Extreme adaptive optics for the detection of extrasolar planets

The aim of the Spectro-Polarimetric High-contrast Exoplanet Research (SPHERE) instrument is to detect extremely faint astronomical sources (i.e., giant extrasolar planets) in the vicinity of bright stars.¹ The detection capabilities of an exoplanet hunter are largely controlled by its adaptive optics (AO) system.^{2, 3} Better AO correction provides improved coronagraph extinction and fewer residual defects. The challenging SPHERE science goals require a very high-order performance AO system to feed a quasi-perfect flat wavefront, corrected for atmospheric turbulence and internal defects, to the scientific instruments.

In May 2014 SPHERE (see Figure 1) was installed on the third unit telescope (Melipal) of the Very Large Telescope (VLT) in Chile.^{3, 4} After four months of extensive and comprehensive tests (for robustness, performance, and ease of use) the instrument is now available to the astronomical community for observations (until April 2015). The AO system on the instrument is known as the Sphere AO for eXoplanet Observation (SAXO)^{2, 3} and is the ‘heart’ of the instrument, which ‘beats’ 1200 times per second to provide unprecedented image quality from a large ground-based telescope operating at optical/near-IR (NIR) wavelengths. As such, SPHERE presents tremendous potential for exoplanet discoveries.⁴

Figure 1. Concept of the Spectro-Polarimetric High-contrast Exoplanet Research (SPHERE) instrument. Top left: Schematic diagram of the instrument's four subsystems, including the main functions of the common path subsystem. Colored arrows indicate optical beams in the near-IR (red), visible (blue), and common-path (orange). Top right: Photograph showing SPHERE on the Very Large Telescope. The instrument has dimensions of 6×4×2.5m and weighs 8 tons. It rests on three anti-vibration pillars. Bottom left: Example J-band saturated, coronagraphic processed images showing a companion (iota Sgr b) to a very low-mass star. The companion is 4000 times fainter than its parent star at 0.24". Bottom right: Image of a dust ring right around the star HR 4796A. SAXO: Sphere Adaptive Optics (AO) for eXoplanet Observation. Vis: Visible. NIR: Near-IR. IFS: Integral field spectrograph. IRDIS: IR dual-imaging spectrograph. ZIMPOL: Zurich Imaging Polarimeter. mas: Milliarcsecond.

To meet the detection requirements for SPHERE, our instrument design is divided into four subsystems, namely, the common path optics and three science channels. The common path includes pupil-stabilizing fore-optics (tip-tilt and derotator) and insertable polarimetric half-wave plates. The common path also contains the SAXO system with a visible wave-front sensor and NIR coronagraphic devices. These feed the IR dual-imaging spectrograph (IRDIS)⁵ and the integral field spectrograph (IFS)⁶ with a highly stable coronagraphic image at NIR wavelengths. The three science channels are used to gather complementary information that maximizes the probability of exoplanet detection. This information is obtained at a large range of wavelengths and includes imaging, spectral, and polarization data.⁷

The SAXO system measures and corrects any wave-front perturbations (rapidly varying turbulence or quasi-static instrumental speckles)⁸ to ensure the unprecedented image quality for the instrument. Our SAXO design therefore incorporates some of the world's most advanced AO components and concepts. We use a fast (800Hz bandwidth) tip-tilt mirror, an active toric mirror (TTM),⁹ and a deformable mirror (DM) with 41×41 actuators (1377 of which are active). We based the wave-front sensing on a filtered Shack-Hartmann (SH) approach, where we use a state-of-the-art electron multiplying CCD (EMCCD)¹⁰ that can operate at up to 1200Hz with less than 0.1e⁻ of equivalent readout noise. We designed the filtering pinhole to remove the aliasing effect^{11, 12} and to be adjusted as a function of atmospheric conditions. The high-performance characteristics of the EMCCD, combined with advanced centroiding techniques (e.g., weighted center of gravity)¹³ allow us to achieve impressive limit magnitudes with SAXO. The initial requirements for SPHERE are met, in terms of wave-front correction, at an ultimate performance limit magnitude (magR) of 9–10. In addition, at a magR of 15–16 our AO system still provides a significant gain (about a factor of 5–10 with respect to a purely turbulent case), as illustrated in Figure 2.

Figure 2. Top: Performance of the SAXO system, as measured in the laboratory (curves) and in on-sky observations (dots). Bottom: Near-IR and visible (I and R band) images of the same object. All the images have the same field of view (1.8 arcseconds). The diffraction limit is reached in the H and I images at 20 milliarcseconds (instead of 18), which was reached in the R band. This represents the best performance for any ground- or space-based monolithic telescope. SR: Strehl ratio.

We use a real-time computer (SPARTA) to control the TTM and DM, with a final latency of 80μs. We incorporate modal gain optimization for DM modes, together with a linear-quadratic-Gaussian law,¹⁴ which was specifically designed for the TTM, to automatically identify and filter out up to 12 vibration peaks that are randomly spread between 10 and 300Hz. With this latter feature, we can reach a final residual jitter of less than 2 milliarcseconds root mean square (i.e., one-twentieth of the H-band diffraction). This is fundamental for ensuring the optimal operation of the various SPHERE coronagraphs. We also include two auxiliary loops in SAXO, for two reasons. First, we ensure fine centering (at 1Hz) on the coronagraph by compensating for chromatic effects. We use a specific IR tip-tilt sensor¹⁵ and a differential tip-tilt mirror for this purpose. In addition, the auxiliary loops ensure pupil centering (at 0.1Hz) to compensate for telescope run-out. We achieve this by analyzing SH edge sub-aperture fluxes, and perform the correction using a small tip-tilt mirror that is located very close to the focal plane.¹⁶

Together, all of our design features produce a very flat wave-front for SPHERE, which corresponds to a Strehl ratio of more than 90% in H- and diffraction-limited images at visible wavelengths. We have also developed SAXO to be as automatic as possible. The system auto-optimizes itself using its own internal data, i.e., without any user intervention. In addition, the calibration and verification (quality check) processes are completely automatic. SAXO—and SPHERE—can therefore be operated by a non-specialist without compromising performance levels.

We have designed an advanced AO system for the VLT's SPHERE instrument. The strong performance capabilities of the instrument permit detection of extremely faint exoplanets close to bright stars. Although our AO and instrument performance fully meet the specifications, we will continue to improve the system and keep it at the cutting edge of ground-based telescope instrumentation and exoplanet hunting. In the coming months and years we plan to test and install new high-order wave-front sensors^{17, 18} and dark hole techniques.