The first well-resolved images of local-galaxy stellar nurseries that are poor in elements heavier than helium give the best picture yet of the conditions in which stars may have formed in the early Universe. See Letter p.218
Astronomers dub elements heavier than hydrogen and helium 'metals', and these make up only trace amounts of all matter by mass. For example, about 2% of the interstellar matter in the neighbourhood of the Sun comprises metals, the most abundant of which are oxygen and carbon. These elements have a role in catalysing the birth of stars that is far out of proportion to their low abundance. On page 218, Rubio et al.1 present the first well-resolved pictures of metal-deficient stellar nurseries found in a nearby dwarf galaxy, by recording the spectral lines emitted by carbon monoxide (CO). The results open up CO spectroscopic imaging as a diagnostic for exploring the relationship between metal content and star formation for substantially metal-deficient systems.
Stars form out of cold, dense clouds of molecular hydrogen (H2). In these clouds, metals act as coolants, helping the gas to reach low temperatures and facilitating its collapse into pre-stellar condensations. Metals also form interstellar dust, which shields stellar nurseries from starlight that would otherwise break molecules apart and heat the gas.
These metals are produced in stellar interiors. When stars die and explode, some of the newly produced metals are mixed into the interstellar gas. Thus, successive generations of stellar birth and death lead to a gradual enrichment of heavy elements in the interstellar medium. These, in turn, aid the subsequent formation of new stars. Following this logic backwards, early generations of stars probably formed in stellar nurseries that contained few metals compared with the Milky Way or similar present-day galaxies. Thus, to understand the build-up of the first stars and galaxies, astronomers must measure how a dearth of metals (low metallicity) affects the star-formation process.
To study star formation in metal-poor gas, astronomers study the least-massive galaxies in the present-day Universe. These dwarf galaxies are not believed to be truly young, and so they are imperfect analogues of distant primordial systems. But because of a combination of their inefficient star-formation activity and weak gravity (exploding stars can blow heavy elements entirely out of a small galaxy), they are deficient in heavy elements. Therefore, researchers use them as local 'laboratories' to investigate how a lack of metals affects the formation of stars in interstellar gas clouds.
Direct observation of the H2 that makes up most of the cold, dense gas in these clouds is difficult. This forces astronomers to observe molecular tracers that are mixed with the H2, and whose spectral signatures are used to infer the abundance of H2 indirectly. The workhorse tracer is CO, the second most abundant interstellar molecule2. CO survives in the interstellar medium mainly in regions where there is enough dust to shield it from starlight, and it is studied through its millimetre-wavelength emission, which is detectable by radio telescopes. In a galaxy such as the Milky Way, dust is plentiful and CO is fairly well mixed with H2. However, the dust (and the CO itself) is made of heavy elements. Emission from CO therefore tends to be faint in galaxies of low metal content.
There is a long history of hunting for CO in dwarf galaxies, with the goal of understanding metal-poor stellar nurseries. For decades, the Small Magellanic Cloud (a dwarf satellite galaxy of the Milky Way) and a few similar galaxies remained the most metal-poor systems in which stellar nurseries had been detected by their CO emissions. A barrier of about one-fifth of the Milky Way's metallicity emerged3 as a practical limit to the detection of CO, and direct knowledge of stellar nurseries in galaxies below this limiting value was largely lacking.
Two years ago, researchers from the same group as Rubio et al. used the Atacama Pathfinder Experiment telescope in Chile to push past this 'metal barrier'4. They observed CO emission from the Wolf–Lundmark–Melotte (WLM) dwarf galaxy, which is part of the same Local Group of galaxies as the Milky Way (Fig. 1). WLM has a metal content5 only about one-tenth that of the Milky Way, and about half that of the previous record holder6 from which CO had been detected. The researchers showed that the CO emission from star-forming regions in WLM was faint compared with that from other tracers of gas and star-formation activity. This implied that CO molecules in WLM traced only the densest, most opaque parts of an extended stellar nursery.
Rubio et al. have now used the Atacama Large Millimeter/submillimeter Array (ALMA), the world's most powerful millimetre-wavelength telescope, to take well-resolved pictures of these regions in WLM. The authors' images of CO emission reveal that the nurseries are confined to stunningly small clumps that presumably represent only the densest parts of the star-forming gas clouds (see Fig. 1 of Rubio and co-workers' paper1). By contrast, CO emission pervades star-forming regions of the Milky Way, such as the Orion molecular cloud7,8. The images enabled the authors to directly measure the CO-emitting clouds' sizes (about 3 parsecs across). They also measured the clouds' kinetic energies, because radio telescopes can track the motion of CO gas by measuring the shift of the frequencies of its emission lines relative to their rest values (the Doppler effect). On the basis of these measurements, Rubio et al. argue that the physical properties (density, pressure and self-gravity) of the CO-containing clouds in the WLM galaxy do resemble those of similarly sized clumps present in metal-rich locales such as the Sun's neighbourhood — even though most of a given star-forming cloud in WLM is invisible in CO emission.
The authors argue that this similarity in physical properties helps to explain why star clusters born in metal-poor galaxies resemble those seen in less-extreme systems. In effect, they propose that the main impact of WLM's lack of metals is to render the bulk of the cloud difficult to see using CO. The lack of dust in WLM means that our best tracer of H2 is present only deep in the cloud, and the behaviour of most of the H2 is perhaps not so different from that in other 'normal' galaxies. This agrees, at least qualitatively, with simulations and theoretical predictions for the behaviour of CO and H2 in metal-poor galaxies9. The authors further speculate that the small size of these dust-enshrouded, CO-emitting clumps may explain the relative paucity of highly massive stellar clusters in small, isolated galaxies.
The current study highlights a changing approach to studying star formation in low-metallicity systems. Modern telescopes have begun measuring the energetics, densities and turbulent character of metal-deficient stellar nurseries. This is a substantial advance on simply hunting for faint CO emission from such systems. ALMA is now operating full-time, so we could see exciting progress in this field in the coming years.
However, the fundamental problem of knowing how much H2 gas is present in metal-poor systems remains daunting, especially given this striking demonstration that CO inhabits only small, dense pockets of gas in the interiors of extended stellar nurseries. High-resolution observations of other gas tracers, including ionized and neutral carbon, and of dust (studied through its infrared emission and the attenuation of starlight that it causes) will be needed to piece together the structure of metal-poor clouds in detail. Rubio et al. have investigated these other tracers, but only at poor resolution that is not well matched to the tiny CO-emitting clouds found in WLM.
Finally, it should be noted that only a handful of clouds have been measured in a single system, but the star-formation process can be violent and random on small spatial scales. It will be fascinating to see if these first results are indeed indicative of a broader population of clouds in other low-metallicity galaxies. Footnote 1
Rubio, M. et al. Nature 525, 218–221 (2015).
Bolatto, A. D., Wolfire, M. & Leroy, A. K. Annu. Rev. Astron. Astrophys. 51, 207–268 (2013).
Taylor, C. L., Kobulnicky, H. A. & Skillman, E. D. Astron. J. 116, 2746 (1998).
Elmegreen, B. G. et al. Nature 495, 487–489 (2013).
Lee, H., Skillman, E. D. & Venn, K. A. Astrophys. J. 620, 223–237 (2005).
Israel, F. P., de Graauw, Th., van de Stadt, H. & de Vries, C. P. Astrophys. J. 303, 186 (1986).
Wilson, B. A., Dame, T. M., Masheder, M. R. W. & Thaddeus, P. Astron. Astrophys. 430, 523–539 (2005).
Buckle, J. V. et al. Mon. Not. R. Astron. Soc. 401, 204–222 (2010).
Glover, S. C. O. & Clark, P. C. Mon. Not. R. Astron. Soc. 426, 377–388 (2012).
Hunter, D. A. et al. Astron. J. 144, 134 (2012).