Feature

Hunting for New Planets

How can we detect planets orbiting distant stars?

by Chris Smyth

Long the domain of science fiction, the search for planets outside our own solar system is now respectable science. Canterbury University astronomer Dr John Hearnshaw has been searching for planets for five years, using the university's observatory at Mt John, near Lake Tekapo. Research in this field has got to the point where he's ready to say there's a good chance planets will be found within the next ten years.

Hearnshaw is one of only half a dozen or so researchers in the world -- and the only one in the southern hemisphere -- using a spectroscopic technique to try and find minute changes in star movement that could be caused by planets.

Motivated by "the challenge of trying to find a planet, something that's never been done before," he says the ultimate reward would be to eventually find life.

For now though, Hearnshaw is content to search for planets. A variety of methods have recently become feasible. They're generally on the very limits of what's technically possible right now, but they've all been tried or at least contemplated by astronomers.

Take a Look!

Simply pointing a telescope at a star and looking to see if it has a planet might seem the most obvious method. Unfortunately, this is much harder than it sounds, or planets would surely have been found by now.

Planets shine only by light reflected from their star and, because they're small, they can't reflect a lot. For example, to actually see the Earth from a long way off, we'd need to be able to detect an object right next to the Sun, but shining with only about one ten-billionth the Sun's apparent brightness. On the basis of a rapid back-of-the-envelope calculation, Hearnshaw estimates the Earth reflects too little light to be seen directly from more than one parsec (3.26 light years or 30,000,000,000,000 kilometres) away. The distance to the nearest star, Alpha Centauri is 1.3 parsecs.

The real limiting factor here is not whether the planet reflects enough to be seen, but whether or not a very faint planetary image can be made out in the glare of the nearby star.

According to Hearnshaw, turbulence in the Earth's atmosphere and limitations in telescopes' optics mean it's unlikely planets will ever be seen directly from the Earth's surface. From space the odds are a bit better. A telescope with a mirror specially designed to reduce blurring diffraction effects, mounted on a satellite above the distorting effects of our atmosphere, could theoretically see Jupiter, which is much bigger and brighter than the Earth, from as far away as ten parsecs.

Another option is to observe in the infrared part of the spectrum. At a wavelength of 20 microns, the Sun emits only one million times more radiation than the Earth, instead of ten billion times more as is the case for visible light. The glare problem would thus be reduced, but unfortunately we would still have to observe from space, because very little infrared radiation gets through our atmosphere.

Through Sunlight and Shadow

The photometric method relies on very careful measurements of the brightness of stars -- and a huge amount of good luck.

If a planet passes between its star and us here on Earth, then the star's apparent brightness would be diminished a little. So, by carefully observing the amount of radiation received from stars and watching for occasional dips, we may be able to infer the presence of planets.

Unbelievable though it may seem, this method might be technically feasible -- if not practically so. Seen from afar, Jupiter would block one percent of the Sun's light. The best stellar photometry techniques available today have random errors of about plus or minus one-tenth of one percent, making observation of any brightness variation of the Sun due to a Jovian transit a reasonable proposition. The Earth, on the other hand, would produce a drop in brightness of a mere 0.01 percent. Hearnshaw says space-based observations might be capable of detecting such a small dimming but because of atmospheric effects this could never be seen from the ground.

However, the odds against finding a planet by the photometric method are quite literally astronomical. Firstly, we need to be lucky enough to be looking at another planetary system virtually exactly edge-on. Secondly, transits only last a few hours out of orbital periods that could be decades in duration. And thirdly, this method allows no way of distinguishing planets from any small, dim companion stars that may be present.

Hearnshaw estimates that if all stars have planets, then round-the-clock monitoring of 2,000 stars would find a transit in about a year, but this is hardly a feasible option!

Wobbling Stars

To most of us the pattern of the stars in the sky seems unchanging--we expect the Southern Cross to retain its shape. However, the stars are actually all moving with respect to each other. This movement will eventually so distort the shape of the Southern Cross, for instance, that by about the year 28,000 the New Zealand flag is going to have to be changed -- regardless of what traditionalists may say.

This movement of the stars is known as their "proper motion" and the study of star position and movement is called astrometry. With accurate telescopic measurements of stars' positions over only a few years, such proper motion is relatively easy to detect, especially for our nearer neighbours.

If a star has an unseen orbiting companion, then the companion's gravity will cause the star to be tugged a little, first one way, then the other, as the planet moves around it. Viewed from Earth, this will cause the star's proper motion across the sky to be "wobbly". Provided we observe a star for long enough, these small wobbles should cause any planets to reveal themselves.

But there are problems with this method of planet detection. Such stellar wobbles would be minute because planets are generally much less massive than their stars and could only have a very small effect on the motion of the star. To get any useful data we need to look for stars with big planets, like Jupiter, and planets like this will be relatively distant from the star concerned. This is because the huge quantities of hydrogen and other gases that make up planets of this type boil away if they are too close to their star. Large distances mean they take a long time to orbit, thus making the star's "wobbles" very slow. In effect there's a trade-off -- the bigger the wobbles, the longer we've got to wait to see them.

Another problem is the likelihood of there being several planets of differing masses and orbital times, making the star's wobbles rather complicated.

Despite all this, the American-based astronomer Peter van der Kamp announced in 1976 that observations since 1938 had given him sufficient astrometric data to indicate the presence of two planets around Barnard's Star -- a small red star that is the second-closest star to the Sun. One of these planets supposedly had a mass about nine-tenths that of Jupiter, and orbited Barnard's Star every 12 Earth-years. The other was said to have four-tenths Jupiter's mass, and orbited every 20 years. Unfortunately no-one has since had the time to repeat his observations and -- worse still -- even using van der Kamp's own data, no one can reproduce his conclusions.

Just Ask Them

By far the most dramatic method of finding an extra-solar planet would be to receive a message from any alien civilisation that may live there. The certain knowledge we're not alone in the universe would profoundly change our civilisation -- but how likely are we, really, to hear from anyone else?

Hearnshaw has his doubts. If there are civilisations out there -- some presumably more advanced than our own -- then how come we haven't heard yet, he argues? As for planets without intelligent life, or with less technologically advanced races: we can't realistically expect to receive a message.

Even if we ever did make contact, the universe's ultimate speed limit -- the speed of light -- is much too slow for effective dialogue to take place across the vast interstellar distances.

There are several programmes under way aiming to pick up signals from alien civilisations, including the Search for Extra-Terrestrial Intelligence and NASA's microwave observing project. Hearnshaw thinks they're worth trying, but says he's "more interested in trying to find planets."

A Spectroscopic Bet

Hearnshaw has his money on the spectroscopic method -- and he's been searching for planets this way at Mt John since 1988.

As with the astrometric method, it takes advantage of stars being tugged back and forth by any orbiting planets they may have. The slight to-and-fro motion of the star with regard to Earth causes the star's light to be Doppler-shifted: as the star moves towards us, its light waves will "bunch up" a little, causing them to have a slightly shorter apparent wavelength. Similarly, as the star moves away, its light gets "stretched out", giving it a wavelength a bit longer than before. (The same effect causes emergency service sirens to have a higher-sounding pitch when approaching than when speeding away.)

Every star's radiation has its own characteristic spectral pattern -- its "fingerprint" -- containing dark lines at specific, well-known wavelengths. As the radiation is Doppler-shifted back and forth, the positions of these dark lines in the spectrum move back and forth too, in a precisely periodic way. By observing this, by noting how long it takes for the motion to repeat itself and by how much the radiation is Doppler-shifted, Hearnshaw hopes to be able to deduce not only the presence of any planets, but also their distances from the star and a lower limit on their masses.

Technical problems abound. The radiation will only be Doppler-shifted by a tiny amount, because planets' small masses mean a star's back-and-forth motion will be very slight indeed. Hearnshaw calculates that a planet of Jupiter's mass, orbiting at a distance five times that of the Earth from the Sun, would cause a star of the Sun's mass to have a velocity variation of plus or minus 13 metres per second (about 47 km/h).

This is on top of a star's typical speed through space of up to about 60 kilometres per second with respect to the Sun. You also have to take into account the fact that the Earth is orbiting, moving with respect to the Sun at about 30 kilometres per second -- a velocity comparable to that of other stars -- with its direction of motion constantly changing. These motions, of other stars and of the Earth, cause much bigger Doppler shifts in stars' light than any possible planet-induced variations. Somehow, these gross effects must be cancelled out so only the tiny sought-after Doppler shifts can be distinguished.

Fortunately this can be done. The Earth's orbit around the Sun is very precisely known, meaning Doppler shifts due to its motion can be quite easily cancelled out. Then, to see any stellar movement caused by planets, we need only look for changes in stars' motion along the line of sight. The up to 60 kilometres per second component of a star's movement is essentially constant, meaning the resulting large Doppler shift is constant too and can be ignored. Small variations are all that count.

To observe such tiny variations in Doppler shift, spectrographs -- the devices that detect the position of lines in stellar radiation -- need to be kept very stable, both mechanically and thermally. Any vibration of the equipment or variation in its temperature will cause spurious changes in the apparent position of detected spectral lines, and thus render useless any results obtained.

The answer lies in optical fibre technology. Prior to the 1980s, spectrographs were physically attached to telescopes. The telescopes were inside domes that had to be opened to the sky and thus were subject to temperature variations during use. They were constantly in motion as particular objects were tracked across the sky. Now, however, an optical fibre can carry light from a telescope to the spectrograph, which is kept separate in a chamber that is thermally stable and kept as vibration-free as possible.

Surprisingly, Hearnshaw says this is one of the few areas of optical astronomy that can be carried out better on Earth than in space. It's easier to control temperature and eliminate vibration here on the ground, and of course it's much cheaper and more convenient. Atmospheric conditions need only be good enough for the star's light to be sufficiently stable for focussing on the end of the optical fibre -- a much less demanding requirement than many others in astronomy.

Equipment at the Mt John observatory currently has the theoretical capability of detecting stellar velocity changes as small as 50 metres per second (180 km/h), but with design improvements, Hearnshaw believes this figure will be down to 20 metres a second (72 km/h) by next year, and in principle this is precise enough to find planets.

So, when can we expect to hear news of a new planet being discovered? Hearnshaw hedges his bets on this question but recalls that he and a former PhD student Kaylene Murdoch spent four years on their previous planet-hunt. At the end of it no planets had been found for certain, but they had found some stars with velocity variations, and the presence of planets could not be ruled out as a possibility.

A new campaign is being launched in 1994 and, based on his past experience, Hearnshaw says results should be available four or five years later. With the improvements at Mt John, he says "There's a good chance that by the end of the century ourselves, or someone, may have success."

What a way to end the millennium.

Mt John

The telescope at the Mt John Observatory, near Lake Tekapo, has a one-metre diameter mirror, making it the largest in New Zealand.

For his planet-hunting work Hearnshaw uses this instrument together with an optical fibre, 25 metres long and 105 microns across to transmit light to a two-ngstrom per millimetre échelle spectrograph. (The "ngstrom", equal to one ten-billionth of a metre, is a convenient length unit for dealing with light. Radiation with a wavelength of 5800, for instance, is yellow light.) So, at Mt John, the spectrograph translates a wavelength change of two ngstrom in the incoming light into a shift of one millimetre on the resulting spectrum -- a very high level of dispersion.

At the heart of the Mt John spectrograph is a grating that reflects light at different angles depending on its wavelength. The greater the wavelength, the larger the angle of reflection. This process, known as "diffraction", will therefore split light into its constituent colours. Any dark lines (ie. specific wavelengths where there's less light) will then be revealed on the resulting spectrum.

An échelle spectrograph uses a grating, with its facets (the parts from which the light is reflected) aligned so light is reflected at a very high angle. This means any small variations in wavelength result in large changes in reflection angle.

Such technology, conceived in the 1940s and in astronomical use since the early 1970s, allows very high precision to be obtained in spectrographic work. In 1976 Mt John was the first observatory in the southern hemisphere and only about the sixth in the world to install a spectrograph of this type.

The spectrograph is stored in a cabinet whose temperature is controlled to a few hundredths of a degree per night. To reduce thermal noise, the detector is cooled by liquid nitrogen.

On any particular observational run a stellar velocity precision of plus or minus 20 metres per second can be obtained. Due to systematic errors incurred in dismantling and reassembling the apparatus however, the long-term precision is currently plus or minus 50 metres per second.

Chris Smyth is a freelance journalist specialising in science stories.