The evolution of galaxies

June 10, 2013

When we look up to the sky on a dark, clear summer night (rare in Germany) we see the shimmering band of the Milky Way stretching from horizon to horizon. This is our home in the Universe, a galaxy consisting of approximately one hundred billion stars.

Computer simulation of the formation of a massive disk galaxy (center) from accreting gas streams along the intergalactic cosmic web (from lower left to upper right, dark blue is cold gas, red is hot gas). The forming young galaxy (light blue) is ejecting gas enriched with heavy atoms back out into circum-galactic space (yellowish & cyan colors).

Text: Reinhard Genzel

Like in most other so called ‘disk’ or ‘spiral’ galaxies, most of these stars, with a mass similar to the Sun or somewhat less, live in a fairly thin rotating disk of diameter 60,000 light years. Two or three more stars are formed from the gravitational collapse of dense interstellar gas clouds every year somewhere in that disk, for which reason our Milky Way is considered to be ‘actively star forming’. At the center of the disk is a spheroidal concentration of mostly older stars, called the bulge, within which resides a massive black hole of about four million times the mass of the Sun. This structure is typical of most other disk galaxies, even if masses, sizes and the relative proportions of bulge to disk range widely.

Disk galaxies make up most of the more massive star forming galaxies in the current Universe but there are numerous other, amorphous, irregular systems of typically smaller mass, as well as an entire second population of massive spheroidal or elliptical galaxies, with little current star formation. However, while galaxies are the most conspicuous ‘island Universes’ of light emitting stars, most normal, baryonic matter is actually located outside galaxies, and is in form of very hot, intergalactic plasma. Furthermore, baryons in turn make up only 20%of the matter content in the Universe; the rest is thought to be ‘dark matter’, which interacts only, or predominantly, via gravity. Dark matter is commonly postulated to be in form of a yet not identified, heavy sub-atomic particle.

Ever since the discovery of galaxies as the building blocks of the stellar extragalactic Universe about a century ago, the key question, for professional astronomers and the interested public alike, has been how and when they might have formed, how they have been evolving, and why there are starforming disks, as well as massive ‘dead’ spheroids. Although we have at this point only partial, incomplete answers to these questions, there has been a lot of progress over the past two decades. Researchers at several Max Planck Institutes (MPE, MPA and MPIA) have made major and in some cases transformational contributions to that progress, both in terms of experimental work and observation, as well as in theory, numerical simulations and modelling.

Basic concepts

From precision measurements of the cosmic microwave background and the large scale structure of galaxy distributions on the one hand, and very large computer simulations on the other, we can broadly understand the formation of galaxies in the ‘cold dark matter model’. Local dark matter concentrations (‘halos’) collapsed by self-gravity in the substrate of the expanding Universe and constituted the locations of baryonic galaxies forming at their cores. These local overdensities are thought to have been already imprinted as quantumfluctuations in the earliest phases after the Big Bang. Theoretical work over the last three decades, pioneered by Simon White and his colleagues at MPA, suggests that baryonic gas present in these dark matter halos was driven inwards in the collapsing halos, and formed concentrations on galactic and sub-galactic scales. Since these baryonic gas concentrations were endowed with an initial angular momentum, the smallest scale of initial collapse was about a tenth of the size of the dark matter halos where the proto-galaxies are expected to form centrifugally supported disks, approximately of the size of modern galaxies. All numerical simulations of large scale structure evolution, such as the large “Millenium Simulation”carried out by Simon White and Volker Springel at MPA, find that large scale structure builds up hierarchically from smaller structures to larger sizes, from smaller mass to larger mass. From the galaxy’s perspective, this means that the embryonic galaxy grew over time from gas streams fed from the cosmic‘web’, including, from time to time, an incoming smaller galaxy/halo in that stream (a so called ‘minor merger’). More rarely (once every 3 years or so for a massive galaxy), there would also occur a highly dissipative ‘major merger’, at the end of which two disk galaxies were permanently transformed to a larger spheroidal galaxy.

Taking to time travel

Astronomy, unlike physics, but like biology, cannot reconstruct time evolution by active experiments or studies of individuals. Astronomy relies on the determination of the evolution from distribution functions of populations, and on archeological evidence of past events still present today, such as the properties of very old stars, or the large scale structure of the Universe. Fortunately, and owing to the finite speed of light and the large size of our Universe, astronomers can also goon time travel backward in time, by using the faint signals of very distant galaxies with big telescopes to observe Milky-Way-mass galaxies when they were still young and in the process of formation. This article tells the story of what we have learned on these time travels.

Cosmological look-back studies of the properties of galaxy populations have revolutionized the empirical knowledge about galaxy evolution over the past two decades. Large surveys have been carried out with broad-band photometry across the electromagnetic spectrum, from X-rays, to the ultra-violet and optical, to the near- and far-infrared, and all the way to the radiobands, exploiting the largest space- and ground-based telescopes available world-wide. The Max Planck groups at MPIA (Hans-Walter Rix), MPA (the groups of Guinevere Kauffmann and Simon White) and at MPE (my group, and the groups of Ralf Bender, Gunther Massinger and Kirpal Nandra) have been actively engaged in these massive surveys, for instance, with the Sloan Digital Sky Survey (SDSS), with the Very Large Telescope(VLT) of the European Southern Observatory, with the Hubble Space Telescope (HST) and, most recently, with the Herschel far-infrared space telescope of the European Space Agency (ESA). It is now clear that the earliest smallish protogalaxies formed already 500 to 800 million years after the Big Bang, less than 5% of the current age of the Universe. As expected from the earlier theoretical work, massive galaxies, and in particular massive spheroidal systems appear in larger abundances later, a few years after the Big Bang, or about 10 years ago, when galactic star formation activity reached abroad maximum.

Throughout cosmic evolution, most (>95%) of the star forming galaxy population shows a fairly well established near-linear relation between their stellar mass and their star formation rate. The functional form of the relation remains approximately constant and only the ratio of star formation rate to stellar mass, at a given stellar mass, increases rapidly as we look further back in time. Instead of two to three stars per year as in the modern Milky Way, a Milky-Way mass galaxy 10 years ago formed stars at a twenty times faster rate. Accurate star formation rates unaffected by interstellar dust extinction were most recently provided by the MPE-developed PACS photometer onboard Herschel (PI Albrecht Poglitsch), allowing the first deep look-back surveys in dusty far-infrared bright galaxies led by Dieter Lutz and his colleagues. These findings, along with statistical studies of the spatial clustering and abundances, suggest that star forming galaxies grow in mass mainly from in-situ steady star formation over a period of several years, rather than from mergers and/or ‘starburst’ events. A comparison of stellar masses and dark halo masses of galaxy populations, feasible since a few years from a combination of the imaging surveys at different cosmic times and the computer simulations, shows that galaxy formation was an inefficient process throughout that entire cosmic time, converting less than 20% of the available cosmic baryons into stars in galaxies.

From gas to stars and back again

The data and theoretical modelling suggest that galaxy grow that the peak of the formation epoch happened in an equilibrium between baryonic gas accreting onto galaxies and promoting star formation in dense molecular gas clouds, the consumption of gas by star formation, and outflows of gas driven out of galaxies by massive stars through winds and supernova explosions. This concept has recently been tested and confirmed by direct studies of the molecular gas reservoirs in the young galaxies. Linda Tacconi and her collaborators at MPE have observed millimeter wavelength emission lines of the carbon monoxide (CO) molecule, as a proxy of the main constituent of dense, star forming molecular gas, molecular hydrogen, with the Plateau de Bure millimeter interferometer of the CNRS/MPG/IGN Institute IRAM, in the first large survey of cold gas in distant star forming galaxies. They find that their molecular gas contents were about 4-5 times greater than in comparable galaxies in the local Universe, as studied in the COLDGASS IRAM CO survey of Guinevere Kauffmann, Amelie Saintonge and their colleagues. Otherwise the physical processes of star formation in the early Universe seem pretty much the same as in the local Universe. This is exactly as expected in the ‘gas regulation model’. While young galaxies were initially provided with large amounts of fresh gas in the dense, early Universe, gas supplies and gas fractions dwindled as the Universe expanded and became more tenuous. In addition, UV and optical observations by several groups, including ours, demonstrate that most of the star forming galaxies 10 years ago exhibited powerful galactic winds, blasting gas enriched with heavy atoms formed during stellar nucleosynthesis back into the circum-galactic halo and beyond. These winds probably played a crucial role in keeping the efficiency of galaxy formation as low as observed, especially at low galaxy masses.

Early disks

The time travel method also allows spatially resolved ‘in situ’ observations of the stellar and gas components within the young galaxies. Adaptive optics assisted integral field spectroscopy on large telescopes, developed in my group at MPE, especially with the SINFONI spectrometer on the VLT (PI Frank Eisenhauer), have for the first time also resolved the ionized gas motions in the young galaxies. From measurements of line of sight velocities from the Doppler shifting of the Hα-recombination line in different parts of these galaxies, Natascha Forster Schreiber and her team have found that more than a third of the young star forming galaxies were rotationally supported disks, as suggested by the earlier theoretical work, although these disks are much clumpier, turbulent and perturbed than the disk of the Milky Way.


Another remarkable and unexpected finding is that actively star forming galaxies grow only until they hit a ‘mass limit’, the so-called Schechter mass, which is near the mass of the Milky Way. Throughout cosmic time, galaxies appear to shut down their star formation activity when they reach this limit, and transit to the ‘dead’ spheroidal galaxy population. The mechanisms responsible for this ‘quenching’ are currently not understood; it may be caused by a sudden drop in efficiency of gas accretion, cloud or star formation, or an increase inefficiency of gas outflows driven by stars or massive blackholes. Because of the implied structural changes from disky to spheroidal morphologies, it is plausible that mergers may also be involved in the process.

What is the role of massive black holes?

Not only the Milky-Way but apparently all other bulged galaxies and spheroids in the local Universe have a central mass concentration that is probably a massive black hole of about one tenth of a percent of the mass of the entire galaxy, as has been shown by surveys co-led by Ralf Bender at MPE. These massive black holes formed at about the same time as their host galaxies. As shown especially clearly in deep X-ray surveys with the NASA Chandra and the ESA XMM observatories (co-led by Gunther Hasinger and Kirpal Nandra at MPE), these massive black holes were rapidly growing and luminous at the peak of the massive galaxy formation epoch 10 years ago. Accreting black holes convert typically ten to thirty percent of the accreted rest mass energy into short-wavelength radiation and into nuclear winds. This astounding efficiency of energy creation may be one explanation for the galactic winds discussed above, and perhaps also for the quenching massive galaxies. It is not clear how black holes of such large masses are formed, since all but 1/108 of the original angular momentum of a gas particle in the disk of a galaxy must be removed for that gas particle to be able to cross the event horizon of a black hole. Mergers have been proposed as one channel of removing angular momentum sufficiently. The large gas fractions of young galaxies might provide another channel, since a marginally stable, gas rich disk naturally has a large internal ‘friction’ mediated by gravitational torques, and may thus promote efficient internal gas accretion onto nuclei.

The impact of MPG research in galaxy evolution

Galaxy evolution is a very large and active field of modern astrophysical research, which was, until recently, completely dominated by groups in the United States and the United Kingdom. The powerful facilities and missions offered by ESO and ESA have given European researchers a tremendous opportunity for playing an ever more significant role in this rapidly developing field of research. As I have shown above, the establishment of several well-funded experimental, observational and theoretical efforts at MPA, MPE and MPIA, have placed MPG-research in galaxy evolution at the frontier of the field, across a broad front of approaches.

Go to Editor View