Polarization Sagnac Interferometer

• Polarization Sagnac interferometer is a common-loop design that splits orthogonal polarization modes to encode relative phase information for applications like entangled-photon generation.
• It utilizes polarizing beam splitters, wave plates, and nonlinear crystals to create counter-propagating paths that cancel common-mode noise and provide passive phase stability.
• The architecture supports diverse applications from fiber-optic sensors and nonlocal interference to gravitational-wave detection, underscoring its versatility in precision optical measurements.

A polarization Sagnac interferometer is a common-loop interferometric architecture in which orthogonal polarization components, or equivalently opposite helicity modes, counter-propagate through substantially the same optical system and are recombined so that the measurable signal is encoded in a relative phase, a polarization rotation, or a two-photon interference term. In practice the loop is typically organized around a polarizing beam splitter or dual-wavelength polarizing beam splitter, with wave plates, Faraday rotators, nonlinear crystals, cavities, or electro-optic modulators inserted so that the two circulations experience the same bulk optics in opposite order. This geometry is used for polarization-entangled photon sources, nonlocal interferometers, fiber and free-space sensors, polarization-independent switching, nanoradian rotation metrology, and polarisation Sagnac speedmeters for gravitational-wave detection (Li et al., 2015, Williams et al., 2014, Alarcón et al., 2020, Spencer et al., 2021).

1. Canonical topology and polarization routing

The defining operation is polarization multiplexing of the clockwise and counter-clockwise paths. In a standard entanglement source, a diagonally polarized pump is separated by a PBS or DPBS so that $H$ propagates clockwise and $V$ propagates counter-clockwise; a DHWP or related wave-plate sequence rotates one circulation so that both directions interact with the same phase-matching configuration before recombination at the same splitter port. In the telecom-band type-II source, a continuous-wave Ti:Sapphire laser at $775\,\mathrm{nm}$ is sent into a DPBS, the $H$-polarized pump travels clockwise, the $V$-polarized pump is rotated to $H$ by a DHWP before entering the loop counter-clockwise, and both directions overlap in a single $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$ type-II PPKTP crystal with poling period $46.2\,\mu\mathrm m$ (Li et al., 2015).

Two canonical output-state families recur. For type-II operation the two amplitudes are

$$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$

with $\phi$ adjusted to obtain $V$0 when $V$1 (Li et al., 2015). For type-I or same-polarization processes the loop generates

$V$2

as in three-color and two-crystal nondegenerate sources (Hentschel et al., 2010, Stuart et al., 2013).

The same routing principle admits substantial architectural variation. One implementation uses a high-extinction Glan–Thompson polarizer and a pair of broadband periscopes so that both counter-propagating pump components are converted into horizontal polarization before entering a single $V$3 ppKTP crystal at $V$4, after which the two SPDC outputs are transformed into $V$5 and $V$6 and recombined into one spatial mode (Hentschel et al., 2010). Another realizes an unfolded Sagnac inside a Mach–Zehnder defined by two $V$7 calcite beam displacers, with a $V$8-HWP balancing temporal walk-off for type-0 spontaneous four-wave mixing in $V$9 of polarization-maintaining fiber (Lee et al., 2020). A plausible implication is that “polarization Sagnac interferometer” designates a family of common-path polarization-multiplexed loops rather than a single optical layout.

2. Phase accumulation, reciprocity, and passive stability

The central dynamical variable is the relative phase between the two counter-propagating polarization amplitudes. In the beam-displacer implementation the phase is written

$775\,\mathrm{nm}$0

where $775\,\mathrm{nm}$1 is the path-length asymmetry and $775\,\mathrm{nm}$2 is a static offset (Lee et al., 2020). In three-color SPDC the loop phase obeys

$775\,\mathrm{nm}$3

with each contribution $775\,\mathrm{nm}$4; the negative sign on the pump term reflects counter-propagation relative to signal and idler, and is the basis of first-order thermal cancellation in the loop (Hentschel et al., 2010). In the telecom type-II source, $775\,\mathrm{nm}$5 arises from differential optical path length in the loop, including crystal birefringence and Gouy phase shift at the focus, together with static phase delay from the input QWP/HWP; the input QWP and HWP and the axial position of the PPKTP crystal provide the control knobs (Li et al., 2015).

Passive phase stability is a direct consequence of the common-path geometry. In the three-color source no active stabilization is needed, and the paper attributes this to the Sagnac-based design and to phase-drift cancellation to first order (Hentschel et al., 2010). In the beam-displacer Sagnac source the long-term phase-stability was characterized by an Allan deviation of $775\,\mathrm{nm}$6 over an integration time of about $775\,\mathrm{nm}$7 hour with no active phase-stabilization, while a 24-hour run with 10 min scans yielded a fluctuation in fitted phase $775\,\mathrm{nm}$8 under lab temperature drift $775\,\mathrm{nm}$9 peak-to-peak (Lee et al., 2020).

A closely related hybrid, the displaced-Sagnac Mach–Zehnder interferometer, illustrates the same common-mode principle in a non-polarization-specific setting. With footprint $H$0, laterally displaced counter-propagating beams, and individually accessible paths, it showed $H$1 at $H$2 and $H$3 for integration times up to $H$4 hours without active stabilization (Micuda et al., 2014). This suggests that the essential stability mechanism is the common traversal of nearly identical glass and air, with polarization routing supplying an especially compact way to realize it.

3. Quantum-light generation and nonlocal polarization interference

Polarization Sagnac interferometers are a standard platform for entangled-photon generation because the two competing pair-creation amplitudes share the same optics and are recombined at a single port. The reported performance spans nondegenerate, degenerate, pulsed, and continuous-wave sources.

Implementation	State or role	Representative reported result
CW-pumped type-II PPKTP Sagnac (Li et al., 2015)	$H$5	$H$6, $H$7, $H$8
Three-color ppKTP Sagnac (Hentschel et al., 2010)	$H$9	$V$0 pairs/s/mW/THz, fidelity $V$1
Pulsed GVM-PPKTP telecom Sagnac (Jin et al., 2013)	$V$2 at $V$3	fidelities $V$4 and $V$5
Two-crystal nondegenerate Sagnac (Stuart et al., 2013)	flexible $V$6 superposition	$V$7, $V$8

The quantitative details differ by phase-matching scheme. The telecom CW source reported raw Bell-basis visibilities $V$9 and $H$0, together with a Hong–Ou–Mandel-derived bandwidth $H$1 at $H$2 (Li et al., 2015). The pulsed GVM-PPKTP source exploited the group-velocity-matching condition

$H$3

giving intrinsic spectral purity $H$4 for a $H$5 crystal and a $H$6 pump bandwidth of $H$7, with Bell-CHSH parameters $H$8 for $H$9 and $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$0 for $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$1 (Jin et al., 2013). The beam-displacer Sagnac based on spontaneous four-wave mixing reported raw visibilities $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$2 and $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$3 in the $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$4 basis, $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$5 and $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$6 in the $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$7 basis, an average visibility $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$8, and $$10\,\mathrm{mm}\times2\,\mathrm{mm}\times1\,\mathrm{mm}$$9 (Lee et al., 2020).

The same interferometric logic can be exported to nonlocal entanglement analysis. A nonlocal polarization interferometer using two spatially separated balanced Mach–Zehnder or Sagnac interferometers suppresses single-photon interference while retaining two-photon interference. In its standard configuration $46.2\,\mu\mathrm m$0, the nonlocal correlation depends only on the anti-diagonal density-matrix elements $46.2\,\mu\mathrm m$1 and $46.2\,\mu\mathrm m$2, with separable-state bounds

$46.2\,\mu\mathrm m$3

whereas Bell states produce unique signatures at $46.2\,\mu\mathrm m$4 (Williams et al., 2014). The same setup yields CHSH-like parameters $46.2\,\mu\mathrm m$5 and $46.2\,\mu\mathrm m$6, and the experiment reported $46.2\,\mu\mathrm m$7 and $46.2\,\mu\mathrm m$8 with single-photon interference $46.2\,\mu\mathrm m$9 visibility (Williams et al., 2014).

4. Rotation sensing, polarization metrology, and non-reciprocal phase

In metrological applications the polarization Sagnac interferometer converts a small non-reciprocal polarization effect into an interference signal while rejecting many reciprocal perturbations. In the static polarization Sagnac for Faraday-rotation measurement, a linearly polarized input set to $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$0 is split by a 50/50 beam splitter, routed through a PBS and two quarter-wave plates so that the sample is probed by opposite circular polarizations, and then recombined. The detector signal obeys

$$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$1

and for small $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$2 the differential signal is

$$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$3

The same analysis showed that a static Sagnac is functionally equivalent to partially crossed polarizers in figure of merit, but, unlike the PCP, it can discriminate rotations with different time and parity symmetries when appropriate modulation is introduced (Rowe et al., 2016).

A distinct recent proposal replaces path-area sensitivity by helicity sensitivity. In a frame rotating about the $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$4-axis with angular frequency $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$5, left- and right-circularly polarized modes satisfy $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$6, so over a one-way path of length $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$7 the polarization Sagnac phase is

$$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$8

For $$|\Psi\rangle=\frac{1}{\sqrt2}\bigl(|H_sV_i\rangle+e^{i\phi}|V_sH_i\rangle\bigr),$$9 effective round trips in a high-finesse cavity the total phase is

$\phi$0

The key contrast with the conventional path Sagnac effect is that the polarization signal is frequency independent, whereas the path-area phase scales as $\phi$1 or $\phi$2 (Fedderke et al., 2024). The proposal explicitly uses this to separate vibration-induced phase from rotation by operating simultaneously at two frequencies $\phi$3, and gives numerical estimates $\phi$4 for $\phi$5, $\phi$6, $\phi$7, and $\phi$8 in an RF cavity (Fedderke et al., 2024).

A different non-reciprocal contribution arises in fiber Sagnac interferometers with polarization-dependent loss. There the measured non-reciprocal phase is

$\phi$9

where $V$00 is induced by the sequence of PDL–birefringence-axis misalignments and is interpreted through continuous weak measurement (2207.13553). Numerically, for total birefringent phase $V$01 and $V$02, the geometric phase remains small until a critical $V$03 $V$04, then abruptly jumps by $V$05 (2207.13553). A common misconception is that any non-reciprocal phase in a Sagnac loop is simply the conventional area phase; the fiber-PDL and helicity analyses show that polarization Sagnac interferometers can realize qualitatively different phase mechanisms.

5. Switching, isolation, and polarization-state engineering

Because clockwise and counter-clockwise directions are already encoded in polarization, the architecture is well suited to polarization-independent routing. In the fiber-optical single-photon switch, an arbitrary polarization state

$V$06

is split by a 50:50 coupler into two counter-propagating amplitudes, and a Mach–Zehnder-like sub-circuit inside the loop routes $V$07 and $V$08 onto two separate LiNbO$V$09 phase modulators driven identically. With a $V$10 fiber delay in one direction, the electrical pulse overlaps only one circulation, so after recombination

$V$11

The experiment at $V$12 reported raw visibility $V$13, average extinction ratio $V$14 over $V$15 inputs, and insertion loss $V$16, dominated by phase-modulator insertion loss (Alarcón et al., 2020).

A Sagnac-type optical isolator can be built from two Faraday rotators, two half-wave plates, an internal horizontal polarizer, and a PBS defining input and output ports. In the ideal Jones description the forward transformation is the identity while the backward transformation is blocked, so the isolation is

$V$17

A bulk-optical prototype at $V$18 recorded isolation between $V$19 and $V$20 for all tested polarizations, including linear, circular, and elliptical states (Al-Mahmoud et al., 2021).

Polarization Sagnac loops also operate in continuous-variable quantum optics. In a photonic-crystal-fiber Sagnac, a 50:50 coupler launches two equal-intensity pulses counter-propagating in the same principal axis of a $V$21 PCF, a $V$22 relative phase shift makes the recombined output nearly circularly polarized, and balanced detection measures squeezing in the dark Stokes plane. The simulations included loss, non-instantaneous Raman response, excess phase noise, second- and third-order dispersion, and self-steepening, and reproduced the experimental $V$23 polarization squeezing; without any loss or phase-noise the modeled $V$24 PCF loop yielded up to $V$25, and optimized length up to $V$26 predicted $V$27 without loss or phase noise and $V$28 once GAWBS is included (Tacey et al., 2013).

6. Polarisation Sagnac speedmeters and limiting nonidealities

At kilometer scale, the same polarization-routing principle underlies the polarisation Sagnac speedmeter proposed for future gravitational-wave detectors. In the ET-oriented topology, a power-recycling mirror and a 50:50 non-polarizing beam splitter feed two arm cavities of length $V$29; immediately before each arm, a QWP converts the linearly $V$30-polarized beam into circular polarization, and after reflection from the arm optics a second pass through the same QWP rotates the returning light by $V$31, so that each counter-propagating beam visits both arms in opposite order and recombines at the beam splitter (Wang et al., 2013). The output phase in the 10 m prototype description is proportional to differential velocity rather than static displacement: $V$32 with $V$33, yielding the low-frequency speedmeter suppression of radiation-pressure noise (Spencer et al., 2021).

The ET study showed that, at low frequency, $V$34, so radiation-pressure noise can be canceled by choosing the homodyne angle $V$35, without a signal-recycling mirror or filter cavity (Wang et al., 2013). Using realistic optical parameters, the computed quantum-noise-limited sensitivity of the polarizing Sagnac matched the design dual-recycled Michelson with two $V$36 filter cavities throughout $V$37–$V$38 and was better below $V$39 (Wang et al., 2013).

The limiting optics are now quantified experimentally. For a PBS at $V$40 incidence, the measured transmission extinction was

$V$41

and the reflection extinction was

$V$42

The QWP birefringent phase was $V$43, and cavity-mirror birefringence in the 10 m prototype lay in the range $V$44 to $V$45 for the ITM and $V$46 to $V$47 for the ETM (Spencer et al., 2021). The paper further states that this birefringence, together with the QWP imperfections, can be canceled by careful adjustment of the QWP angle, to the extent that the PBS extinction ratio becomes the leading polarization limitation (Spencer et al., 2021).

These limitations clarify two points that are often conflated. First, passive phase stability does not imply immunity to all polarization defects; PBS extinction, QWP retardance error, PDL, and cavity birefringence enter as systematic leakage channels or parasitic phases. Second, a Sagnac geometry need not be a conventional area gyroscope. Depending on the inserted polarization optics, the observable can be a Bell-state phase, a helicity-dependent rotation phase, a speedmeter response, a geometric phase from continuous weak measurement, or a polarization-independent routing phase. The unifying feature is not the application but the counter-propagation of polarization-defined modes in a common loop.