Exoplanets at a discount

Fig. 1: An image of Beta Pic's companion taken with the apodising phase plate on VLT/NaCo, after processing. The light from the central star was blocked out (in processing). Credit: ESO

Astronomers have many ways of spotting exoplanets round far away stars – but getting a direct look at them, especially with ground-based telescopes, remains a difficult job. With a planet emitting very little light of its own, and appearing to us essentially on top of the host star, its radiation is completely drowned in the image of the star. Catching those few photons and separating them from the flood of light from the star requires some clever observational tricks. To do this with ground-based telescopes, we at the very least need adaptive optics, to prevent the atmosphere from creating a blurry mess and  keep the image nice and sharp, and often some sort of mask that will block out as much as possible of the stellar light. But an upgrade to one of ESO’s near-infrared workhorse imager NaCo on VLT’s 4th Unit Telescope has just made it a whole lot easier.

In June, a team led by Anne-Marie Lagrange confirmed the discovery of an exoplanet in the dusty disk surrounding young star  β Pic using NaCo. The story of this discovery is interesting in itself: in 2008 this same team reprocessed and analysed NaCo data from 2003, stacking together the best images from a large set, and spotted what looked like a companion to β Pic. As they couldn’t tell for sure that this was an actual planetary companion rather than just a spurious background source, they pointed VLT and NaCo back at the star. But the new images showed nothing, and the planet’s existence could not be confirmed. In late 2009, at last, Lagrange and her collaborators tried again, and this time the spot had reappeared on the other side. The planet, β Pic b, is definitely there, and we now have images of it moving round its host star. This was a very neat result that was covered quite extensively on blogs (here, here) and in the media.

Lagrange’s detection method was essentially brute force: stare at the star for a long time, many times, and use only the best data. The light from the star itself was removed from the images by imaging a second star for the same lengthy amounts of time to get a reference of the instrument’s characteristic point spread function – or, what an image of a perfect point source looks like with NaCo in these particular observing conditions. With this light pattern removed from the β Pic images, the companion can come out of hiding. Telescope time is money, so this work is expensive stuff.

At the same time as Lagrange’s team were analysing their latest data, instrumentalists from  ETH Zurich, Leiden, Arizona, ESO and Heidelberg were testing a new optic they’d designed specifically for imaging exoplanets with NaCo. While most coronagraphs use a mask to place a solid physical barrier between the starlight and the detector, this coronagraph, called an apodising phase plate (APP), doesn’t actually block any of the light coming through the system. Instead it modifies the phase of the light waves to create a large D-shaped dark region around the star, where a companion is easier to see.

Phase masks are quite commonly used in applications where we’re interested in studying the faint surroundings of a bright object, like disks. Clever design of the optic can allow us to manipulate the image such that more extended patches around the star are also darkened, making any close companion objects in that region more visible. This doesn’t produce particularly “nice” images – in fact, you’d barely recognise that you’re looking at an actual star (see e.g. Fig. 2) – and the technique requires quite a bit of post-processing.

This week, Sascha Quanz of ETH Zurich and collaborators report on the results of this initial observing run with the APP in ApJ Letters. During the commissioning phase they pointed the telescope at β Pic – and completely independently found the planetary companion with the new instrument configuration. What’s more, in terms of actual on-source observing time, the new detection of β Pic b was around 4-6 times faster than Lagrange’s method. NaCo in its APP configuration needed just 20 minutes of on-source observation time to bring out the signal of the planet over the noise.

Getting new instruments is always fun of course, but this result shows nicely how just a small well-designed optic can make our existing telescopes and instruments a whole lot more efficient. Telescope time is expensive, so efficiency in time equals efficiency in cost. We don’t always need big new instruments to make new discoveries. The continued scientific productivity of our smaller 2-4-m class telescopes, such as the William Herschel Telescope in La Palma, is a solid proof of that, and our current 8-m-class facilities will continue to produce great science in the ELT era. So it’s always a little sad to see good telescopes and instruments shut down, despite their potential, to make room for a new generation – especially when the decision is purely driven by financial needs. Inevitable, but sad nonetheless.

NaCo itself will be in operation for a further 2 years at VLT, before making way for mega-spectrograph MUSE. The APP is available to the community for general use.

Fig. 2: Image of a single star with the APP (without exoplanet). Diffraction rings on one side, a dark hole on the other. Credit: ESO


Sascha P. Quanz, Michael R. Meyer, Matthew Kenworthy, Julien H. V. Girard, Markus Kasper, Anne-Marie Lagrange, Daniel Apai, Anthony Boccaletti, Mickael Bonnefoy, Gael Chauvin, Philip M. Hinz, & Rainer Lenzen (2010). First Results From VLT NACO Apodizing Phase Plate: 4-micron Images of the Exoplanet beta Pictoris b accepted in ApJ Letters arXiv: 1009.0538v1

A.-M. Lagrange, et al (2010). A giant planet imaged in the disk of the young star Beta Pictoris Science, 329 (5987), 57-59 arXiv: 1006.3314v1

Matthew A. Kenworthy et al. (2010). An apodizing phase plate coronagraph for VLT/NACO Proc. SPIE, 7735 arXiv: 1007.3448v1

(Thanks to fellow Leidener Matt Kenworthy for explaining the methods to me)


  1. Thanks a lot for this explanation. Explanations in media are often too short and too thin.