The chemical composition of a massive galaxy in the early Universe reveals an extremely short period of star formation. This result could challenge our ideas about the evolution of galaxies and of the Universe itself. See Letter p.248
Stars are fossils that retain the history of their host galaxies. At the end of their lives, they explode as supernovae, producing heavy elements that are distributed into the surrounding interstellar gas. New stars that are created from this gas contain the elements that were produced from the previous generations of stars. By analysing the abundance patterns of the elements, it is therefore possible to determine how many and what kind of supernovae exploded in the past. On page 248, Kriek et al.1 estimate elemental abundances of a massive galaxy, as seen 11 billion years ago, which suggest that this galaxy formed by a short and intense burst of star formation and then suddenly 'died' without producing more stars. The authors' findings could reshape our theories of galaxy formation2.
Explaining the origin of the elements is one of the scientific triumphs at the interface of nuclear physics and astrophysics. As the astronomer Fred Hoyle predicted3, carbon and heavier elements were not produced during the Big Bang, but are instead created inside stars. The α-elements — oxygen, magnesium, silicon, sulfur and calcium — are produced by massive stars (those approximately ten times more massive than the Sun) before being ejected in supernovae. Conversely, iron (Fe) is mainly produced by a different type of stellar explosion called a type Ia supernova. Theoretical models can successfully reproduce the observed abundances of these elements in nearby stars of the Milky Way4. Elements heavier than zinc can form by neutron-capture processes in stars, or by other exotic astronomical events5.
This knowledge of nuclear astrophysics has been used to determine the formation and evolutionary histories of nearby (low redshift) galaxies, using an approach called galactic archaeology. Kriek and colleagues show that this approach can also be applied to distant (high redshift) galaxies. The authors study a massive galaxy that has a relatively large redshift of 2.1 and find a surprisingly high ratio of α-elements to Fe. This result suggests that the galaxy had an intense period of star formation that lasted for only between 0.1 billion and 0.5 billion years (approximately 10% of the age of the Universe at that time) before type Ia supernovae began to occur.
The nature of the physical mechanism that causes such quenching of star formation has been one of the big questions in astronomy for more than 20 years. One suggested mechanism is feedback from the supermassive black holes at the centres of massive galaxies (Fig. 1) — observed as active galactic nuclei (AGN). AGN-driven winds would have removed the surrounding interstellar gas, preventing new stars from forming. However, the high α/Fe ratio in massive galaxies has not yet been fully explained by numerical simulations of galaxies that include this feedback mechanism6.
Kriek and collaborators' results make the situation even more complicated. Their galaxy is observed as it was when the Universe was only 3 billion years old. At this time, black holes might not have been large enough to produce sufficient feedback to prevent star formation, because black holes are expected to co-evolve with galaxies. This co-evolution is suggested by the tight correlation between the masses of galaxies and their black holes. It is also consistent with the space density of quasars — optically bright AGN — and cosmic star-formation rates, both of which decrease sharply7 beyond redshifts of 2, where the authors' galaxy is found.
Thanks to quantum mechanics and atomic spectroscopy, elemental abundances can be estimated from absorption lines in stellar spectra. However, it is not straightforward to estimate these abundances even in the spectrum of a single star because of complex fluid mechanics (hydrodynamic motions) and the lack of local thermodynamic equilibrium in stellar atmospheres. Such estimates are even more difficult for the spectra of galaxies, which comprise many stars that have different masses, ages and chemical compositions. In galaxies, spectral absorption lines are broadened because of the motion of stars and weakened by emissions of young stars or AGN. A further problem for spectroscopy is that in galaxies younger than 1 billion years, about 40% of the light comes from asymptotic giant branch (AGB) stars8, which are not well understood.
Despite these difficulties, Kriek and colleagues use spectroscopy to find a high α/Fe ratio in a galaxy that has a relatively high metallicity — the ratio of Fe to hydrogen is half that of the Sun's. The combination of these two abundance ratios has never been seen in nearby stars: stars in the Galactic halo have high α/Fe ratios at less than one-tenth of the Sun's metallicity, those in the Galactic disk have lower α/Fe ratios at higher metallicities, and those in the Galactic bulge have high α/Fe ratios but only at one-third of the Sun's metallicity9.
What studies should be carried out to understand Kriek and collaborators' galaxy spectra? First, high-resolution spectra of metal-rich and α-enhanced stars — as in the authors' galaxy — need to be obtained for the Milky Way and should be compared with the authors' galaxy spectra. Such stars could be discovered using ongoing galactic-archaeology surveys such as the Apache Point Observatory Galactic Evolution Experiment (APOGEE) in New Mexico. Second, it is important to understand the evolution and spectra of AGB stars, both observationally and theoretically. If these stars are members of binary systems, their evolution and spectra would be very different. Finally, because the time at which type Ia supernovae start to occur is key to interpreting the authors' observations, it is crucial to study the progenitor systems, which are also binary systems, that cause these supernovae. This is closely related to the evolution of the Universe itself — type Ia supernovae were used in the 2011 Nobel-prizewinning discovery that the Universe's expansion is accelerating10,11.
If the α/Fe ratio in the authors' high-redshift galaxy is confirmed, this would challenge our current models of galaxy formation and raise many questions. For example, what kind of physical process can stop star formation on such a short timescale? Star formation in this galaxy might have been boosted and then quenched so that α/Fe is higher than usual. Alternatively, material produced in type Ia supernovae might have been efficiently removed by galactic winds. If this ejection was caused by AGN, what is the origin of the corresponding supermassive black holes?
Finally, how did the authors' galaxy evolve after star formation stopped? Is there a present-day counterpart of this galaxy, and how common are such galaxies? Even with the largest telescopes and the best detectors that astronomers have at present, it is difficult to measure the elemental abundances of a large sample of galaxies at high redshifts. However, such measurements will be possible in the near future with the James Webb Space Telescope and with extremely large (25–40 metres in diameter) ground-based telescopes. Meanwhile, theorists will need to predict elemental abundances across cosmic time in the quest to understand the formation and evolution of galaxies.