A New Scenario For The Formation Of Globular Clusters

Globular clusters are densely packed, spherical agglomerations of dim, very old, small stars. They reached some notoriety in the eighties because, according to some account, they seemed to be older than the time elapsed since the Big Bang, which was obviously suspicious. In the nineties, distance measurements, stellar models, cosmological measurements improved and the mystery of the age of globular clusters (GCs from now on) was solved. Was the interest in GCs then over? Not at all! These objects had many more surprises in store for astronomers.

For one thing, there seems to be a bit more iron than expected in them. Iron is relatively easy to measure in stars, and it is commonly considered as a good proxy of the overall content of heavy elements (curiously, astronomers refer to all elements other than hydrogen and helium – including oxygen and noble gases – as ‘metals’). The iron fraction (or ‘metallicity’) of some very old star in the Milky Way is ridiculously low (less than a millionth of the metallicity of the sun). The metallicity of the GCs is also low, but not ridiculously so: it is always at least several thousandths of the sun’s metallicity. Something prevents the GC’s metallicity to get depressingly low.

However, the big surprises emerged at the turn of the century, with the advent of a new generation of instruments, which allowed the detailed study of individual GC stars. Plotting together colors and luminosities of stars in a GC (this is the so-called Hertzsprung-Russell diagram, a fundamental tool in stellar astronomy), researchers noticed that these stars distributed in parallel stripes, indicating that they belonged to different families. The era of the study of multiple stellar populations in GCs begun. Even more puzzling was the discovery of the so-called anticorrelations among heavy elements.

If a star was rich in sodium, it was at the same time unavoidably poor in oxygen (and vice-versa). The same occurred for magnesium and aluminum.  These anticorrelations are typical of GCs and are not observed in stars of similar ages and masses. After years of investigations, astronomers reached a conclusion: some process of self-enrichment was going on inside GCs: some primordial group of stars was evolving and, during its evolution, aged or dying stars were expelling gas.

The ejected gas was the raw material for the formation of new stars. Self-enrichment thus means that the enrichment process occurs within the GC, and does not necessitate of external influences. The ejecta of exactly what kind of stars were and still is hotly debated. The problem was complex because some elements (noticeably iron), unless oxygen or magnesium, showed remarkable uniformity among different stars in almost all GCs. This fact immediately excluded a possible source of self-enrichment: the final stage of evolution of massive stars, characterized by a colossal explosion, a supernova (supernovae produce a lot of iron).

Some scenarios were proposed, but they seemed to have a common problem, dubbed mass budget problem. We do see in GCs signs of the first generation of stars, whose ejecta contributed to the formation of younger, self-enriched stellar populations. But they are not the majority of observed GC stars. Self-polluted stars are at least as numerous as the non-self-polluted ones. This was odd, because stellar ejecta, in all proposed models, represented only a small fraction of the total stellar mass. To finish with our survey of GC oddities, we do see nowadays young massive stellar clusters with masses similar to our GCs, but they do not seem to share their properties. One of the leading figures in this field  (Alvio Renzini) thus wrote: “special conditions encountered only in the early Universe appear to be instrumental for the occurrence of the GC multiple population phenomenon”.

Proposed Four Stages Of Globular Cluster Formation

Our group in Prague started thinking about these puzzling facts and came out with a possible scenario made of four stages. We tried to identify the special conditions in the early Universe Alvio Renzini alluded to and realized that, at this time, the sky was full of very special stars, commonly referred to as population III (PopIII) stars. For one thing, PopIIIs are almost completely metal-free. This is unavoidable because the Big Bang synthesizes only hydrogen, helium, and a minimal amount of light elements. All heavy elements are synthesized later on by stars. Metals play an important role in the star formation process as we observe it in the present Universe. Some of the metals are very efficient in cooling the gas out of which the stars forms. This cooling process facilitates the star formation. The gas out of which PopIII star formed did not cool easily. Consequently, at least for a fraction of these stars, a lot of gas had to accumulate before the star formation process could begin. These stars grew big (many tens, in some cases a hundred times the mass of the sun) and ended their short life with an immense explosion. These catastrophes in the early cosmos represent the first stage of our proposed scenario.

Figure provided with permission from Springer.

Similar to the aftermath of a nuclear explosion, a gigantic blast wave expanded away from the disaster zone, sweeping up everything it encountered. A huge, relatively thin shell of gas (primordial gas and ashes of the PopIII explosion) collected behind this shock wave. It turns out that this gas is gravitationally unstable: small irregularities in its distribution (unavoidable, after such a mess) grew bigger by gravitationally attract surrounding material until the shell fragmented in small clumps, most of which ended up forming stars (second stage of our proposed scenario). The very first stellar population in today’s GCs, if our story is on track, is made of these stars.

These stars, formed in this expanding shell of gas (dubbed by us super-shell stars, or SSSs in short) have interesting properties. They are formed in part from the ashes of the PopIII explosion and these ashes, in turn, contain a lot of iron. That is why the stars of GCs are not as iron-poor as other stars with similar ages. As time went by, the SSSs evolved and, as any other star, they started releasing some material into the interstellar medium through the so-called stellar winds (third stage). In particular, massive stars among the SSSs ejected copious amounts of gas, in a relatively short time.

This ejecta expanded in any direction. Part of this ejecta propagated thus inwards, towards the center of the shell, where the PopIII explosion took place. This is the interesting part of our proposed scenario because this ejecta created in turn a dense, relatively thin shell, somewhat similar to the shell produced by the PopIII explosion, but propagating in the opposite direction. The attentive reader has surely already understood what happened afterward: this inward-propagating shell is also gravitationally unstable and fragmented, forming new stars (fourth and final stage). These are the evolved, self-enriched stars we observe nowadays in GCs. Moreover, the SSSs inherited the velocity of the PopIII-driven shell of gas, which was relatively large. This implies that they were receding from the location where the PopIII explosion took place. Although their expansion was slowed down by their mutual gravitational attraction, a large fraction of them might have ended up too far away from the other stars of the nascent GC to be captured again.

Here comes, therefore, our explanation of the mass-budget problem introduced before: the very first population of stars in a nascent GC is big. Much bigger than what we see nowadays. A large fraction of these stars, however, got lost. Before getting lost, these stars managed to expel gas through stellar winds. A large fraction of this ejecta propagated inwards, fragmented, and created a new population of stars. Also, most of the stars responsible for the self-enrichment escaped the GC, leaving however behind a large amount of gas ejected through stellar winds. That is why there is an equal amount of pristine and polluted stars in GCs.

End of story? Not really: we left a shell of gas, mainly made of ejecta of SSSs, propagating inwards and forming new, polluted stars. What happens when these stars evolve? Could their ejecta form a new shell, which will fragment and form a further population of stars? In principle, yes, the cycle might start again and we can witness the formation of more than two stellar populations in a young GC. Indeed, the Hertzsprung-Russell diagram of some GCs indicates the presence of more than two populations (in one case, perhaps, up to seven!)

Does our recently published paper demonstrate that GCs really formed as we propose? Not really; what we have shown in our recent publication is only the layout of our scenario, and some simplified simulations demonstrating that this mechanism could really work. In the paper, we did not hide difficulties and open questions. For instance, we are at the moment unable to reproduce the very ‘heavyweight’ GCs we see in the sky, but only the small/medium ones. We think that open questions and unsolved problems are the salts of science, as they can inspire other researchers to make a step ahead and improve the proposed mechanism, or to reject it outright, which would be a progress for science anyway (it would restrict the range of possibilities).

For the moment, we hope we have convinced the reader that GCs are complex objects, with a very interesting history to tell, and for sure still many surprises in store.

These findings are described in the article entitled Globular Cluster formation in a collapsing supershell, published in the journal Astrophysics and Space Science. This work was led by Simone Recchi from the Astronomical Institute of the Czech Academy of Sciences.