Cool Planet Imaging Coronagraph (CPI-C)

• CPI-C is a high-contrast imaging instrument on CSST that suppresses stellar diffraction and speckle noise to directly image mature exoplanets and circumstellar material.
• It employs pupil modulation, deformable mirrors, and multi-roll strategies to achieve contrasts near 10⁻⁸ across 600–900 nm with inner working angles <0.55 arcseconds.
• The instrument bridges current coronagraphic capabilities and future Earth-twin missions by pioneering reflected-light exoplanet detection and atmospheric retrieval techniques.

The Cool Planet Imaging Coronagraph (CPI-C) is the dedicated high-contrast imaging instrument of the China Space Station Telescope (CSST), developed to suppress stellar diffraction and speckle noise sufficiently to detect and characterize faint companions and circumstellar material in close angular proximity to nearby bright stars. In the CSST literature, CPI-C is consistently presented as the mission’s direct-imaging facility for mature, low-temperature exoplanets in reflected light, with associated programs on circumstellar disks and zodiacal dust, and as a space-based coronagraphic pathfinder operating in the approximate $10^{-8}$ contrast regime rather than the $>10^{-10}$ regime associated with Earth-analog imaging (Collaboration et al., 7 Jul 2025, Dou et al., 12 Dec 2025).

1. Position within CSST and instrument-level role

CPI-C is one of the five scientific instruments carried by CSST and shares the telescope’s 2 m off-axis three-mirror anastigmat primary optical system, whose focal length is 28 m and whose primary-system dynamic image quality is quoted as $R_{\rm EE80}<0.15''$. The absence of central obstruction in the off-axis CSST architecture is treated as a major coronagraphic advantage, because it reduces diffraction structure and improves achievable throughput and contrast at small working angles. Within the overall payload, CPI-C is described as the only CSST instrument capable of direct imaging of mature, low-temperature “cold” exoplanets around nearby FGK stars, while also complementing survey imaging and spectroscopy from the other instruments through high-contrast imaging of disks and circumstellar dust (Collaboration et al., 7 Jul 2025).

A later dedicated CPI-C description divides the payload into two bays: Module 5, the Scientific Detection Unit, and Module 6, the Control and Power Supply Unit. Module 5 houses the coronagraph optics, visible camera, Shack–Hartmann wavefront sensor, kilo-actuator deformable mirror, tip–tilt mirror, and local electronics; Module 6 contains the near-IR camera and filter wheel, instrument computers, and thermal and power-control subsystems. The same source describes CPI-C as receiving roughly 10% of CSST’s total observing time over a 10-year mission, whereas the earlier CSST overview states that CPI-C observes for about 4.5 months during the scientific operation period, excluding parallel observations. This suggests that the public programmatic description evolved between the mission-level review and later instrument-specific planning documents (Dou et al., 12 Dec 2025, Collaboration et al., 7 Jul 2025).

2. Scientific scope and target population

CPI-C’s core exoplanet objective is to search for mature Jupiter-like planets and super-Earths around nearby solar-like stars from the habitable zone to the snow line, with the principal orbital range stated as $0.8$–$5$ AU in the CSST review and $0.5$–$5$ AU in the dedicated instrument paper. The principal host-star population is nearby FGK stars within 40 pc. One CSST review passage refers to “thousands of FGK-type stars within 40 pc of the Sun” as the main target set, while another states that CPI-C is expected to conduct a high-contrast imaging survey of “1,000 nearby stars”; later catalog-building work describes a preselected sample of roughly 700 nearby FGK stars, mostly with $V\sim 0$–7 mag, assembled from Hipparcos or SIMBAD-based criteria (Collaboration et al., 7 Jul 2025, Dou et al., 12 Dec 2025, Zhu et al., 13 Nov 2025).

The reflected-light regime is central to this science case. In the CPISM simulation framework, the planet/star contrast is written as

$$\text{contrast}_p = \frac{f_{\text{planet}}}{f_{\text{star}}} = A_g \left( \frac{R_p}{r} \right)^2 \left[\frac{\sin (\theta)+(\pi-\theta) \cos (\theta)}{\pi}\right],$$

with $A_g$ the geometric albedo, $>10^{-10}$0 the planet radius, $>10^{-10}$1 the 3D star–planet separation, and $>10^{-10}$2 the phase angle. This formulation encodes the basic scientific premise of CPI-C: mature cool giants are faint thermally but can remain accessible in reflected light if stellar suppression, angular separation, and photometric precision are sufficient (Zhu et al., 13 Nov 2025).

The non-planetary science program is also fundamental rather than auxiliary. Mission-level descriptions state that CPI-C will carry out high-contrast imaging detection and studies on circumstellar disks and zodiacal dust, providing observational evidence relevant to planet formation and evolution. In the broader CSST exoplanet program, CPI-C is positioned as the direct-imaging counterpart to microlensing, transits, and slitless or integral-field spectroscopy, with particular leverage on non-transiting outer planets and on disk morphology in scattered light (Collaboration et al., 7 Jul 2025).

3. Coronagraphic architecture and wavefront-control concept

The public description of CPI-C’s coronagraphic principle is consistent at a high level: the instrument adopts pupil modulation and high-precision wave aberration correction technology to suppress diffraction photons and quasi-static speckles generated by the primary optical system. The CSST review ties this architecture to pupil-modulation approaches associated with Ren and Dou and emphasizes that the combination of pupil modulation and precise wave-aberration control is what produces the ultra-high-contrast imaging workspace required for exoplanet detection (Collaboration et al., 7 Jul 2025).

The most detailed optical description appears in the dedicated instrument paper. There, CPI-C is implemented as a multi-pupil relay in which the beam is first folded and calibrated, then passed through a tip–tilt mirror at Pupil 1, a kilo-actuator deformable mirror at Pupil 2, a 31-step transmission apodizer at Pupil 3, and a Shack–Hartmann wavefront-sensing branch with a microlens array at Pupil 4. The science channel includes an intermediate image-plane mask and a visible/IR split that feeds a visible EMCCD camera and a near-IR InGaAs camera. The step-transmission apodizer is the principal diffraction-suppression element, while the deformable mirror and tip–tilt mirror provide active correction of residual phase errors and image motion (Dou et al., 12 Dec 2025).

The simulation paper "CPI-C -- Instrument Simulation" describes the high-contrast optics in HCIPy as a transmission apodizing filter, a 952-actuator MEMS deformable mirror, and a focal-plane mask, propagated with Fraunhofer optics. In that implementation, the apodizer is a stepped-transmission pupil mask with 32 strips, and the dark-hole region is generated numerically by Electric Field Conjugation rather than by the laboratory optimization method cited for the physical instrument. The same paper explicitly states that with aberrations but without DM correction, contrast degrades to $>10^{-10}$3 at 662 nm, whereas DM shaping via EFC restores a simulated dark zone at $>10^{-10}$4 between about $>10^{-10}$5 and $>10^{-10}$6 (Zhao et al., 12 Nov 2025).

Wavefront control is described as two-layered in the later instrument paper. First, the Shack–Hartmann wavefront sensor and the deformable mirror remove low-order aberrations, with the paper specifying correction of the first 66 orders of low-order aberrations, excluding tilt and piston, and reduction of absolute phase aberration to below $>10^{-10}$7 RMS. Second, a dark-hole optimization loop between the science camera and the deformable mirror suppresses residual quasi-static speckles by about two additional orders of magnitude. This produces the stated transition from a static $>10^{-10}$8 level at $>10^{-10}$9 to the $R_{\rm EE80}<0.15''$0 design regime (Dou et al., 12 Dec 2025).

4. Published performance envelope, band definitions, and observing modes

Several public CPI-C descriptions report the same contrast ambition but not a single frozen parameter set. The CSST review states that CPI-C achieves ultra-high contrast imaging of exoplanets “$R_{\rm EE80}<0.15''$1 in 600–900 nm,” with “the high-contrast inner working angle (IWA) not greater than $R_{\rm EE80}<0.15''$2 at 633 nm.” The later instrument paper states a contrast requirement better than $R_{\rm EE80}<0.15''$3 at an IWA of $R_{\rm EE80}<0.15''$4–$R_{\rm EE80}<0.15''$5 in the visible, and gives a representative design point of $R_{\rm EE80}<0.15''$6 at 661 nm for $R_{\rm EE80}<0.15''$7, with an outer working angle near $R_{\rm EE80}<0.15''$8 at the same wavelength. CPISM-based simulation papers model the dark-hole region as approximately $R_{\rm EE80}<0.15''$9–$0.8$0, or equivalently $0.8$1–$0.8$2 depending on the band and simulation setup. This suggests that the public description moved from top-level mission requirements toward more implementation-specific parameterizations as the design matured (Collaboration et al., 7 Jul 2025, Dou et al., 12 Dec 2025, Zhu et al., 13 Nov 2025, Zhao et al., 12 Nov 2025).

The sensitivity statements are similarly layered. The CSST review gives a concise reflected-light benchmark: with an exposure time of more than 1 hour, CPI-C can achieve $0.8$3 for a 25 mag planet around a 5 mag star. The dedicated instrument paper quotes point-source sensitivities of $0.8$4 in 30 s for a 20th magnitude source in the visible and a 15th magnitude source in the near-IR, under stated throughput and background assumptions. These numbers serve different purposes—deep coronagraphic planet detection versus top-level point-source sensitivity—and therefore should not be conflated (Collaboration et al., 7 Jul 2025, Dou et al., 12 Dec 2025).

The observing mode is explicitly a staring mode rather than scanning. Multiple spacecraft roll angles are part of the baseline speckle-calibration strategy: after one staring sequence, the telescope can be rotated around the visual axis so that astrophysical sources move relative to the quasi-static speckle field, enabling ADI-like self-referencing in space. Mission descriptions further note that observation scheduling must account for these continuous multi-roll sequences and their field-of-view constraints (Collaboration et al., 7 Jul 2025).

Public band definitions are not completely uniform. CPISM target-simulation papers describe eight imaging bands centered at 520, 662, 720, 850, 940, 1265, 1425, and 1542 nm. The dedicated instrument paper instead defines a compact seven-band photometric system with visible bands F661, F729, F877 and near-IR bands F1040, F1265, F1425, F1532. A plausible implication is that the filter set remained under optimization across successive public releases, even though the scientific logic—sampling optical methane-sensitive continuum and absorption features plus near-IR continuum and the $0.8$5 water/methane complex—remains consistent (Zhu et al., 13 Nov 2025, Dou et al., 12 Dec 2025).

5. Detectors, calibration chain, and simulation ecosystem

CPI-C’s visible science channel is modeled and described as an EMCCD-based camera, while the near-IR channel uses an InGaAs detector. The instrument-simulation paper gives a visible EMCCD format of 1024$0.8$61024 pixels with a plate scale of $0.8$7 pix$0.8$8, a unity-gain read noise of 160 e$0.8$9 pix$5$0, dark current of $5$1 e$5$2 pix$5$3 s$5$4, and clock-induced charge of 0.2 e$5$5 pix$5$6 frame$5$7. The dedicated instrument paper quotes essentially the same visible scale and adds near-IR values: a 640$5$8512 InGaAs array at $5$9 pix$0.5$0, with 50 e$0.5$1 pix$0.5$2 read noise and 10 e$0.5$3 pix$0.5$4 s$0.5$5 dark current (Zhao et al., 12 Nov 2025, Dou et al., 12 Dec 2025).

The wavefront sensor uses a separate EMCCD whose calibration has been reported in detail. That detector is an 8-channel EMCCD system with 14-bit digitization, an exposure range of 3 ms to 5 s, and an operating temperature study at $0.5$6C, $0.5$7C, and $0.5$8C. The recommended baseline is $0.5$9C, where the usable EM-gain range is 1–150$5$0, inter-channel non-uniformity at maximum gain is about 9.35%, and the minimum reported Noise-Equivalent Photon Count is $5$1. That study ties detector performance directly to the $5$2 wavefront-sensing regime required by $5$3 contrast imaging (Dou et al., 25 Nov 2025).

A substantial part of the CPI-C development environment is the end-to-end simulator CPISM. In the target-selection paper, CPISM is organized into five modules: target simulation, imaging simulation, observational effects, camera simulation, and data-product generation. It uses CK04 stellar atmospheres, Batalha et al. (2018) reflected-light models, Fourier optics for the coronagraph, EMCCD detector physics, HST-like cosmic-ray models, and FITS-based Level 0 outputs. In the instrument-simulation paper, the same framework uses HCIPy for the coronagraph and EFC for simulated dark-hole creation, and it includes detailed modeling of striping, interference patterns, bias drift, CIC, multiplication noise, and temperature-dependent EM gain (Zhu et al., 13 Nov 2025, Zhao et al., 12 Nov 2025).

Engineering validation has also been reported at the structural level. Mechanical-test-platform work for the scientific probe module gives a measured test-platform fundamental frequency of 436.2 Hz, exceeding a design requirement of 300 Hz, with vibration and response-surface analyses intended to support launch-environment qualification of the coronagraph hardware (Kong et al., 20 Nov 2025).

6. Data products, reduction logic, and scientific exploitation

At the mission-pipeline level, the public CSST review defines CPI-C Level-1 data succinctly: instrumental-effect correction, background-light removal, cosmic-ray removal, and related processing are used to generate calibrated image products. The review does not enumerate a standard Level-2 planet-catalog product for CPI-C, unlike some of the other CSST instruments, and therefore leaves higher-level high-contrast reductions to more specialized analysis stages (Collaboration et al., 7 Jul 2025).

Those specialized stages are described in the simulator and science-case papers. CPISM generates Level 0 raw frames, while downstream science analysis uses roll-based ADI-like subtraction, reference differential imaging, and optimized subtraction algorithms. Public CPI-C texts explicitly mention ADI/LOCI-like methods, O-IRS, and G-RDI; the Alpha Centauri mock-observation study performs RDI following Ren & Chen (2021), using an SVD-based global optimization to build a stellar background model from star-only frames. In that simulation, a hypothetical $5$4 planet at $5$5 from Alpha Centauri B is recovered with measured contrasts from $5$6 to $5$7 and $5$8 values between 4.451 and 8.870 across four optical bands (Zhu et al., 13 Nov 2025, Dou et al., 12 Dec 2025).

The scientific interpretation stage is explicitly framed as multi-band reflected-light retrieval. CSST documents state that CPI-C can obtain images of exoplanets in multiple bands from optical to near-IR, and then derive atmospheric spectra and physical properties through spectral fitting, including cloud layer, abundance, radius, effective temperature, surface gravity, and mass. They also state that multiple observations at different times can be used to fit planetary orbits and determine dynamical masses (Collaboration et al., 7 Jul 2025).

Disk science follows a parallel path. The $5$9 Eridani simulations use MCFOST-generated scattered-light disk images passed through CPISM and show that CPI-C can resolve debris-disk structures down to approximately 3 au, recover inclinations and radial extents close to the input models, and distinguish a narrow ring from a broad continuous inner disk. In the same study, $V\sim 0$0 Eri b is not detected in conservative total-intensity simulations, but polarimetric methods are explored using two fixed linear polarizers at $V\sim 0$1 and $V\sim 0$2 with 1% instrumental-polarization leakage in the simulation model. The outcome is deliberately cautious: polarimetry may help, but robust planet detection would require longer exposures, more sophisticated post-processing, or improved calibration (Bao et al., 20 Sep 2025).

7. Broader coronagraphic context and significance

CPI-C belongs to the same technological class as other visible-wavelength space coronagraph efforts, but its public positioning is specific. The Roman Coronagraph Instrument white paper describes space-based visible coronagraphy at raw contrasts of $V\sim 0$3–$V\sim 0$4 as a “critical intermediate step” between ground-based systems fundamentally limited to about $V\sim 0$5 at small working angles and future flagship missions targeting $V\sim 0$6 contrast for Earth analogues. CPI-C’s quoted contrast floor, wavelength coverage, and reflected-light cool-planet science case place it squarely in that intermediate regime: beyond current ground-based small-angle visible performance, but not a direct Earth-twin mission (Bailey et al., 2019).

The subsystem logic is also consistent with the general coronagraph literature. Reviews of coronagraphic instrumentation stress that the limiting floor comes from residual speckles induced by wavefront aberrations and that practical high-contrast performance depends on the integrated design of telescope pupil, coronagraph, wavefront sensing, deformable mirrors, and differential post-processing. CPI-C’s published architecture—off-axis unobstructed aperture, pupil apodization, focal masking, deformable-mirror dark-hole creation, EMCCD-based sensing and science imaging, and roll-based plus reference-based speckle subtraction—fits exactly within that systems-level framework (Galicher et al., 2023).

Historically, CPI-C also occupies the same lineage as earlier “cool planet” coronagraphic concepts that combined coronagraphy, active wavefront correction, and multi-channel or multi-band analysis to reach methane-bearing or reflected-light giant planets. Earlier high-contrast programs such as Project 1640 demonstrated the instrumental logic of coronagraph + wavefront calibration + IFS-based speckle suppression for cool companions, although at less demanding contrast levels and in a ground-based near-IR context (Hinkley et al., 2010).

Within CSST, however, CPI-C has a more specific significance. Mission documents describe it as the first space-based instrument capable of directly imaging the reflection light from cool exoplanets in visible wavelengths, while also establishing a direct link to future Earth-mass planet imaging by developing $V\sim 0$7-class space coronagraphy, exozodiacal-dust characterization, and multi-band atmospheric retrieval of mature giants. The numerical gap between $V\sim 0$8 and $V\sim 0$9 remains substantial, but the CPI-C literature consistently treats the instrument as a technological and scientific bridge toward that deeper regime rather than as a substitute for it (Dou et al., 12 Dec 2025, Collaboration et al., 7 Jul 2025).