Adaptive Optics

Figure 1: Top : Simplified illustration of AO systems architecture. Bottom : Images (log scale) of HD 119181 taken on 2013 at 1.65 µm with the CANARY instrument installed at the William Hershel Telescope. The use of adaptive-optics allows to unveil the presence of a tight binary (400 mas separation) and improves the angular resolution and contrast by a factor 10.

Adaptive Optics (AO) systems enable high angular resolution ground-based observations.

Why do we need such systems for ground-based astronomy ?

Ground-based and observations suffer from disturbances introduced by the turbulences that form in the Earth atmosphere and blur stronomical images. We talk about seeing-limited images, where seeing refers to the full width at half maximum of the instrument optical response (the PSF) and depends on the atmospheric conditions. AO systems are designed to compensate for the effect of those turbulence in real-time and restore the angular resolution close to the diffraction-limit, as ilustrated in figure 1 on a binary.

How do they work ?

The atmospheric turbulence disturbs the propagation of the electric field, and particularly its phase by introducing an Optical Path Difference (OPD) over the telescope pupil by 1 up to 10 µm rms. AO rely on a wave-front sensor (WFS) to convert the phase variations over the telescope pupil into a digital signal thanks to an optical element (micro-lens array, pyramid, axicon,...) and a detector. Such WFSs are usually sensitive to the first or second order derivative of the OPD, which allows to reconstruct the wave-front disturbances from the detector image.

Once estimated, the impact of the atmospheric turbulence is corrected for by using a deformable mirror (DM), whose shape can be controlled using piezo-stack actuators for instance, that introduces the opposite reconstructed OPD in the optical path towards the science instrument.

The disturbances brought by the atmospheric turbulence vary rapidly, every milli-seconds basically. The process of measurements then correction must be repeated at an high rate and this is the role of the Real Time Computer (RTC) to update the DM commands from the WFS measurements through a control law (PI, LQG, ...) that optimizes the trade-off rapidity/stability of the loop.

What level of correction can they reach ?

Several physical effects limit the AO correction, such as the DM fitting error due to the limited number spatial frequencies the DM can compensate for. Moreover, the correction is applied with a delay from the acquisition query to the DM commands update, which introduces a servo-lag error as well. Finally, the WFS measurements are affected by noise and aliasing effects that contaminate the final AO-delivered image.

In order to assess AO efficiency to restore the diffraction-limited angular resolution, we usually use the Strehl-ratio (SR), which is a scalar number defined between 0 (no angular resolution) and 1 (diffraction limit). The SPHERE system that equips the Very Large Telescope at Chili can reach a SR up to 0.9 in K-band (imaging wavelength: 2.2 µm). However, this high performance is localized in the specific sky location of the guide star that serves the WFS to analyze light perturbations. The correction quality degrades rapidly within a few arcseconds due to the anisoplanatism effect.

I'm involved in the developement of future AO system such as HARMONI and MOSAIC on the 39m ELT and MAVIS on the VLT.