Planet finder interferometers "DARWIN" and "OASES" have been described by Leger et al (Icarus in press) and by Angel and Woolf (Astrophysical Journal submitted). Although these devices have some differences, they also have features in common. There are four (or five) mirrors of 1m diameter. The devices are about 50m across, and have an angular resolution of about 0.03 arc seconds. They can observe and take spectra of Earth-like planets out to a distance of about 10pc, in the 7-17 m region and they are just sensitive enough to detect water, carbon dioxide and ozone in an Earth-like planet. The specific needs in making these minimal finding and spectroscopic observations are:

1. The starlight is interferometrically nulled so as to be no more than the signal from a single planet.

2. The telescopes and all components in the optical path are radiatively cooled to a temperature where their thermal emission is negligible.

3. The devices operate at about 5AU from the Sun, where the thermal emission of our own system's zodiacal dust is negligible.

4. The detected noise is the photon shot noise of the zodiacal dust of the external system.

5. Signals are very feeble and so detectors of exceptionally low noise are needed.

6. The signals from all planets are multiplexed by the interferometer, so that all planets of a system are observed simultaneously. Reconstruction produces a 3-dimensional result with two-dimensional maps of the system (which shows the planets revolving around their star), and the third dimension shows intensity against wavelength for each source, i.e. it is the spectrum, with a spectral resolution built into the device.

7. The spectral observations take a long time to observe (weeks to months per system) to achieve adequate signal/noise.

8. The technological requirements to make these observations are harder to meet than they have been for any spacecraft to date. There are very tight tolerances for pointing, phasing and amplitude control that require new techniques to be developed. And all systems must operate where no human help can be made available.
   

Seeing further, and the exo-zodiacal dust limit
The limitations of the system are first, that an "Earth" has a standard luminosity, and so can only be seen out to a standard distance. Secondly, the device has a limited angular resolution. If the planet is too close to the star, it cannot be resolved. A system with 1m mirrors is just sensitive enough to see "Earths" out to 10pc. If the zodiacal dust of other systems has more emission than our solar system, then the sensitivity drops. If there is four times as much dust emission, then we can only see out half as far, to 5pc. To bring back sensitivity, we can increase the size of the telescopes, two-meter mirrors can see twice as far as one-meter mirrors.

For the second limitation, an interferometer with doubled linear dimensions has double the angular resolution. A ~50m device can resolve an "Earth" around a "Sun" at 10pc. Intrinsically fainter stars will have their "Earth" closer in, and so require higher angular resolution. A 100 meter interferometer can resolve an "Earth" around a "Sun" out to 20pc.

Seeing CH₄ and N₂O
Methane would be meaningful to observe because it is indicative of the anaerobic decay of organic matter. Nitrous oxide is produced on Earth by the Nitrogen cycle, and as such indicates an atmosphere of oxygen and nitrogen, the fixing of nitrogen, presumably by life processes, and oxidation of ammonia to N₂O. Bands of methane and nitrous oxide, most visible near 7.8m and 8.6m respectively are harder to see than ozone for two reasons. First they are narrow, and secondly methane occurs at the short wavelength end of the spectral band where an Earth-temperature planet has a reduced flux (also water absorbs there). The attached spectrum of part of Earth, from the Nimbus satellite, show bands about 10 times narrower than Ozone. (The individual lines have a width of about 0.002m). The methane has a comparable central intensity, but the nitrous oxide is much weaker. The spectral resolution to see them has to be about 0.05m Also, the flux at the Methane band is reduced by about 2.5 compared with Ozone. For Methane we are trying to detect a roughly 25 times smaller amount of light removed from a planet spectrum, and this requires about 25 times larger collecting areas to achieve the same detectability as Ozone. This would require the 1-meter telescopes to be replaced by a set of four 5-meter class telescopes. If 1.5-meter telescopes were considered for a "Mark I" planet finder, then 7.5-meter class telescopes are needed for "Mark II". For N₂O the band is about 5 times more shallow, and would need about 250 times the collecting area, or about 15m apertures.

Simple substitution of larger round mirrors in a planet finder would not be enough, because the angular resolution of the mirrors would cut out signals from outer planets. A preferred scheme would have rectangular mirrors, preferably of about 2x10 m. This would be fine for methane. Nitrous Oxide needs a different solution. To observe Methane, it would be very helpful for design studies of a Next Generation Space Telescope to be coordinated with planet finder studies. A form of NGST in which the aperture is elongated would be helpful for fitting into a planet finder, and this option is being considered by the NGST designers.

Detection of molecular Oxygen
Molecular oxygen does not have strong bands in the 7-17 m spectral region. It seems likely that the optimum wavelength for detection is the A band at the long wavelength (red) end of the visible spectrum. At those wavelengths, photon fluxes are about 20 times less than in the 7-17 m region. Also, the A band is relatively narrow (Dl/l) compared with the breadth of the IR ozone band. Thus about 200 times as much collecting area will be needed to see it as to see Ozone - all else being equal. A roughly 15-20m class space telescope is needed.

Observing a planet with a large telescope in this region has been discussed by Sandler and Stahl. In their simulations of a 6 meter aperture, the planet image sits on a pedestal of light from the central star, which is 1000 times brighter than the planet. A 15-20m telescope would have a sharper diffraction pattern, and the planet/"background" signal would be rather similar to that for a planet finder mission. Detailed simulation it is needed to determine the minimal telescope size, but reduction much below 15m seems unlikely.

None of the spectroscopic observations discussed above is remotely as difficult as the simplest imaging of an external planet's surface. The optimum study to follow the first spectroscopic observations of external planets is - more spectroscopy. An optimum sequence would seem to be, first a planet finder with small mirrors, then a planet finder with large mirrors, then a third generation Space Telescope. The benefit is that it should be possible to start exploring those cyclic processes which are chemical indicators of life on Earth.