The search for Earth-like exoplanets

Given the age and size of our universe ­– astronomers are wondering, just where are all the extra-terrestrials? Well if life on another world is to be found, first we need to speed up the rate at which we can detect possible life-sustaining exoplanets. Here we learn how a group of Danish astronomers intend to do just that…

 Perhaps the greatest paradox thrown up by recent advances in astronomy is why visitors from alien civilisations are not sampling the delights of ours.

In 1950, the Italian physicist, Enrico Fermi, suggested that, even with the slow space travel that we’ve managed so far, it should take no more than 10 million years to colonise our Milky Way galaxy. Given that it is around 13.7 billion years old and contains hundreds of billions of stars, the chance of our race being the galaxy’s first astronauts seems very small and ‘someone’ should have been knocking on our door by now (Ian Crawford, Scientific American, 2000).

However, we haven’t heard any knocking and there are no confirmed traces of extra-terrestrial life or artefacts on Earth. So, given that there’s been the time and the opportunity, is it that very few of these billions of stars have planets? Or, that even fewer have planets with clement temperatures capable of sustaining life? Or, that the warm planets are bone dry? Could it be that there are no other Earth-like planets in our galaxy?

It is only in the last 20 years that scientists have been able to make the transition from supposing the existence of exoplanets - planets around other stars - to wondering how common and how similar to the planets of our own solar system they may be. In 1992, radio astronomers Wolszczan and Frail announced the discovery of two planets orbiting the pulsar PSR B1257+12 but it was not until 1995 that Mayor and Queloz from the University of Geneva discovered a planet orbiting 51 Pegasi, a main-sequence star. 51 Pegasi b, as it was called, provoked initial scepticism because of its massive size and very short orbital period of 4.2 days, which implied that it was too close to its parent star for conventional theories. However, other discoveries of similar planets were quickly made and a new class of astronomical objects was named – Hot Jupiters, colloquially known as Roaster Planets.

As of January 15^th this year, 859 exoplanets have been identified, including 128 multiple planetary systems. The majority of them are Hot Jupiters with poor prospects of life but this should not discourage us from supposing that Earth-like exoplanets may be just as common. For almost all of the planet hunting period, telescopic detection of small Earth-like objects capable of supporting life was virtually impossible and we have discovered these huge gas planets simply because they are giants and the easiest to detect. However, this may be about to change through the work of the Stellar Observation Network Group (SONG) and their adoption of ultra-sensitive EMCCD technology.

The two most common exoplanet detection methods to date have been the transit and radial velocity methods. The radial velocity method relies on detecting the small orbit of the star itself around the solar system’s centre of mass caused by the gravitational pull of the exoplanet. The star will move toward or away from us and the radial velocity may be deduced from displacement in the star’s spectral characteristics due to the Doppler Effect. It detects the mass of close-orbiting gas giants easily but larger orbits and smaller planets are more problematic. However, it has been the most productive of the planet-hunting techniques.

Unlike the radial velocity method that provides information on a planet’s mass, the photometric transit method provides information about a planet’s radius. If a planet crosses, or transits, in front of its parent star as observed from Earth, then the brightness of the star will be reduced by a small amount. The amount the star is perceived to dim is directly related to the relative sizes of the star and the planet. However, the method has two major drawbacks. Firstly, planetary transits are only observable where the planets orbit is perfectly aligned to our observation point here on Earth or, in the case of the Keppler space observatory, a telescope in near-Earth orbit. Only about 10% of small orbits are so aligned and this percentage decreases for larger orbits. Secondly, it produces many false positives – perhaps as high as 35% - and confirmation is necessary, typically from a radial velocity measurement.

Even with a combination of these methods, it is much easier to detect heavy planets in small orbits and that explains the bias towards the detection of Hot Jupiters. It would be very hard to detect the Earth from a remote observation point and even more so to discern the structure of our solar system. And, given that the transit and radial velocity methods both detect effects of the orbital motion of the exoplanet, observation of two or three orbits is necessary to increase the confidence level. Given that the orbital time of our outer planets is of the order of a few decades, detecting three orbits would take the better part of a century or more.

The December 1936 issue of the periodical Science contained a brief note written by Albert Einstein entitled “Lens-Like Action of a Star by the Deviation of Light in the Gravitational Field”. In this note Einstein describes how the gravitational field from a star can deviate the light from a background source star and, thereby, increase its apparent magnitude if the two stars are near-perfectly aligned on one line of sight. The required degree of alignment is actually so great that Einstein’s article concludes that, “Therefore, there is no great chance of observing this phenomenon.”

Fortunately Einstein was, for once, wrong. Unable to predict the enormous developments in computers, detectors and optics of the late 20^th Century, he could not envisage today’s automated telescopes surveying several hundred million stars in the Milky Way daily. The sheer scale of the exercise means that we can, and do, observe these very rare gravitational microlensing events. And, since microlensing observations do not rely on the radiation received from the lens object, astronomers can study objects no matter how faint. Therefore, it is an ideal technique to study the galactic population of faint or dark objects, such as brown dwarfs, red dwarfs, white dwarfs, neutron stars, and black holes.

The interest for planet hunters is that gravitational microlensing events are also capable of revealing exoplanets in orbit around the lensing star and, since it provides a snapshot of the microlensed system, there is no need to wait up to a 100 years for a few orbits to go by. Neither is it as biased towards massive planets in close orbits. We can also discern the structure of an extra solar system and, therefore, investigate the frequency of Earth-like planets in Earth-like orbits and giant gas planets in outer orbits, just like Jupiter and Saturn in our own solar system. This may be especially important in the search for other life in the universe as many scientists believe that heavy planets in the outer reaches of a solar system are important for bringing water to the inner, hotter parts of a solar system. This is caused by their gravitational fields capturing comets and other ice-containing space bodies and causing heavy bombardment events on the inner planets.

This theory is supported by Herschel Space Observatory measurements, which show that the Hartley 2 comet contains water with the same chemical signature as Earth's oceans. It is located in the remote Kuiper Belt along with icy, rocky bodies, including Pluto, other dwarf planets, and innumerable comets. If we can find other solar systems with a similar make up of large gas giants in outer orbits with smaller, rocky planets in closer orbit, the same mechanism may have brought water to those inner planets as well.

Einstein may have been wrong in not being able to predict the rate of our technological advances but he was right in that the likelihood that two random stars will become sufficiently aligned due to their proper motion is vanishingly small. Therefore, nearly all microlensing events are observed towards the centre of the Galaxy in the densest fields in the night sky, some 20,000 light years distant. And, this is the downside of gravitational microlensing - these fields are very crowded and the images of the stars in our detectors appear as a continuum where only the brightest giant stars can be distinguished. To discern main sequence stars, whose angular diameter on the sky is much smaller we need higher resolution. This is important as there is a relation between the low mass detection limit and the angular diameter of the source stars. So, to push the limit down towards Earth-mass we need to observe events in the fainter main sequence stars.

[caption id="attachment_32336" align="alignleft" width="200" caption="Figure 1 – Proposed placement of the 8 SONG telescopes. The green lines signify the time span that a star on the celestial equator can be observed with the individual telescopes and the significant overlap gives round the clock coverage."][/caption]

SONG, the Stellar Observations Network Group, is an initiative of Danish astronomers from Aarhus University and the Niels Bohr Institute in Copenhagen. Our aim is to build a global network of at least eight robotic one-metre plus telescopes. These are being placed around the globe in two rings: one in the northern and one in the southern hemisphere. The idea is to be able to observe microlensing events from any star around the clock, since the telescopes succeed each other as the star passes overhead (Figure 1). In a microlensing event, the signal from an Earth-mass planet has a duration of about half an hour. It is vital, therefore, to make very dense observations of an event all the time if one wants to detect such planets.

The CCD (charge-coupled device) is ubiquitous in astronomy for the detection of visible and UV light where their high quantum efficiency and low dark current is ideal. However, at higher readout speeds in excess of one million pixels per second, readout noise is a serious hindrance for faint targets. In our work to image fainter main sequence stars, we use the ‘Lucky Imaging’ method to increase the resolution. This depends critically on a high frame rate and high pixel speeds are vitally important. Moreover, when taking images at high speed of faint targets, the detector is essentially photon counting, as most of the frames will contain zero photons.

In our quest for speed and resolution we decided to try EMCCD (electron multiplying CCD) technology, which promises fast readout times and negligible readout noise.  This improvement in resolution would potentially improve significantly the photometry of faint stars in extremely crowded fields. However, photometric stability is vital and the EMCCD has sources of noise not found in conventional CCDs. Therefore, we initially tested a number of EMCCD cameras in the laboratory to determine whether they could give us the high frame rate and pixel speeds we needed together with low noise at very low light levels.

[caption id="attachment_32337" align="alignleft" width="200" caption="Figure 2 – The Lucky Imaging system on the base of the 1.54m Danish Telescope"][/caption]

Based on our experience we chose the iXon 897 EMCCD from Andor, which is capable of producing images at very high readout speeds and negligible readout noise, even at very low light levels. We had to adjust conventional procedure for photometric data extraction to take account of the exponentially distributed EMCCD output and the bias variation, but worries about whether the stochastic nature of the EM amplification itself hinders accurate photometry over long time scales have been rebutted and we have developed an algorithm that elegantly reduces EMCCD data and produces stable photometry. In fact, the photometric scatter is close to the theoretical limits over most of the explored range of magnitudes.

Our Lucky Imaging system based around the Andor EMCCD has now been implemented on the 1.54m Danish Telescope (Figure 2) at La Silla, Chile, and has lived up to expectations in this real world environment on the outskirts of the Atacama Desert (Figure 3). The iXon 897 back-illuminated EMCCD has an 512 x 512 array of 16 micron pixels and the firmware has been specially modified by Andor to also read out the overscan regions; 20 columns to the left and 6 columns to the right of the image area.

We are at the very early stages of the search for life in the galaxy and habitable planets for our outward expansion but the signs are good.

Astronomers now predict that one in six stars hosts an Earth-sized planet in a close orbit, creating a pool of tens of billions of planets in our galaxy alone. Already, with less than 1,000 exoplanets discovered, four new planets ranging between 1.5 and twice the size of the Earth have been shown to be orbiting a star like our own Sun in a location where they could potentially have liquid water to sustain life. Given that we found a fraction of one millionth of the potential target planets, our chances of finding hundreds, perhaps thousands, of life-sustaining planets appears very high.

[caption id="attachment_32338" align="alignleft" width="149" caption="Figure 3 - The La Silla Observatory in Chile. In the southern Atacama Desert and at a height of 2,400 metres, La Silla is home to the European Southern Observatory and hosts several other national and project telescopes, including the 1.54m Danish Telescope."][/caption]

The significant improvement in resolution, fast readout times and negligible readout noise is a prerequisite for successfully observing gravitational microlensing events. After proving itself in Chile, the new imaging system is now to be fitted to the SONG telescopes, such as the Teide Observatory on Tenerife and Qinhai observation station on the Tibetan Plateau so that SONG’s quest to find small, earth-like exoplanets capable of supporting life can go forward with confidence.

Author: Kennet Harpsøe, Centre for Star and Planet Formation and Niels Bohr Institute, University of Copenhagen