ASTRONOMY: Coronagraph can image an exo-Earth

Is there life beyond Earth? This is one fundamental question that science may one day be able to answer.

Th 0706lfw N6extrasolar

Is there life beyond Earth? This is one fundamental question that science may one day be able to answer. Efforts such as NASA’s Terrestrial Planet Finder (TPF) mission, designed to find Earth-like extrasolar planets and capture their light for spectroscopic analysis, could sense faraway life. Although spending limits passed by the U.S. Congress earlier this year have effectively killed the TPF, scientific curiosity is not vanquished so easily, and so research aimed at future missions goes on.

In a prime example, researchers at the Jet Propulsion Laboratory, California Institute of Technology (Pasadena, CA) have demonstrated a coronagraph in the laboratory that, if combined with a space telescope, would be capable of imaging an Earth-like planet orbiting another star.1 The setup, a Lyot coronagraph with wavefront-correcting adaptive optics, relies on multiple images combined with image processing to reduce the background noise level to 0.1 × 10-10 times the intensity of the central star, substantially lower than the 1 × 10-10 intensity of an “exo-Earth.”

The test “star” for the setup was a 785-nm-emitting laser diode (while its monochromatic wavelength does not represent the broadband conditions of an actual space coronagraph, modeling and some initial experimentation with broadband light point the way to good performance under real-life requirements). The star is imaged via a pair of paraboloids to a first image plane containing a centrally obscuring graded-transmission coronagraph mask; a deformable mirror is located between the paraboloids. A second image plane holds a Lyot mask (two partially overlapping circular apertures); a camera is located at the third and final image plane.

The deformable mirror (DM) is necessary to reduce speckle, which arises from tiny mid-spatial-frequency (a few to several tens of) ripples in the other mirrors; in a space telescope, the speckle would arise mostly from the primary mirror. The DM, which has 32 × 32 actuators, must correct the wavefront to λ/10,000 (root mean square) to reduce speckle to acceptable levels.

The Lyot configuration results in a D-shaped low-noise portion of the image field to one side of, and blocking, the central star. A single static image of this field has a noise level of 6 × 10-10 the intensity of the central star. To reduce this to 0.1 × 10-10, the researchers rely on “roll deconvolution,” in which the D-shaped field is rotated about the central star (imitating a space telescope rotating on its axis), carrying the speckles along with it. The resulting annular image field is a composite of many individual images. Test “planets” consisting of highly attenuated versions of the laser star were created, mimicking an exo-Earth (1 × 10-10), an exo-Jupiter (10 × 10-10), and half an exo-Jupiter (5 × 10‑10). The roll-deconvolved image clearly showed all three planets (see figure).

Th 0706lfw N6extrasolar
Three “planets” (attenuated versions of a test “star”) are projected into a Lyot coronagraph that contains a deformable mirror to reduce speckle (left). The coronagraph’s D-shaped field is rotated and many images of a field filled with residual speckles are taken (center). Roll deconvolution removes the speckles (which rotate with the field), revealing the three planets, one with the intensity of an exo-Earth (right). (Courtesy of Jet Propulsion Laboratory)
Click here to enlarge image

Phase and amplitude correction

The DM actually corrects for phase and amplitude variations. “This is possible because phase ripples, introduced by the DM, will become a mixture of phase and amplitude ripples as the wavefront propagates forward through the optical system,” says John Trauger, one of the researchers. “The reason is elegantly buried in the hermitian property of the Fourier transform, which is used to describe the propagation of the complex electric field from the pupil (where the deformable mirror is located) to the focal plane. We can always represent the phase corrections on the deformable mirror as the sum of even and odd functions, which transform as even-real and odd-imaginary functions. From this, we can construct a set of phase ripples on the DM that will correct both amplitude and phase errors in the focal plane with a single DM, if we restrict ourselves to one half (one side) of the field of view.”

Trauger notes that it is also possible to correct the full field of view (that is, both sides of the star) by using a pair of DMs suitably placed in the optical system. “This experiment is planned for later this year,” he says.

The field of view is roughly 1 arcsecond around the star. “Nearly all exoplanets (Earth-twins and Jupiter-twins) will be well within this radius,” says Trauger. The speckle background at the image plane can be pushed even lower with any combination of more DM actuators, lower-noise DM drivers, and better speckle-nulling algorithms, he adds.

John Wallace


1. J.T. Trauger and W.A. Traub, Nature 446, 12 (April 2007).

More in Optics