Stress Engineered Optic: Principles & Applications

• Stress engineered optics are optical elements designed with controlled elastic stress to produce spatially varying birefringence and polarization transformations.
• They enable spectropolarimetric imaging by converting input polarization and wavelength into distinct point spread functions for single-shot Stokes metrology.
• SEO techniques extend to integrated photonics and stress mirror polishing, offering tunable control over resonance, wavefront, and spin–orbit coupling.

Searching arXiv for recent and foundational SEO papers to support the article. arXiv search query: "stress engineered optics spectropolarimetry Shack-Hartmann wavefront" A stress engineered optic (SEO) is an optical element whose polarization, phase, resonance, or surface figure is deliberately shaped by controlled elastic stress. In the narrow sense used in recent bulk-optics work, an SEO is a plane-parallel stressed birefringent window that functions as a space-variant waveplate and converts input polarization and wavelength into distinctive point spread functions (PSFs); in broader usage, the same design principle appears in spin–orbit coupling elements, stress-optic integrated modulators, soft photoelastic media, and stress mirror polishing (Spiecker et al., 24 Sep 2025, Liang et al., 2019, Wang et al., 2022, Lemared et al., 2020).

1. Definition and constitutive optical principle

In bulk SEO implementations for polarimetry, the optic is a plane-parallel optical window, typically fused silica, with a deliberately engineered mechanical stress distribution. The stress distribution induces stress birefringence: a spatially varying difference in refractive index for orthogonal polarization eigenmodes, together with a spatially varying fast/slow axis orientation. A common loading geometry has trigonal symmetry, with three tightly localized stress regions around the edge separated by $120^\circ$. Near the center of the aperture, this produces a birefringent structure in which the retardance varies approximately only with radius and the fast-axis orientation varies approximately only with azimuth. The local phase retardance is written as

$$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$

and in the central region of a trigonal SEO it is approximated by

$\delta(\rho)=c\,\rho,$

where $c$ is the dimensionless stress parameter (Spiecker et al., 24 Sep 2025).

This central-region form makes the SEO a space-variant waveplate. A useful Jones representation is

$$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$

with $R(\theta)$ the Jones rotation matrix. For fixed applied force, the stress-induced birefringence is roughly wavelength-independent, so $c\propto 1/\lambda$; shorter wavelengths therefore experience larger retardance gradients (Spiecker et al., 24 Sep 2025).

A second canonical formulation appears in work that uses a piece of BK7 glass radially compressed by a metal ring. There the SEO is written in the circular basis as

$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$

The diagonal terms preserve input circular polarization, while the off-diagonal terms flip circular polarization and add azimuthal phase factors $e^{\pm i\phi}$. In that form the SEO is simultaneously a stress-induced, space-variant wave plate and a spin–orbit coupling device (Liang et al., 2019).

2. Pupil-plane PSF encoding and single-shot Stokes metrology

When an SEO is placed in the pupil plane of a 4F imaging system, its space-variant birefringence modifies the pupil function so that the image-plane PSF becomes polarization-dependent. With an analyzer $A$ placed before the sensor, the pupil-plane field is modeled as

$$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$0

and the image-plane field is the Fourier transform of this pupil field. The detected PSF intensity

$$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$1

therefore depends on the input polarization state through the space-variant Jones operator. This is the basis of star-test polarimetry: the shape and structure of the PSF encode the Stokes parameters of the input polarization (Spiecker et al., 24 Sep 2025).

For a monochromatic input, the SEO-plus-analyzer system is calibrated as a linear mapping

$$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$2

where $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$3 is the Stokes vector and $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$4 is a $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$5 measurement matrix built from PSFs recorded for known calibration states such as H, V, P, M, R, and L. For multiple wavelengths $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$6, the model becomes

$$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$7

with wavelength-specific blocks concatenated in a $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$8 matrix. This construction exploits the fact that $$\delta(\rho,\phi,\lambda)=\frac{2\pi}{\lambda}t\,\Gamma(\rho,\phi),$$9, so distinct wavelengths produce distinct PSF morphologies even at fixed polarization. In experiments with $\delta(\rho)=c\,\rho,$0, $\delta(\rho)=c\,\rho,$1, and $\delta(\rho)=c\,\rho,$2 nm lasers, median angular errors on the Poincaré sphere for polychromatic retrieval were reported in the range from $\delta(\rho)=c\,\rho,$3 rad to $\delta(\rho)=c\,\rho,$4 rad depending on wavelength and $\delta(\rho)=c\,\rho,$5, and the work states that experiments show angular errors as small as $\delta(\rho)=c\,\rho,$6 mRad, with red and blue measurements outperforming green (Spiecker et al., 24 Sep 2025).

The same encoding principle has been integrated with a Shack–Hartmann wavefront sensor. In that geometry, each lenslet produces a local point source whose SEO-modified PSF carries polarization information in its shape, while the PSF displacement carries local wavefront gradient information. The augmented parameter vector is

$\delta(\rho)=c\,\rho,$7

and a calibrated measurement matrix yields both local Stokes parameters and local wavefront gradients from a single frame. The reported performance was measurement of wavefront gradients as small as $\delta(\rho)=c\,\rho,$8 with a polarimetric angular error of approximately $\delta(\rho)=c\,\rho,$9 radians on the Poincaré sphere in a single frame measurement (Spiecker et al., 29 Sep 2025).

3. Spin–orbit coupling, orbital angular momentum, and quantum random walks

In the quantum-random-walk formulation, the SEO couples polarization, treated as the coin space, to orbital angular momentum (OAM), treated as the walker space. Under the circular-basis Jones operator, an input $c$0 evolves into a superposition of $c$1 and $c$2, while $c$3 evolves into a superposition of $c$4 and $c$5. The radial dependence $c$6 modulates the coupling strength across the beam, so the QRW amplitudes depend explicitly on radius as well as OAM index (Liang et al., 2019).

This behavior differentiates SEOs from q-plates. A q-plate is purely off-diagonal in the circular basis and always flips circular polarization while adding or subtracting OAM by $c$7. An SEO, by contrast, has non-zero diagonal terms $c$8, so part of the field remains in the same polarization and OAM while another part flips polarization and shifts OAM. The result is a richer coin–shift structure, multi-path QRW dynamics, and probability distributions in which many intermediate OAM values can have non-zero probability. The radial dependence also introduces the stress parameter $c$9 as a direct tuning parameter for the walk (Liang et al., 2019).

A further design result is the high-$$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$0 asymptotic regime. Because the number of oscillations of $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$1 across the aperture grows with $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$2, the paper shows that for sufficiently large $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$3, probability densities become insensitive to further increases in $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$4. The stated design guideline is to choose $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$5 to enter a robust asymptotic regime. For small $$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$6, by contrast, the spin–orbit coupling is weak over most of the aperture and the QRW has smaller OAM spread. This establishes SEO not only as a passive patterned retarder but as a tunable spin–orbit platform (Liang et al., 2019).

4. Integrated, acousto-optic, piezo-photonic, and soft-material realizations

The SEO principle extends beyond bulk stressed windows into integrated photonics, acousto-optics, piezo-photonic detection, and soft photoelastic media. In all of these cases, stress is treated as a design degree of freedom rather than as a parasitic.

Platform	Stress mechanism	Optical function
SiN micro-ring resonator	PZT-actuated stress-optic effect	Resonance tuning and modulation
Thin-SiN spiral AO modulator	Traveling acoustic strain field in oxide/SiN stack	Broadband phase modulation
ZnO thin-film MSM detector	Residual and interfacial stress, enhanced by Si$$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$7N$$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$8 pillars	Zero-bias UV detection
Gelatin under Hertzian contact	Three-dimensional stress field in soft material	Retardation and azimuth mapping

A silicon nitride stress-optic microresonator modulator uses a PZT actuator to apply strain through the oxide cladding into an ultra-low-loss Si$$J(\rho,\phi,\lambda)=R(-\theta(\phi)) \begin{pmatrix} e^{+i\delta(\rho,\lambda)/2} & 0\ 0 & e^{-i\delta(\rho,\lambda)/2} \end{pmatrix} R(\theta(\phi)),$$9N$R(\theta)$0 core. The stress modifies the effective index through the photoelastic effect and shifts the ring resonance condition $R(\theta)$1. Reported device metrics include low optical loss of $R(\theta)$2 dB/cm in a $R(\theta)$3m resonator at $R(\theta)$4 nm, DC to $R(\theta)$5 MHz 3-dB bandwidth, power consumption of $R(\theta)$6 nW, $R(\theta)$7 dB extinction ratio, $R(\theta)$8 million $R(\theta)$9 over a $c\propto 1/\lambda$0 GHz tuning range, and a tuning efficiency of $c\propto 1/\lambda$1 MHz/V. Two control applications were demonstrated: PDH laser stabilization reducing laser frequency noise by $c\propto 1/\lambda$2 dB, and laser carrier tracking (Wang et al., 2022).

A distinct integrated realization uses traveling-wave acousto-optic stress fields on thin silicon nitride. There the governing stress-optic relation is

$c\propto 1/\lambda$3

with an AlN transducer launching acoustic waves that modulate the optical phase in a spiral waveguide architecture. The spiral repeatedly weaves the light through the acoustic field up to $c\propto 1/\lambda$4 times over spirals up to $c\propto 1/\lambda$5 cm in length, enabling a $c\propto 1/\lambda$6 of $c\propto 1/\lambda$7 V at $c\propto 1/\lambda$8 MHz with $c\propto 1/\lambda$9 dB of insertion loss. The work emphasizes that the design avoids heterogeneous integration, release processes, complicated fabrication procedures, and modifications of the commercial foundry fabricated photonic layer stack (Kenning et al., 6 May 2025).

In a broader piezo-photonic usage of the term, a zero-bias ZnO ultraviolet detector uses in-device stress engineering rather than a bulk birefringent window. The active ZnO layer is stressed by the device stack, and Si$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$0N$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$1 pillars are introduced specifically to increase residual stress. Grazing-incidence X-ray diffraction showed approximately $$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$2 higher residual stress in the pillar-based device, and the detailed analysis reports approximately $$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$3 higher compressive residual stress for Sample II than for Sample I. At room temperature Sample II showed approximately $$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$4 higher UV-induced voltage magnitude, and at cryogenic temperatures it consistently showed higher SNR. The paper interprets the improved performance as a consequence of a larger piezopotential generated by the engineered stress field (Sau et al., 12 Nov 2025).

Soft-material photoelasticity supplies the constitutive building blocks of SEO from another direction. In gelatin subjected to a three-dimensional Hertzian stress field, the integrated stress–optic law is

$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$5

where $$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$6 is the difference of the secondary principal stresses. The measured retardation and azimuth showed reasonable agreement with analytical prediction, and the stress-optic coefficient of the gelatin gel used was reported as $$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$7. This demonstrates that controlled 3D stress in a transparent soft medium can produce predictable birefringence and optic-axis orientation, which is the same forward map used in more conventional SEO design (Tung et al., 2022).

5. Stress-engineered optical surfaces and stress mirror polishing

SEO also includes optical components whose final surface figure is produced by controlled elastic deformation during manufacture. In stress mirror polishing (SMP), a mirror blank is designed so that under a prescribed load it deforms into an inverse shape, is polished with a full-size spherical tool, and then elastically relaxes into the desired asphere when the stress is released. The optical target is expressed as a conicoid

$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$8

and the deformation required during polishing is written as

$$\hat{J}_\text{SEO}(\rho,\phi)= \begin{bmatrix} \cos[\Delta(\rho)] & e^{i\phi}\sin[\Delta(\rho)]\ e^{-i\phi}\sin[\Delta(\rho)] & \cos[\Delta(\rho)] \end{bmatrix}, \qquad \Delta(\rho)=\frac{c\rho}{2}.$$9

the difference between the spherical substrate and the target asphere (Lemared et al., 2020).

The crucial SEO variable is the variable thickness distribution. For a target dominated by third-order spherical aberration, an analytical VTD provides a first approximation, and finite-element shape optimization is then used to refine the back surface so that the loaded deformation matches the target aberration mode. In the reported design study, NASTRAN optimization solver and a finite element model were used to support the process, with Python orchestration of the optimization loop. The specific example concerns a large lightweight Zerodur mirror with an initial outer mechanical diameter of $e^{\pm i\phi}$0 mm, a clear aperture diameter of $e^{\pm i\phi}$1 mm, a central thickness of $e^{\pm i\phi}$2 mm, and an edge thickness of $e^{\pm i\phi}$3 mm. The target deformation during polishing had SA3 amplitude $e^{\pm i\phi}$4m RMS (Lemared et al., 2020).

The reported numerical results show the ability of the shape-optimization process to support SMP for a peculiar aspherical shape generated from a spherical optical surface thanks to thickness distribution reshaping. After several optimizations on the reduced pupil, the final residual without piston was approximately $e^{\pm i\phi}$5 nm RMS over the clear aperture, high-order spherical modes after removing the first $e^{\pm i\phi}$6 Zernike polynomials were approximately $e^{\pm i\phi}$7 nm RMS, and the maximum Von Mises stress was $e^{\pm i\phi}$8 MPa, within the stated acceptable limit of $e^{\pm i\phi}$9 MPa. In this form, SEO is not primarily a polarization device; it is an active-optics manufacturing method in which geometry, load, and elastic response are co-designed to produce an optical figure (Lemared et al., 2020).

6. Common misconceptions, practical constraints, and research directions

A recurrent misconception is that SEO denotes accidental or parasitic stress birefringence. Across the cited work, the defining feature is the opposite: the stress field is deliberately engineered so that the optic performs a known transformation. In bulk pupil-plane systems that transformation is a calibrated mapping from Stokes parameters, and in some cases wavelength, to PSF structure; in integrated photonics it is a calibrated or modeled mapping from stress to effective index and resonance shift; in SMP it is a mapping from load and thickness distribution to optical sag (Spiecker et al., 24 Sep 2025, Wang et al., 2022, Lemared et al., 2020).

A second misconception is that SEO is intrinsically broadband or intrinsically wavelength-agnostic in every implementation. The recent spectropolarimetric work explicitly assumes discrete, narrow-linewidth wavelengths and models the composite PSF as a linear superposition of wavelength-specific PSFs. It also identifies practical constraints such as PSF size and cross-talk, the slowly varying polarization assumption across each lenslet aperture, and sensitivity to alignment and calibration drift. In the combined Shack–Hartmann implementation, dynamic range is limited at the high end by PSF overlap and gradient-induced PSF-shape changes, and at the low end by localization noise and calibration accuracy (Spiecker et al., 24 Sep 2025, Spiecker et al., 29 Sep 2025).

A third misconception is that SEO and q-plates are interchangeable. The QRW analysis shows that the distinction is structural: q-plates are purely off-diagonal in the circular basis, whereas SEOs generally retain non-zero diagonal terms and radial dependence. This difference produces richer dynamics, but it also makes SEO behavior more calibration- and geometry-dependent. Similarly, integrated stress-optic devices trade some simplicity for other benefits: the SiN microresonator modulator exhibits a $A$0-versus-bandwidth trade-off and ferroelectric hysteresis in PZT, while the acousto-optic spiral modulator trades area and insertion loss for lower drive voltage and optical broadbandness (Liang et al., 2019, Wang et al., 2022, Kenning et al., 6 May 2025).

The current literature suggests a convergence of formerly separate research directions. Bulk SEO star-test polarimetry, simultaneous polarimetry and wavefront sensing, stress-optic integrated modulators, acousto-optic spiral architectures, and mechanically programmable photoelastic media all use stress as an optical design parameter. A plausible implication is that future SEO systems will continue to combine modalities—spectropolarimetry, wavefront sensing, spin–orbit conversion, and integrated control—provided that calibration, stress stability, and cross-talk remain within the bounds set by the underlying elastic and photoelastic models.