6 Prospects and conclusion

As discussed in Boccaletti et al. (1998b), the sensitivity of dark-speckle imaging is currently limited by the presence of a few fixed speckles in the cleaned image which results from small permanent residual phase errors on the wavefront, escaping correction by the adaptive optics. A local bias in the error signal or actuator response, causing the average phase value to be off-set from zero, generates such fixed speckles which survive the local intensity modulation in the multiple exposures. Subtraction of the cleaned image from a similar image obtained on a reference star improves the darkening, but removing most of the fixed speckles in the first place is highly desirable. In coronagraphic images these defects become dominant.

A possible way of reducing the effect of these fixed aberrations consists in boiling the static speckle pattern. This can be achieved in principle, if offset voltages on the adaptive mirror are randomly modulated around their average value with appropriate amplitude and higher frequencies, the stroke of the actuators being within the range of the AO servo noise. Although such a process smoothes the residual speckles and then increases the average halo intensity on the long exposure, it does not really affect the dark-speckle analysis which remains "not sensitive'' to fluctuating speckles.

The most critical device for the starlight cancellation is the coronagraph itself. New concepts in coronagraphy have been recently suggested: the achromatic interfero-coronagraph (AIC) (Gay & Rabia 1996), and the phase-mask coronagraph (Rodier & Roddier 1997). Both enhance faint companion imaging close to the Airy peak. The capability of such devices combined with a dark-speckle analysis remains to be thoroughly investigated.

As far as the detector is concerned the main present limitation is the saturation of the centroiding processor which implies low photon fluxes (<300 ph/frame). Higher photocathode efficiency and larger CCD chips, together with more powerful processors should improve the situation. Also important for high-contrast imaging is the abscence of internal scattered light within image intensifier tubes and CCDs. In addition, the feasibility of dark-speckle imaging with a near IR detector featuring a 38 e^-/pixel read-out noise is being studied.

Finally, these laboratory tests are considered a preliminary attempt in the context of NGST studies to design a concept of visible/nearIR and multi-purpose coronagraphic capability (Gezari et al. 1997; Moutou et al. 1998; Rabbia et al. 1998; Lefevre et al. 1998). Exoplanets detection is presently a key goal of the design reference mission. Requirements and specifications for such an instrument are currently under evaluation (Labeyrie et al. 1999). In this context, intensive laboratory simulations will be required in the next few years and coronagraphic prototype should enable to assess the best solutions for future missions.

Acknowledgements

We are grateful to the ONERA team for their efficient support during this 15 days run. We especially thank B. Fleury (ONERA) for the control of the AO loop, A. Labeyrie (Observatoire de Haute-Provence) for helpful discussions and R. Burnage (Observatoire de Haute-Provence) for careful reading. We also congratulate V. Thevenet (Paris Observatory) for the precise realisation of the Lyot stops.