Roman Science

Exoplanets : Direct Imaging

banner credit : NASA/W. Stenzel

Our understanding of the internal structure, atmospheres, and evolution of planets was originally developed through models that were tuned to explain the detailed properties of the planets in our own solar system.

Beta Pictoris b An image of Beta Pictoris b, a giant planet several times larger than Jupiter, from the Gemini Planet Imager (GPI), a coronagraph on the Gemini Observatory ground based telescope. This young, hot planet glows in infrared light as heat is released from its formation. GPI is sensitive to large,Jupiter-size planets beyond ~5 AU. The Roman Space Telescope provides the first opportunity to observe and characterize older planets,like those in our solar system, in reflected starlight down to Neptune-size planets lying from 3 to 10 AU from their parent star. (GPI/Gemini Observatory/AURA, Image processing by Christian Marois, NRC Canada.)

Surveys of exoplanetary systems have led to the realization that there exists a diversity of worlds with very different properties and environments than those in our solar system. Models of planet formation and evolution have had to be expanded and generalized to explain the properties of these new worlds, often including new and uncertain physics. Our understanding of these new worlds therefore remains primitive. The best hope of understanding the physical properties of this diversity of worlds is through comparative planetology: detailed measurements of, and comparisons among, the properties of individual planets and their atmospheres. Understanding the structure, atmospheres, and evolution of a diverse set of exoplanets is an important step in the larger goal of assessing the habitability of Earth-like planets discovered in the habitable zones of nearby stars. It is unlikely that any such planets will have exactly the same size, mass, or atmosphere as our own Earth. Measuring a large sample of systems with a range of properties will be necessary to understand which properties permit habitability and to properly interpret these discoveries.

Direct imaging provides the critical approach to studying the detailed properties of exoplanets. Images and spectra of directly imaged planets provide some of the most powerful information about the structure, composition, and physics of planetary atmospheres. This information can in turn help scientists better understand the origin and evolution of these systems. The direct imaging technique is also naturally applicable to the nearest and brightest, and thus best-characterized, solar systems.

Advancing the technology for direct imaging of exoplanets was the top priority medium-scale space investment recommended by NWNH. Coronagraphy on the Roman Space Telescope will be a major step towards the long-term goal of a mission that can image habitable Earth-mass planets around nearby stars and measure their spectra for signs of life.

Exoplanets are orders of magnitude fainter than their parent stars. A coronagraph is a set of optical elements that suppresses the star's light to create a region where a dim planet can be extracted. Due to optical imperfections in any system, all coronagraphs must be designed together with wavefront control via one or more deformable mirrors (DMs).

The combined coronagraph and wavefront control system is characterized by the contrast, inner working angle, and stability achieved. Contrast is the degree to which the instrument can suppress scattered and diffracted starlight in order to reveal a faint companion. Inner Working Angle (IWA) is the smallest angle on the sky at which it can reach its designed contrast. This angle is typically only a few times larger than the theoretical diffraction limit of the telescope. The resulting residual stellar halo must also be stable over the time scale of an observation, so that the halo can be subtracted to reveal an exoplanet or disk.

In principle, with a well-characterized and stable point spread function, various subtraction methods that have been developed and used on both ground and space images can be employed to average the background photon noise and extract faint planets that are below the raw contrast level. The recent history of planet imaging shows that recovering planets with up to factors of 10 fainter contrast than the background is regularly accomplished, both on large ground telescopes and from the Hubble Space Telescope. We thus characterize the coronagraph instrument by its detection limit, that is, the limiting magnitude of a recoverable planet relative to the star from the combination of the coronagraph and wavefront control and data processing.

Simulated CGI images with two Jupiter-size exoplanets

Simulated CGI images with two Jupiter-size exoplanets detected as point sources around a Sun-like star at a distance of 14 pc. These simulations incorporate structural-thermal-optical performance (STOP) models of the observatory, and instrument control loops compensating for wavefront errors and line-of-sight pointing jitter. They represent the data expected from CGI’s Hybrid Lyot Coronagraph mode operating in a bandpass centered at 575 nm, with 30-hour integration times, and 3 observations spaced apart at 2-year intervals. The dashed curves trace the orbital trajectories of the two planets; the flux scale is in units of co-added photoelectrons.

Coronagraph Instrument Context and Performance Predictions

Only a handful of exoplanets have been imaged to date from the ground and with the Hubble Space Telescope. We have obtained spectra on an even smaller number. All of them are gas giants more massive than Jupiter that reside at great distances (> 10 AU) from their parent stars. Additionally, observational biases have restricted those detections to very young planets — typically less than a hundred million years — that shine brightly at infrared wavelengths through their interior heat. This warm phase constitutes only a small fraction of a planet’s lifetime immediately following its formation. Therefore, the far more numerous population of mature, evolved planets has eluded the current generation of high-contrast imaging instruments. The Roman Space Telescope provides the first opportunity to observe and characterize planets physically resembling those in our Solar System, spanning ages up to several billion years. The coronagraph on the Roman Space Telescope will operate in visible wavelengths at flux ratios down to a few parts per billion and an inner working angle of less than 0.2 arcseconds. It will be capable of imaging a dozen known radial velocity planets in reflected starlight at orbital separations ranging from 2 to 10 AU.

Predicted CGI performance in the context of known planets and existing and planned high-contrast instruments

Predicted Roman Space Telescope CGI performance in the context of known planets and existing and planned high-contrast instruments. The y-axis indicates the flux ratio between a planet and its host star (for individual planets) or between the star and the dimmest source detectable at 5s after post-processing (for instrument performance curves). The lower x-axis is projected separation in arcseconds, and the upper x-axis is the corresponding physical separation for the Tau Ceti system. Points and lines are color-coded by wavelength of observation. Solid and dashed lines are 5s point source detection limits versus separation from the host star; these limits are calculated from post-processed data. The predicted performance for the future observatory, JWST, is plotted as a dashed line. Lines labeled "CGI req" are the Roman Space Telescope level 2 requirements. Lines labeled "CGI pred." are current best performance predictions. Black triangular points are estimated reflected light flux ratios for known gas giant radial velocity-detected (RV) planets at quadrature, with assumed geometric albedos of 0.5. Red squares are 1.6 mm flux ratios of known self-luminous directly-imaged (DI) planets. Dotted lines connect each DI planet’s known 1.6mm flux ratio to its predicted flux ratio at 750 nm (yellow diamonds) or 550 nm (blue circles), based on COND or BT-Settl planet evolutionary models. Cyan points represent the reflected light flux ratios of Earth and Jupiter at 10 pc as well as super-Earths Tau Ceti e and f. Figure created by Vanessa Bailey; for plot source code and further details on the input assumptions, see

Spectroscopic Characterization

Spectroscopic observations of directly imaged exoplanets are challenging, partly due to the extremely low irradiances of planets observed in reflected starlight. Nevertheless, because exoplanet characterization is the chief scientific motivation for advancing high-contrast imaging technology, the Roman Space Telescope Coronagraph Instrument is designed to include a modest spectroscopy capability. The prism spectroscopy mode on CGI will disperse the point spread function of an individual exoplanet over the wavelength range 675—785 nm with a spectral resolving power of l/Dl³50. For Jupiter-sized planets with flux ratios in the detection range of CGI, this spectroscopy mode will measure the shape and depth of the methane absorption feature at 720 nm. In conjunction with atmosphere modeling, such observations can place rudimentary constraints on methane gas abundances and cloud properties.

Model spectra of gas giant exoplanets at visible wavelengths

Model spectra of gas giant exoplanets at visible wavelengths, illustrating the effect of different metallicity levels and cloud layer species on the bulk atmosphere albedo (courtesy of Natasha Batalha and Mark Marley).

This data simulation shows the noisy recovered spectrum for a Jupiter-size planet at flux ratio 5´10-9 observed for 400 hours

The spectroscopic observing mode of the Roman Space Telescope CGI will enable characterizing the 720 nm methane absorption feature at a spectral resolving power of l/Dl=50. This data simulation shows the noisy recovered spectrum for a Jupiter-size planet at flux ratio 5´10-9 observed for 400 hours.

Circumstellar Debris Disks

Simulated images of faint debris disks

Simulated images of faint debris disks observed with CGI’s Hybrid Lyot Coronagraph observing mode. Left: a disk with a surface brightness of 7x the zodi (SB~19 mag/sq. arcsec). Right: a an annular disk similar to t Ceti with an inner radius of 0.3” and a peak SB of ~22 mag/sq arcsec. Figure credit: John Debes and Ewan Douglas.

The coronagraph will also be sensitive to debris disks with a few times the solar system's level of dust in the habitable zones of nearby (~10 pc) sun-like stars. The high sensitivity and spatial resolution (0.05 arcsec is 0.5 AU at 10 pc) of the Roman Space Telescope CGI images will map the structure of these disks, revealing asymmetries, belts, and potentially gaps due to unseen planets. CGI will make the most sensitive measurements yet of the amount of dust in or near the habitable zones of nearby stars. This is important for assessing the difficulty of imaging Earth-like planets with future missions as well as for understanding nearby planetary systems.

NASA logo


NASA Official: Julie McEnery
Media Inquiries: Claire Andreoli
Website Curator: Jennifer Brill

Privacy Policy privacy

Stay Connected