Space Coronagraph Optical Bench (SCoOB)

• SCoOB is a vacuum-compatible high-contrast imaging testbed designed to mature coronagraphic technology and enable direct exoplanet detection.
• It integrates a vector vortex coronagraph with a Boston Micromachines deformable mirror and employs advanced wavefront sensing and control methods like iEFC and LOWFSC.
• The platform demonstrates raw contrast levels as low as 10⁻⁹ in thermal-vacuum conditions and supports studies ranging from visible to ultraviolet coronagraphy.

The Space Coronagraph Optical Bench (SCoOB) is a vacuum-compatible high-contrast imaging testbed at the University of Arizona developed to mature coronagraphic hardware, wavefront sensing, and wavefront control for future spaceborne direct imaging and spectroscopy of exoplanets. Inspired by the Coronagraphic Debris and Exoplanet Exploring Payload (CDEEP), SCoOB combines a vector vortex coronagraph with a Boston Micromachines Kilo-C deformable mirror in a thermal-vacuum environment and has subsequently served as a platform for vacuum contrast demonstrations, Mueller-matrix polarimetry, end-to-end contrast modeling, self-coherent camera implementation, and ultraviolet coronagraphy studies (Ashcraft et al., 2022, Gorkom et al., 2022, Gorkom et al., 2024, Ashcraft et al., 2024, Gorkom et al., 11 Sep 2025).

1. Origins, objectives, and flight-like environment

SCoOB was conceived as a high-contrast imaging testbed for operation in a thermal vacuum chamber in order to advance spaceborne coronagraphic technology for visible-light detection of habitable exoplanets and other faint circumstellar targets. The design goal was a raw contrast of $10^{-8}$ or better, in a configuration explicitly intended to bypass atmospheric limitations and to test deformable mirrors, coronagraphs, and focal-plane wavefront sensing and control algorithms in a flight-like environment (Ashcraft et al., 2022).

The environmental architecture is a central part of the testbed definition. The bench is mounted on a $1 \times 2$ meter pneumatically isolated Newport optical breadboard inside a Rydberg Vacuum Sciences 104430 thermal-vacuum chamber. The chamber dimensions are 1.2 m diameter by 2.2 m length; the design vacuum is $1 \times 10^{-8}$ torr; the temperature range is $-150^\circ$C to $+150^\circ$C; and the quoted thermal stability at room temperature is $\pm 2^\circ$C. Vacuum-compatible opto-mechanics, nichrome heaters, and gaseous $\mathrm{N}_2$ were adopted for thermal control, while the underground location and pneumatic isolation were chosen to reduce vibrational disturbance (Ashcraft et al., 2022).

Later operation established the distinction between design specification and achieved operating condition. In vacuum performance demonstrations, SCoOB operated at approximately $10^{-6}$ Torr during non-cryogenic testing, with additional mitigation of jitter, stray light, and detector noise through a Minus K passive vibration isolation table, a field stop, a precision-fabricated source pinhole, and a low-noise vacuum-compatible CMOS sensor (Gorkom et al., 2024).

2. Optical architecture and principal hardware

The optical layout is an adapted version of the CDEEP design. A point-source simulator based on either a point-source microscope or a single-mode fiber tip with $<5\,\mu\mathrm{m}$ mode-field diameter illuminates the system. The beam is collimated by OAP0, relayed by OAP1 and OAP2 to the deformable mirror, passes through polarization optics, and is then focused by OAP3 to the focal-plane mask at approximately $F/48$. Downstream optics reimage the Lyot pupil and form the final science image (Ashcraft et al., 2022, Doelman et al., 2023).

The deformable mirror is a Boston Micromachines Kilo-C MEMS device with 1.5 $1 \times 2$0m stroke. Different SCoOB papers report the hardware in slightly different operational descriptions: one gives approximately 1020 actuators across a 9.2 mm aperture, whereas another describes 952 active actuators, one non-responsive actuator, and a $1 \times 2$1 mm beam footprint with 32 actuators across the pupil. In both descriptions, the Kilo-C is the high-order wavefront control element central to dark-hole generation (Ashcraft et al., 2022, Gorkom et al., 2022).

At the coronagraphic focal plane, SCoOB has supported both a knife-edge Lyot mode and a vector vortex coronagraph (VVC) mode. The VVC is a charge-6 liquid-crystal vector vortex waveplate; in later vacuum demonstrations, the installed mask was manufactured by Beam Engineering Company and optimized for a 20% band centered at approximately 635 nm. The Lyot train includes a stop of about 95% throughput, and some configurations incorporate an SCC reference pinhole outside the geometric pupil (Ashcraft et al., 2022, Gorkom et al., 2024, Derby et al., 2 Sep 2025).

The bench has also undergone targeted hardware refinements driven by contrast-limiting diagnostics. Reported upgrades include a precision 4 $1 \times 2$2m source pinhole, a field stop after the Lyot stop for mitigation of stray and scattered light, strategic Acktar-lined baffling, upstream placement of polarizers and quarter-wave plates to suppress ghosts, and a vacuum-packaged Sony IMX571 CMOS detector with $1 \times 2$3 pixels, $1 \times 2$4 read noise, and a 16-bit ADC (Gorkom et al., 2024).

3. Coronagraphic modes and wavefront control

SCoOB’s initial coronagraphic development emphasized two complementary focal-plane masks. The knife-edge Lyot coronagraph functioned as a comparatively simple baseline mode, while the vector vortex coronagraph was selected for achromaticity, high throughput, and small inner working angle. In early in-air operation, phase retrieval using deformable-mirror-generated diversity drove the system to approximately 10 nm RMS residual wavefront error and a Strehl ratio of about 0.95 before dark-hole digging (Gorkom et al., 2022, Ashcraft et al., 2022).

High-order control has centered on electric-field-conjugation variants, especially implicit Electric Field Conjugation (iEFC). In SCoOB, iEFC is a data-driven focal-plane wavefront control method that uses empirically measured response matrices rather than a detailed optical model. Early in-air results with a knife-edge focal-plane mask and Lyot stop reached a mean raw contrast of $1 \times 2$5 over $1 \times 2$6 at 633 nm. Subsequent VVC-based work established sub-$1 \times 2$7 visible-light dark holes in SCoOB’s $1 \times 2$8 half-sided region (Gorkom et al., 2022, Derby et al., 2 Sep 2025).

Low-order and high-order loops were later combined on the same bench. A Lyot-based low-order wavefront sensing and control loop used the starlight rejected by the vortex coronagraph and reflected by a reflective Lyot stop to a dedicated camera, while iEFC provided the concurrent high-order control. In air, this combined LOWFSC+iEFC system maintained $1 \times 2$9 contrast levels; the best reported contrast was $1 \times 10^{-8}$0, and a mean contrast of $1 \times 10^{-8}$1 was maintained for two hours with iEFC maintenance every 10 minutes. The same study reported that if the reference-offset compensation between the two loops was omitted, contrast degraded to $1 \times 10^{-8}$2 because the low-order loop attempted to undo the high-order dark-hole solution (Milani et al., 2 Sep 2025).

SCoOB has also been used to implement a self-coherent camera (SCC). In this configuration, a small off-axis Lyot-plane pinhole leaks a fraction of starlight so that it interferes with residual speckles and produces fringes at the science camera. The reported SCC Lyot stop used a 9.1 mm pupil, a 300 $1 \times 10^{-8}$3m pinhole corresponding to 3% of the pupil, and a 6 mm pinhole offset. On SCoOB, SCC dark-hole digging reached a mean contrast of $1 \times 10^{-8}$4 in a $1 \times 10^{-8}$5 half-annulus, compared with $1 \times 10^{-8}$6 for iEFC in a similar region; the stated limiting factors were calibration speed and testbed drifts, especially those associated with mechanical pinhole modulation (Derby et al., 2 Sep 2025).

4. Performance metrics and demonstrated contrast

Coronagraph studies associated with SCoOB have adopted the standardized metrics advocated in the Optimal Optical Coronagraph review: raw contrast, throughput, inner and outer working angles, detection significance, spectral bandwidth, and sensitivity to aberrations and alignment errors. That framework was proposed precisely because the community lacked a general set of standard performance metrics applicable to theoretical designs, testbeds, and deployed instruments, and because such quantities permit comparison of coronagraph concepts and direct connection between optical performance and detection significance (Ruane et al., 2018).

Within SCoOB-specific papers, contrast is reported as the mean intensity in the dark hole normalized by the off-axis or stellar PSF peak. The principal reported dark-hole region for the mature VVC configuration is $1 \times 10^{-8}$7 in a half-sided D-shaped geometry. In vacuum, the measured science-camera jitter is $1 \times 10^{-8}$8 RMS (Gorkom et al., 2024).

Configuration	Bandpass and region	Mean contrast
In air, knife-edge FPM + iEFC	633 nm, $1 \times 10^{-8}$9	$-150^\circ$0
Vacuum, VVC + iEFC	630 nm, $-150^\circ$1 BW, $-150^\circ$2	$-150^\circ$3
Vacuum, VVC + iEFC	630 nm, 2% BW, $-150^\circ$4	$-150^\circ$5
Vacuum, VVC + iEFC	630 nm, 5% BW, $-150^\circ$6	$-150^\circ$7
Vacuum, VVC + iEFC	630 nm, 10% BW, $-150^\circ$8	$-150^\circ$9
Vacuum, VVC + iEFC	630 nm, 15% BW, $+150^\circ$0	$+150^\circ$1

At shorter wavelengths, out-of-band operation degrades as expected from VVC retardance error: the reported 2% results are $+150^\circ$2 at 543 nm and $+150^\circ$3 at 493 nm. The 2024 vacuum study explicitly associated these performance gains with the introduction of the field stop, improved source pinhole, reduced jitter, and lower-noise detector chain (Gorkom et al., 2024). In that sense, SCoOB functions not only as a coronagraph testbed but also as a controlled environment for isolating which hardware changes materially lower the contrast floor.

5. Polarization aberrations and end-to-end contrast limits

A major later use of SCoOB has been the direct measurement and modeling of instrument-level polarization aberrations. To do this, a dual-rotating-retarder polarimeter was built around the testbed so that the full Mueller matrix could be measured at the coronagraphic exit pupil. The measurements were made at 525 nm and 630 nm and reduced with the Python package katsu. The reported pupil-averaged diattenuation is about 16.5% with a standard deviation of 7.3%, the retardance spatial standard deviation is approximately $+150^\circ$4 at 525 nm, and the depolarization index is very near 1 across the pupil. Mid-frequency structure at roughly 4–5 cycles per pupil appears in both diattenuation and retardance maps and was attributed to instrument-specific effects such as coating non-uniformity or anisotropy rather than simple low-order aberrations (Ashcraft et al., 2024).

End-to-end polarization simulations clarified the operational significance of those measurements. Using a VVC with crossed circular polarizers, the dominant polarization aberrations were identified as retardance defocus and tilt from the off-axis parabolas and fold mirrors. In the $+150^\circ$5 dark hole, these polarization aberrations did not noticeably degrade the mean contrast, but they did produce brighter speckles at $+150^\circ$6 with intensity up to $+150^\circ$7. The same simulations found that SCoOB’s mean contrast is more sensitive to VVC and quarter-wave-plate retardance errors than to polarization aberrations from the mirror train (Anche et al., 2024).

A broader end-to-end numerical model then incorporated measured VVC retardance maps, modeled polarization aberrations, measured surface and reflectivity errors, diffuse and specular reflectivity, deformable-mirror quantization and noise, and beam jitter. The reported total incoherent raw contrast floors in the $+150^\circ$8 dark hole are $+150^\circ$9, $\pm 2^\circ$0, and $\pm 2^\circ$1 at 630 nm for 2%, 5%, and 10% bandwidths, and $\pm 2^\circ$2, $\pm 2^\circ$3, and $\pm 2^\circ$4 at 543/525 nm for the same bandwidths. In that model, narrowband performance is ultimately limited by diffuse reflectivity, especially from the Aeroglaze Z306 stop at the fast steering mirror, whereas broadband performance is dominated by chromatic EFC residuals associated with surface and coating errors and Talbot cycling (Anche et al., 2 Sep 2025).

A recurrent misconception in high-contrast testbeds is that polarization aberration is necessarily the immediate dominant floor. In SCoOB, the measurements and simulations indicate a more specific hierarchy: mirror-induced polarization aberrations are measurable and relevant, particularly at small inner working angle, but the dominant limits in the benchmark $\pm 2^\circ$5 dark hole are more strongly tied to VVC/QWP retardance quality, diffuse reflectivity, and chromatic EFC residuals (Anche et al., 2024, Anche et al., 2 Sep 2025).

6. Ultraviolet extension and use as a component-validation platform

SCoOB has increasingly been used as a platform for subsystem validation beyond its baseline visible-light vacuum demonstrations. In ultraviolet coronagraphy studies motivated by the Habitable Worlds Observatory, the bench is described as a flexible, vacuum-compatible testbed equipped with a deformable mirror, interchangeable vector vortex and shaped-pupil masks, and component-level metrology. The UV program emphasizes the difficulty of extending coronagraphy to the 200–400 nm regime, where tighter wavefront sensing and control, optical surface quality, scattered light, and polarization aberrations all become more severe; the reported analytic and numerical modeling shows that contrast for first-order phase aberrations, scatter, and beam walk can scale as $\pm 2^\circ$6, and the dark-hole contrast in realistic regimes was described as scaling as $\pm 2^\circ$7. Preliminary SCoOB predictions for that program are $\pm 2^\circ$8 contrast in a 2% band at 300 nm and $\pm 2^\circ$9 in 5% bandwidth, presently limited primarily by chromatic EFC residuals (Gorkom et al., 11 Sep 2025).

The same UV effort has introduced carbon-nanotube shaped-pupil coronagraph masks, described as ultra-low-reflectivity components for stray-light suppression. A prototype CNT-based shaped-pupil mask with a $\mathrm{N}_2$0 dark hole has been fabricated and is undergoing visible-light validation on SCoOB before ultraviolet deployment (Gorkom et al., 11 Sep 2025).

SCoOB has also been used to qualify source optics. Microfabricated pinholes intended to replace commercial laser-drilled pinholes were designed in the 2–8 $\mathrm{N}_2$1m diameter range, fabricated in a 1 $\mathrm{N}_2$2m thick $\mathrm{N}_2$3 membrane overhanging a silicon substrate, and coated with 200 nm of aluminum. A physical-optics model of the source chain coupled to 3D finite-difference time-domain simulation showed that these pinholes behave as short cylindrical waveguides, including cross-polarization coupling. Experimentally, the microfabricated devices produced at least six visible Airy rings, whereas the laser-drilled reference did not clearly show the first Airy ring (Jenkins et al., 2023).

Another device-level study used SCoOB and the NASA JPL In-Air Coronagraphic Testbed to evaluate the triple-grating vector vortex coronagraph. On SCoOB, operating at $\mathrm{N}_2$4 with iEFC and polarization filtering at 633 nm, the tgVVC prototype reached a mean contrast of $\mathrm{N}_2$5 between $\mathrm{N}_2$6 after masking a ghost from the quarter-wave plate. Comparison with the JPL testbed indicated that the SCoOB speckle field was flatter, supporting the interpretation that some stripe-like residuals seen elsewhere were model-control artifacts rather than intrinsic to the device. The need for polarization filtering in this prototype was traced to manufacturing inaccuracies, grating misalignment, and non-ideal retardance rather than to a conceptual failure of the tgVVC architecture (Doelman et al., 2023).