News & Views | Published:


How to make a massive star

Two competing theories have been applied to the formation of high-mass stars. Observations of two stellar systems now suggest that the accretion model has a weightier claim than its rival merger model.

How do high-mass stars form? Through accretion — the gravitational collapse of a dense cloud of gas and dust1 — as do better-understood stars of lower mass like the Sun? Or do they form through the collision or merging of smaller stars? Two papers2,3 in this issue find evidence for dense disks of dust and gas around high-mass stars, adding to the mounting evidence for the first of the two options.

The origin of stars of high mass (ten or more times that of the Sun) is one of the most interesting unsolved questions in astrophysics. It has long been considered difficult, if not impossible, for these stars to form through accretion like their smaller cousins, because the intense radiation pressure from their central source halts the inflow of material. This, and the fact that young high-mass stars are often observed in very dense clusters of stars, led to the idea that they could have formed through the collision or merging of mid-sized stars4.

Both the accretion and merger models make predictions that can be tested observationally. In the accretion model, the effects of angular momentum in a star's progenitor cloud cause the formation of a disk and highly aligned, or ‘collimated’, supersonic outflows that clear out matter to form tunnels of lower density (Fig. 1a). According to the collision model, mergers are more likely to disrupt disks than to create them, and outflows are not likely to be collimated5 (Fig. 1b). The presence or absence of these features can thus be used to distinguish between the two models.

Figure 1: Rival formations.

a, A dense, rotating central disk of dust and gas, and a highly collimated supersonic outflow, are characteristic of the accretion model of high-mass star formation. An infalling envelope of gas also rotates, forming a torus around the disk. Outflows carve cavities (tunnels) of less dense material in this envelope. b, In the alternative merger model, many stars (including binary systems) are contained in the middle cloud centre, some of which collide and merge to form high-mass stars. Individual outflows and stellar winds from massive stars combine to give a large, uncollimated flow out from the system. Scales of a and b are about 1.5 light years across and 15 light years across, respectively.

Patel et al. (page 109 of this issue)2 and Jiang et al. (page 112)3 take very different approaches to find disks around two different massive stellar objects. In both cases, the disk in question is concealed inside a larger, less dense envelope of matter that obscures it from view at many typically observed wavelengths in the visible and near- to mid-infrared regions. The opacity of this dusty envelope decreases with increasing wavelength; so an obvious way round the problem is to observe at longer wavelengths.

This is the approach taken by Patel et al.2. They observed an object called Cepheus A — with a mass about 15 times that of the Sun and a luminosity some 12,000 times greater — using the recently commissioned Submilli-meter Array (SMA)6 telescope on Hawaii. The SMA allows unprecedented resolution at wavelengths sensitive to the dust and molecular emission from dense regions such as the disk around a star. Patel and colleagues observed in the far infrared at a wavelength of 917 µm, and found a disk with a radius of 300 AU (1 AU is the distance from the Sun to Earth), and a mass somewhere between one and eight times that of the Sun. The authors scanned across the surface of the disk, measuring changes in a characteristic wavelength of radiation emitted by the gas methyl cyanide, and concluded that the disk is rotating. The basis for this conclusion is a cosmic consequence of the well-known Doppler principle: things moving away emit sound or light waves at lower frequencies, and vice versa.

Observations of Cepheus A by the Very Large Array (VLA) telescope in New Mexico2 at a longer radio wavelength of 3.6 cm supply further compelling evidence for accretion. They show a highly collimated outflow in a direction perpendicular to the disk's plane, energized by hot, free electrons. Such mechanisms are ubiquitous in low-mass star formation, where to conserve angular momentum and allow most of the infalling gas and dust to accrete onto the star, some material is ejected out towards the polar regions.

The complementary observations of Jiang et al.3, made at a wavelength of 2 µm, concern the Becklin–Neugebauer (BN) object. At the centre of this object, familiar as one of the first big finds of infrared astronomy and as the brightest object in the sky (apart from the Sun) when observed at near-infrared wavelengths, is a star with seven times the mass of the Sun and 2,500 times its luminosity. The authors used a technique known as polarimetry to measure the scattering and polarization of light emitted by the central star. Because dust absorbs a large fraction of incident radiation at each point of scattering, denser matter in a region will absorb more light, making a polarized image appear darker. The image of the BN object made by Jiang and colleagues shows a dark lane suggestive of a disk, accompanied by a brightening in two lobes, indicative of reduced scattering in less dense bipolar cavities.

Detailed models of radiative transfer can extract the geometry of the material surrounding the star from such pictures. These models include the effects not just of the scattering properties of dust, but also of the alignment of dust grains in magnetic fields, which impart polarization to radiation merely passing through and not scattering. Radiative transfer models of the BN source find a disk of radius 800 AU embedded inside an envelope roughly twice that size.

Because massive stars are typically more distant than low-mass stars, most previous observations of disk-like structures around massive stars did not have the resolution to distinguish objects of small angular extent. These observations most probably included the larger-scale ‘envelope tori’ surrounding the disk, or were taken at radio wavelengths that are sensitive to emission from outflows as well. They were thus not as convincing evidence for disks, and failed to clinch the argument for the accretion model.

Do the observations of Patel et al.2 and Jiang et al.3 prove that the accretion model is correct? Perhaps — for the objects observed, at least. The observations provide a consistent picture of accretion that includes not just disks, but also rotating, infalling envelopes and outflows. These are just the ingredients that would, according to recent theories of accretion, allow the formation of massive stars: material that accretes into the cooler equatorial regions can be shielded from the extreme radiation of the star that normally halts the flow, whereas intense stellar radiation is channelled into the outflow cavities7,8.

It may also be that both models of star formation operate, but in different circumstances: accretion in more isolated systems; and accretion, with merging, in dense clusters. Recent dynamical simulations indicate that the formation of close binary stars is a likely outcome of accretion in a cluster9. If the binaries are as close as 1 AU in separation, it is likely that some will merge.


  1. 1

    Shu, F. H., Adams, F. C. & Lizano, S. Annu. Rev. Astron. Astrophys. 25, 23–81 (1987).

  2. 2

    Patel, N. A. et al. Nature 437, 109–111 (2005).

  3. 3

    Jiang, Z. et al. Nature 437, 112–115 (2005).

  4. 4

    Bonnell, I. A. & Bate, M. R. Mon. Not. R. Astron. Soc. 336, 659–669 (2002).

  5. 5

    Bally, J. & Zinnecker, H. Astron. J. 129, 2281–2293 (2005).

  6. 6

    Ho, P. T. P., Moran, J. M. & Lo, K. Y. Astrophys. J. 616, L1–L6 (2004).

  7. 7

    Yorke, H. W. & Sonnhalter, C. Astrophys. J. 569, 846–862 (2002).

  8. 8

    Krumholz, M. R., McKee, C. F. & Klein, R. I. Astrophys. J. 618, L33–L36 (2005).

  9. 9

    Bonnell, I. A. & Bate, M. R. Mon. Not. R. Astron. Soc. (in the press); preprint available at

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.