Roman Coronagraph: High-Contrast Exoplanet Imaging

• Roman Coronagraph is a space-based high-contrast instrument that uses dual deformable mirrors and multiple coronagraph modes for direct exoplanet imaging.
• It implements advanced wavefront control techniques such as EFC and iEFC, achieving raw contrasts near 10⁻⁹ and precise spectral characterization.
• Validated through simulations and thermal-vacuum tests, its design underpins future flagship missions targeting Earth-like exoplanets and faint circumstellar structures.

The Roman Coronagraph is an extreme-contrast, space-based high-order wavefront control system aboard the Nancy Grace Roman Space Telescope, designed to demonstrate the technologies and methodologies required for direct imaging and spectral characterization of exoplanets and circumstellar debris disks at visible wavelengths. With raw contrast goals reaching $\sim10^{-9}$ and inner working angles down to $3~\lambda/D$, Roman Coronagraph’s multiple observing modes, two deformable mirrors (DMs), and photon-counting detectors enable system-level validation of methods crucial for future flagship missions (e.g., Habitable Worlds Observatory) targeting Earth-like worlds and faint circumstellar structures (Milani et al., 6 May 2024, Llop-Sayson et al., 2 Dec 2025, Cady et al., 31 Jul 2025, Krist et al., 2023).

1. Instrument Architecture and Operating Modes

The Coronagraph Instrument (CGI) architecture consists of an on-axis 2.4 m telescope primary, dual 48×48 actuator DMs for phase and amplitude correction, fast-steering (tip/tilt) and focus mechanisms, precision filter and mask wheels, and a photon-counting electron-multiplying CCD (EMCCD) detector. The optical train incorporates both Hybrid Lyot Coronagraph (HLC) and Shaped Pupil Coronagraph (SPC) modes, with respective focal-plane and pupil-plane mask architectures. Key specifications are:

• HLC: Imaging mode with IWA $\sim2.8~\lambda/D$, OWA $\sim9.7~\lambda/D$, 10.1% bandwidth at 575 nm, raw contrast floor $\sim3\times10^{-10}$ (no aberrations), core throughput $\sim4.2\%$.
• SPC-WFOV: Wide-field imaging at 825 nm, IWA $\sim5.6~\lambda/D$, OWA $\sim20.4~\lambda/D$, 11.4% bandwidth, core throughput $\sim4.2\%$, mean raw contrast $\sim9\times10^{-10}$.
• SPC-Spec: Spectroscopy at 660 nm, IWA $\sim2.6~\lambda/D$, OWA $\sim9.4~\lambda/D$, 17% bandwidth, core throughput $\sim3.0\%$ (Krist et al., 2023, Milani et al., 2021).

Imaging, low-resolution $R\sim50-70$ spectroscopy, and dual-beam polarimetry (using sequential Wollaston prism assemblies) are all supported, with four contiguous photometric bands spanning $575-825$ nm.

2. Starlight Suppression and Wavefront Control Algorithms

Active starlight suppression relies on high-order wavefront sensing and control (HOWFSC) using the two DMs to create regions of destructive interference (dark holes) in the coronagraphic image plane.

Electric Field Conjugation (EFC) and Implicit EFC (iEFC)

• Model-based EFC employs a linearized optical model with Jacobian matrices $G_1$, $G_2$ mapping DM actuator heights into complex focal-plane electric field amplitudes, minimizing $\|\mathbf{E}\|^2 + \lambda\|\mathbf{A}\|^2$ for DM commands $\mathbf{A}$ (Milani et al., 6 May 2024, Cady et al., 31 Jul 2025).
• Implicit EFC (iEFC) calibrates the Jacobian $G_\mathrm{IEFC}$ in situ by applying DM modal probes and measuring the linear response in pairwise probe-difference images, thus removing explicit reliance on optical modeling and achieving model-error robustness. The control update solves

$$\mathbf{m}_c = -\left(G_\mathrm{IEFC}^\top G_\mathrm{IEFC} + \lambda I\right)^{-1} G_\mathrm{IEFC}^\top \boldsymbol{\delta}$$

where $\mathbf{m}_c$ denotes the DM mode amplitudes, and $\boldsymbol{\delta}$ the probe-difference measurements. Commands are mapped to DM actuators via a modal matrix. The iEFC method has been demonstrated in end-to-end simulations for both monochromatic and broadband (up to 10%) cases on the SPC-WFOV mode, achieving mean normalized intensities down to $\sim 1\times10^{-10}$ in monochromatic control and $\sim 6\times10^{-9}$ in 10%-bandwidth scenarios (Milani et al., 6 May 2024).

Calibration and Stability

Robust calibration of the system PSF and the DM influence function is crucial. For iEFC, achieving a $1\times10^{-8}$ contrast floor requires $\sim$6.8 hours of reference star calibration (Hadamard DM basis, SNR sufficient to suppress probe-noise), rising to $\sim$20 hours for multi-band operation. The direct Jacobian measurement absorbs misalignments and model errors (e.g., lateral mask shear) that would otherwise degrade model-based EFC performance (Milani et al., 6 May 2024).

Thermal-vacuum testing with the flight CGI hardware confirmed contrast below $5\times10^{-8}$ in both HLC and SPC-WFOV over 360$^\circ$ dark holes. The stability requirements (rms drift $<1\times10^{-9}$ over $\sim$44 hours) have been validated in time-resolved tests, demonstrating technical feasibility for future, deeper-contrast missions (Cady et al., 31 Jul 2025).

3. Reference Star Selection and Observation Planning

High-contrast performance depends critically on the properties of reference stars used for wavefront calibration and differential imaging. The formal criteria for reference stars are:

• $V < 3$ (brightness for fast SNR accumulation)
• Angular diameter $\mathrm{UDD}_V < 2$ mas (minimizes incoherent mask leakage due to stellar size)
• No companions within the field at contrasts brighter than $\sim10^{-7}$
• Slew angle $|\Delta\mathrm{pitch}| \leq 5^\circ$ (to limit thermally induced wavefront drift during science-reference star switching)

Preliminary vetting has yielded a sample of 40 primary and 18 reserve references, with ongoing high-resolution AO and speckle imaging campaigns probing for unresolved multiplicity down to $\Delta\mathrm{mag}\sim 6-7$ at $0.6''$ at both 562 nm and 832 nm. These constraints govern the sky coverage and calibration efficiency of Roman and will be directly relevant to future Habitable Worlds Observatory operations (Hom et al., 12 Nov 2025, Wolff et al., 26 Nov 2024).

4. Detector Systems, Calibration, and Performance Limits

The EMCCD detector enables photon counting with negligible read noise. Calibration methodology includes absolute flux calibration (using a set of HST CALSPEC standards, yielding a final error of 1.94% for the planet-to-star flux ratio) (Payne et al., 2022), flatfielding via matched-filter raster scans of extended sources (e.g., Uranus/Neptune) to achieve pixel-to-pixel gain residuals of $\sim0.5\%$ rms (Maier et al., 2022), and periodic trap-pumping/dark-current mapping to monitor radiation-induced charge transfer inefficiency.

The overall error budget allots $<5\%$ fractional rms for the planet/star flux ratio, with dominant contributions from calibration, photon shot noise, and contrast-stability noise (low- and high-order speckle drift). Mean residual contrast floors after calibration and RDI/ADI post-processing are forecast at $10^{-8}$–a few $\times 10^{-9}$ at separations of $4$–$9~\lambda/D$ (Nemati et al., 2023, Llop-Sayson et al., 2 Dec 2025).

5. Simulations, Post-Processing, and Science Yield

Physical optics modeling (PROPER, POPPY) and end-to-end mission simulations have validated all functional instrument modes. Angular and reference-differential imaging, plus Karhunen-Loève (KLIP) and non-negative matrix factorization (NMF) post-processing, are used to suppress residual speckles and extract planet or disk signals. Pointing jitter (nominally $\sim0.3$ mas rms) dominates near the inner working angle, with performance degrading if jitter exceeds $\sim$1 mas rms. Simulated limits on exozodiacal dust detection yield median $5\sigma$ sensitivity to $\sim$12 zodi per resolution element (face-on orientation, $V\leq5$) (Llop-Sayson et al., 2 Dec 2025, Douglas et al., 2021).

Roman’s polarimetry module achieves contrast improvements by a factor of $\sim6$ via polarimetric differential imaging (PDI) and attains polarized contrast floors near $8\times10^{-11}$ in simulations, with $~2\%$ accuracy on $Q/U$ Stokes vector recovery at point-source contrasts of $10^{-8}$ (Doelman et al., 2023, Anche et al., 2022). Debris disks and warm dust are accessible to detailed polarized-scattering analysis.

6. Implications for Future High-Contrast Missions

The Roman Coronagraph establishes the end-to-end methodologies—instrument alignment, wavefront control with two DMs, calibration, and data reduction—required to reach contrast levels necessary for direct imaging of sub-Jovian exoplanets and faint circumstellar structures. Its validation of HOWFSC techniques (including iEFC), in the presence of real instrumental and environmental errors, forms the technical basis for scaling to lower contrast ($<10^{-10}$), smaller inner working angle ($<3~\lambda/D$), and larger apertures in next-generation exo-Earth imaging observatories. The instrument’s architecture, calibration pipeline, and operations planning are directly extensible to missions such as Habitable Worlds Observatory (Krist et al., 2023, Hom et al., 12 Nov 2025, Cady et al., 31 Jul 2025).

---

Summary Table: Core Roman Coronagraph Parameters

Architecture	IWA ($\lambda/D$)	Bandwidth (%)	Core Throughput	Raw Contrast (no aberrations)
HLC	2.8	10.1	4.2%	$3\times10^{-10}$
SPC-WFOV	5.6	11.4	4.2%	$9\times10^{-10}$
SPC-Spec	2.6	17.0	3.0%	$1.1\times10^{-9}$ (bow-tie ROI)

---

Roman’s coronagraphic performance, calibration strategies, and robust error budgeting represent the critical stepping-stone to direct imaging of true Earth analogs in reflected light (Krist et al., 2023, Llop-Sayson et al., 2 Dec 2025, Cady et al., 31 Jul 2025, Milani et al., 6 May 2024).