we ignore the different material characteristics of the jacket, buffer and cladding of the fiber, as well as the actual geometry of the contact points of the IPC (a flat contact pad of ∼15mm\sim15\mathrm{mm} length on one side, and cylindrical section on the other). Rather, we consider that the IPC applies force at two opposite points of the homogeneous fiber, so that we can model its action by considering a diametrically compressed disk (see Ref. [37], or Example 8-10 in Ref. [38]), which gives

σxx=−2fπ((R−y)x2r14+(R+y)x2r24−12R)(3)\sigma_{xx}=-\frac{2f}{\pi}\left(\frac{(R-y)x^2}{r_1^4}+\frac{(R+y)x^2}{r_2^4}-\frac{1}{2R}\right) \tag*{(3)}
σyy=−2fπ((R−y)3r14+(R+y)3r24−12R)(4)\sigma_{yy}=-\frac{2f}{\pi}\left(\frac{(R-y)^3}{r_1^4}+\frac{(R+y)^3}{r_2^4}-\frac{1}{2R}\right) \tag*{(4)}
σxy=2fπ((R−y)2xr14−(R+y)2xr24)(5)\sigma_{xy}=\frac{2f}{\pi}\left(\frac{(R-y)^2x}{r_1^4}-\frac{(R+y)^2x}{r_2^4}\right) \tag*{(5)}

where ff is the applied force per unit length, defined as f=F/Lf=F/L, with FF representing the total applied force and LL the contact length. Here by symmetry we have σyx=σxy\sigma_{yx}=\sigma_{xy} and the distances from the points of contact to point (x,y)(x,y)

r1=x2+(R−y)2,r2=x2+(R+y)2.(6)r_1=\sqrt{x^2+(R-y)^2},\qquad r_2=\sqrt{x^2+(R+y)^2}. \tag*{(6)}

Figure 2 shows the principal axes of compressive stress where the load is applied in the vertical direction. These define the fast and slow axes of birefringence at each point in the fiber. Also shown in the figure are the principal axes corresponding to compressive stress as a function of the position (x,y)(x,y).

Figure 2. Principal axes of compressive stress for an optical fiber with force applied along the vertical (yy) direction. The red circles correspond to positions of the MCF cores. The simplified model assumes a homogeneous fiber material. The arrows are included to aid visualization only.

The resulting effect of the IPC depends greatly upon the geometry of the cores and their radial position r=x2+y2r=\sqrt{x^2+y^2} relative to RR. For the IPC considered here, the dimensions are such that we can consider the small-rr limit r≪Rr\ll R, where the radial position of the cores is much smaller than the pressure point radius RR (r≪Rr\ll R), we expand Eqs. (3)-(5) to leading order in x/Rx/R, y/Ry/R, which gives

σxx=fπR(1−4x2R2)(7)\sigma_{xx}=\frac{f}{\pi R}\left(1-\frac{4x^2}{R^2}\right) \tag*{(7)}
σyy=−fπR(3+4y2R2−8x2R2)(8)\sigma_{yy}=-\frac{f}{\pi R}\left(3+\frac{4y^2}{R^2}-\frac{8x^2}{R^2}\right) \tag*{(8)}
σxy=fπR(8xyR2)(9)\sigma_{xy}=\frac{f}{\pi R}\left(\frac{8xy}{R^2}\right) \tag*{(9)}

where we note that the first correction is second order in r/Rr/R (via x,yx,y). For the jacketed fiber used in this work, R=450 μmR=450\ \mu\mathrm{m} (half the 900 μm900\ \mu\mathrm{m} jacket diameter at the IPC contact point) and r=252≈35 μmr=25\sqrt{2}\approx35\ \mu\mathrm{m} (the distance from the fiber axis to each core, given the 50 μm50\ \mu\mathrm{m} square core spacing [33]). The second-order terms are at most about 2.5%2.5\% of the leading terms at the four core positions, and can be safely ignored. Thus, the stress tensor for all cores reduces to that of a single-mode fiber (SMF) with core at the origin:

σSMF=(fπR00−3fπR)(10)\sigma_{\mathrm{SMF}}= \begin{pmatrix} \frac{f}{\pi R} & 0 \\ 0 & -\frac{3f}{\pi R} \end{pmatrix} \tag*{(10)}

where the eigenvectors correspond to the xx (perpendicular to direction of force) and yy (parallel to direction of force) directions.

Let us consider first that the IPC is aligned to apply force in the vertical (yy) direction. Using (10), the effect of the IPC on the polarization can be described by the Jones matrix

P(0)=(eiϕh00eiϕv)(11)\mathrm{P}(0)= \begin{pmatrix} e^{i\phi_h} & 0 \\ 0 & e^{i\phi_v} \end{pmatrix} \tag*{(11)}

written in the basis defined by the horizontal (HH) and vertical (VV) polarization directions. The phases ϕh\phi_h and ϕv\phi_v depend on the force applied by the IPC. We can apply a rotation matrix to (11) to write the Jones matrix describing the IPC aligned at angle θ\theta, giving:

P(θ,δ)=(eiδcos⁡2θ+sin⁡2θsin⁡θcos⁡θ(eiδ−1)sin⁡θcos⁡θ(eiδ−1)eiδsin⁡2θ+cos⁡2θ)(12)\mathrm{P}(\theta,\delta)= \begin{pmatrix} e^{i\delta}\cos^2\theta+\sin^2\theta& \sin\theta\cos\theta\left(e^{i\delta}-1\right) \\ \sin\theta\cos\theta\left(e^{i\delta}-1\right) & e^{i\delta}\sin^2\theta+\cos^2\theta \end{pmatrix} \tag*{(12)}

where we define the differential phase δ=ϕh−ϕv\delta=\phi_h-\phi_v. Without loss of generality, we use Dirac notation to describe the polarization state of light passing through the fiber. Considering an initial state ∣ψ0⟩|\psi_0\rangle, the final polarization state is

∣ψf⟩=P(θ,δ)∣ψ0⟩.(13)|\psi_f\rangle= \mathrm{P}(\theta,\delta)|\psi_0\rangle. \tag*{(13)}

The output polarization state can be measured experimentally via the Stokes parameters, which are defined in terms of the entries of the density matrix ρ=∣ψf⟩⟨ψf∣\rho= |\psi_f\rangle\langle\psi_f| as

S1=ρ11−ρ22,S2=2Re⁡(ρ12),S3=2Im⁡(ρ21).(14)\begin{aligned} S_1 &= \rho_{11}-\rho_{22},\\ S_2 &= 2\operatorname{Re}(\rho_{12}),\\ S_3 &= 2\operatorname{Im}(\rho_{21}). \tag*{(14)} \end{aligned}

The theoretical evolution of the Stokes vector S⃗=(S1,S2,S3)\vec{S}=(S_1,S_2,S_3) for right-circular and left-circular input polarization states are shown in Figure 4. Expressions for the Stokes parameters for several common input states and IPC orientations are given in Appendix A.

Bloch sphere trajectories of the polarization state for all four MCF cores under IPC action

Figure 4. Bloch sphere trajectories of the polarization state for all four MCF cores under IPC action, for an initial right-circular (R) input polarization (top row) and left-circular (L) polarization (bottom row). The leftmost sphere shows the theoretical prediction based on the small-rr model (§ 2), which is the same for all cores. Colors indicate the IPC orientation: green – vertical (v); blue – diagonal right (rd); light blue – diagonal left (ld); red – horizontal right (rh); pink – horizontal left (lh). Deviations from the theoretical trajectory are attributed primarily to residual fiber-induced polarization rotations in the ∼11\sim11 m of MCF, DEMUX, and SMF after the IPC (see text).

Intra-core and inter-core phases

The birefringent stress described above governs the polarization transformation within a single core. In the MCF, the stress experienced by core jj is the cladding stress field evaluated at the core center

σab(j)≡σab(rj),a,b∈{x,y}.(15)\sigma_{ab}^{(j)} \equiv\sigma_{ab}(\mathbf{r}_j),\qquad a,b\in\{x,y\}. \tag*{(15)}

Let us consider stress applied in the vertical (yy) direction, and that the geometry is such that two separate cores share common principal stress directions. This holds for the small-rr condition, as well as cores aligned along the horizontal or vertical directions in Figure 2. For light polarized along the horizontal (HH, xx) and vertical (VV, yy) direction, the absolute phase accumulated by core jj over the stressed length LL is set by the stress-optic indices,

ϕjh=kL(n0−C1σxx(j)−C2σyy(j)),ϕjv=kL(n0−C1σyy(j)−C2σxx(j)).(16)\begin{aligned} \phi_{jh} &= kL\left(n_0-C_1\sigma_{xx}^{(j)}-C_2\sigma_{yy}^{(j)}\right),\\ \phi_{jv} &= kL\left(n_0-C_1\sigma_{yy}^{(j)}-C_2\sigma_{xx}^{(j)}\right). \tag*{(16)} \end{aligned}

where C1C_1 (C2C_2) is the stress-optic coefficient for light polarized parallel (perpendicular) to a stress axis. The intra-core retardance is the difference of these,

δj=ϕjh−ϕjv=kL(C2−C1)(σxx(j)−σyy(j)),(17)\delta_j=\phi_{jh}-\phi_{jv}=kL(C_2-C_1)\left(\sigma_{xx}^{(j)}-\sigma_{yy}^{(j)}\right), \tag*{(17)}

consistent with Eq. (1). This is the birefringent phase δ\delta that appears in Eq. (12), and can be accessed experimentally by polarimetry on each core (Sec. III`).

When the IPC acts on two cores, the relevant inter-core observable is the difference of absolute phases between cores ii and jj for a fixed input polarization, ϕs=ϕis−ϕjs\phi_s=\phi_{is}-\phi_{js} (s=h,vs=h,v). Writing Δσab=σab(i)−σab(j)\Delta\sigma_{ab}=\sigma_{ab}^{(i)}-\sigma_{ab}^{(j)} for the core-ii/core-jj stress difference, Eq. (16) gives

ϕh=−kL(C1Δσxx+C2Δσyy),ϕv=−kL(C2Δσxx+C1Δσyy).(18)\begin{aligned} \phi_h&=-kL\left(C_1\Delta\sigma_{xx}+C_2\Delta\sigma_{yy}\right),\\ \phi_v&=-kL\left(C_2\Delta\sigma_{xx}+C_1\Delta\sigma_{yy}\right). \tag*{(18)} \end{aligned}

These separate naturally into a polarization-independent (common) part and a polarization-dependent (birefringent) part. Defining Σ=Δσxx+Δσyy\Sigma=\Delta\sigma_{xx}+\Delta\sigma_{yy} and D=Δσxx−ΔσyyD=\Delta\sigma_{xx}-\Delta\sigma_{yy},

ϕh=−kL2([C1+C2]Σ−(C2−C1)D),ϕv=−kL2([C1+C2]Σ+(C2−C1)D).(19)\begin{aligned} \phi_h&=-\frac{kL}{2}\left([C_1+C_2]\Sigma-(C_2-C_1)D\right),\\ \phi_v&=-\frac{kL}{2}\left([C_1+C_2]\Sigma+(C_2-C_1)D\right). \tag*{(19)} \end{aligned}

Equation (19) is the central structural result of the model. The first term is the inter-core phase common for both input polarizations. The second term is the birefringent part, governed by the difference C2−C1C_2-C_1 and by the stress-anisotropy difference DD. The half-sum 12(ϕh+ϕv)=−kL2(C1+C2)Σ\frac{1}{2}(\phi_h+\phi_v)=-\frac{kL}{2}(C_1+C_2)\Sigma isolates the first, and the difference isolates the second,

ϕh−ϕv=kL(C2−C1)D=δ1−δ4.(20)\phi_h-\phi_v=kL(C_2-C_1)D=\delta_1-\delta_4. \tag*{(20)}

The last equality holds for any mechanism, photoelastic or otherwise, provided only that HH and VV are principal axes at both cores. The two observables are complementary: the intra-core retardance δj\delta_j [Eq. (17)] is measured by polarimetry, while the inter-core phase ϕs\phi_s [Eq. (18)] is accessible by two-core interferometry (Sec. III B). Eq. (20) therefore provides an alternative route to probing the relative retardance imparted to different cores.

A remark on the geometry is in order for what follows. For diametric compression along yy, the exact solution Eqs. (3)–(5) gives σxx(0,y)=f/πR\sigma_{xx}(0,y)=f/\pi R independent of yy, σyy(0,y)\sigma_{yy}(0,y) even in yy, and σxy(0,y)=0\sigma_{xy}(0,y)=0. The model predicts identical stress tensors, so Σ=D=0\Sigma=D=0 and both inter-core phases vanish. Thus, measuring any difference in intra-core birefringence or inter-core phase in cores 1 and 4 is an indication of the departure of the real device from the simplistic model.

Experiment

Single-core Polarimetry

The experimental setup for the characterization of the IPC on the four-core MCF is shown in Figure 3. A 1550 nm continuous-wave laser sends light through a single-mode fiber (SMF), which is connected to a fiber (de)multiplexer (DEMUX) to couple the signal into a multi-core fiber (MCF). A polarization controller (PC) is used on the SMF to optimize the input polarization (see below) before coupling the light into free space using an objective lens.

Figure 3. Set-up for the characterization of the polarization in multi-core fiber. PC: polarization controller, SMF: single-mode fiber, MCF: multi-core fiber, DEMUX: fiber (de)multiplexer for single-core SMF to MCF, PBS: polarizing beam splitter, QWP: quarter-wave plate, HWP: half-wave plate.

The free-space section is used to set the input polarization and to guarantee that it is the same for all cores of the fiber. A polarizing beam splitter (PBS) followed by a quarter-wave plate (QWP) and a half-wave plate (HWP) are used to prepare the same polarization state ∣ψin⟩|\psi_{\mathrm{in}}\rangle at the entrance to all MCF cores at the MCF connector. We can thus ignore all polarization transformations occurring before the polarizer, though the fiber PC was used to optimize transmission. After the free-space polarization optics, the light passes through another objective lens which forms an image of the exit face of the previous MCF on a second MCF patchcord (5 m length), as has been done in previous work [15]. The IPC is placed immediately after the FC/PC connector, to ensure that the cores are arranged close to the “+” geometry shown in Figure 1, and to minimize any fiber-induced polarization changes due to propagation in the fiber. The IPC implements the transformation P(θ,δ)P(\theta, \delta) on core jj, where θ\theta is the direction along which the mechanical stress is applied. After the IPC, the light travels through the MCF patchcord, through a DEMUX into a SMF and then to a polarimeter. A 3-paddle polarization controller M is used on the MCF patchcord to correct polarization rotations due to propagation within the MCF. The output polarization state going into the polarimeter can be described by

∣ψ⟩=MjUjP(θ,δ)∣ψin⟩,(21)|\psi\rangle= M_j U_j P(\theta,\delta)|\psi_{\mathrm{in}}\rangle, \tag*{(21)}

where UjU_j is the polarization transformation due to propagation through MCF core jj, as well as the DEMUX and SMF (losses are ignored). Here MjM_j is the transformation due to the PC acting on core jj. Ideally, when testing core jj, we can adjust Mj=Uj−1M_j = U_j^{-1}, so that

∣ψj⟩=P(θ,δ)∣ψin⟩.(22)|\psi_j\rangle= P(\theta,\delta)|\psi_{\mathrm{in}}\rangle. \tag*{(22)}

This was done in an initial polarization calibration stage by preparing several non-orthogonal input states (e.g. HH, VV, RR, LL) and adjusting the PC so that the corresponding states were observed at the polarimeter.

Polarimetry measurements were performed for all cores with the IPC aligned in five different directions: vertical, two horizontal orientations, and two diagonal orientations, as shown in Figure 1(b). A NEMA 17 stepper motor at quarter-microstep resolution (800 pulses per revolution) was coupled to the actuator screw and used to control the force applied by the IPC. The motor was configured to complete one revolution every 10 s. To ensure consistent mechanical pressure during measurements, the number of screw turns was recorded starting from a pre-defined zero position. Polarization changes appear around turn 12, and measurements were conducted from turn 10 to turn 15.5. No further pressure beyond turn 15.5 was applied to the fiber to avoid the risk of damage to the patchcord.

The polarimetry results are shown in Figure 4 for initial right-circular and left-circular polarization states. The Bloch spheres for each figure show the trajectory of the polarization state as the fiber is acted on by the IPC. We also show the trajectory corresponding to the theoretical model based on the small-rr approximation given in Sec. II, which applies to all cores. The agreement is qualitative rather than quantitative. All trajectories run from the circular pole toward the linear states, as the model requires, but they depart visibly from the predicted great circle and from one another, following curves of smaller radius that do not close on the equator at the predicted point. This is due primarily to the fiber-induced polarization rotations that occur in the MCF patch cord, DEMUX and SMF (∼\sim11 m length in total) after the IPC. Though these were minimized using the PC, even small deviations from Mj=Uj−1M_j = U_j^{-1} lead to considerable changes in the trajectory of the polarization state. A secondary contribution might arise from movement of the glass fiber inside the jacket material during the IPC actuation, though we consider this effect to be minimal in comparison to the residual fiber rotations.

The second physically relevant question concerns the magnitude of the retardance in each core. Two additional limitations affect our ability to make a quantitative comparison via single-core polarimetry. First we observed run-to-run variation between separately acquired data sets, even for similar operating conditions. This could be due to hysteresis and mechanical memory of the jacket and/or buffer material. Second, imprecision in determining the zero position of force applied by the tension screw adds additional uncertainty. Thus, for the polarization measurements we can make rough estimate of the magnitude of the retardance, and resort to interferometric measurements for the inter-core comparison.

A linear retarder acts on the Poincaré sphere as a rotation through the retardance δ\delta, and the circular and linear states are separated by 90∘90^\circ, so the fraction of the polar distance traversed gives a rough estimate of the retardance directly off the sphere. Every core in Figure 4 covers at least half of that distance, placing δj≳45∘\delta_j \gtrsim45^\circ for all four. This is a conservative reading: the measured trajectories depart from the great circle expected for an input exactly at the pole, and for a tilted rotation axis a given polar displacement corresponds to a larger retardance than the great-circle reading implies. By comparison, applying the same measure to a pre-onset window of equal length, in which the actuator contacts the jacket but does not yet load the fiber, returns between 0.6∘0.6^\circ and 5∘5^\circ, and no response is seen below turn 12 in any run. Each core therefore acquires on the order of one radian of birefringent phase, and does so only when the actuator is loading the fiber.

Independently, the right- and left-circular input states constitute two measurements of the same intra-core retardance, and they agree to a median of 13∘13^\circ, which is due most likely to the fact that measurements are performed independently. We therefore expect a resolution of no better than ∼15∘\sim15^\circ for retardance differences between cores, which prevents us from predicting any meaningful core-to-core retardance differences using the polarimetry data. Moreover, this prohibits testing any of the reflection symmetry predicted by the simplified model for our specific core geometry, which is most likely violated by the actual asymmetric device. It also prohibits verification of predicted relation of Eq. (20), which requires core-to-core resolution well below the value quoted above. We expect that more quantitative statements could be made using a bare fiber and simultaneous polarimetry on cores.

For the interferometric measurement reported in section III B, it is essential that the IPC act only on the inter-core phase, without rotating the polarization within each core, which would itself modulate the observed fringe. When the input polarization is along a principal stress axis (HH or VV for θ=0\theta= 0) the Jones matrix Eq. (12) is diagonal in the H/VH/V basis, so each core acquires an absolute phase while its polarization state is unchanged. This holds for any value of the retardance, so the operating condition is set by the input polarization alone, and not by the screw position or by the details of the stress model.

not by the screw position or by the details of the stress model. Figure 5 confirms this, with S1≈+1S_1 \approx+1 (HH input) and S1≈−1S_1 \approx-1 (VV input) held constant across the actuator range for both cores 1 and 4. Cores 1 and 4 therefore acquire different absolute phases at fixed polarization, and the resulting inter-core phase is read out directly in the output intensities. The residual polarization rotations downstream of the IPC, which are responsible for the departure of the trajectories in Figure 4 from the model, do not affect this condition. Because the interferometer contains no polarizing element and these rotations are static, they can at most change the static fringe visibility, which is included in the fringe range 2I02I_0 measured with the motor off (Sec. III B).

Figure 5. Stokes parameter S1S_1 as a function of screw step for core 1 (Brown) and core 4 (Blue), with horizontal and vertical input polarization under vertical IPC pressure. In both cases S1S_1 remains nearly constant, confirming that the IPC acts as a pure phase retarder when the input polarization is aligned with the principal stress axis. Discontinuities are due to manual (non-motorized) application of pressure during these particular measurements.

Two-core Interferometry

To probe the inter-core phase directly, a two-core interferometer was constructed, as shown in Figure 6. The setup is similar to Figure 3, but with a 2×22 \times2 fiber beam splitter (BS) used to divide the incoming laser light into cores 1 and 4, and a multi-core fiber beam splitter (MBS) [22] used to recombine the light, which was then sent into the DEMUX and SMFs. Two single-photon detectors (ID Qube series, efficiency 10%, dead time 5 μs5\,\mu\mathrm{s}) were used to record photocounts at the two output ports of the MBS while force from the IPC was applied. The use of single-photon detectors required attenuation of the input laser, and before each measurement run the total count rates were set to approximately 30 000 counts per second.2 The inter-core phases ϕh\phi_h and ϕv\phi_v for H and V input polarizations can then be extracted from the fringe data.

Figure 6. Experimental setup for the two-core interferometric phase measurements. PC: polarization controller, SMF: single-mode fiber, MCF: multi-core fiber, DEMUX: fiber (de)multiplexer, PBS: polarizing beam splitter, QWP: quarter-wave plate, HWP: half-wave plate, BS: 2×22 \times2 fiber beam splitter, MBS: multi-core fiber beam splitter.

Figure 8. Same as Fig. 7 but with horizontal input polarization. The oscillation amplitude is observably smaller than for V polarization (compare Fig. 7), which also makes the phase drift more noticeable.

Time-domain oscillations

Figures 8 and 7 show representative time traces of the photon counts at detectors D1 and D2 for H and V input polarizations, respectively. In both cases, oscillations are visible and the two detectors are anti-correlated, as expected. These anti-correlated oscillations are interference fringes between the light from cores 1 and 4 at the MBS, driven by the IPC-induced inter-core phase. The oscillation amplitude is consistently larger for V than for H input. The IPC screw is ramped in and out over each 60 s cycle and the fringe follows at fmotf_{\mathrm{mot}} over several successive cycles, so the response is parabolic in shape resulting from the repeated ramp-in/ramp-out cycle.

Inspection of the time traces shows that successive fringe extrema grow slowly over a run, with later cycles reaching larger amplitudes than earlier ones, which is consistent with viscoelastic settling of the polymer jacket and coating under repeated load. Mechanical memory of this kind is expected for a polymer-buffered fiber and we have not characterized it here. It is, however, largely irrelevant if the IPC is used for stabilization of the sub-hertz inter-core phase drift of an all-MCF interferometer, where the actuator would be driven by a feedback loop tracking an error signal. With regards to reproducibility, the observed modulation frequency followed the actuator frequency at all three motor speeds tested (see below). In the polarimetry runs, no response was seen below turn 12. In contrast, the screw position at which the response begins is not reproducible, because the force zero cannot be set precisely, and the fringe extrema grow slowly over a run. We therefore do not claim open-loop repeatability of the phase. The intended closed-loop use requires only that the inter-core phase increase monotonically with the applied force, which is consistent with the parabolic fringe traces of Figure 7.

Figure 7. Photon count rates at detectors D1 (blue) and D2 (orange) as a function of time, with vertical input polarization in cores 1 and 4 and the IPC applying vertical pressure (motorized actuator cycling through three revolutions/30s and back). The anti-correlated oscillations between the two output ports of the MBS are the interference fringes produced by the IPC-induced inter-core phase modulation. Deviations are due to phase drift from ambient effects.

Visible in the time traces is also the irregular background modulation in the count rates due to the phase instability introduced by the 2×22 \times2 SMF BS that splits the laser beam before coupling into cores 1 and 4, which occurs on timescales comparable to the IPC cycling period. To separate the IPC-driven signal from this background, we performed a spectral analysis using the fast Fourier transform.

Frequency analysis and phase extraction

Figure 9 shows the power spectra obtained via Fast Fourier Transform (FFT) of the time traces averaged over all repeated runs for each polarization. A clear peak appears near the expected motor frequency fmot=1/60 Hz≈0.016 Hzf_{\mathrm{mot}} = 1/60\ \mathrm{Hz} \approx0.016\ \mathrm{Hz} for V input, and a smaller peak for H input. The slight shift from fmotf_{\mathrm{mot}} to the slightly higher ∼0.018\sim0.018 is most likely due to errors from our homemade stepper motor controller. A small residual drift of the operating phase point φ0\varphi_0 over the measurement most likely contributes to the peak width through a weak amplitude modulation of the fringe, though φ0\varphi_0 was approximately constant over the few cycles used here. The peak amplitude provides a cleaner, more noise-robust estimate of the IPC-induced fringe amplitude AA, separated from the low-frequency phase noise previously mentioned. The spectra sit on a sloped background that is largest at low frequency and falls toward higher frequency, the signature of the slow environmental phase drift in the separate SMFs before the first DMUX. The V-input peak stands well above this background and its height is consistent with the oscillation amplitude seen directly in the time trace, whereas the H-input peak is comparable to the local background level, so the H amplitude is best regarded as an upper bound (see below).

Power spectra of photon-count oscillations for detectors D1 and D2

Figure 9. Power spectrum of the photon-count oscillations at detectors D1 (top figure) and D2 (bottom figure), averaged over all runs, for V input (solid lines) and H input (dashed lines). The expected motor frequency fmot=1/60 Hz≈0.016 Hzf_{\mathrm{mot}} = 1/60\ \mathrm{Hz} \approx0.016\ \mathrm{Hz} (one full screw load/unload cycle in 60 s) is indicated by the vertical red dashed line. The V-polarization spectrum shows a clear peak near fmotf_{\mathrm{mot}}, while the H-polarization peak is barely above the natural phase fluctuations of the interferometer.

To confirm that the modulation is driven by the IPC rather than by ambient drift, we repeated the measurement at different motor speeds. Cycling the screw at 5, 10, and 15 s/revolution moves the drive frequency in proportion, and the fringe peak follows: the 15 s/rev data peak near 0.013 Hz (expected 1/90 Hz ≈0.011\approx0.011 Hz) and the 5 s/rev data near 0.038 Hz (expected 1/30 Hz ≈0.033\approx0.033 Hz). The peaks therefore lie 8 to 17% above the corresponding fmotf_{\mathrm{mot}}, including the 10 s/rev data, a shift that we attribute to the stepper-motor controller. The spectral analysis of the phase modulation is consistent across different motor speeds, and the fiber showed no physical damage after multiple measurements. The 10 s/rev rate used for the data reported here gave the best tradeoff between the fiber’s limited mechanical bandwidth at faster rates and increasing phase-drift contamination at slower rates.

To extract the inter-core phase difference from the fringe amplitude, we use the two-port interference relation for a single detector,

I=Iˉ+I0cos⁡φ,(23)I = \bar{I} + I_0 \cos\varphi, \tag*{(23)}

where Iˉ\bar{I} is the mean count rate, I0I_0 the fringe amplitude (so the full fringe peak-to-peak is 2I02I_0), and φ\varphi the inter-core phase difference. Before each run the static phase was set to an easily reproducible operating point by adjusting the input BS fibers to minimize the count rate at one detector, corresponding to φ0≈π\varphi_0 \approx\pi. As the IPC screw drives the inter-core phase from 0 to δφ\delta\varphi, the count rate responds quadratically, as can be seen in Figure 7. The peak-to-peak modulation of the count rate is

G=I0(1−cos⁡δφ),(24)G = I_0 \left(1-\cos\delta\varphi\right), \tag*{(24)}

where 2I02I_0 is the full fringe peak-to-peak count range, obtained with the motor off while the environmental phase drifts freely over several oscillation periods, and reported in Table I.

D1D2
Vertical (V)
A1A_1 (counts)2063.5±66.62063.5 \pm66.61534.5±78.81534.5 \pm78.8
2I02I_0 (counts)23733.3±185.223733.3 \pm185.219939.6±176.319939.6 \pm176.3
δφ\delta\varphi (deg)54.7±1.054.7 \pm1.051.3±1.451.3 \pm1.4
δφ\delta\varphi (rad)0.95±0.020.95 \pm0.020.90±0.020.90 \pm0.02
δφmin⁡\delta\varphi_{\min} (deg)24.4±0.424.4 \pm0.421.6±0.621.6 \pm0.6
Horizontal (H), upper bounds
A1A_1 (counts)≤794.0±88.5\le794.0 \pm88.5≤629.7±58.9\le629.7 \pm58.9
2I02I_0 (counts)23924.0±187.523924.0 \pm187.518079.8±170.818079.8 \pm170.8
δφ\delta\varphi (deg)≤33.2±1.9\le33.2 \pm1.9≤34.0±1.6\le34.0 \pm1.6
δφ\delta\varphi (rad)≤0.58\le0.58≤0.59\le0.59
δφmin⁡\delta\varphi_{\min} (deg)≤9.3±0.5\le9.3 \pm0.5≤9.7±0.5\le9.7 \pm0.5

Table I. TABLE I: inter-core phase modulation depths extracted from the mean of three independent experimental runs, for the two detectors (D1, D2) and both input polarizations (V, H). A1A_1 is the amplitude of the spectral peak at fmotf_{\mathrm{mot}} and 2I02I_0 is the full fringe interference range. The value δφ\delta\varphi follows from Eq. (25) and the operating-point-independent lower bound δφmin⁡\delta\varphi_{\min} from Eq. (26). Uncertainties are statistical, propagated from A1A_1 and 2I02I_0. The HH entries are background-limited and thus are taken as upper bounds.

(24) relates the inter-core phase to GG, the peak-to-peak modulation of the count rate driven by the IPC. However, the spectral analysis gives A1A_1, the amplitude of the Fourier component at fmotf_{\mathrm{mot}}, and the two differ by a factor that depends on the shape of the driven waveform. The IPC pad advances and retracts linearly in time, so the inter-core phase follows a triangular ramp between 0 and δφ\delta\varphi. At the operating point φ0=π\varphi_0=\pi, the count rate responds quadratically, giving a parabolic waveform. Computing its Fourier fundamental gives

A1=I02δφsin⁡δφπ2−δφ2,(25)A_1 = I_0 \frac{2\delta\varphi\sin\delta\varphi}{\pi^2-\delta\varphi^2}, \tag*{(25)}

which we solve numerically for δφ\delta\varphi. In the small-phase limit (25) reduces to A1=(4/π2)G≃0.405GA_1=(4/\pi^2)G\simeq0.405G, and it departs from that limit by less than 1.5% over the range of phases measured here, so the ratio A1/GA_1/G is essentially constant.

(24) assumes that the background phase φ0\varphi_0 is a multiple of π\pi. A conservative bound that is independent of the operating point follows by noting that, for a fixed count difference GG, the smallest possible phase modulation occurs at φ0=π/2\varphi_0=\pi/2, giving

δφ≥δφmin⁡=2arcsin⁡(G2I0),(26)\delta\varphi\ge\delta\varphi_{\min}=2\arcsin\left(\frac{G}{2I_0}\right), \tag*{(26)}

with G=A1π2/4G=A_1\pi^2/4 from the shape factor above. Equations (25) and (26) are applied independently to each detector and each input polarization.

The resulting inter-core phase modulation depths are summarized in Table I. For VV input the extraction gives δφ=0.95\delta\varphi=0.95 and 0.90 rad at the two detectors (54.7° and 51.3°), with an operating-point-independent lower bound above 0.38 rad. The values of δφ\delta\varphi assume that the static phase was set to φ0≈π\varphi_0\approx\pi before each run. For HH input the values are 0.58 and 0.59 rad. The HH spectral peak, however, sits at a height comparable to the local drift background (Figure 9), so A1A_1 for HH is biased upward by the background and the HH entries must be read as upper bounds rather than as measurements with the quoted precision.

We note that polarization modulation from the IPC cannot account for the observed modulation for VV input. The interferometer contains no polarizing element, so a mismatch between the polarization states of the two cores can only reduce the fringe visibility, V=12(1+S⃗(1)⋅S⃗(4))=∣cos⁡(Θ/2)∣\mathcal{V}=\sqrt{\frac{1}{2}\left(1+\vec{S}^{(1)}\cdot\vec{S}^{(4)}\right)}=\lvert\cos(\Theta/2)\rvert, with Θ\Theta the angle between the two Stokes vectors. At an extremum a change in visibility displaces the count rate toward the mean by (1−V)I0(1-\mathcal{V})I_0. The polarimetry of Figure 5 gives ∣S1∣≥0.95\lvert S_1\rvert\ge0.95 for each core throughout the actuation. Assuming the extreme case of opposite rotation (which is contrary to the common evolution seen in Figure 4), gives Θ≤36∘\Theta\le36^\circ and Vmin⁡=0.95\mathcal{V}_{\min}=0.95, and a bound of 0.05I00.05I_0 on any purely polarization-driven effect. For VV input the driven modulation depth is G/I0=0.43G/I_0=0.43, which exceeds this worst-case bound by a factor of 8.6. For HH input, G/I0≤0.16G/I_0\le0.16 is itself only an upper bound, dominated by the background, so the comparison is not meaningful and we use the HH entries of Table I only as upper bounds on the inter-core phase.

As a final, preliminary test we applied two IPCs in series, modulated in unison. The fringe amplitude grew: from the VV-polarization time trace we obtain A1≈3000A_1\approx3000 and a full peak-to-peak 2I0≈160002I_0\approx16000 counts, giving δφ≈1.5\delta\varphi\approx1.5 rad (≈84∘\approx84^\circ), roughly 1.5 times the single-IPC value. This measurement was considerably noisier than the single-IPC data, plausibly from slight asynchrony between the two independently driven actuators which were placed side by side. We therefore present it only as preliminary evidence that the phase accumulates across devices. In principle, there is no reason why concatenation of IPC devices should not increase the imprinted phase, so that phases of π\pi or more may be within reach.

Physical interpretation and limits of the model

The polarimetric and interferometric measurements access complementary information. The dominant, polarization-independent part of the inter-core phase is governed by Σ\Sigma, the difference in mean transverse stress between the cores. A term of this form shifts nhn_h and nvn_v together at a given core, so it produces no retardance for any input polarization and is inaccessible to polarimetry.

It is worth being explicit about what the appearance of an inter-core phase implies. For diametric compression along yy, the solution of § 2 gives σxx(0,y)\sigma_{xx}(0,y) independent of yy, σyy(0,y)\sigma_{yy}(0,y) even in yy, and σxy(0,y)=0\sigma_{xy}(0,y)=0, so cores 1 and 4 see identical stress tensors and the inter-core phase vanishes. This is exact, to all orders in r/Rr/R, and follows from the mirror symmetry of the idealized loading rather than from any approximation. The measurement is therefore a null test of that idealization, and the ≈1\approx1 rad phase difference observed for V polarization is a direct measure of the departure of the real device, a flat pad bearing on one side of a jacketed fiber seated in a cylindrical bore, from the symmetric configuration assumed in the model. Characterizing that geometry is a separate problem from the one addressed here.

Beyond the model, the measured ordering of the phases may give some insight into the physical origin of the observed phase difference. Squaring and subtracting Eqs. (18), the cross terms cancel and give the exact relation

φv2−φh2=(kL)2(C22−C12)(Δσxx2−Δσyy2).(27)\varphi_v^2-\varphi_h^2=(kL)^2\left(C_2^2-C_1^2\right)\left(\Delta\sigma_{xx}^2-\Delta\sigma_{yy}^2\right). \tag*{(27)}

Since ∣C2∣>∣C1∣\lvert C_2\rvert>\lvert C_1\rvert [C1=0.65×10−12 Pa−1C_1=0.65\times10^{-12}\ \mathrm{Pa}^{-1} and C2=4.2×10−12 Pa−1C_2=4.2\times10^{-12}\ \mathrm{Pa}^{-1} [29]], the observation ∣φv∣>∣φh∣\lvert\varphi_v\rvert>\lvert\varphi_h\rvert implies ∣Δσxx∣>∣Δσyy∣\lvert\Delta\sigma_{xx}\rvert>\lvert\Delta\sigma_{yy}\rvert, so that the inter-core stress difference is greater in the direction transverse to the loading axis than along it. This conclusion rests on three assumptions: the inter-core phase is entirely photoelastic (so that polarization-independent contributions, such as a path-length difference due to bending, are absent), that both cores experience the same stressed length LL, and that HH and VV are principal stress axes at both cores. The noise limit of the HH channel does not weaken the ordering, provided that the HH and VV runs were taken with the same static phase φ0≈π\varphi_0\approx\pi, and because the background can only bias the HH amplitude upward. Given these assumptions and the limited precision of the HH data, this interpretation should be regarded as indicative only, and a more precise measurement of φh\varphi_h would be required to confirm it.

Finally, one consequence matters directly for use of the device. Writing the measured phases as a common part and a polarization-dependent part, and assuming that φh\varphi_h and φv\varphi_v have the same sign, the latter amounts to (φv−φh)/(φv+φh)≈0.2(\varphi_v-\varphi_h)/(\varphi_v+\varphi_h)\approx0.2 of the former (0.24 and 0.21 for detectors D1 and D2, respectively). Because φh\varphi_h is only an upper bound, this ratio is a lower bound on the polarization-dependent fraction. For VV input, at most about 80% of the imparted phase is therefore independent of the input polarization. The IPC is thus not fully polarization independent, and the true polarization dependence may be larger than estimated here. In an application where the input polarization is neither known nor fixed, this residual dependence sets the accuracy with which the inter-core phase can be controlled. Operating with the input aligned to a principal axis, as here, removes the ambiguity entirely.

Conclusions

We have shown that a commercial inline polarization controller functions as an inline inter-core phase shifter for multi-core fiber. Two-core interferometry gives a reversible inter-core phase of 0.95±0.020.95\pm0.02 rad for vertical input polarization, with an operating-point-independent lower bound of 0.380.38 rad, and at most 0.590.59 rad for horizontal input. The screw is ramped in and out over each cycle and the fringe follows the drive frequency across several successive cycles and across three actuation rates.

We emphasize that here we demonstrate the repurposing of a device for phase control rather than characterization of stress-induced retardance in the MCF. An IPC is normally adjusted by rotating the clamped section about the fiber axis while squeezing, so that the birefringence axis is steered and the fiber twisted. Here the rotational degree of freedom instead fixes the load axis at a chosen angle relative to the core lattice, a setting that has no meaning for single-mode fiber, and the force alone is varied. The result is an off-the-shelf, low-loss, inline component that delivers close to a radian of inter-core phase without fanout to SMF or free-space optics. We observe mechanical memory effects, even under sub-hertz modulation. However, as the intended application is not high-speed modulation but stabilization of the slow, sub-hertz inter-core phase drift of an all-MCF interferometer, any slowly varying mechanical response can be absorbed by a tracking loop, provided that the inter-core phase increases monotonically with the applied force.

A simple diametrically compressed disk model is used to predict that for an input polarization along a principal stress axis determined by the Jones matrix, each core acquires an absolute phase at fixed polarization and the device acts as a pure inter-core phase shifter. The model ignores key characteristics of the IPC and fiber, and is used as a guide only to identify the operating conditions. Indeed, at leading order it predicts no inter-core phase at all between the pair of cores measured. Thus the observed signal is a direct indication of the deviation of the real contact geometry from the simplified model. We expect that each device and mounting will behave somewhat differently, similar to the paddle polarization controllers typically used in fiber optics.

There are several avenues open for future work. The phase range can plausibly be extended: the screw travel limit was not reached in any measurement, and preliminary data with two devices in series reach ≈1.5\approx1.5 rad. Faster actuation would require replacing the manual IPC with a piezoelectric or electro-optic polarization squeezer while retaining the geometry. Also interesting is to apply the same load to a bare, unjacketed MCF, where the stress components then vary appreciably across the core positions, and the Jones matrix becomes core dependent. This is the regime already exploited for multi-axis load and bend sensing with Bragg gratings in bare MCF [23, 25]. Used as an actuator rather than a sensor, and with the load applied at an angle neither along nor perpendicular to the core axes, it would couple the spatial and polarization degrees of freedom of the guided light within the fiber itself, opening a route to combined path-polarization operations in an all-fiber geometry.

Stokes parameters for special cases

The Stokes parameters (14) follow directly from Eqs. (12) and (13) for arbitrary θ\theta and δ\delta. Table 2 collects the results for the six cardinal input states at the IPC orientations used in this work. For θ=0\theta= 0 the first two rows confirm that HH and VV are eigenstates of P(0,δ)P(0,\delta) and remain unchanged under vertical pressure for any δ\delta, which is the pure-phase-shifter condition exploited in Sec. III B. The remaining rows show that the diagonal and circular inputs evolve with a sensitivity that grows with δ\delta, which is why circular input is used for the polarimetry of Figure 4.

Statesθ=0\theta= 0θ=π/2\theta= \pi/2θ=π/4\theta= \pi/4θ=−π/4\theta= -\pi/4
(H)(1,0,0)(1, 0, 0)(1,0,0)(1, 0, 0)(cos⁡δ,0,−sin⁡δ)(\cos\delta, 0, -\sin\delta)(cos⁡δ,0,sin⁡δ)(\cos\delta, 0, \sin\delta)
(V)(−1,0,0)(-1, 0, 0)(−1,0,0)(-1, 0, 0)(−cos⁡δ,0,sin⁡δ)(-\cos\delta, 0, \sin\delta)(−cos⁡δ,0,−sin⁡δ)(-\cos\delta, 0, -\sin\delta)
(D)(0,cos⁡δ,−sin⁡δ)(0, \cos\delta, -\sin\delta)(0,cos⁡δ,sin⁡δ)(0, \cos\delta, \sin\delta)(0,1,0)(0, 1, 0)(0,1,0)(0, 1, 0)
(A)(0,−cos⁡δ,sin⁡δ)(0, -\cos\delta, \sin\delta)(0,−cos⁡δ,−sin⁡δ)(0, -\cos\delta, -\sin\delta)(0,−1,0)(0, -1, 0)(0,−1,0)(0, -1, 0)
(R)(0,sin⁡δ,cos⁡δ)(0, \sin\delta, \cos\delta)(0,−sin⁡δ,cos⁡δ)(0, -\sin\delta, \cos\delta)(sin⁡δ,0,cos⁡δ)(\sin\delta, 0, \cos\delta)(−sin⁡δ,0,cos⁡δ)(-\sin\delta, 0, \cos\delta)
(L)(0,−sin⁡δ,−cos⁡δ)(0, -\sin\delta, -\cos\delta)(0,sin⁡δ,−cos⁡δ)(0, \sin\delta, -\cos\delta)(−sin⁡δ,0,−cos⁡δ)(-\sin\delta, 0, -\cos\delta)(sin⁡δ,0,−cos⁡δ)(\sin\delta, 0, -\cos\delta)

Table 2. Theoretical Stokes vectors (S1,S2,S3)(S_1, S_2, S_3) as a function of the IPC orientation angle θ\theta, for horizontal (H), vertical (V), diagonal (D), anti-diagonal (A), right-circular (R), and left-circular (L) input polarization.

Acknowledgments

This research was funded by ANID FONDECYT-Regular Grants No. (1230796, 1231940, 1230897, 1240746, 1260111), ANID Millennium Science Initiative Program ICN17−012, ANID Anillo Project ATE250003, and ANID AC3E CIA250006.

References

  1. [1]G. B. Xavier and G. Lima, Communications Physics 3, 9 (2020).
  2. [2]D. J. Richardson, Science 330, 327 (2010), https://science.sciencemag.org/content/330/6002/327.full.pdf.
  3. [3]D. J. Richardson, J. M. Fini, and L. E. Nelson, Nature Photonics 7, 354 (2013).
  4. [4]B. J. Puttnam, G. Rademacher, and R. S. Luís, Optica 8, 1186 (2021).
  5. [5]C. Melo, M. Reyes F, D. Arroyo, E. S. Gómez, S. P. Walborn, G. Lima, M. Figueroa, J. Cariñé, and G. Saavedra, Communications Engineering 4, 77 (2025).
  6. [6]C. Deakin, M. Enrico, N. Parsons, and G. Zervas, Journal of Lightwave Technology 37, 1954 (2019).
  7. [7]P. J. Winzer and D. T. Neilson, J. Lightwave Technol. 35, 1099 (2017).
  8. [8]K. Saito and S. Matsuo, Journal of Lightwave Technology 34, 55 (2016).
  9. [9]Y. Yao, Z. Zhao, and M. Tang, Sensors 23, 3436 (2023).
  10. [10]Z. Zhao, M. Tang, and C. Lu, Opto-Electronic Advances 3, 190024 (2020).
  11. [11]J. Song, S. Sun, C. Jiang, H. Chen, X. Zhu, J. Ren, and S. Wang, Optical Fiber Technology 71, 103296 (2023).
  12. [12]L. Suo, Y.-P. Peng, M. Wang, W.-Z. Zhao, and L.-C. Gao, Sensors (Basel) 22, 5630 (2022).
  13. [13]N. Cuando-Espitia, M. A. Fuentes-Fuentes, A. Velázquez-Benítez, R. Amezcua, J. Hernández-Cordero, and D. May-Arrioja, Scientific Reports 11, 18345 (2021).
  14. [14]J. Cariñé, M. N. Asan-Srain, G. Lima, and S. P. Walborn, npj Quantum Information 7, 172 (2021).
  15. [15]M. M. Taddei, J. Cariñé, D. Martínez, T. García, N. Guerrero, A. A. Abbott, M. Araújo, C. Branciard, E. S. Gómez, S. P. Walborn, L. Aolita, and G. Lima, PRX Quantum 2, 010320 (2021).
  16. [16]D. Martínez, E. S. Gómez, J. Cariñé, L. Pereira, A. Delgado, S. P. Walborn, A. Tavakoli, and G. Lima, Nat. Phys. 10.1038/s41567-022-01845-5 (2022).
  17. [17]B. D. Lio, L. K. Oxenlowe, D. Bacco, D. Cozzolino, N. Biagi, T. N. Arge, E. Larsen, K. Rottwitt, Y. Ding, and A. Zavatta, IEEE J. Sel. Top. Quantum Electron. 26, 6400108 (2020).
  18. [18]E. Gómez, S. Gómez, I. Machuca, A. Cabello, S. Pádua, S. Walborn, and G. Lima, Phys. Rev. Applied 15, 034024 (2021).
  19. [19]Q. Wu, C. Wang, H. Zhang, Z. Yu, Y. Li, Y. Guo, H. Wang, D. Zhang, T. Wang, Z. Chen, and X. Chen, Light: Science & Applications 13, 19 (2025).
  20. [20]W. Xiong, C. W. Hsu, Y. Bromberg, H. M. Oz, N. M. Lerman, S. J. B. Chang, and H. Cao, Light: Science & Applications 7, 47 (2018).
  21. [21]G. H. dos Santos, K. B. Sawada, N. Villalba, C. Jara, N. Guerrero, C. Melo, M. H. Magiotto, D. Martínez, G. B. Xavier, J. Cariñé, G. Saavedra, E. S. Gómez, S. P. Walborn, and G. Lima, High-dimensional quantum key distribution via full core-mode encoding over deployed multicore fibers (2026), arXiv:2606.04211 [quant-ph].
  22. [22]J. Cariñé, G. Cañas, P. Skrzypczyk, I. Šupić, N. Guerrero, T. Garcia, L. Pereira, M. A. S. Prosser, G. B. Xavier, A. Delgado, S. P. Walborn, D. Cavalcanti, and G. Lima, Optica 7, 542 (2020).
  23. [23]M. Silva-López, C. Li, W. N. MacPherson, A. J. Moore, J. S. Barton, J. D. C. Jones, D. Zhao, L. Zhang, and I. Bennion, Optics Letters 29, 2225 (2004).
  24. [24]M. Silva-López, W. N. MacPherson, C. Li, A. J. Moore, J. S. Barton, J. D. C. Jones, D. Zhao, L. Zhang, and I. Bennion, Applied Optics 44, 6890 (2005).
  25. [25]G. M. H. Flockhart, W. N. MacPherson, J. S. Barton, J. D. C. Jones, L. Zhang, and I. Bennion, Opt. Lett. 28, 387 (2003).
  26. [26]J. Villatoro, A. Van Newkirk, E. Antonio-Lopez, J. Zubia, A. Schülzgen, and R. Amezcua-Correa, Opt. Lett. 41, 832 (2016).
  27. [27]J. P. Moore and M. D. Rogge, Opt. Express 20, 2967 (2012).
  28. [28]Z. Zhao, Y. Dang, and M. Tang, Photonics 9, 381 (2022).
  29. [29]G. Ocampo and K. Saitoh, Opt. Express 33, 16052 (2025).
  30. [30]H. H. Diamandi, Y. London, and A. Zadok, Optica 4, 289 (2017).
  31. [31]Thorlabs, Inc., In-line fiber optic polarization controllers, https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=2161 (2014), last edited March 5, 2014; accessed June 8, 2025.
  32. [32]OptoSigma, In-line fiber polarization controllers / ipc-250 (2024).
  33. [33]Fibercore, SM-4C1500(8.0/125)/001 Multicore Fiber Product Page, https://fibercore.humaneticsgroup.com/products/multicore-fiber/multicore-fiber/sm-4c150080125001 (2026), accessed: 2026-05-11.
  34. [34]M. Johnson, Applied Optics 18, 1288 (1979).
  35. [35]A. M. Smith, Electronics Letters 16, 773 (1980).
  36. [36]M. M. Frocht, Photoelasticity, Vol. 1 (John Wiley & Sons, New York, 1941).
  37. [37]P. D. Giannino and B. Bendow, Appl. Opt. 20, 430 (1981).
  38. [38]M. H. Sadd, Elasticity: Theory, Applications, and Numerics, 1st ed. (Elsevier Academic Press, 2005).

Paper details

vlnc:W_qz54pysn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents