First author

Designing a space telescope to detect Earth-like planets poses a number of technical challenges, most notably achieving contrast levels high enough to image a planet that appears 10 billion times fainter than the star it orbits. John Trauger, an astrophysicist at the Jet Propulsion Laboratory of the California Institute of Technology in Pasadena, and his colleagues have now reached those imaging limits in the laboratory using a coronagraphic instrument with a telescope that uses masks and a deformable mirror to suppress the star's glare (see page 771).

How does your model work?

If we can separate a planet's reflected light from that of its star we can not only detect that planet, but also analyse its spectrum to learn what is in its atmosphere. This process poses two main problems. One is the diffracted light caused by the edges of the telescope's primary mirror. Adding a Lyot-type coronagraph to the telescope is the simplest way to remove this. The other problem is fixing the irregularities in the light wavefront that come from flaws on the mirror's surface. Even tiny ripples will create a 'speckle' in the image that looks just like the signal from a planet. Because there are many randomly ordered errors on any mirror, the true signal is overwhelmed.

How do you resolve this problem?

We use a deformable mirror to correct aberrations to a really high accuracy, within one ten-thousandth of a wavelength. Such mirrors can change shape very precisely to perfectly correct for flaws in a primary mirror, making the number of speckles manageable.

How would your model scale up for use in space?

The apparatus is the same size as it would be in a flight instrument, about 1.5 by 2 metres. But in terms of accuracy it's not quite ready. It is a very simple system, with no more than six or seven mirrors between the star and the coronagraph mask. It could be launched with a telescope as one unit, and could be tested on the ground for performance. This contrasts with other systems that would require precise formations of several spacecrafts, which has never been done.

What do you think our chances are of finding Earth-like planets?

Our guess is that about 10% of the stars in our Galaxy probably have Earth-like planets. But we are only guessing. The only other planet systems we know of at present are unusual, with giant, Jupiter-sized planets orbiting a star every four days or so. How unusual is Earth and our system? It's a burning question.