Worlds Collide

Thousands of years ago, a convocation of astronomers on a world like ours might have awaited anxiously for their two suns to set, so that they could monitor the motions of their sister planet. They had been keeping track of its path across the sky studiously for years, and its irregular motions had defied explanations. Over the last decade, it had become brighter and fainter on a regular basis. People were worried that it could be a sign of doom. Although it would take hundred of years for this civilizations to develop a mathematical model for its motion, as it turns out, it was indeed a sign of doom. Their sister planet was in an eccentric orbit about their twin suns, and would eventually collide with their own world.

…

The star BD +20 307 first came to the attention of astronomers on Earth in the early 1990s. Although it was classified as a sun-like star, observations with the IRAS satellite revealed that it was unusually bright in the infrared (Stencel, R. E. & Backman, D. E 1991, ApJS, 75, 905 ; Whitelock, P. A. et al. 1991, MNRAS, 250, 638; Oudmaijer, R. D. et al. 1992, A&AS, 96, 625). This drew the attention of the group that I, later on, had a bit part in. A star will emit plenty of infrared light on its own, but most of its light emerges at higher energies, in the optical. When a star is seen with a large amount of infrared light, it is a sign that something much larger and cooler is near the star. In this case, the amount of infrared light could only be explained if that something was distributed in something like a cloud that was roughly the size of the Earth’s orbit around the sun. Inseok Song, Ben Zuckerman, and their collaborators concluded that the star was surrounded by a large amount of debris, which astronomers generically call dust (Song, I. et al. 2005, Nature, 436, 363). Dust is important to astronomers, because in the right situations, it can become one of the building blocks of planets.

Planets, and the stars that they surround, form when clouds of gas and dust collapse. Clouds of gas and dust exist throughout our Galaxy, slowly churning as they are heated by gravitational interactions with stars and with other clouds. When a cloud is disturbed enough, fragments of it start to collapse. However, turbulent motion within the cloud — whirls and eddies that are present everywhere in the tenuous matter of interstellar space — works against this  collapse. All of the clouds will end up rotating to some degree, and eventually the collapse of the cloud will be slowed when the cetrifugal force of the rotating gas and dust counters the inward gravitational pull. The center of a star finds a way to collapse (perhaps the jets of matter launched by collapsing stars can carry away enough angular momentum), but some of the material ends up forming a disk around the star. It is out of this disk that planets forms.

Although many stars are identified as being very young because they are surrounded by large amounts of dust, after examining BD +20 307, astronomers concluded that it wasn’t likely to be among the youngest stars. Its dust disk was not as large as infant stars, a few hundred thousand years old, like those seen in the near the “Pillars of Creation” in the Eagle Nebula, or at the center of Bok Globules near Orion. There was no evidence that BD +20 307 was still collecting material from its surrounding disk, as seen in stars only a few million years old (named after their prototype, T Tauri). In fact, BD +20 307 looked like it was at least tens or hundreds of millions of years old. Although this would make BD +20 307 young compared to the Sun (which is about 5 billion years old), finding this much dust around a juvenile star like BD +20 307 is unusual.

Dust is unusual around older stars because of two competing effects. The most important one is that young stars rotate quickly, is thought to drive powerful winds that clear away free dust. However, some of the dust gets gathered together to form planets. Over the first few million years of a star’s life, small flecks of dust collide and stick together, slowly growing from the size of pebbles, to that of rocks and mountains. Around our Sun, the remnants of these objects are asteroids, comets, and Kuiper belt objects. A rare handful of these objects grow large enough that their gravity takes over, and they collect enormous amounts of matter as they orbit the star. These become planets. At an age of a few million years, the inner planets should already have formed around BD +20 307.

However, in some cases, dust is only locked up in planets temporarily. In the early stages of a planetary system, the orbits of planets and asteroid-like objects aren’t yet settled. The planets are bombarded by smaller bodies for millions of years, and, occasionally planets will collide with each other. Our own moon is thought to be the product of a collision between the young Earth and a planet the size of Mars. Many of these collisions reduce the planets and asteroids back into dust. After the last major collision occurs, the dust is cleared away by the star’s light and wind relatively quickly. What is left behind is a stable system with a few well-separated planets that is relatively free of dust, like our own.

Astronomers are unsure how long it takes for a planetary system to settle down, and the major collisions to cease. The large amount of dust in BD +30 207 suggested to us that the process had not yet stopped. In fact, the amount of dust around that star suggested to us that a collision between two planet-sized objects must have occurred recently. Therefore, we decided to study BD +30 207 more, in order to determine how old it was, and obtain an estimate of how long the epoch of planetary collisions lasted around a Sun-like star.

We were in for a surprise.

…

Telling the age of a star like our Sun turns out to be a difficult task. (Actually, it is only one of three difficult tasks, telling how old a star is, how far away it is, and how massive it is. I will get back to the last two in a minute). We know the age of our Sun because we’ve measured radioactive isotopes on Earth and in meteorites. Although each radioactive atom decays into another element at a random time, large numbers of them decay with well-measured mean rates. As a result, by measuring the ratios of different isotopes — those of the isotopes that decay, and the isotopes they produce — one can estimate the age of a rock. Measurements on Earth suggest that it, and therefore also the Sun, formed about 4.5 billion years ago. Unfortunately, we don’t yet have the ability to measure these
isotopes in distant stars, so that tool isn’t available to tell us the age of BD +30 207.

To determine the age of a star, we need to develop a mathematical model for how they age. For a very young star, this isn’t too hard. A star like our Sun will start burning hydrogen about a million years after the cloud it forms from begins to collapse. However, the star isn’t done collapsing — it will slowly contract further for about ten million years. So, we can tell a star is young because it is still a bit swollen compared to an older star of the same mass. Unfortunately, this begs two questions. First, how do we tell that another star has the same mass? Second, even if we know the masses of two stars, how do we tell that one star is bigger than the other, unless we also know their relative distances?

To tackle all three questions at once, astronomers look for groups of stars that are at about the same distance from us, which are referred to as star clusters. These star clusters are thought to be fragments of single, large clouds that collapsed all at once, so that all the members of the cluster will have about the same age. The stars won’t all have the same mass, but that turns out to be helpful. Astrophysicists can make models for how stars with the same age but different masses will appear, and compare them to star clusters to determine the ages of the groups of stars. They find that many stars in Orion are still contracting, which according to their models implies that Orion 3-8 million years old (several generations of stars have formed, and are found in distinct groups within the nebula of left-over gas). In star clusters like the Pleiades and Hyades, however, the stars are all finished contracting. To determine their ages, astrophysicists use the fact that massive stars burn through their fuel much faster than less massive ones. From the masses of stars that remain, astronomers estimate that the Pleiades is 80 million years old, and the Hyades is 650 million years old.

So astronomers can guess the ages of star clusters, but how do they know the age of a star that is not a member of a large group, like BD +20 307? Frankly, it is very hard, and there are no real satisfying answers. Astronomers can only gather clues, and see if the pieces fit together reasonably. With estimates of the ages of the stars in Orion, they Pleiades, the Hyades, and maybe a dozen other nearby clusters of stars based on their models, astronomers have proceeded to look for other properties of stars that are related to their ages.

One might guess that one can use the fact that hydrogen is converted into helium in a star as a clock to measure its age — the more helium there is relative to hydrogen, the older it should be, right? Well, that is true on average, but it turns out that all of the hydrogen gets burned in the core of the star, and the helium that results doesn’t really make it to the surface. Therefore, we can’t see the helium.

However, there is another element we can use as a clock, lithium. Almost all of the lithium in the universe was formed, along with hydrogen and helium, in the big bang. Unlike helium, however, lithium is not produced in stars. In fact, lithium is so easy to burn in nuclear reactions, that even the lithium in the outer layers of stars will get consumed. Indeed, astronomers have measured the amount of lithium in the outer layers of stars in Orion, the Pleiades, the Hyades, and other star clusters, and found that the younger stars tend to contain more lithium. It isn’t a perfect test, because different stars in the same cluster have differing amounts of lithium. However, it is a place to start, so astronomers apply this test to other stars. When we measured the lithium in BD +30 207, we were surprised, because there wasn’t much. It suggested that the star was billions of years old, which is much older that we expected for a star in which planets had recently collided. Not knowing how to reconcile the two facts, we decided (too hastily, in retrospect) that the lithium measurement was probably not that useful, and looked for other ways of estimating ages.

As I’ve mentioned, as an interstellar cloud collapses to form a new star, its whirls and eddies get transferred to the nascent star. The star is born rotating quickly, like the archetypal ice skater that spins faster as she draws her arms in close. This rotation in turn drives a magnetic dynamo in the heart of the star, creating magnetic storms that burst out of its surface. You may be familiar with the manifestations of these storms on our Sun — dark, cool sunspots on its surface, dramatic solar flares, and a steady wind of plasma. The first two of these phenomena can be studied on other stars, and the third can be inferred indirectly.

We sought sunspots by looking for subtle, regular variations in the brightness of the star that should occur as the dark spots move across the face of the rotating star. With careful measurements, we measured how fast BD +30 207 was rotating. By comparing stars of different ages, we find that they slow down as they age, presumably because the star’s wind carries away its rotational energy. With this in mind, we measured the rotation rate of BD +30 207, and found that it was rotating quickly, once every 3.5 days (for comparison, the Sun rotates once every 27 days at the equator). This suggested that it was young — a few hundred million years old.

We next searched for evidence of magnetic storms. As a star ages and its rotation slows, the dynamo powering the magnetic storms loses power. This causes the stellar flares to become less intense. Astronomers measure these flares by looking at the highest-energy light from stars, X-rays. This is a rather crude tool, because in addition to decreasing over millions of years as a star ages, the magnetic storms also increase and decrease over cycles that last decades. Therefore, if a star is bright in X-rays it can be taken as a sign of youth, but if it is faint it is hard to tell whether it is a 100 million-year-old star in a lull of a decade-long cycle, or a billion-year-old star at the peak of a decade-long cycle. We requested that the Chandra X-ray Observatory observe BD +30 207, hoping that we would find that it was bright in X-rays, which would confirm our hunch that the star was young. We did find that it was relatively bright in X-rays, which suggested that it was only a few hundred million years old.

So, now we had three measures of the age of BD +30 207 — its fast rotation and bright X-ray emission suggested it was young, although its lack of lithium suggested it was old.

We could have looked at a fourth fact, but our preconceived ideas prevented us from considering it. I have already mentioned using lithium as a clock, because it is formed in the Big Bang, and consumed within stars. At the same time, as a star burns hydrogen and helium, it produces heavier elements, like carbon, nitrogen, and iron. Over time, stars release this material back into space, to populate interstellar clouds that then collapse to form the next generation of stars. Each generation of stars tends to contain more of these heavy elements, so from the amount of metals, we can get a rough estimate of the generation to which a star belongs.  BD +30 207 does not contain that much in the way of metals, which suggests that it is billions of years old.

I don’t think our group really considered this in our earlier analysis. Perhaps we did, and decided it was unimportant. Anyway, it is a moot point now.

As it turned out, we had missed something much more important.

…

As we were trying to puzzle out how to present our disparate measurements, we received a message from another collaborator, Alycia Weinberger. She had obtained spectra of BD +30 207, and found something puzzling. Before getting into that, though, I should explain the significance of a “spectrum.”

A spectrum is one of the main tools that astronomers use. A spectrum is examined by splitting the light from a star up into its composite colors, much like a prism can split the light of the Sun up into a rainbow. The main reason astronomers do this is to identify the elements that compose a star. Each element in the periodic table responds to specific wavelengths of light. For example, neon emits an orange light when it is heated. Similarly, sodium emits yellow light, and copper a green light. Conversely, cool atoms tend to absorb light of the same specific frequencies that they can emit. Therefore, when an atom lies in the atmosphere of a star, it absorbs specific frequencies of light, which creates dark bands in the spectra of stars. Looking for these dark bands is how we measured the lithium abundance in BD +30 207.

Astronomers can do even more with the spectrum. If a star is moving, for instance, the lines will be shifted in wavelength. This is referred to as the Doppler effect. On Earth, we are familiar with the Doppler effect on sound, which causes a fire engine’s siren to have a higher pitch when it is approaching a listener than when it is moving away. By measuring the amount by which the spectral absorption bands produced by various elements are shifted from their known frequencies, we can measure how fast a star is moving. This is important to astronomers for two reasons. First, all stars are moving within the Galaxy, so measuring the velocities of many stars allows us to estimate how fast the Galaxy rotates. Second, most stars are found in pairs, orbiting each other like the earth orbits the sun. Measuring how the velocities of pairs of stars changes as they orbit each other allows us to estimate their masses. Finally, for single stars, the measurements of the frequencies of spectral lines have become accurate enough in recent years that the tug of planets can be measured.

There were two surprises in Alycia’s spectrum. The first was that lines from each element showed up twice in her spectrum. The second was that the frequencies of the elements shifted periodically (Weinberger, A. J. 2008, ApJ, 679, L41). BD +30 207 was not a single star, but two nearly-identical stars orbiting each other every 3.5 days. (We had, in fact, checked once before to ensure that BD +30 207 was not a pair of stars. However, we got unlucky, and only saw evidence for one star. This was presumably because the stars were at the place in their orbit where they have the same velocity, so their spectral lines overlapped. We only took one observation, and, in retrospect, there was a 10% chance that that would happen.)

This turned our conclusions on their head. Two Sun-like stars orbiting each other every 3.5 days will only be separated by 8.5 million kilometers. This is roughly 1/8th of the distance between Mercury and the Sun, and only about 12 times the radius of the Sun. Two stars orbiting this close will exert enormous tidal forces on each other, which will cause their rotation to be synchronized with the orbit (the same thing has happened with the Moon, which causes the same side to always face the Earth). This will keep both stars rotating rapidly throughout their lives. This explained two of our observations: the stars are rotating fast, which in turn will power a strong magnetic dynamo, making the pair bright in X-rays.

From our study (Zuckerman et al. 2008, ApJ, 688, 1345), we only had two measurements of the age of the star. The lack of lithium in its atmosphere implied that it had been consumed over at least a half-billion years. The lack of heavy elements in the star implied the star was a member of a generation that formed billions of years ago.

We were left with a puzzle: for some reason, after a billion years orbiting without drama around this distant pair of stars, two planets suddenly collided. Was it because planets around a pair of stars is unstable? Astrophysicists’ best calculations suggests that they shouldn’t be, so long as the planets orbit far outside that of the two stars. However, chaos exists at some level even in our own Solar system, and it is conceivable that having two stars at the center of a panetary system would exacerbate the chaos.

Or, perhaps the fact that there is a pair of stars in BD +30 207 is unimportant, and its planets were sent on a collision course because a third star disturbed the system. Some have suggested that gravity of passing stars could disturb the most tenuously-held members of our own solar system, sending comets hurtling toward the sun.

This question remains to be answered.

. . .

In my idle time, my mind wanders along more speculative avenues, and I begin to imagine what it would have been like to live on the planets around BD +30 207. That someone might have watched these events unfold is far-fetched, but not impossible. On Earth, life probably first formed within about 1 billion years after our planet first condensed from circum-solar dust. It took another 3.5 billion years for multi-celled life to emerge, 4 billion years before dinosaurs roamed the Earth, and 4.5 billion years for humans to emerge and contemplate our world. We don’t know whether life forms more or less readily on planets around other stars. Nonetheless, there is a chance that life could have existed on the planets around BD +30 207, before they collided. It might even have realized its fate.

What would they have seen? Once one of the planets was disturbed, its orbit would have become eccentric. An astronomer on the other planet would have seen the disturbed planet brightening and fading, as it alternately approached and receded from the central pair of suns. If they developed a theory of gravity, like Newton did, they might have questioned whether their planetary system could survive, or whether some invisible hand would have to intervene to set it right. Newton wondered the same.

Indeed, from Earth, in the 18th century, it was realized that over the last few hundred years, Jupiter had been moving closer to the Sun, and Saturn farther away. This caused some concern among natural philosophers and mathematicians of the day. However, Pierre Simon de Laplace, through arduous and innovative computations, showed that the motions of Saturn and Jupiter was part of a long, periodic process, and that ultimately their orbits were stable (from the Smithsonian Annual Report, 1874, pp. 129-168, as presented in The Golden Age of Science, edited by Bessie Zaban Jones, 1966, Simon and Schuster, pp. 71-73). Unfortunately, our supposed astronomers on the planet around BD +30 207 would find no such comfort in mathematical calculations.

Eventually, the disturbed planet would disrupt the orbit of the home planet of our ill-fated astronomers. The poles of their home world might shift, throwing half the world into a harsh, hot eternal day, and half into a frozen night. Perhaps the orbit of their home world would itself become disturbed, causing further wild variations in the seasons, as the planet plunged toward the pair of suns, or was cast outward toward the blackness of space.

This could go on for tens of thousands of years. If life were not extinguished by the cruel variations of the climate, there can be little doubt that civilization would be unable to survive. Crops would fail, rivers change their course, and continents would become covered in glaciers, or broiled by a runaway greenhouse effect.

Still, I persist in imagining that intelligent life would find corners of the world to hold on to, perhaps huddled underground. In dim artificial light, they might recall stories of the day that their ancestors first realized that a wandering star in their sky had foretold their doom. Perhaps some geometers, part of an ancient monastic order, had worked out how the orbits of their planets would progress. On the final day they would emerge, and look up to watch the face of their sister planet fill their sky, and bear their end with stoic resignation.