LOW IN THE SKY on a clear winter鈥檚 night you will see the brilliant stars of the constellation of Orion. Some of these stars are only a few million years old 鈥 newborns in stellar terms 鈥 part of a vast conglomeration of recently formed stars. What is going on here to make amorphous clouds of interstellar material give birth to stars? And how will the stars that are formed turn out? Will they develop planetary systems like our own? Over the past few years we have come a lot closer to finding out.
Astronomers believe that stars like the Sun start life in the cores of giant clouds of dust and molecular hydrogen measuring several light years across. At some point, one of these clouds can be triggered to collapse under its own gravity. Just what this triggering mechanism is remains a puzzle, but most of the current explanations involve turbulence and collisions between clouds.
These collisions leave the collapsing cloud spinning, albeit very slowly. As the cloud contracts it inevitably starts to spin faster, its angular momentum being conserved 鈥 just as ice skaters spin faster if they pull in their outstretched arms. However, a cloud that simply carried on contracting without losing any angular momentum would never be able to form a stable star. The central star would end up spinning at an enormous rate 鈥 so fast that it would tear itself to pieces.
Advertisement
But the cloud does condense into a star because it is, in effect, viscous. As particles bump against slower ones slightly farther out, they lose energy, and as a result those closest to the star spiral down onto it. So, at the centre, the material should accumulate in a protostar 鈥 a dense ball of gas with a hot core that will eventually become a new star 鈥 and the remains of the spinning cloud should form an orbiting disc around it. This picture was backed up last year by dramatic images taken with the Hubble Space Telescope, which showed protostars embedded in disc-like structures some ten times the size of our Solar System, silhouetted against the glowing backdrop of the Orion Nebula.
But while the inner parts of the disc are being emptied in this way, angular momentum is gradually exported towards the outer regions of the disc. So moving material towards the central star becomes more difficult the farther out you go, and eventually these outer parts with their high angular momentum are left behind. Ultimately they may coagulate into lumps that go on to form planets, as in our own Solar System. Around 99 per cent of the angular momentum in the Solar System lies in the orbital motion of the planets, mostly in the giant outer planets Jupiter and Saturn.
Until recently, astronomers thought that material at each point in the disc would circle at the natural orbital speed: just fast enough to stay in orbit, and not so fast that it is flung out of the system. The disc would merge with the stellar surface, with the star鈥檚 equator spinning at exactly the same rate as gas just above its surface.
Magnetic braking
But early measurements of the spin rates of protostars showed no sign of stars spinning at anything like this extreme rate. In 1991, Arieh K枚nigl of the University of Chicago suggested that this might be because the strong magnetic fields generated in the interior of many stars might be exerting a 鈥渂raking鈥 effect. These fields are themselves ultimately generated by the spinning of the star, via Coriolis forces acting on gas flows deep inside it 鈥 the faster the spin, the stronger the magnetic field generated by the Coriolis forces.
A proportion of the matter in the disc is ionised, and K枚nigl reasoned that this could interact with the magnetic field to slow down the spin of the protostar. Although particles in the outer part of the disc have a higher angular momentum than their counterparts closer in, their angular velocity 鈥 the number of orbits they complete in a given time 鈥 is lower, just as our Moon takes longer to orbit the Earth than a low-orbit satellite. As the protostar spins, its magnetic field sweeps round with it. And because the protostar is spinning faster than the outer part of the disc, dragging its magnetic field through the ions of the outer disc puts a brake on the spin of the star.
However, this is not the only effect. Close to the star, K枚nigl reasoned, the ionised gas in the disc that was being drawn by the star鈥檚 gravitational field onto its surface would not be able to flow freely across the magnetic field lines. Instead, the particles would be forced to flow to the surface along the magnetic field lines, just as charged particles from the Sun strike the Earth鈥檚 atmosphere only in the auroral arcs around the magnetic poles. Particles sliding down the field lines would have little chance to lose their excess angular momentum en route, and so would carry it into the star, making it spin faster. However, the braking effect leaves the star spinning more slowly than the innermost part of the disc. So ions in this part of the disc would be overtaking the star鈥檚 magnetic field, making the star spin faster.
K枚nigl has worked out that the balance between the different effects should depend on the strength of the magnetic field, which in turn depends on the star鈥檚 spin rate. Taken together, he predicts, these effects act as a built-in regulator that stabilises the star鈥檚 speed of rotation. If the star is rotating fast, the field will be strong and the drag from the outer disc should brake the star鈥檚 rotation. As its spin slows, the magnetic field becomes weaker (because the Coriolis forces inside the star are weakened), so the drag from outer regions decreases. In time, K枚nigl calculates, the star should reach an equilibrium spin speed of about a tenth the extreme rate predicted by the simpler accretion theory.
K枚nigl鈥檚 theory has been put to the test by a team led by Jerome Bouvier of the University of Grenoble. Between 1990 and 1993 the team-studied several dozen stars of a type called T Tauri. These stars have masses similar to the Sun, and are still in the throes of formation. Many are still accreting material from a protostellar disc. They also have powerful magnetic fields 鈥 a fact that is revealed by dark spots on their surface similar to sunspots on the Sun. These spots appear at points where intense parts of the magnetic field erupt through the surface, inhibiting the normal convection currents. This blocks the flow of heat from the interior, and so makes this part of the surface cooler than its surroundings.
The starspots were used by Bouvier and his colleagues to measure the spin rate, because as they move into and out of view they change the brightness of the star by up to 10 per cent. From these regular changes in brightness, the researchers found that the time for one rotation was, on average, ten times slower than the expected orbit rate 鈥 just as K枚nigl predicted in his disc-braking theory. Moreover, in 1993 Suzan Edwards and her team at the University of Massachusetts used infrared observations to show that, although some T Tauri stars were spinning faster than this rate, all of these had little or no disc around them, lending further weight to the braking theory (see Diagram).
Middle age
Disc braking can only occur in very young stars. T Tauri stars are only found in young star-forming regions 鈥 up to a few million years old. This means that the disc accretion phase cannot last for longer than a few million years. After that, the disc runs out of material to supply to the star, or is blown away by the star鈥檚 radiation field, or starts to coagulate into lumps that will form planets 鈥 or perhaps a combination of all three. Whichever mechanism applies, it kills off the viscosity that drives the accretion, and breaks the mechanical link between the star and the disc. So as the protostar enters its final stage of contraction, it naturally starts to spin faster 鈥 skaters retracting their outstretched arms again 鈥 and settles down to a stable middle age. At an age of a few tens of millions of years, it should be burning hydrogen in its core and looking very much like the Sun does now.
Three years ago, Chris Campbell of the University of Newcastle upon Tyne, Hannah Quaintrell of the University of Sussex and I realised that if this picture was right, we should be able to use it to work backwards. By applying it to stars about the age of the Pleiades 鈥 around 70 million years 鈥 we hoped to work out the sizes of the discs that they once possessed, and so get some insight into the early life of the stars, when planets may have formed. Our computer model took into account all the major effects: the accretion process, the changing magnetic field and the gravitational contraction, as well as K枚nigl鈥檚 braking effect.
To test the model, we started it running with 200 stars that had the same distribution of disc masses that is observed for T Tauris: up to a tenth of the mass of the Sun. When we simulated their evolution over 70 million years 鈥 from the T Tauri phase to the beginning of core hydrogen burning 鈥 we ended up with a distribution of spin rates that looked very much like the observed spin rates of real stars of similar mass and age in the Pleiades. This match between observation and our model gives welcome support to K枚nigl鈥檚 braking.
In our simulation, stars with massive discs ended up spinning fairly slowly, about once every seven days. Those with progressively less massive discs decoupled from them earlier as the accretion flow weakened. Then they began to spin faster as they contracted, and finished up as fast spinners with rotation periods ranging from a few hours to a week.
This link between the spin of the star and the mass of the disc that surrounded it in its distant past allows us to estimate how many stars should have planetary systems like the Sun鈥檚. The total mass of the planets in our own Solar System is about a tenth of the lowest disc mass seen for T Tauri stars, so some mass must have been lost by the Sun鈥檚 disc during its T Tauri phase. Even so, this implies that the Sun started off with a relatively small disc, which means that when it reached the age of the Pleiades stars it would have been spinning relatively rapidly: once every few days. So the stars in the Pleiades that rotate more slowly than this 鈥 more than half of them 鈥 would probably have had much more massive discs, and built more massive planetary systems than the Sun鈥檚.
Of course, there are some essential pieces that will have to be put in place before we know for sure that this picture is correct. We don鈥檛 know how much mass was lost from the Sun鈥檚 original disc before and during the planet-building phase. And so far, the disc-braking model and its predictions have been based on very simple magnetic fields, shaped like the dipole field of a bar magnet. For more convincing evidence of magnetic braking, astronomers will have to work out exactly what is going on in the region where the disc cuts into the magnetosphere of a T Tauri star. The geometry of the stellar magnetic field is crucial, because it determines how close to the star the disc is disrupted.
But measuring the shape of a stellar magnetic field is not easy. Not even the largest optical telescopes can resolve the discs of stars beyond the Sun, so a star鈥檚 surface can only be mapped by indirect means. The technique we use relies on the Doppler effect: if a light source emitting a fixed frequency is moving towards you, the light will appear to have a slightly higher frequency than usual; if the source is receding, frequency is lower. On a rapidly rotating star, a point on the equator is approaching the Earth as it rotates into view, then recedes again a few hours later as it disappears around the other side of the star. Light from that point is first Doppler-shifted to a higher frequency then moves to a lower than normal frequency as the point recedes. Because light comes to us from all parts of the disc simultaneously, the overall effect is that spectral lines appear 鈥渟meared out鈥 over a range of wavelengths: the lines appear both broader and shallower than they would if the star did not rotate.
Now suppose there is a starspot on a star鈥檚 surface 鈥 less bright than the rest of the star, but giving out some light nonetheless. At any given moment, the light from the spot will show a Doppler shift that corresponds to the velocity of the spot towards or away from us. Because the spot is discrete, its spectral line is much sharper than that of the whole star, and stands out against the stellar background. So in the starlight we see the faint spectral 鈥渇ingerprint鈥 of a spot moving towards us then away. By taking a series of snapshots over the rotation period, the velocities of individual spots can be estimated from their maximum shifts. Those with low velocities must be near the poles, while the features with the highest shift must correspond to the fast-moving spots at the equator.
At present, groups led by Klaus Strassmeier in the University of Vienna, Jean Fran莽ois Donati at the University of Toulouse, Martin K枚rster at the Max Planck Institute for Extra-Terrestrial Physics in Garching near Munich and myself at St Andrews are using this technique to map starspots on the surfaces of discless T Tauri stars and rapidly rotating stars in slightly older populations. The field geometries of these stars ought to be similar to those of the T Tauri stars with discs, and can be studied without the additional complications caused by infalling and outflowing material from the disc.
My own group and the researchers at Munich are concentrating on a star known as AB Doradus, which is slightly less massive than the Sun and roughly the same age as the Pleiades stars. It spins on its axis relatively fast 鈥 once every 12.3 hours 鈥 so we can observe most of its surface in a single night. Its pattern of magnetic activity has turned out to be radically different from the Sun鈥檚. Instead of clustering in belts near the equator, the starspots occur at all latitudes, and are concentrated in a crown encircling the rotation pole.
Other spectral details are giving us clues about the shape of the magnetic field just above the surface of the star. As AB Doradus rotates, the spectrum shows a 鈥渇ingerprint鈥 of neutral hydrogen that exactly matches the periodic pattern you would expect to see if clouds of hydrogen were trapped just above the surface of the star, and rotating with it.
The clouds are being whirled around the star faster than the stable orbital speed, so something must be anchoring them to one point above the stellar surface or they would be flung away from the star. The most likely reason they stay in place is that they are held there by the closed magnetic loops lying far above the stellar surface. Thanks to new developments in instrumentation these 鈥渟lingshot prominences鈥 can now be mapped simultaneously with the underlying starspots. Like the spots, the slingshot prominences appear to be concentrated at high latitudes on the star. The maps show that the magnetic fields of these young stars form giant loops up to four or five times taller than the biggest loops seen on the Sun.
The next step will be to find out exactly where material falling onto the star is landing. As it slams into the surface, travelling at speeds of several hundred kilometres per second, such material can increase the local temperature by several thousand degrees. It should be possible to map these hotspots by looking at the spectra, in the same way as for the dark spots. If the material is found to be streaming onto young stars along the vast magnetic loops, that will be powerful evidence that magnetic fields are disrupting their discs, and that disc braking is indeed occurring. We will then be one step nearer to knowing how many planetary systems share our Universe.