'More than half the stars have planets'

Print edition : April 19, 2013

David W. Latham: 'We would really like to find evidence that there is life on one of these planets by finding a ‘biosignature’.' Photo: By Special Arrangement

An artist's illustration of the variety of planets being detected by the Kepler Mission. Photo: Harvard mithsonian Centre for Astrophysics

Interview with David W. Latham of the Harvard-Smithsonian Centre for Astrophysics.

DAVID W. LATHAM is Senior Astronomer with the Harvard-Smithsonian Astrophysical Observatory and the Centre for Astrophysics (CfA). He is a key member of and one of the leading investigators in the science team of the National Aeronautics and Space Administration’s (NASA) Kepler Mission to discover and characterise habitable earth-like planets. He also has the responsibility for follow-up observations and characterisation of transiting planets. He has also been designated as the Chief Mission Scientist for the proposed Transiting Exoplanet Survey Satellite (TESS) project, an all-sky survey from space. He is a member of Harvard University’s Origins of Life Initiative, working to characterise the bulk properties and geochemistry of exoplanets. Latham was recently in India to attend a workshop on Exoplanets and Recent Trends in Radial Velocity Method organised by the Physical research Laboratory (PRL), Ahmedabad. Excerpts from a wide-ranging interview he gave Frontline:

Where are we in terms of understanding exoplanets and the formation of planetary systems?

That’s a big question. Well, we have made enormous progress in the last 10-20 years because we now know of hundreds and hundreds of planets orbiting other stars. For example, we know that little planets—planets as small as the earth or a few times the size of the earth—are much more common than giant planets like Jupiter, at least, 10 times more common. That’s very interesting because it means that there are lots of small planets around stars. And more than half of the stars have planets; that’s a very recent result from the Kepler Mission announced at the American Astronomical Society meeting in January this year. That’s one of the most important insights that we have, that small planets are very common and systems of planets are very common. But so far we have not found very many systems exactly like our own solar system, with the giant planets out in very wide orbits and the terrestrial planets like the earth and Venus inside in shorter-period orbits. There may be many systems like that, but we haven’t found them yet because the facilities that we use are best at finding short-period orbits. That’s because we use the “transit technique”, which is much more likely to see short-period orbits. But we are pushing in that direction.

We have learnt a lot about the evolution of planetary systems. We now know that planets have to migrate. They form in one part of the area around the star, far out from the star typically. When the system is still very young, they migrate inwards. They go into much closer orbits with much shorter periods. That’s something that we had no good idea about 15 years ago except for one or two theoreticians who had predicted that something like that might happen. But most of us started with the solar system as the architecture that would be used for all of the planetary systems out there. So those are some of the things that we have learned—there are lots of small planets and there is in-migration.

What leads to this migration? Is there evidence of that in the solar system?

Well, actually, there is some pretty good evidence that something dramatic happened at a relatively early age of the solar system, maybe 200 million years after the solar system formed. There was an onset of chaotic rearrangement of the planets. The evidence for that are the craters that we see on bodies like the moon. The dating of the creation of those craters, the timing of when they formed, shows that there was a period of heavy, intense bombardment, and there are now pretty convincing models for the history of the solar system which show that things started out in a reasonably stable configuration but the orbits of the big planets evolved a little bit and then somebody got too close to Jupiter. There was a strong gravitational interaction and things were sent all over the solar system, and little planets and moonlets crashed into each other and some of the planets even swapped their positions.

That’s not exactly migration but that’s what the solar system probably had. What probably happened instead in many of the systems that we see where the planets formed early in the history of the system when there was still material in a disc, which provided drag and so the planets spiralled in and the disc disappeared. That’s one of the ideas for migration.

The other idea mostly involves the more massive planets, the bigger planets, where you might have several big planets forming out in the region which is cold enough for the material to condense and they interact gravitationally with each other—a little bit like the solar system had in its chaotic phase—and some of the planets are knocked out, while some go into orbits that reach into the inner part and they eventually circularise because of tidal mechanisms. But I think for many of the systems of small planets that we see with Kepler, where the systems have to be flat, because we see transits by two, three, four planets, which means they all have to have more or less the same orbital plane.

Otherwise, we wouldn’t see them all transiting. Those must have had very sedate histories. They must have migrated. They formed farther out and they have come in more or less together in formation, kind of shepherding each other. There are good theories of how that would work.

In your talk you mentioned that when Kepler started out you were picking up Jupiter- and Neptune-like systems and only of late you have been discovering more earth-like systems.

That’s because we were using ground-based techniques that were not very sensitive to small planets, and it was easy to find the big ones. We have to go to space to find transiting systems for very small ones, or people doing “radial velocity” measurements have to build instruments with much better performance or much better velocity precision to push down to smaller masses. So it’s really the evolution of the technology that has enabled the study of small systems.

So the discovery of smaller planets has so far been only from space and not from the ground. Is that right?

Well, velocity people work from the ground, but the real progress in finding transiting systems has been a result of going to space. Kepler is not the only space mission. The French-led COROT (COnvection ROtation et Transits planétaires) mission, which was launched a couple of years before Kepler, with a less capable spacecraft, led the way. That was pioneering. It didn’t have the sensitivity of Kepler. For example, they didn’t find the multiple transiting systems because it’s only small planets that are transiting and they didn’t have good enough sensitivity to see small enough planets. So they didn’t find any of those. But still they pioneered the technique. Actually, there was a Canadian mission with a little satellite the size of a suitcase, called MOST, that did some very early work; not very sensitive because it only had a small telescope.

But Kepler has made major breakthroughs because it’s a big telescope with a very good photometric performance. It’s a one-metre class telescope. It’s not an inexpensive project. The cost to launch, well officially it says $550 million on the website; it was really close to $650 million.

You mentioned that Kepler is good at finding short-period planets.

Those are the easiest ones.

But why do you say they are the easiest because I would imagine that, while they are transiting, it would be easier to pick larger planets instead of smaller ones because of greater dipping of light during the transit?

Well, it’s also the short-period ones that are easy to find because the transits happen more often, and you don’t have to wait as long to see them. Of course, the big ones are much easier to detect. For smaller planets, you have to have much better photometric performance to see the very shallow dips when only a tiny fraction of the light is obstructed.

Does the number of planets you have been able to see with Kepler tell us something?

Yes. If you figure out how many planets there are in our galaxy, almost every star has at least one planet. This is the conclusion that we get from the analysis of the Kepler data. We have only looked at a small part of the sky and nearby stars, so we are extrapolating for the whole galaxy. It looks like more than half the stars have at least one planet.

Of these, you mentioned only 100 or so have been confirmed.

Yes. Well, more than 100 now have been properly confirmed. By that I mean their masses have been measured accurately enough so that you have very high confidence that they have to be planets, which is to say that they have masses that we understand for their sizes—they are either rocks or they are icy or they are gas giants so that they fit our theoretical ideas for the structure of planets. Most of them have not been confirmed by dynamical mass measurements because we don’t have the technical capability to measure the velocities with enough precision to determine such small masses yet. That’s why the next generation of radial velocity machines are being designed for the really large telescopes where you need lots of light. These are very expensive instruments—G-CLEF [GMT Consortium Large Earth Finder] is a $25-million instrument—and hopefully they will allow us to push to the determination of masses for things that are as small as the earth.

But doesn’t the transit technique have a limitation? Only if you happen to be in the plane of the planet’s orbit will you be able to look at it. So how constrained are you even with Kepler in detecting planets?

That’s right. It’s a very serious constraint. At the shortest periods, only one out of 10 is going to be aligned properly. And if you go to longer periods—for example, if you are on a nearby star and looking back at the solar system—there is only one chance in 200 that you will be in the right plane to see the earth transit. At a one-year period, it’s about half a per cent! So we are seeing only a tiny fraction of the actual planets there, but transiting planets are very special because you get both the size and the mass, [and] you get the radius and the mass. [Kepler can measure only the size of the planet; only when this is combined with a follow-up ground-based radial velocity measurement can one get the radius and the mass]. So you can derive the bulk property; the density, for example, and also the surface gravity. So you know a lot about transiting planets if you have both the transits and the mass.

In terms of being able to pick up a transiting object, what is the limiting dipping of light that you are able to pick up?

The Kepler Mission was designed so that it could detect a planet like the earth passing in front of a star like the sun, and that signal is about 80 parts per million. So Kepler was specified to perform four times better; that is, 20 parts per million. So if you want to find something of the size of Mercury around a star like the sun, that would be very hard for Kepler.

But if you look around a small star—something that’s half the size of the sun—it would be four times easier because the area is four times smaller. So we do have a paper coming out very soon on Mercury-size transit detection, but that is the limit. The smallness of the planet that you can detect is set by how precise the photometry is.

And for velocity measurements, the sensitivity would arise from how precisely you can measure the Doppler shift. What kind of resolution would you actually require there to confirm a planet and correctly determine the mass?

Well, the earth is a good example. The motion of the sun in response to the pull of the earth is about 9 centimetres/second. You can walk faster than that. That’s pretty amazing that we are trying to measure the motion of a star at the speed that you can walk. The state-of-the-art is like half a metre per second, or maybe a little bit better; maybe a factor of two better depending on how you look at it.

What fraction of the planets that Kepler has picked up lie in the so-called “habitable zone”, or the “Goldilocks zone”?

Not very many. Only a couple of dozens. And we have not yet confirmed any rocky planet in the habitable zone although we are pretty close to that. That will happen pretty soon.

Is there a correlation between a large-sized planet system being a multi-planet system and a small-sized planet system being a multi-planet system?

Well, that’s one of the interesting results. The giant planets in short-period orbits seem to be lonely. We don’t see other nearby giant planets in the same system. It’s very different for the small planets. With the small planets, we see many systems which are compact, with closely packed orbits of two, three, four and even five and six planets. I have a “just-so story” (with apologies to Kipling) to tell.

The just-so story is that the giant planets are too unruly, boisterous and rambunctious, and they interact with each other because they have a lot of mass and they scatter and throw the other planets out of the system. Maybe there are other giant planets in the same system, but they are not in the same plane and we don’t see them. But with the small planets, they are much better behaved and they stay in a nice flat geometry. So we can see the multi-planet system because of that.

Does this also tell you something more about the formation of planetary systems itself?

I think so. Yeah. One might think, well, if you see all these small planets, they have stayed together and have ended up as the compact systems that we observe with Kepler, they must not have been a great big planet to stir things up because if there had been a great big planet, it would have messed everything up. Not necessarily though because we have Jupiter in our system and our system is pretty flat. It is not as flat as most of the systems that we find with Kepler because orbits are sometimes misaligned by more than 2° in the solar system; I think the maximum is 5°.

Why does the flatness occur?

Well, it takes us back to the original formation out of a flat disc. When you still have dust and gas, it’s natural to form a flat disc. You have some rotation, but the material perpendicular to the rotation comes down. And, if it’s still small particles, they collide with each other and so it naturally forms a flattened disc. If you formed the disc first and then formed the planets out of a flat disc, quite naturally you have a flat geometry. That’s an old idea; it dates back to 1750. It’s taken over 250 years to figure out some of the features.

Does the structure of the galaxy itself determine the geometry of the planetary formation?

It looks like nature found it easy to make planets around stars within a few thousand light years of the sun. There are even planets around pulsars, and pulsars must have blown up as supernovae. Somehow either planets formed after the supernovae or they survived. It’s pretty easy to extrapolate and assume that planets are common everywhere. So planets seem to be an almost universal phenomenon. Formation of planets must be a natural part of star formation, at least around stars like the sun.

The chemical characteristics of the stars around which planets are found seem to be a major focus of research in the understanding of these planets. Are there correlations between the characteristics of the planets and those of the host stars?

Yes. There are some interesting correlations emerging. The most important one, or at least the best established one, is that giant planets like Jupiter are found more commonly around stars that are rich in heavy elements. The richer the abundance of heavy elements, the more likely you are to find a giant planet around that star.

One of the usual interpretations is that you need raw materials to make planets, and the most popular idea on how to make giant planets is that you have to form a rocky core first, and once the rocky core gets big enough, then it can collect gas from the disc and build up to a giant planet. So it is easy to suppose that planets will be more likely to occur around metal-rich stars and it wouldn’t matter whether they were giant planets or little planets. But it turns out that little planets don’t require metal-rich environments to form. That’s one of the recent results that have emerged in the last year or two, especially from the Kepler Mission.

So I have a “just-so story” to tell about that one too. If you don’t have metal-rich environments, maybe you don’t form those rocky cores fast enough and the gas goes away—it’s been actually a long-standing problem for models of giant planet formation. The disc seems to dissipate too quickly. If you look at young stars you see, after five or, at most, 10 million years, the discs are gone. You don’t see the excess infrared emission from them. The people building the rocky cores had problems making them that quickly.

So if you have lot of raw material, you can make the cores fast and you can make giant planets. If you don’t make the rocky cores fast enough, then they don’t grow up to be giant planets. They might collect a little gas or they might just be left as true rocky planets, like the earth, with iron cores and the original atmosphere is gone. Maybe that fits in with the standard model of planet formation, the so-called core accretion model, that you could have lots of small planets. They just didn’t succeed in becoming giant planets because they didn’t do it fast enough.

But you also mentioned that some of the planets found by Kepler are not rocky.

Yes. A lot of these planets of the size of Neptune and smaller are low density. So they can’t just be rocks. Probably, they have extended atmospheres. They did get some gas, some hydrogen and helium. And that makes them look bigger.

Coming to the Indian team’s proposal to put up a ground-based observation system for exoplanets. Given what has been achieved so far by both ground-based and space-based observations, how do you see the potential for this programme? You did mention in your talk that there is still a lot of opportunity for one-metre-class telescopes.

I think the Indian initiative can make important contributions because we are badly limited in the number of radial velocity follow-up facilities we have. We have done only a hundred follow-ups out of about 3,000 Kepler Objects of Interest. That’s partly because we don’t yet have good enough instruments but is also partly because we don’t have enough telescope time. You need hundreds of observations sometimes to figure out the mass of the planet. So it takes lots of telescope time, and it’s good to see the performance that they are getting already with PARAS [PRL Advanced Radial-velocity All-sky Search] at Mount Abu. I think there is a significant role to be played by even a 1.5-metre telescope; 2.5 m is, of course, better.

But you had mentioned that by the time that comes up, much of the follow-ups would have been done.

For Kepler, yes. But there is the proposed follow-on mission TESS that will provide plenty of new candidates. That will be the next step. And in many ways the TESS discoveries will be more interesting.

The reason is that they will be nearby and brighter because it is an all-sky survey; we won’t miss the nearby bright ones.

So those will be good targets for mass determinations with PARAS and will be the best for the follow-up of the planetary atmospheres by missions like the James Webb Space Telescope. That will be the leading facility for studying the atmospheres of planets, not just planets like Jupiter, but we hope planets enough like the earth that you could even imagine living there. That’s the long-range goal.

And, of course, the question: “Are we alone in the universe?” We would really like to find evidence that there is life on one of these planets by finding a “biosignature” by seeing molecules in the spectrum of their atmospheres that have to be created by life.

What in your opinion are the real outstanding questions that need the immediate attention of exoplanet programmes?

We have to understand how common rocky planets, enough like the earth, are. Are any of them good targets for further study to see if they show any biosignatures? Do their surfaces have the right temperature for liquid water? Have they shed their original hydrogen and helium atmospheres and now have secondary atmospheres with the right molecules for the formation of life—the invention of life I call it, molecules like carbon dioxide or water.

To be interesting the planet doesn’t have to be an exact twin of the earth. We think that there are rocky planets that are two to three times more massive than the earth, they’re called super-earths.

Originally, people talked about maybe they could be 10 times as massive, but I don’t think we are finding any that big. That’s a big open question: what’s filling in the parameter space between planets like the earth and ones the sizes of Neptune and Uranus, which are four times the size of the earth and 14 and 18 times the mass of the earth. There is a big gap in the solar system all the way down to Venus and the earth. What does nature have in that big range? We need to find that out. That’s the transition region between Neptune and the earth that we need to study. That’s where I am focussing my attention at present.

Do you see the same gap in other planetary systems?

No. We see plenty of planets between those sizes and masses. It’s just some accident that we don’t have them in the solar system. We don’t understand that in any detail. Maybe they used to be there and they got busted up into asteroids. I don’t know. Makes for good science fiction.