Spatially Variant Half-Waveplate (S-HWP)

• S-HWPs are birefringent devices with uniform π retardance and spatially varying fast-axis orientation that enable local polarization control.
• They are analyzed using Jones, SU(2), and Stokes formalisms, facilitating geometric phase modulation and SAM-to-OAM conversion, as seen in q-plates.
• Integrated in QHQ configurations or cascaded systems, S-HWPs serve as key modules for arbitrary polarization shaping in optical skyrmions and mode multiplexing.

to=arxiv_search 北京赛车女json {"query":"Spatially variant half-waveplate q-plate structured light arXiv", "max_results": 10} to=search_arxiv 天天彩票网json {"query":"Spatially variant half-waveplate q-plate structured light", "max_results": 10} A spatially variant half-waveplate (S-HWP) is a singly inhomogeneous linear retarder with uniform retardance $\delta(x,y)=\pi$ and spatially varying fast-axis orientation $\alpha(x,y)$ across the transverse plane. In structured-light optics, it is the geometric-phase waveplate that locally manipulates polarization and, for circular inputs, imposes a position-dependent Pancharatnam–Berry phase while flipping helicity. The canonical q-plate is the azimuthally patterned special case $\alpha(x,y)=q\phi+\alpha_0$. Within the recent literature, S-HWPs appear both as fundamental spin–orbit photonic elements and as modular constituents of more general doubly inhomogeneous waveplates, arbitrary polarization shapers, and contactless virtual polarization elements realized at a remote free-space plane (B et al., 2021, Wang et al., 5 Jun 2025, Lei et al., 4 Jul 2026).

1. Taxonomy and defining characteristics

Within the terminology used for spatially varying birefringent optics, an s-plate is a singly inhomogeneous waveplate: one parameter is spatially uniform while the other varies over the aperture. Two classes are distinguished. In one class, the retardance is uniform and the fast axis varies, $\alpha=\alpha(x,y)$. In the other, the fast axis is uniform and the retardance varies, $\delta=\delta(x,y)$. In the structured-light literature, and in the principal use relevant here, “s-plate” usually denotes the first class: uniform retardance with spatially varying fast-axis orientation (B et al., 2021).

An S-HWP is the half-wave specialization of that first class. Its defining condition is $\delta=\pi$, so the element behaves locally as an ideal HWP at every point, but with a locally rotated fast axis. This spatial variance makes polarization rotation position dependent and turns the device into a geometric-phase modulator (B et al., 2021, Wang et al., 5 Jun 2025).

A d-plate, by contrast, is doubly inhomogeneous: both $\alpha(x,y)$ and $\delta(x,y)$ vary spatially. The distinction is operationally important. An S-HWP alone provides local half-wave action with a prescribed rotation axis, whereas d-plates provide simultaneous spatial control of polarization and phase and, when used with polarization projection, also amplitude (B et al., 2021).

A common conflation is to identify S-HWPs with q-plates. The q-plate is only the canonical azimuthal subset, defined by $\alpha(x,y)=q\phi+\alpha_0$. More general S-HWPs need not have azimuthal symmetry and need not be restricted to scalar vortex generation. This broader view becomes explicit in the SU(2) treatment of spatially varying retarders and in recent work on arbitrary polarization textures and virtual waveplates (B et al., 2021, Lei et al., 4 Jul 2026).

2. Jones, SU(2), and Stokes-space descriptions

In the Jones formalism, a general linear retarder with fast axis $\alpha$ and retardance $\alpha(x,y)$0 can be written as

$\alpha(x,y)$1

with

$\alpha(x,y)$2

For an HWP, $\alpha(x,y)$3, this reduces to

$\alpha(x,y)$4

The spatially varying case is obtained by replacing $\alpha(x,y)$5 with $\alpha(x,y)$6 (Wang et al., 5 Jun 2025, Lei et al., 4 Jul 2026).

The same transformation admits an SU(2) representation particularly useful for geometric-phase analysis:

$\alpha(x,y)$7

For an S-HWP, $\alpha(x,y)$8, this becomes

$\alpha(x,y)$9

In this form, the retarder acts as a rotation of the Stokes vector by angle $\alpha(x,y)=q\phi+\alpha_0$0 about the equatorial axis $\alpha(x,y)=q\phi+\alpha_0$1 on the Poincaré sphere (B et al., 2021).

The Stokes-space description emphasizes the local polarization action. An ideal linear retarder effects a proper rotation of $\alpha(x,y)=q\phi+\alpha_0$2 by angle $\alpha(x,y)=q\phi+\alpha_0$3 around the equatorial unit axis $\alpha(x,y)=q\phi+\alpha_0$4. For a half-waveplate, the rotation angle is $\alpha(x,y)=q\phi+\alpha_0$5; for a quarter-waveplate, it is $\alpha(x,y)=q\phi+\alpha_0$6. This makes the S-HWP the local $\alpha(x,y)=q\phi+\alpha_0$7-rotation element in cascaded polarization-field synthesis (Lei et al., 4 Jul 2026).

This mathematical equivalence between Jones, SU(2), and Stokes descriptions is not merely formal. It clarifies that the S-HWP is simultaneously a local birefringent retarder, a polarization-space rotation operator, and a geometric-phase element. A plausible implication is that different communities—Jones optics, geometric-phase optics, and full-Stokes beam engineering—are often describing the same object in different languages.

3. Geometric phase, q-plates, and spin–orbit conversion

For circularly polarized inputs, the action of an S-HWP is especially compact:

$\alpha(x,y)=q\phi+\alpha_0$8

Thus, the device flips spin angular momentum and imparts a spatially varying $\alpha(x,y)=q\phi+\alpha_0$9 Pancharatnam–Berry phase (B et al., 2021, Wang et al., 5 Jun 2025).

When the fast-axis pattern is azimuthal,

$\alpha=\alpha(x,y)$0

the output acquires a helical factor $\alpha=\alpha(x,y)$1, corresponding to orbital angular momentum transfer. In the notation of the virtual-polarization-element work, for helicity $\alpha=\alpha(x,y)$2 the output OAM charge is

$\alpha=\alpha(x,y)$3

This is the standard SAM-to-OAM conversion mechanism of a q-plate-like S-HWP (Wang et al., 5 Jun 2025).

For linear inputs, the same HWP rotates the polarization azimuth locally by twice the optical-axis angle. In the remote-plane VPE formulation, a linear input at angle $\alpha=\alpha(x,y)$4 is transformed to $\alpha=\alpha(x,y)$5, where $\alpha=\alpha(x,y)$6 denotes the local optical-axis orientation of the virtually realized HWP (Wang et al., 5 Jun 2025).

The literature treated here also pushes beyond the circular-input q-plate paradigm. A generalized q-plate-like gadget can transform a polarization state $\alpha=\alpha(x,y)$7 to its enantiogyre while imparting prescribed scalar phases $\alpha=\alpha(x,y)$8 and $\alpha=\alpha(x,y)$9 through a $\delta=\delta(x,y)$0 operator. In the special circular-input limit, the construction reduces to a spatially variant half-waveplate plus a phase plate, thereby recovering the J-plate concept as a generalized SAM-to-OAM converter (B et al., 2021).

A recurrent misconception is that the S-HWP is only a vortex generator. In fact, the vortex case is one important subset. The broader role of the S-HWP is to encode a local equatorial rotation axis, which can be used for scalar phase sculpting, vector-beam generation, polarization-state conjugation, and more general spin–orbit mappings (B et al., 2021).

4. From S-HWPs to doubly inhomogeneous waveplates

A central analytical result is that any d-plate can be realized exactly by three s-plates in a QHQ configuration: two quarter-wave s-plates sandwiching a half-wave s-plate. If the quarter-wave s-plates share fast-axis distribution $\delta=\delta(x,y)$1 and the central S-HWP has fast axis $\delta=\delta(x,y)$2, then the effective single retarder has parameters

$\delta=\delta(x,y)$3

$\delta=\delta(x,y)$4

Conversely, for a target d-plate $\delta=\delta(x,y)$5, the inverse design is

$\delta=\delta(x,y)$6

The corresponding SU(2) factorization is

$\delta=\delta(x,y)$7

This establishes the S-HWP as the central module in a universal local decomposition of spatially varying birefringent retarders (B et al., 2021).

The same work shows that a single physical QHQ triplet can realize a family of d-plates through controlled relative rotations of the component s-plates. This means that refabrication is not always necessary to obtain new effective $\delta=\delta(x,y)$8 maps. This suggests a reconfigurable experimental architecture based on standard s-plate hardware rather than bespoke d-plate fabrication (B et al., 2021).

The d-plate formalism also supports complex amplitude shaping when combined with a polarization projection. For a scalar input $\delta=\delta(x,y)$9, passage through a spatial SU(2) operator $\delta=\pi$0 followed by projection onto analyzer $\delta=\pi$1 yields

$\delta=\pi$2

Using this design, the paper explicitly simulates carving higher-order modes from a Gaussian input. With $\delta=\pi$3 and $\delta=\pi$4, the constructed d-plate followed by projection produces LG$\delta=\pi$5, LG$\delta=\pi$6, HG$\delta=\pi$7, and HG$\delta=\pi$8 target fields in the $\delta=\pi$9 projection, while the orthogonal polarization carries the remnant (B et al., 2021).

In this architecture, the S-HWP is no longer just an isolated device; it is the indispensable half-wave element in a constructive factorization of arbitrary spatially varying retarders.

5. Cascaded S-HWPs and arbitrary polarization fields

Recent work on optical skyrmions places the S-HWP in a different role: not primarily as a phase-conversion element, but as the carrier of the target polarization-orientation map $\alpha(x,y)$0. The target state is parameterized by polarization orientation $\alpha(x,y)$1 and ellipticity angle $\alpha(x,y)$2, with normalized Stokes parameters

$\alpha(x,y)$3

A convenient normalized Jones vector is

$\alpha(x,y)$4

This parameterization supports direct inverse design of spatially varying waveplate cascades (Lei et al., 4 Jul 2026).

For horizontal input polarization, an arbitrary output field can be synthesized with an S-HWP followed by an S-QWP using the closed-form rules

$\alpha(x,y)$5

The S-HWP writes the full azimuth map $\alpha(x,y)$6, while the S-QWP sets the ellipticity map. The stage order can also be swapped, with $\alpha(x,y)$7 and $\alpha(x,y)$8, yielding the same target field (Lei et al., 4 Jul 2026).

A fabrication-oriented alternative replaces the S-HWP by a second S-QWP. For right-hand circular input, the two-QWP design rules are

$\alpha(x,y)$9

The reported motivation is practical: S-HWP writing doubles processing time relative to S-QWP and can introduce a dark horizontal artifact, whereas the two-QWP implementation reduces writing time and avoids that artifact (Lei et al., 4 Jul 2026).

Using these design rules, the work realizes Néel-, Bloch-, and anti-skyrmions, including second- and fourth-order structures and $\delta(x,y)$0 and $\delta(x,y)$1 textures, as well as $\delta(x,y)$2 arrays of identical or hybrid skyrmions. Polarization-resolved measurements of the six standard components $\delta(x,y)$3, $\delta(x,y)$4, $\delta(x,y)$5, $\delta(x,y)$6, $\delta(x,y)$7, and $\delta(x,y)$8 are reported to be in excellent agreement with simulations (Lei et al., 4 Jul 2026).

An important conceptual point follows. An S-HWP alone controls local half-wave action and therefore local orientation structure, but simultaneous arbitrary control of orientation and ellipticity generally requires at least one additional stage. This directly addresses the limitation, emphasized in the skyrmion work, that spatially resolved control of both $\delta(x,y)$9 and $\alpha(x,y)=q\phi+\alpha_0$0 is nontrivial with single-stage devices (Lei et al., 4 Jul 2026).

6. Realization platforms, remote implementations, and application domains

Conventional S-HWPs are contact-based devices. Reported implementations include photoaligned liquid-crystal polymer films, direct writing of alignment patterns, patterned birefringent films, and Pancharatnam–Berry metasurfaces. In these architectures, the Jones transformation is applied at the physical surface carrying the fast-axis map $\alpha(x,y)=q\phi+\alpha_0$1 or $\alpha(x,y)=q\phi+\alpha_0$2 (B et al., 2021, Wang et al., 5 Jun 2025).

The fabrication-agnostic view in the d-plate factorization work is that any method capable of producing high-fidelity s-plates can support the QHQ synthesis of arbitrary d-plates. Photoaligned liquid crystals offer controlled retardance and large apertures; direct writing supports arbitrary axis patterns; metasurfaces provide subwavelength anisotropy but face chromatic-dispersion and bandwidth constraints. In all cases, orientation errors map directly into phase errors $\alpha(x,y)=q\phi+\alpha_0$3 for S-HWPs and into incorrect effective $\alpha(x,y)=q\phi+\alpha_0$4 in QHQ stacks (B et al., 2021).

The 2025 virtual-polarization-element framework extends S-HWPs into a non-contact regime. A metasurface at $\alpha(x,y)=q\phi+\alpha_0$5 implements a designed local Jones operator $\alpha(x,y)=q\phi+\alpha_0$6 such that, after free-space propagation, a target matrix-valued function

$\alpha(x,y)=q\phi+\alpha_0$7

is realized at a remote plane $\alpha(x,y)=q\phi+\alpha_0$8. For an S-HWP target, $\alpha(x,y)=q\phi+\alpha_0$9. The field propagation is modeled by the angular-spectrum method with evanescent components excluded, and the inverse design computes a metasurface-plane matrix from the matrix Fourier coefficients of $\alpha$0 (Wang et al., 5 Jun 2025).

That platform uses anisotropic $\alpha$1 rectangular nanopillars of height $\alpha$2 on glass with period $\alpha$3 at $\alpha$4. For a single-function virtual HWP over a central $\alpha$5 region at $\alpha$6, the realized Jones-matrix fidelity is reported as $\alpha$7, polarization fidelity as $\alpha$8, and conversion efficiency as approximately $\alpha$9 of the theoretical maximum for the limited target area; enlarging the functional area increases the conversion efficiency to approximately $\alpha(x,y)$00. Virtual vortex waveplates were also demonstrated for $\alpha(x,y)$01, $\alpha(x,y)$02, $\alpha(x,y)$03, and a dual-mode configuration combining $\alpha(x,y)$04 in an inner ring with $\alpha(x,y)$05 in an outer ring (Wang et al., 5 Jun 2025).

A separate fabrication route writes spatially variant retarders directly inside silica using ultrafast laser direct writing. The reported conditions are Yb:KGW at $\alpha(x,y)$06, $\alpha(x,y)$07 repetition rate, $\alpha(x,y)$08 pulse duration, aspheric focusing with $\alpha(x,y)$09, translation speed $\alpha(x,y)$10, raster line interval $\alpha(x,y)$11, pulse energy $\alpha(x,y)$12, and depth around $\alpha(x,y)$13. A single modification layer provides approximately $\alpha(x,y)$14 retardance, so two layers yield a QWP of approximately $\alpha(x,y)$15 retardance at $\alpha(x,y)$16. Unit devices are about $\alpha(x,y)$17, and arrays extend to $\alpha(x,y)$18 (Lei et al., 4 Jul 2026).

Across these platforms, the application envelope is broad. The sources explicitly connect S-HWPs and their extensions to spin–orbit coupling and vortex beams, matrix Fourier optics and mode multiplexing, microscopy, optical trapping, quantum photonics, non-invasive beam engineering, contactless polarization manipulation, and topological light fields such as optical skyrmions (B et al., 2021, Wang et al., 5 Jun 2025, Lei et al., 4 Jul 2026). The resulting picture is that the S-HWP is both a specific optical element and a general design primitive: a local half-wave polarization rotator whose spatially patterned axis can be used directly, cascaded with other variant retarders, or synthesized virtually at a remote plane.