From HARPS to CODEX: exploring the limits of Doppler measurements

Only 3–6 years ago, the Doppler technique was believed to have reached its final limitations in measuring stellar velocities and finding extra-solar planets. The 3–4 m s −1 precision level achieved, at that time, by various teams, was certainly limited by instrumental performances, but also constrained, as believed by a part of the community, by intrinsic stellar limitations. The advent of HARPS drastically changed this view. The instrument demonstrated, through its recent discoveries, that stars more ‘stable’ than 1 m s −1 actually exist, and that their radial velocity (RV) can be measured at that level of precision. Short-term precision of 20 cm s −1 rms and long-term precision of the order of 30–60 cm s −1 rms have been actually achieved on real stars, showing that RVs still harbor a great potential, and not only in the domain of extra-solar planets. Indeed, HARPS inspired the CODEX@ELT experiment for the direct determination of the expansion of the Universe, measuring the Doppler shift of Ly-α forest lines as a function of time. This experiment calls for a Doppler precision as low as 1 cm s −1 , which in turn inspires new possibilities in the domain of extra-solar planets. We will investigate the obstacles on the way to cm s −1 precision. The discussion presented here will be based on our experience with HARPS and what we consider to be the current limitations set by the instrument, telescope, atmosphere and star. Finally, we will also provide an outlook on possible improvements and expected performances, which will finally define new scientific opportunities.