Blocking starlight

15 chapters, each an animated explainer with its physics and sources. Every drawing says whether it is an illustration, a model from stored values, or real data, and by how much it is exaggerated. Only the transit chapter is something SetuHQ itself does.

Chapter 7 of 15

Blocking starlight

To photograph a planet, the starlight must be blocked. A Jupiter twin reflects about a billionth of its Sun's visible light; an Earth twin about 10⁻¹⁰ (Lawson et al. 2013). Young giant planets glow in the infrared and are far easier, which is why imaging finds young, wide, massive planets.

Illustration
Starlight blocking, in four scenes. A sun-like occulter (an illustration based on SOHO LASCO); a coronagraph mask sliding over a star so the glare collapses into residual speckles and the four HR 8799 planets appear on schematic orbits; a starshade flying far in front of a telescope, not to scale; and a two-telescope nulling interferometer whose dark fringe cancels the star.

Blocking starlight

  1. History. Bernard Lyot invented the coronagraph in 1930 to study the Sun's corona without an eclipse; a focal-plane mask blocks the disc and a "Lyot stop" removes the diffracted light (Lyot 1939). SOHO's LASCO coronagraphs show the same idea every day (Brueckner et al. 1995). The Sun scene above is an illustration based on LASCO, not a LASCO image.
  2. Coronagraph families. Lyot-type amplitude masks, apodised pupils and phase masks such as the vortex. The key figure is the inner working angle, in units of λ/D (Mawet et al. 2012). Earth at 10 pc sits 0.1″ from its star: about 5.3 λ/D for a 6 m telescope at 550 nm [calc].
  3. Adaptive optics and speckles. A deformable mirror flattens the turbulent wavefront hundreds of times a second. What remains are speckles; angular differential imaging lets the sky rotate while speckles stay fixed, so they can be subtracted (Marois et al. 2006).
  4. Starshade. A flower-shaped screen flown far in front of a telescope (Cash 2006). The Roman Starshade Rendezvous concept: 26 m at 26,000 km; HabEx: 52 m at about 76,600 km (HabEx Final Report).
  5. Nulling interferometry. Two telescopes combined half a wave out of phase cancel the star on a dark fringe while an off-axis planet sits on a bright one (Bracewell 1978; the LIFE concept, Quanz et al. 2022).

Coronagraph families, plane by plane

Each family below is drawn as four planes of one optical train. The intensities are computed, not painted: Fraunhofer optics of an ideal, monochromatic telescope with a clear circular pupil, shown on a log stretch from the star's peak down to a millionth. The mask and stop sizes, the apodiser and the planets' brightness are illustrative and say so.

A · objectiveB · occulting discC · field lensD · Lyot stopE · camera lensF · imagecompanionstarlightB: star image, disc on itD: bright rim; stop insideF: the companion survivesA · objectiveB · occulting discC · field lensD · Lyot stopE · camera lensF · imagecompanionstarlightstar image, disc on itbright rim; stop insidethe companion survives
Illustrationnot to scale; after Lyot 1939
Plate: Lyot's coronagraph (1930), stylised. Starlight (amber) is stopped at the disc (B); the light the disc's edge diffracts (dashed) reaches the rim of the pupil image and is caught by the stop (D); a companion's light (blue) misses the disc and passes. Ray paths are schematic.
Illustration
Four coronagraph families as optical trains, each in four planes with a log-intensity profile below. Lyot: an occulting disc blocks the star's image, the light it diffracts lands on the rim of the re-imaged pupil, and a closing Lyot stop removes that rim so the starlight beside the mask falls by a factor of hundreds. Apodised pupil: the pupil's transmission tapers to its edge, the image's rings fade and the core widens. Vortex: a phase ramp that winds twice around the centre throws all of an on-axis star's light outside the pupil, where the stop removes it. Adaptive optics with angular differential imaging: synthetic frames in which speckles stay fixed while the planet turns with the sky; the median of the frames is subtracted, each frame is turned back and the stack reveals the planet.
  1. Lyot coronagraph. An opaque disc in the focal plane blocks the core of the star's image. What it does not remove is the light its edge diffracts, which the next lens sends to the rim of the re-imaged pupil; a Lyot stop slightly smaller than the pupil catches that rim (Lyot 1939). With a 4 λ/D disc, closing the stop from 150% to 80% of the pupil cuts the starlight at 8 λ/D from 6.7×10⁻⁵ to 1.8×10⁻⁷ of the star's peak (no coronagraph: 7.8×10⁻⁵) [calc], while a planet well outside the disc keeps 64% of its light, the stop's area.
  2. Apodised pupil. Instead of a mask, the pupil's transmission tapers to its edge, so the image has no sharp edge to ring from. For A(ρ) = (1 − ρ²)² the brightest ring near 5 λ/D falls from 7.8×10⁻⁴ to 1.2×10⁻⁶ of the clear pupil's peak [calc], but only 1/(2n + 1) = 20% of the light gets through and the core is wider. Real apodisers (shaped pupils, prolate and APLC designs) are optimised numerically; this taper is illustrative.
  3. Vortex phase mask. A transparent mask whose phase winds around its centre (e^(2iθ), charge 2) moves an on-axis star's light outside the re-imaged pupil: for an ideal charge-2 vortex on a clear circular aperture the field there is exactly zero inside the pupil and (1/ρ)² outside it [calc], so a stop just inside the pupil removes the star. Phase masks are the family that works closest to the star (Mawet et al. 2012); real vortices leave residuals from wavefront errors and the star's finite size.
  4. Adaptive optics and ADI. A deformable mirror turns the seeing halo into a sharp core, but quasi-static speckles λ/D wide remain. With the instrument's pupil fixed, the sky and the planet turn through the night while the speckles stay; the median of the sequence models the speckles, and each frame minus it, turned back by its field rotation and stacked, keeps the planet (Marois et al. 2006). The frames on the stage are synthetic. On the sky, GPI reaches about 10⁻⁴ raw contrast at 0.4″ in 60 s and more than ten times deeper after an hour of ADI (Bailey et al. 2016).

The contrast ladder

  1. JWST MIRI coronagraph 2e-04at 1″, 11.3 µm (HIP 65426) · achieved · Carter et al. 2023, ApJL 951, L20
  2. Ground AO (GPI), raw 1e-045σ at 0.4″, H band, 60 s · achieved · Bailey et al. 2016
  3. JWST NIRCam coronagraph 1e-055σ at 1″, 4.4 µm (HIP 65426) · achieved · Carter et al. 2023, ApJL 951, L20
  4. Roman Coronagraph 1e-07requirement at 6–9 λ/D; goal 10⁻⁹ · requirement · Roman Coronagraph requirement
  5. Earth twin, reflected light 1e-10Sun-like star, visible light · target · Lawson et al. 2013, SPIE 8864
  6. HWO 1e-10Earth twins around Sun-like stars · goal · Lawson et al. 2013, SPIE 8864

HR 8799's planets on the canvas use exoplanet.eu's semi-major axes and eccentricities; their orientation on the sky is illustrative, and time is compressed. Contrasts from different instruments are measured at different separations, wavelengths and significances, so the rungs are not directly comparable.

1,216 of the 8,049 planets with a stored distance in exoplanet.eu were found by imaging (the first method listed). Show them on the Galaxy map ›

peak 118 (2024)

19882026

Planets found by imaging per year, 1988–2026: 1,216 in all; the most in one year was 118 in 2024. Source: exoplanet.eu, fetched 2026-10-18.
Sources

↑ ↓ to move · Enter to open · Esc to close