Here's a fun paper that I need to talk to my globular cluster expert-friends about. The idea is that some of them are actually the remnants of satellite galaxies which have had their dark matter halos tidally disrupted. I get the strong impression that this isn't nonsense, but I don't know enough about globular clusters to judge exactly where it is on the credibility scale. But instinctively, I like the idea very much. Expect an update when I get the chance to check with those in the know.
First, the author notes that globular clusters show a linear relation between their total mass and that of their parent halo (i.e. the total dark and visible matter of their nearest galaxies), suggesting their formation is more related to the dark matter than stellar evolution. This extends across a huge range of sizes, so huge I'm a little worried... for example a globular cluster in one galaxy has a stellar mass of a thousand solar masses, whereas its parent galaxy has a total mass of about a billion. This smacks of the "big things are bigger" bias : of course you see trends if you extend things this far. You can't really avoid it. But so long as the relationship is sufficiently tight, it's still interesting.
She also says that whereas faint galaxies have just a few, old clusters, the bright galaxies and mergers have lots of young clusters, suggesting that they're destroyed over time. This could mean that some have different origins (some being primordial and long-lived, others more transient) as she suggests, but I don't see why it couldn't be that they all form in the same way and the surviving clusters in faint galaxies are just those lucky enough to have avoided destruction. I suppose this depends on how short the destruction timescale is though, and from what I can remember from innumerable meetings, I would guess it's expected to be short. So this is probably a fair point.
A really interesting detail she mentions is that some globular clusters are now known to have multiple stellar populations, that is, having formed their stars in bursts at different times. Which she says shouldn't be possible, because the first generation of stars ought to have strong winds and explosions that blast out all the remaining gas and prevent any further star formation. If, however, the clusters have plenty of dark matter, then their mass would be enough to prevent the gas from escaping, allowing it to fall back in and form more generations of stars.
This paper is only submitted, not yet accepted, and in my opinion it's still a little rough around the edges. I would have liked a bit more information about the simulation setup because as described it seems too good to be true. Following the formation of individual stars ?? How the hell does it manage that ? Is it hydro, n-body, what ? I mean, I could look up the cited papers, but I'm not going to. And some terms are used a bit inconsistently, e.g. "environment" is used both on the large scales of whole galaxies (e.g. whether they're in groups or clusters) but also to the situation of the star clusters (whethere they're in dwarf or giant galaxies). Which can be somewhat confusing.
Another rather intriguing point she notes from her earlier simulations is that the minimum mass a dark halo needs to allow star formation is environmentally dependent. This is counter-intuitive but it makes sense. If a massive halo starts forming stars, the chemical processing enhances the metallicity and the supernovae and winds can disperse this enriched gas into nearby halos. The higher metallicity allows it to cool more easily, making it less resistant to gravitational collapse and so more prone to forming stars. And I think that's pretty neat. Sometimes in my public talks I go on a protracted, deliberately-breathless rant about the complexities of star formation and feedback, but this just isn't something that's ever occurred to me.
Until recently I've thought of galaxies and star clusters as really qualitatively different, discrete objects, with only hints of a more continuous categorisation being given at the NAM conference in the summer. But the author says there are four basic classes forming an elegant sequence : pure dark halos, dark-matter dominated satellite galaxies which have some stars, stripped halos which are stellar-dominated but with some dark matter, and genuine star clusters which completely lack dark matter. This has a very strong instinctive appeal to be; intuitively it just makes sense that things shouldn't be so strictly one thing or another as in the classical picture.
And how the dark matter can be stripped from a small galaxy to turn it into a star cluster has in fact already been addressed in relation to much larger objects. In simulations, dark matter is structured in a spheroidal halo, with its particles in random, radial orbits. Unlike the disc, even particles which are sometimes found in the central regions can also sometimes be found much further away. So the dark matter is continuously vulnerable to tidal stripping, whereas the central stars and gas are much more resilient because they're always deep in the gravitational potential well.
Much of what follows is a detailed, quantitative examination of the simulations. The upshot is that it broadly seems to work : it's not perfect but the basics are there (and the imperfections seem to depend on missing physics not included in the algorithms). It even includes globular clusters with multiple stellar populations. And best of all, it includes the testable prediction that if this is what's really going on, we should see many more globular clusters around the most distant galaxies, those in the early universe.
Personally I think this idea is very intriguing. But there's an elephant in the room which is deafening by its absence : what the hell does this mean for the missing satellite problem ? Pure dark matter simulations predict way more galaxies than we actually observe – could it be that when you include the baryonic physics, most of them just dissolve ? That'd be extremely satisfying, much more so than trying to get the feedback precisely balanced so that just the right number happen to form stars. If in fact most of the globular clusters we see are the surviving satellites after all, this would both unify the studies of star clusters and galaxies and solve a major problem with the Standard Model of cosmology.
It's too good to be true, of course, and the author doesn't mention this at all. But I like it.
EDIT : After discussions with the experts, this paper definitely isn't nonsense. The idea that some globular clusters could be remnants of disrupted galaxies isn't new, so here the novelty appears to be that this can potentially explain all or a very large fraction of them. We noted that the objects explored here are all very small, typically a lot smaller than Milky Way globular clusters – this probably is only because of the use of dwarf galaxies in the simulations, and they do lie neatly on the scaling relations for much larger objects. That's only a minor caveat, but it does explain why the author doesn't look at the missing satellite problem as the objects being studied are just too small.
A potentially bigger issue is that all models apparently predict more globular clusters in the past, so more quantitative comparisons are needed to show how this can be distinguished from alternative ideas. And why all the objects in the simulations appear to lie below the line indicating they should be tidally disrupted, we're not sure. There seems to be a bimodality in the simulated objects : either they're dark matter dominated or have very little, which is not much discussed. Presumably this reflects the vulnerability of these objects to stripping : if they fall into a situation where tidal forces can remove the dark matter at all, then this will generally remove almost all of it, whereas if they don't, they won't – they won't lose any through internal processes. Finally, the experts mention that the big challenge for globular clusters is explaining anti-correlations seen in their chemical abundances, though that one's beyond my pay grade.
I've also made minor edits to the rest of the text and corrected my mistaken pronouns.
No comments:
Post a Comment