Direct Imaging of Exoplanets

Discovering and characterising alien worlds

The majority of planets outside our Solar System (exoplanets) have been detected indirectly - through the reflex motion on their parent star (radial velocity) or by the regular dimming as they pass between us and their star (transit).

My research tackles three approaches to directly imaging exoplanets around nearby stars - by finding the best target stars to carry out direct imaging searches, by minimising the glare of the parent star by using coronagraphs, and finally measuring and removing optical errors introduced by the telescope optics by using focal plane wavefront sensing.

Direct Imaging of Exoplanets

Graduate student Alex Bohn is carrying out a survey for directly imaged exoplanets using the VLT instrument SPHERE, called the Young Sun Exoplanet Survey (YSES). We have a sample of 72 young stars from the Sco-Cen association that were identified by Eric Mamajek and Mark Pecaut.

Starting in 2017, we have surveyed all the stars for a first epoch measurement and are now starting the second epoch observations. Amongst the candidate exoplanets Alex discovered a new circumstellar disk, detailed in Bohn et al 2019 A&A.

Making Coronagraphs

The coronagraph was invented by Bernard Lyot in 1930 to enable observations of the Sun's outer atmosphere. The faintly glowing extended gas in the sun's corona was only visible during a Solar eclipse by the Moon, and Lyot wanted to carry out observations without waiting years at a time.

The optical layout of his Lyot coronagraph is shown below:

Optical layout of a Lyot coronagraph. Sketch of a Lyot coronagraph

The telescope focuses light from both the star and the planet at the Focal Plane, where a mask then blocks out light from the star, but the planet (which is in a slightly different place on the sky) is not blocked by the mask. A lens takes the light from the FP and forms a Pupil Plane (PP). Here, the planet's pupil is an even disk of light, but for the star, the presence of the metal disk in the FP causes light to be redistributed into a series of bright rings. A mask with a diameter smaller than the full telescope pupil image diameter then blocks most of this scattered star light - this is called a Lyot stop. A second lens then reimages the pupil to form the final sky image on the detector.

The Lyot coronagraph is simple, but it has several disadvantages: (i) If the star image is not kept centred on the metal disk, starlight "leaks" around the edge of the disk and into the rest of the coronagraph, increasing unwanted light and decreasing the signal to noise at the location of the planet. (ii) The Lyot stop blocks the outer edge of the telescope pupil, and so the angular resolution of the telescope is reduced - the planet image gets larger in diameter, and (iii) the total throughput of the coronagraph decreases due to the Lyot stop's smaller diameter, and there is a trade-off between throughput and contrast.

Many coronagraphs are sensitive to pointing errors, and so require fast closed tip-tilt loops to ensure that the target star stays in the same position in the FP and so that the suppression works. By designing an optic that modifies only the phase in a pupil plane of a telescope, we remove this constraint. This Phase Apodization Coronagraphy (Codona and Angel 2004) is realised in the form of an Apodizing Phase Plate (APP) (Kenworthy et al. 2007) coronagraph:

Optical layout of an APP coronagraph. Sketch of an APP coronagraph

By having no FP mask, there is no spurious diffraction in the PP. Instead there is the APP Coronagraph, which modifies the point spread function (PSF) of all objects in the field of view of the coronagraph.

The APP is completely insensitive to vibrations in the telescope and the AO system tip-tilt correction, it is a very simple coronagraph, consisting of only one opti, and the phase pattern is a smoothly varying function of position, with no discontinuities or vortex singularities. Mathematically ideal coronagraphs with these types of discontinuity are challenging to manufacture and can result in degradation of the coronagraphic null and reduced imaging performance.

The suppression in the dark zone, the Inner Working Angle and the encircled energy transmission are related, and have to be traded off against each other. For most designs we keep 60 to 70% of the encircled energy within the Airy disk of the PSF.

The original APP coronagraph introduced phase variations in the wavefront using differing Optical Path Differences (OPDs) of Zinc Selenide. Since this is an inherently chromatic process, it restricted the bandwidth of the APP.

Panchratam phase can be introduced using the fast axis orientation of a birefringent liquid crystal, which then introduces a phase retardance depending on the orientation of the fast axis. Working with Michael Escuti at NCSU, we can then realise complex phase patterns that are not possible using the diamond turned Zinc Selenide technique.

The liquid crystal layers are still chromatic, but multiple layers of liquid crystal can be added on top of each other and the layers align themselves automatically, building up a more achromatic sandwich. Typically three layers will make a 100% bandwidth coronagraph.

We have tested this new Vector APP coronagraph in the laboratory at Leiden and we have demonstrated its performance over a 50% bandwidth with two layers in Otten et al. (2014), Optics Express.

Vector APP coronagraph viewed between two crossed polarisers under white light. Credit: Gilles Otten. Photograph of a vector APP coronagraph between two crossed polarisers

By using a combination of quarter wave retarder and Wollaston prism, two simultaneous PSFs with complementary dark holes are produced, along with the broadband performance, making this a highly robust coronagraph for exoplanet detection and characterization.

The addition of a phase ramp in the APP phase pattern causes the two APP PSFs to split, with an additional third non-coronagraphic PSF appearing between the two. This PSF is called the leakage term, and is due to any deviations in performance from the liquid crystal layers acting as a half-wave plate. This optic therefore does not require any quarter wave plates or Wollaston prism, and is therefore very simple to install in any exoplanet camera.

We now build grating vector APPs (gvAPPs) for many large telescopes around the world, designed by Emiel Por, David Doelman, and Christoph Keller. We are carrying out observational programs with these high contrast coronagraphs.

Phase Sorting Interferometry

The average complex halo derived from 67 2.5 second Clio image cubes and 43,550 WFS measurements - from Codona and Kenworthy (2013). The image shows the amplitude of the focal plane PSF as height, and the colour is the phase. The alternating rings of red and blue colour show that we are correctly sensing the complex PSF, where you would expect alternating rings of phase 0 (blue) and 180 (red) degrees respectively.Complex point spread function reconstructed using the PSI technique

Several methods have been developed to estimate the point spread function (PSF) using the science camera images themselves, notably Angular Differential Imaging (ADI), Spectral Differential Imaging (SDI), Locally Optimized Combination of Images (LOCI), Principal Component Analysis (PCA), and combinations of all the above.

Johanan Codona (University of Arizona) and I developed a method for estimating the Science camera PSF directly using the Wavefront Sensor Camera telemetry on ground-based adaptive optics systems. It is called Phase Sorting Interferometry (PSI) and is described in Codona and Kenworthy (2013) ApJ, 767, 100. In this paper we demonstrate on-sky measurement of the PSF at the science camera, and also derive several measures of the atmospheric turbulence.

We are implementing PSI for the ERIS camera and it is considered as the baseline focal plane wavefront sensing method for the ELT instrument METIS.

Last updated around mid-June 2019.