Observing the formation of stars and planets W. R. F. Dent The formation of stars and planets is a problem which has intrigued astronomers for two hundred years. Laplace and Kant are acknowledged as the first to discuss the idea of a flattened nebula from which our Solar System formed. However, it is only in the past 20 years that it has been possible actually to observe the many phenomena which occur during this violent process. These studies have recently taken a significant step forward with the advent of a new generation of telescopes which operate in the submillimetre band, at wavelengths between infrared and microwaves. One of the largest of these is the 15metre diameter James Clark Maxwell Telescope, situated 4200 men-es above the Pacific Ocean on Mauna Kea, Hawaii. It is clear that the lOI stars in our Galaxy must have formed from the clouds of gas and dust found in the spiral arms. But star formation takes 106-10’ years, so it is impossible to follow the complete process as it happens. What we can observe, however, are regions in the Galaxy which contain stars caught at different stages in the process. From these snapshots, it is possible to generate a story of how a cloud might collapse to form proto-stars at dense spots within its envelope. Following the theoretical evolution of these proto-stars, we can trace examples of the gas clearing phase, where the vigorous young star is first revealed, as well as the stage where planets may be condensing. One of the main problems until recently was that, as one might expect, most of these processes take place deep within the cores of the collapsed parent clouds. The visual extinction may be 1000 magnitudes or more (that is, a reduction by a factor of 10Z5”in the optical flux). This extinction is due to a small fraction of dust grains (about the size of those in tobacco smoke) mixed in with the gas in the clouds. However, the absorption through such material is proportional to h-*, so moving to longer wavelengths - such as the i&a-red and millimetre waves - allows us to observe even the most deeply obscured stages in the evolutionary process.
W. R. F. Dent, MSc.,
After gaining an M.Sc. degree at Jodrell Bank, Manchester, he did postgraduate research on star formation at the University of Kent. Since 1988 he has worked at the Royal Observatory, Edinburgh, and the JCMT in Hawaii. He is currently an astronomer and support scientist based at the Joint Astronomy Centre.
Endeavwr. New Smia Volume 16, No. 3.1992. 0160-9327/92 6mo + 0.00. Pergamon Press Ltd. Printed in Great Britain.
Between the near infra-red at wavelengths of a few microns and microwaves at around a few centimetres, the atmospheric transmission is highly wavelength-dependent. In the far infra-red, between 30 and 300 microns, satellites or high-flying aircraft are required to get above most of the atmosphere. However, between 0.3mm and 1.5mm, on the Earth’s best ground-based observing sites, the atmosphere becomes relatively transparent (figure 1). Mauna Kea, a 4200 metre volcano on the island of Hawaii, is one of the few such sites which is accessible (Antarctica being another). In 1987 the United Kingdom, Canada, and Holland opened the James Clerk Maxwell Telescope (JCMT), a 15 metre diameter dish which was specifically designed to observe in these atmospheric windows (figure 2). This is currently the largest telescope of its kind. Millimetre and sub-millimetre observing techniques rely on a combination of both radio and optical methods. The telescope itself is built as a parabolic dish. However, the accuracy of the parabola has to be very high - generally better than one-tenth of the shortest observing wavelength. On the JCMT, this accuracy is maintained by specifically designing the mount and support structure to flex in such a way that the surface is always a parabola, but with a focal length that changes in a well-defined way. In addition, the surface is constructed of multiple panels, each of which can be set under computer control. The current surface accuracy is 30-35 microns rrns, although it is hoped to improve on this in the near future. The detectors mounted on the telescope are mainly of two types: (1) heterodyne mixers, which are used mainly for observing transition lines from rotating gas molecules; and (2) bolometers, which are used to detect broad-band
continuum emission from warm dust, Brehmsstrahlung radiation from ionized gas, and Gyrosynchrotron radiation. The heterodyne receivers in use employ low-noise cooled Schottky diode and SIS (Superconductor Insulator Superconductor) mixers; the development of these is taking place in collaborative efforts between several laboratories in the partner countries. The current bolometer is cooled to less than 1 K to improve sensitivity. Both these types of detectors are used for observing a wide rang of phenomena, but the operating wavelengths make them particularly suitable for observing interstellar gas and dust at temperatures of lo-100 K. As will be described later, this is just the temperature range seen in clouds where star formation is occurring. The process of star formation itself can be broken down into three main evolutionary stages: the initial collapse of the dense cloud to form a protostar; the initial ignition of nuclear reactions at the core; and the ‘unveiling’ of the star from the parent cloud. On this basis the following sections are believed to be in chronological order. However, neither theory nor observations have yet reached a stage where the whole process is exactly understood, and in some cases, the description given is of our current best guess Initial cloud collapse The spiral arms of our Galaxy contain many giant clouds, composed mainly of molecular gas and grains of dust. Although the density in the interiors of these clouds may be lo3 times greater than the surrounding interstellar space, they are prevented from gravitational collapse simply by the internal gas thermal and turbulent pressure. However, circumstances can conspire to cause the densest spots to start collapsing further
f’ ’ \
I I I I I
i\ ’I?\ L! \ I 4c 0
Figure 1 Atmospheric transmission between a wavelength of 300 pm (1000 GHz) and 400 pm (70 GHz) from Mauna Kea under good conditions. By comparison, at sea level the transmission is more typically given by the lower dashed line.
(figure 3a). Impact shocks from expanding supernovae, from nearby star formation (see below), or from Galactic spiral density waves may all do this. These giant molecular clouds are well known, and have been observed in rotational transitions of CO and many other molecules for several years.* Small denser clumps within the envelope have also been detected - particularly in molecules such as CS, HCO’, and NH, (which are only excited at higher densities). Despite strenuous efforts, however, it has proved difficult conclusively to observe one of these cores actually in the process of collapse. This is partly because such a core would inevitably have extremely high intervening extinction, and partly because the cores are cool due to isolation from the warming effect of photons from surrounding stars. Furthermore, theoretical models suggest that this collapse stage occurs very quickly compared to later stages of stellar birth, so a region containing a large number of more evolved young stars would, on average, have very few of these pre-stellar cores actually at the point of contraction. Programmes are underway at the JCMT and at other telescopes to search for such collapse.
compressed by several orders of magnitude) now plays a major role in shaping the evolution of the clump. In addition, the angular momentum of the original large cloud will be conserved, resulting in rotation velocities of 50-500 km s-’ in the collapsed core. A combination of
these effects is believed to result in a further release of energy through the twisting and compression of the magnetic field lines trapped in the cloud (see figure 3b and c). The JCMT has proved to be an important tool in the search for objects at this stage of evolution. Numerous surveys of the sub-millimetre continuum radiation from warm dust in regions of star formation have been carried out. Frequently, compact clumps of dust are found which are not prominent, or sometimes are completely undetectable at other wavelengths. Figure 4 shows one of the most extreme of these, known as IRAS4, near to the well-studied nebular NGC1333. The two bright objects in this 1100 pm continuum image are both unusually massive dense cores which we are observing through emission from the warm (2WO K) dust. The JCMT data indicate the optical extinction towards IRAS4A is - 10Q105 magnitudes, making it one of the most deeply embedded cores known; this explains why it has not been seen at shorter wavelengths. Theoretical models of protostars predict that such massive cores would have relatively high luminosities due to the release of energy as described previously. But IRAS4A and
Formation of a protostar As the mass of the core increases, so does the gravitational potential, and the velocity of the accreting gas and dust. Violent shock fronts will heat the incoming material, and the core temperature will start to increase through gravitational accretion. It is believed that at the same time, the interstellar magnetic field (which, through the collapse, has been
*Molecular hydrogen, having no dipole moment, is not detectable under these conditions, although it is the most common molecule in the clouds. CO is the second most common molecule.
Figure 2 The James Clerk Maxwell Telescope. The entrance aperture is normally kept fully covered by a large protective screen of woven PTFE. This is used to protect against wind shake, uneven solar heating of the surface, and dust contamination.
--,:..:: . .
:...:. . ...,.. . . . ... . . ’
Figure 3 Cartoon depicting the main stages in the formation of a single star (not to scale). (a) Initial collapse of a dense region in a larger cloud. (b) Protostellar accretion; the inner rotating core is being heated through accretion, has collapsed to a flattened disc (shown edge-on), and may be responsible for bipolar outflowing jets. (c) Formation of the young star at the core may trigger a new outflow; much of the surrounding material is now in the form of a disc which has flattened at the centre. (d) Most of the parent cloud has been dispersed, leaving a main sequence star, plus a flat disc with possibly planetary formation.
Figure 4 False colour image of part of the low mass star formation region NGC1333 IRAS4 in dust continuum emission (from Sandell et a/. 151). The brightest object is IRAS4A. The axes are in arcseconds, where 60 arcsec is 0.1 parsec (or 3 x 1O’5 metres).
17 parsec on a side.
B both have relatively modest luminosities (only about 20 times that of the Sun). This perhaps suggests that they are at a very early stage of their evolution, before accretion or nuclear burning have become significant. IRAS4A and B may be two of the current best candidates for young protostars. Figure 5 shows a more distant region of the star formation known as ON2. The area has been exstensively studied by the VLA (Very Large Array, Mexico) and other telescopes and arrays. All the individual peaks on this false colour map are due to separate sources of energy, with luminosities of lo”-lo4 times that of the Sun. The upper objects are known young stars, the uppermost being the most evolved. The lower double source has not been detected at other wave-
lengths, and is likely to be the most recently evolved part of this young highluminosity cluster. Thus this is a possible example where star formation in the upper (northern) region has triggered cloud core collapse and a series of further star formation events to the south. One of the current ‘holy grails’ of stellar formation is to show conclusive evidence of a protostellar accretion disc. In a few objects the JCMT has detected a flattened dust distribution (figure 6), and in some of the more massive and luminous cases the object is rotating about the minor axis. These may, therefore, be the best examples of ‘solar nebulae’, as originally proposed by Laplace [ 11. However, only the most extended, massive, and hence fast-rotating discs are
large enough to be resolved by current telescopes. There is, as yet, no conclusive direct evidence for discs around progenitors of stars like that of the Sun. Although continuum emission shows compact luminous cloud cores, one of the first dynamical pieces of evidence that star formation is taking place in a core is usually not infall, as might be expected, but gas actually flowing out at high velocities. The JCMT has been used to study several of these outflow objects. Figure 7 shows a typical example. In this, and many other cases, there are two opposing jets, giving rise to a so-called bipolar outflow. In this source, the jets are inclined at about 45” to the line of sight, giving rise to the receding and approaching lobes. The origin of
flows are very common phenomena; at least 70 per cent of cloud cores are associated with them . It is likely that they are a ubiquitous part of almost all stages of star formation.
Figure 6 Image of the source G35.2N at a wavelength of 1100 urn. The elongated structure is believed to be due to a disc viewed edge-on of diameter 0.2 parsec.
these flows is currently not well understood. Theoretical models suggest either magnetohydrodynamic acceleration of gas off the surface of the accretion disc (see figure 3c), or gas driven by a wind from the surface of the young star. However, in both cases, this would mainly occur on a size scale too small to be resolved by current systems. One presently unanswered question is
whether these outflows start before or after the protostellar phase has ended; in other words, must a young star be already formed at the centre before a wind can start? The possible detection of an outflow near IRAS4 suggests that such flows can occur at the very earliest phase of evolution, before the star itself has even formed. Surveys by JCMT and other telescopes have shown that out-
Figure 7 The high velocity CO gas around the young star GGD12-15. Gas receding at 5-10 km s-’ is shown in red, and gas moving towards us around velocity range is shown in blue emission.
The first stages of nuclear burning The currently accepted definition of a protostar is one in which the release of energy from nuclear burning plays a minor role compared with gravitation accretion. However, at some point, the central core of the collapsing clump will reach a density and temperature sufficient for deuterium and eventually hydrogen fusion to commence. The exact stage at which this occurs is unclear. It may even start up in a series of ‘kangeroo hops’, where the initial burst of energy temporarily blows material back out and dampens the process. The presence of shocks, magnetic fields, and rotation are all likely to make this a highly non-linear startup. As far as the observer is concerned, the only definite proof that nuclear burning is occurring is when a relatively normal optical star is observed. It is now clear that many cases of what were thought to be ‘protostars’ a few years ago are actually embedded stars which have already started nuclear burning. But by observing at sub-millimetre wavelengths, we can now observe the earliest phases; that is, the ‘true’ protostars. The clearing phase Stars on the main sequence part of their life cycle, such as the Sun, are not generally surrounded by massive clouds of dust and gas. The total mass of material around the Sun, mostly in the form of planets, is = O.OOl& (where ~MQ is the mass of the Sun). Somehow, therefore, the star must remove most of the parent cloud during part of its early evolution. The JCMT has been used to survey a number of solar mass stars which are optically visible, but which show signs of youth (such as high velocity outflow). It was found that most of these stars had surrounding clouds of mass O.Ol-O.l& - considerably more than that of the solar system, but much less than seen around objects such as IRAS4 (see above). The angular resolution available at the shortest observing wavelengths on the JCMT means it is possible to resolve the closest of these clouds, and the results indicate assymetrical dust distributions, with major axes of 1000-2000 AU (1 AU is an Astronomical Unit, or the distance between Sun and Earth). By comparison, the size of the solar system is = 70 AU. These stars are presumably still in the process of clearing away their parent cloud . JCMT data have also shown a strong correlation between the presence of outflows and dense cores. It is, therefore, very plausible that these high velocity outflows, seen at almost all stages of stellar birth, are responsible for
the clearing of the cores (figure 4d). As a star evolves, the surrounding cloud is eaten away, until by the time it becomes a ‘normal’, or main sequence star, much of the material is gone. However, in a few cases of apparently normal stars (such as Vega and Formalhaut), the IRAS satellite detected a small amount of surrounding dust. Results from JCMT observations indicate that these clouds have a size of a few hundred AU . It is likely that this is dust in a large Oort Cloud, similar to that around our own Solar System. It is also very likely that a small fraction of the original dust and gas will have collapsed further into sub-condensations in orbit around the central star. These will eventually form planets. However, much higher resolution and sensitivity will be required before we can detect such objects.
Conclusion The current state of observations of star formation is determined to a large extent by advances in instrumentation. For this reason, the JCMT is soon to begin work on a collaboration with the nearby Caltech Observatory. The two telescopes will be linked as an interferometer, thus increasing the spatial resolving power by several times. In the future, it is hoped that the JCMT will form a major part of a larger submillimetre interferometer array on Mauna Kea. The resulting data with an order of magnitude increase in resolution will allow us to observe on scales where both accretion and planetary formation are occurring. Acknowledgements The James Clerk Maxwell Telescope was built in collaboration between the UK SERC, and the Dutch ZWO (see C.
M. Humphries, Endeavour, New Series, 4, 132, 1980). Project scientist for construction was Prof. R. Hills, of MRAO, Cambridge. The JCMT is now operated by the Royal Observatory, Edinburgh, on behalf of the SERC, ZWO, and the Canadian HIA, which joined the collaboration in 1988.
References [l] Laplace, ‘Exposition du System du Monde’, 1797.  Parker et al. Monthly Notices Royal Astronomical Society, 234, 67 pp., 1988.  Weintraub et al. Astrophys. J. Lett., 340, L69, 1989.  Becklin et al. In: ‘Submillimetre Astronomy’, p. 147, ed. Watt & Webster, Kluwer, 1989.  Sandell et al. Astrophys. J., 376, L17, 1991.