I set PEST up to discover planets. That was the Big Hairy Goal. And because these planets would be mine – all mine – I would be able to call them PEST-1b etc. That would have been seriously cool. It hasn’t worked out that way. The reasons why are instructive.
The probability that a given star has a transiting planet is pretty small
The original search strategy was to look for transit signals around nearby stars. I compiled a list of target stars of brightness between mag 7 to 12 – bright enough to allow good precision, but not to saturate the CCD. There were 371 stars on this list.
PEST spent the first few months of its existence looking at about 5 different star fields each night, cycling through each in turn. Each field contained at least one of my 371 target stars. This was a lot of moving around (re-pointing) for the telescope. I soon got worried about mechanical wear – it’s an off-the-shelf beast and not the world’s most robust telescope.
A transit needs to be recorded at least 3 times in order to confirm that the signal repeats at a particular period. To have a reasonable chance, I’d have to observe the same fields for months, until Earth in its glide about the sun turns its night face towards other star-fields. Then repeat for the next couple of years. Only about 10 to 20 stars in each field were bright enough to search for transits. So within a reasonable survey duration I would be able to cover only several hundred stars. The KELT planet survey observes several hundred thousand stars in search of transits. Even if all stars have a planet, many things have to be just right to detect transits. The planet’s orbit has to be oriented pretty close to edge-on, and the transit has to be deep enough, and the period has short enough to see multiple transits, and – well you get the idea. With my sole venture, transit detection was unlikely.
Transits are not only caused by planets
But say I did see what looks like a planetary transit, i.e. a not-too-deep, flat-bottomed repeating dip in the light. Would I be able to announce this as a planet? No, because it might not really be a planet. The star could be really large, a giant, and the transiting object a small star rather than a planet, blocking just a small part of the light and mimicking a planetary signal. Or the transit could actually be quite deep, caused by twin stars dancing around and periodically eclipsing each other (eclipsing binary), but if there there happens to be a star in the background along the same line of sight, the total blended light might exhibit a shallow planet-like transit.
To eliminate these possibilities I’d need to take and interpret the spectra of the star – spreading out its light into constituent colours – giving good indications of temperature, luminosity, mass and radius. The spectra may also tell us if the star is a binary. It takes a large telescope to do these sorts of observations. For the KELT-10b discovery, observations with the WiFeS spectrograph on the 2.3m ANU telescope at Siding Spring, Australia gave us key stellar parameters, confirming the star is not a giant. Also, it told us that the transits are not due to a star, because the mutual gravitational pull between members of a binary star induces to-and-fro wobbles (radial velocity, RV) that show up in the spectra. A stellar companion would have induced large RV variations of over 1 kms-1, which we did not see. Incidentally, as of 2015, KELT had vetted 105 candidates with WiFeS, of which 67 (62.6%) were eliminated as eclipsing binaries. As nebulae and galaxies were to the comet-hunter Charles Messier in 1771, so eclipsing binaries are to the modern exoplanet hunter – vermin of the skies.
One way to check if there is another star in the background diluting the transit, is to image the star and its immediate surrounds at exquisite resolution, using a technique called Adaptive Optics (AO). For KELT-10b, AO imaging was done with one of the telescopes at the VLT (Very Large Telescope) facility in Chile. These four silver sky sentinels sit almost 3km above sea level and have mirrors 8.2m in diameter. Each collects about 750 times more light than PEST.
Not all small objects are planets
Say I’ve worked out the type, mass, and therefore radius of the potential planetary host star. Knowing the transit depth then tells me the radius of the transiting object. If this radius is consistent with a planet – say it’s about the radius of Jupiter – does that mean it’s definitely a planet? No, because brown dwarfs, strange creatures that were born like stars, but not massive enough to have ignited, may have similar radii. We need to know the mass of the object.
The radial velocity (wobble) induced by a planet on a star is small. KELT-10b, a planet of about 0.7 times the mass of Jupiter, causes its host star to move towards and away from us at speeds of up to only 77 ms-1. Very high precision is needed to measure this in spectra. The CORALIE spectrograph on the Swiss 1.2m telescope at La Silla, Chile, used to confirm KELT-10b, is able to measure RV down to 3 ms-1 – about jogging pace.
Today, PEST observes transits. It’s only a part of the planet discovery process, albeit an important one. But discovery is only possible with contributions from a host of specialists – in stars, spectroscopy and spectra interpretation, planet formation and evolution, AO, precision RV, and the people who run the giant telescopes on mountaintops.