The Life of Super-Earths (6 page)

Fortunately we still had five candidates left for the observations at Keck. We were optimistic, but it was already clear that the initial high expectations, that more than half of the transiting candidates could be planets, were severely dashed; what's more, at least two more tests remained. It wasn't impossible that all the candidates would have to be struck from the list. In retrospect, we were very glad to see that our transiting method was working so far—for the first time I felt that we had a clear edge and were ahead of the competition, who were mired in a heap of false positives. If false positives had not been such a major problem, it was conceivable that any of the other competing teams would stumble on a planet from the list by chance and beat us by a month or less in announcing the first one!

The Keck observations went fine, and we seemed to have a clear winner among the four candidate stars we had observed: OGLE-TR-33. It was given this less than poetic name because none of the stars on the list had been observed before, so it was named for its place in the catalog of the team that first observed it. Star 33 had a clear wobble, with an amplitude that corresponded to a very large planet or, more likely, a brown dwarf, the name for a small failed star. Finding a transiting brown dwarf was almost as exciting back then as finding a large planet, so we rushed testing OGLE-TR-33, only to find that it failed the last test. We could not believe it—we had even started writing a paper to the journal
Nature
, while doing our test and models for a second time. Now we had to abandon it. In the meantime, another
one of our top candidates had passed all its tests with flying colors. Initially we had neglected it because OGLE-TR-33 had a clear large wobble and had seemed an easier nut to crack. This star was OGLE-TR-56, and it looked like a Jupiter-mass planet.

As November was ending, we had finally discovered the nature of OGLE-TR-33: it was a system of three stars, two of which orbit each other very closely and eclipse each other, while a large star nearby, the big and brightest in the system, washes out the deep eclipse of the other two. The third star does not have a wobble of its own, but a large wobble of another star in the system (its spectral lines have a large Doppler shift) causes a small distortion in the spectral lines of the third star. Because the third star rotates fast and its spectral lines are broad, that small distortion was just enough to give us the impression of a small wobble, as if due to a planet that matches the shallow washed-out eclipse. OGLE-TR-33 was the ultimate tricky false positive!
¹²

Now we focused our full attention on OGLE-TR-56. It had passed all our tests, including the spectral lines distortion test that had uncovered OGLE-TR-33 as a false positive. We felt very confident that OGLE-TR-56b was a planet precisely because of our experience with OGLE-TR-33 and the other false positives our new transiting method had helped uncover. The method was working, and we quickly got a paper accepted for publication in
Nature.
¹³
In the first week of January 2003 I flew to Seattle to present our discovery to the meeting of the American Astronomical Society. Just like
Captain James Cook 235 years earlier, we had crisscrossed the Pacific Ocean to catch a glimpse of a transit, and we had succeeded.

To top it off, OGLE-TR-56b was an exotic planet—a record holder in several ways: the shortest known orbital period (only twenty-nine hours), hence the hottest known planet (close to 2000 K), as well as the most distant extrasolar planet (at about 5,000 light-years from Earth). The exotic properties caught the attention of the media, while for us and the planet hunters the biggest excitement was that the transiting method for planet discovery was finally figured out. In short, the unexpected large fraction of false positives had been the major obstacle, and our set of tests and use of stellar models solved the problem. Within the next three years we and several other teams would use our approach successfully to confirm more than a dozen new transiting planets. The path to discovering a true Earth was now open. A new age of exploration was upon us.

The first age of exploration began in the fifteenth century. In 1484 one of the men whose efforts would define the era, Christopher Columbus of Genoa, was trying to convince King João II of Portugal to finance his expedition to cross the Atlantic. The king was reluctant, not because he thought Earth was flat, but because Columbus insisted that it was only 10,000 miles around the equator, and that the westward route to India and the Spice Islands would be short. Portuguese sailors (who, thanks to the support of Prince Henry the Navigator earlier in the century, had sailed up and down the Atlantic by the African coast) had estimated a much
larger size for the planet, pole to pole, and had gotten a number much closer to the actual value of about 25,000 miles. (Back in the third century BC, Eratosthenes, a Greek mathematician, had already estimated the same size.) As a result, King João II did not fund Columbus, who then left for Spain. He had the wrong number, but luck was on his side and he stumbled upon the New World.
¹⁴
The Portuguese, in the meantime, went to India sailing around Africa.

The search for new Earths is no different. Since the late 1990s we have had the knowledge and the technology to do it. We have debated numbers and methods. Now we are sailing and waiting for the day when one of us will shout “Terra!”

 

Just like the Portuguese sailors, who started exploring nearby areas, planet hunters must do the same. The sailors would venture into the Atlantic and find islands like the Azores or farther down the coast of Africa, or perhaps even get an early glimpse of South America off the coast of modern Brazil. Our team also began small and cheap during the initial “gold rush” on discovering transiting planets. Most notable of our early stakes is the Hungarian-made Automated Telescope, known as the HAT project or network (HATNet, for short). HATNet is led and literally put together by a young colleague of mine, Gaspar Bakos. Gaspar came to the Harvard-Smithsonian Center for Astrophysics as a graduate student at the enthusiastic recommendation of one of my mentors (and Gaspar's mentor too), Bohdan Paczynski (1940–2006) of Princeton. The rationale for the recommendation was that Gaspar was the person to take advantage of a revolution in
digital imaging (cheap CCDs, such as you can find in any digital camera) and precise image processing.
¹⁵
Together, the two technologies meant we could look at “all the sky, all the time.” Or at least lots of the sky, very often. That, Bohdan believed, could bring a revolution in astronomy. As so often before, Bohdan turned out to be right.

There are many applications for the “all the sky, all the time” approach, but discovering planets by transits is an obviously good one. My colleague Robert Noyes and I thought so and convinced Gaspar as well. Thus HATNet was born on two continents and on a shoestring budget—with photography equipment for telescopes and amateur-grade CCDs, but with professionally designed and machined hardware (in Hungary, Gaspar's home country) and software.

HATNet comprises six small telescopes (not much different from the large cameras with zoom lenses used by professional photographers), four on Mount Hopkins in southern Arizona and two on Mauna Kea.
¹⁶
They are automated, following a cleverly written computer program that receives inputs (e.g., priorities for what should be observed) from the astronomers during the day; then they work all night like robots. HATNet shuts down in case it detects too many clouds or inclement weather, and Arizona communicates important updates to Hawaii. Remember that Hawaii experiences sunset and sunrise later than Arizona. Therefore, the two HATNet telescopes in Hawaii begin their work night later and take over fields that the Arizona telescopes can no longer see. By having “eyes” in Arizona and Hawaii, HATNet effectively extends its work night from twelve to fifteen hours, and can catch and see
more transits.
¹⁷
HATNet was designed to discover transiting planets like Jupiter and Saturn at nearby stars. It has found thirty so far, with some of them (e.g., HAT-P-11b) the size of Neptune; what's more, HATNet and other projects like it have set off a transiting “gold rush.”

Just as anyone could set off for California with a sluicing pan and some grub, the would-be astronomer of today can set herself up to find a new planet simply by maxing out a credit card. (I hope nobody does that literally.) It hasn't all been shoestring budgets, though, as the big guns, such as NASA and the European Space Agency, did not wait long to join the rush. Although a transit-hunting operation can be set up on a limited budget, the big agencies have a real advantage—they can get things lifted into space. Stars twinkle when seen from Earth because the air moves constantly, and not just in one direction; there are different currents at different altitudes. The air currents act as sheets with multitudes of lenses that shift the point-like images of stars, just like the sunlight playing on the bottom of a swimming pool. Much of this twinkling happens at high frequencies—about once every hundredth of a second. And, as it turns out, all this makes detecting a transit of an Earth in front of a Sun-like star practically impossible even with the largest telescopes. A telescope in space has its own challenges (although as the Hubble space telescope taught us, they are not insurmountable), but to find small planets, it's the only way to go.

The Europeans already had a mission in the pipeline that could easily accommodate the transit hunting. Hunting the
co
nvection and
rot
ation of stars (known as COROT, and
referring to the famous pointillist impressionist painter), the telescope was to obtain thousands of images of stars, resembling pointillist paintings, in order to study their subtle light variations and help understand basic things about stars, like their rotation and internal convection. Of course, COROT could do an excellent job detecting planet transits in the process, with its name now standing for convection, rotation, and transits (CoRoT).

In the meantime, NASA had missed an early opportunity to fund a space telescope dedicated to discovering transiting planets, one with a stated goal to discover planets as small as Earth and determine how common planets like ours actually are. William Borucki of the NASA Ames Research Center in California had been trying to convince the agency that his experiment would succeed. Even a null result, meaning that no transiting Earths were discovered, would be meaningful, implying that planets like ours are very rare. NASA panels had turned him down before, but in 1999 Bill Borucki was assembling a crew to propose again; he asked me and a dozen more colleagues to join. The successful discovery of more and more exoplanets with the Doppler shift method was a powerful new motivation.

Even though I hadn't done much work on the problem at the time, I had first thought of discovering such planets in 1999, when a group of us, mostly at the Harvard-Smithsonian Center for Astrophysics, and mostly observers and engineers, came together to propose to NASA an innovative space telescope design for planet detection—with a square mirror, as opposed to a round one. My colleagues Costas Papaliolios and
Peter Nisenson had invented this unusual design in order to minimize stellar glare and allow glimpses of planets huddled close to their stars. With a team of about twenty and led by our experienced space mission scientist Gary Melnick, we prepared a detailed scientific and engineering proposal.

My job on the team was to work out what kind of planets our telescope might be able to discover. It seemed then that super-Earths were in reach. (I liked to call them super-Earths and super-Venuses for short, as it had been common in astronomy to use the adjective “super” for newly discovered or hypothesized objects that are larger in size or energy than known ones. For example, stars that are larger than giant stars are called supergiants, explosions that are stronger than novae are called supernovae, and so on.) The shorthand stuck, as you've probably already surmised.
¹⁸

Finally, in December 2001, the Kepler mission, as it was now known, was approved. Seven years later, in March 2009, Kepler was launched from Cape Canaveral in Florida and two months later beamed down to Earth images and exquisite measurements from stars and planets hundreds of light-years away. Kepler is NASA's first mission capable of finding Earth-size and smaller planets within the habitable zone of stars similar to the Sun (a topic we will return to later). It is a fairly modest telescope, about half the size of Hubble, but with a lens that allows a wide field of view, captured by the largest ever camera built for a NASA science mission—a 95 megapixel monster that can image millions of stars in a single shot.

The NASA Kepler mission will use the transiting method to discover planets like Earth—of the same size and in similar
orbits. The goal of the mission is to do so in a systematic and comprehensive manner, so we can find out what fraction of stars have Earth-like planets. The mission is designed to avoid most of the pitfalls of the standard transiting method by investing in several years of preparation—a comprehensive and carefully vetted input catalog of 200,000 stars that the Kepler telescope is going to search for planets. After checking most of these over the initial few months, Kepler is to settle on about 100,000 to 120,000 stars for the life of the mission, about three to four years. All the advance work should mean that we will get very few false positives among the planet candidates from the Kepler photometry; at the Kepler team we have a plan how to weed out the remaining false positives, using the same methods we applied to the OGLE catalog. As I write this the approach is clearly paying off; our preliminary evaluation is that the fraction of false positives is very small.

Ultimately, the transiting method is more than just a successful technique to discover planets, including Earth-like ones, as valuable as this may be. Transiting planets are the best ones for us to study in the short term. Because we know their masses and radiuses precisely, we can infer their bulk composition from the mean density. In addition, observing transits enables us to remotely analyze a planet's atmosphere. During a transit, the light of the star is colored by the presence of the planet's atmosphere; we can compare the spectra then to the spectra observed when the planet is not transiting, and use spectrographic techniques to identify the chemicals—such as water, methane, or carbon dioxide—present in a planet's atmosphere.