Weaponising dark matter

Stephen Baxter's Xeelee sequence revolves around a war between baryonic and non-baryonic life forms. One memorable sequence features a pulsar being hurled at the Great Attractor (because reasons). Today's paper feels like it could easily fit within such a realm of the gloriously far-fetched, albeit it's not without some reasonable evidence too.

This is just a four-page letter so I'll keep this one very short indeed. Like the last paper, they claim to have found the signature of a very small dark matter halo, but this one's even smaller. The last one was about a billion solar masses, extremely small by galaxian standards but not outrageously so... this one, by contrast, is probably no more than a few tens of millions of solar masses, with a lower limit of just a few thousand.

Such features are certainly predicted in cosmological simulations. Basically, the higher the resolution, the more small dark halos result. But below a certain limit, nobody ever expected to have much chance of ever detecting them, since they'd have so little gravity they'd never attract enough gas to form a single star. And once you add in all the baryonic matter to the simulations (the boring normal matter of stars and gas), presumably most of the smallest ones would be disrupted.

The claim here is they've found a bullet wound in the Milky Way resulting from a collision with one of these minihalos. Actually, again like the last paper, this is not a discovery announcement so much as an independent confirmation by a different method. The original discovery came back in 2017 in the form of a molecular gas cloud with an unexpectedly high line width. 30 km/s is small by the standards of galaxies (the Milky Way would be more like 400 km/s), but with no stars to drive the motion, dynamics have an obvious appeal. Without any other visible material (at least nowhere near enough), dark matter is at least heavily implied.

Here they use Gaia data to look at the velocity of the stars in the vicinity and discover they have a vertical velocity anomaly : in a small region of the disc, the average velocity of the stars perpendicular to the disc drops, even while their dispersion increases. The original CO blob is slap-bang in the middle of this VVA, with the VVA being very much larger than the CO blob. Which would be an awfully suspicious coincidence.

A lump of dark matter colliding with the disc could certainly cause this. Its small size is certainly consistent with a very small halo. But I have no familiarity with stellar dynamics on these scales at all, so I can't tell you how unusual such features are and their figures don't really give much of an indication. They also don't even consider other explanations, which I suppose is fair given the limited available space, but it would have been nice to have mentioned something (besides the obvious impossibility of stellar winds and the like, there apparently being no stars here). And how much such features would we expect to find, if these minihalos exist in the numbers predicted by simulations ? Finally, their tie-in to Ultra Compact Galaxies more generally – which are largely thought to be the stripped cores of more massive galaxies – is just too speculative even for a letter.

In short, it's definitely a very interesting feature to report, but it's going to take a lot more work to say anything definitive about what it actually is. 

A dark RELHIC of an earlier age

The last time I tried to count the number of times objects had been claimed to be the first dark galaxy candidates, I stopped at ten because I got bored. Today's paper adds another one to the list.

To be fair, not all dark galaxy claims are equal. Some would say a galaxy only counts as dark if it really has no stars at all, others that it just needs to be sufficiently dominated by its gas and/or dark matter. Others would insist that it has to have certain dynamics or only be found in the nearby universe*. Most would probably demand it had a primordial origin rather than just being stripped out of a galaxy, but not everyone would agree. So that list of ten could very plausibly be extended or contracted considerably.

* Pretty much everyone agrees that galaxies started off as dark, so we accept that dark galaxies did exist at one point. The controversy is over whether any still remain dark today.

This paper concerns a very particular type of dark galaxy they annoyingly dub a RELHIC. Why annoying ? Because we also have radio relics, which are completely different beasts : they're incomparably larger and more diffuse, and have little or no direct relation to individual galaxies. These objects, on the other hand, are Reionisation Limited HI Clouds, a term coined by one of the authors that I'd urge them to stop using. 

But no matter. What they report on is a very interesting object that in some ways is the sort of dark galaxy candidate everyone wants to find. One of the major reasons to suppose such objects exist at all is that they would solve the long-standing missing satellite problem, whereby simulations produce far more galaxies than are actually observed. The idea is that while the physics of gravity is pretty simple, the physics of star formation is anything but. So maybe these objects do exist, it's just that they've never formed stars. This is now the widely-accepted explanation for missing satellites – indeed, arguably there isn't such a problem at all any more, as (in some models) the only such "dark galaxies" are now so small and so lacking in stars that we wouldn't detect them.

One of the complexities of the physics behind these is that of the Epoch of Reionisation. The first stars as thought to have been super-powerful monsters powerful enough to ionise most of the gas in the early Universe, heating it to the point where it would be driven out of the smallest galaxies completely. Models show that below a critical mass threshold, galaxies would lose all of their gas and never form any stars at all. It's not quite a sharp cut-off, with some near or slightly above the threshold able to form some stars before reionisation brought the process to a permanent halt, but it's close.

Such objects are thought to be extremely difficult to find. Their HI masses should be only a few million times the mass of the sun, about a hundred times less than a typical dwarf galaxy. And their line widths might be only 20 km/s or less, barely wider than the HI line itself . In principle these objects could be almost numberless, just bloody hard to spot. In contrast, most dark galaxy candidates that hit the headlines are much bigger, and usually by the author's own admissions fairly exceptional – massive dark hulks that are relatively easy to find despite being so rare as to indicate little or nothing about how most galaxies form.

Here the authors present a candidate discovered with China's mighty FAST telescope. In contrast to this awful RELHIC term, I can't fault them for the name of this particular object : Cloud 9. Yes, really. Apparently this was first reported in 2023 but I seem to have missed that paper when it came out. 

Here they report on deep Hubble images and confirm that it's really, really dark, with no more than a few thousand solar masses of stars against it's million or so of gas. That, together with its line width of just 12 km/s (!) and small size (1.4 kpc radius), with an estimated halo mass that's extraordinarily close to the mass threshold, make it a compelling RELHIC candidate. It's certainly one of the darkest objects ever found, which is always a pretty cool thing to find.

But just how good, exactly ? My verdict would be... yeah, this one's pretty interesting. Is it definite ? By no means. But it's a good candidate, and absolutely needed to be published.

Glancing at the discovery paper, it seems that Cloud 9 is a little over 100 kpc from M94 itself, with HI clouds closer to the galaxy that are clearly some form of debris. 100 kpc is quite far, but only a few times the size of a large galaxy, and certainly there are many extended streams known which are much larger than this. So this cloud could be a leftover far-flung bit of debris as well, but like Cloud 6 in Leo, it doesn't really fit the general pattern of the other clouds. 

A perhaps more serious difficulty would be that estimating the total mass of the feature must be extraordinarily difficult. HI tends to be found at ~10,000 K, corresponding to a line width of 10 km/s. A width of 12 km/s tells you pretty much nothing beyond the thermal state of the gas, so inferring its dynamics from this is... well, my worry is that you simply can't get a meaningful estimate when things get this narrow. Even if there was no dark matter here at all, the line width wouldn't get much narrower because this is about as narrow as the line can get. It's not a matter of better data in this case : nothing will help, at least not very much.

Another issue is the question of how long the cloud could survive, and conversely, how long it's been in existence. Currently its gas density is well below the threshold for star formation. Assuming it began life so small that the density would reach the threshold (so as to have always been dark), and given its expansion velocity and small size, it would have taken perhaps a hundred million years to reach its present size (without dark matter, as expected if it's just debris). In a another hundred megayears or so it'll double its and likely be undetectable. So if it's debris, we're detecting it at an unusual point in its existence, but sadly this doesn't constrain things too much.

I think this is a case where what's needed is a good set of simulations, especially given the timing constraints from the size of the cloud. What kind of interactions could affect M94 that would produce debris like this ? How often are such things formed, and are those simulations compatible with all the other data of the system ? What happens to existing minihalo RELHICS like this one in a system like this, where there's clear evidence that M94 experienced a merger – can they survive in such a place ?

This one's going to take a lot of work to answer. It would be easy to dismiss this as just another bit of HI fluff... and it might be. But it's so close to what we expect minihalos to be like that the workload might be worth it. And perhaps, just maybe, some other clouds already lurking in the data aren't the boring bits of debris we all thought they were.

The galaxies that seemed magical are actually just very lazy

Today, two papers for the price of one ! The one is a sequel to the other and they're both quite technical, so let's knock off two birds with one stone in the bush and other mixed metaphors. Paper I is here and paper II is here.

The papers in essence address two related questions on Ultra Diffuse Galaxies – the big faint fuzzy things that often seem to lack dark matter, which I've covered here ad nauseum. Neither paper much addresses the dynamics (i.e. total mass) of the objects, but rather the other fun aspect of these galaxies : why are they so wretchedly bad at forming stars ? Many of them have tonnes of gas, so why aren't they forming stars like normal galaxies are ? Why are they so large and yet so faint ?

The first paper deals with the basics. It tries to determine if the star formation efficiency of UDGs really is weirdly low, or if this is only a selection or measurement effect. Spoiler alert : it really is low. So the second paper then address why this might be – is there something missing in our basic model of star formation, or is it just due to the peculiars and particulars of of these particularly peculiar systems ?

Paper I begins by collecting a sample of 22 UDGs and 35 more typical dwarfs of comparable mass.  The UDGs all have atomic HI gas detections, though low resolution so essentially all we know is the mass of gas, nothing about its structure. But it seems that at first glance, UDGs are indeed of systematically lower star formation efficiencies : their star formation rate is less than that of other galaxies of similar gas masses, and given their stellar masses they have more gas than expected. Both of these effects are modest though. It's quite apparent that the population as a whole is systematically offset from the rest, but they're all still within the general scatter. Interestingly, they also note that the properties of the HUDGs (HI-detected UDGs) aren't much affected by environment.

The first question they tackle is whether these objects are really different in terms of their gas content, or if this is just a selection effect. That is, the HI observations might be limiting what can be detected at all. It could be that those of lower gas fractions do exist, it's just that the data isn't sensitive enough to show them. But they find that's not the case : they should be able to find considerably less gassy-objects, so these don't seem to exist at all. These HUDGs are not the tip of a less-gassy iceberg.

Next, their main topic. While the average star formation efficiency of HUDGs appears low, this could just be due to the statistics from using the total gas mass and stellar mass, which smooths over the whole structure of the galaxy. It could be that the star formation efficiency is actually quite normal, just restricted in area. For example the gas could all be concentrated in the centre and forming stars pretty normally, but when averaged over the whole galaxy, this would be "washed out" and it would look like the galaxy was rubbish at forming stars. Overall, they would be, but locally, they wouldn't be that bad. This would imply a significant population of older stars outside the gas-dominated regions.

To test this they use spectral energy distribution fitting. Basically what this does is use many different data points across the optical, UV and IR spectrum to estimate the stellar ages as accurately as possible – this is the best we can do in terms of estimating a galaxy's star formation history. Ideally we'd also like to have resolved measurements of the HI, which sadly they don't have here. But they do SED-fitting for many different points per galaxy, so they can see if the low SFE is something that varies throughout each object or if it's low everywhere.

I'll skip over the details of the SED fitting because I don't understand any of it; I only note that they stress they aren't constructing detailed star formation histories here, just enough to answer their main questions. Their main result is that the low SFE is true on all scales from big to small. It's not just that the gas could be more extended, it's that UDGs are bad at forming stars full stop, even if the gas density is higher. While UDGs are large given their stellar masses, they're not especially large considering their HI masses.

There are a lot of different scaling relations to juggle here, but the end result is very simple : UDGs aren't good at converting gas to stars even when they've got plenty of it. The obvious next question is, of course, why are they so bad at this ? For this we need the sequel.

Paper II takes quite a different approach and is all about modelling. There's a really popular and widely-used relation between the surface density of gas and its star formation activity, but there have been indications for many years that we ought to be using the true, volumetric (3D) density instead. This is harder to measure directly so some assumptions have to be made, but it can be done. 

Here they consider a particular version of the volumetric star formation law that depends on the different components of a galaxy together : the gravity from the atomic and molecular gas, the stars, and the dark matter. All have different gravitational contributions. For instance in the centre of a galaxy everything might be extremely dense, whereas further out their might still be lots of atomic gas and stars but less molecular gas, and on the very outskirts only atomic gas and dark matter. So even if the atomic gas has the same surface density, it may need the extra gravity of the stars to help pull it together and collapse.

Again I shall spare you the technical details of the model. This time my essential note is that they consider UDGs to be rather more dark matter dominated than ordinary galaxies, which flies against the prevailing winds in the last few years. More on that in a moment.

They find that this more complex model... works ! It can explain the star formation rates of both normal galaxies and UDGs under very reasonable assumptions : there's no need for any additional physics, no weird mechanism or alien interference that suppresses the star formation. There are many uncertainties, but their sensible default assumptions are enough to give a good result without any sort of fine-tuning needed. So UDGs are, in a sense, pretty normal.

They also find that the model isn't very sensitive to how much molecular gas the UDGs are assumed to contain. That's bad news for anyone trying to detect their molecular content : essentially it implies they could well have very little of it. They can even estimate just how much molecular gas they expect them to have, given their estimated star formation rates and the (rather surprisingly but repeatedly established) independent finding that molecular gas generally has a constant depletion timescale of about 2 Gyr. In short, their conclusion is that detecting the molecular component will be Bloody Difficult. Dwarf galaxies are already hard to detect, but UDGs will be even worse.

What about that choice of a relatively massive dark matter halo ? They explore this, and rather surprisingly it doesn't matter much. It seems that the density of the dark matter is anyway assumed to be low so that reducing the total mass doesn't make a great deal of difference. In fact this can give slightly better agreement with the measured star formation activity, but unfortunately, they say there are too many other uncertainties to say if their model prefers no dark matter at all to a normal mass halo.

So, there we have it. There's no need for any weird physics here : star formation in UDGs apparently follows the same laws as in every other galaxy. The difference isn't so much the gas content itself as the state it happens to find itself in, just as a puppy can be an energetic ball with the madness of a thousand caffeinated suns or the sleepiest thing since Slothy the Sleepy Sloth swallowed an entire bottle of Nytol after a mug of warm milk in a comfy armchair. 

Does this mean UDGs aren't weird at all though ? Nope ! You might remember that there were previous claims that UDGs are actually just normal galaxies if you redefine how to measure their radius. That was true, but doesn't mean that the other radius estimates were wrong : it still points to UDGs being anomalous, just not in the way we'd understood. 

Here we still don't know what sets the initial conditions of UDGs. Why do they start out so differently to normal galaxies ? Is it their dark matter halos (or lack thereof) and if so, is this compatible with our simulations of the sort of halos we expect to actually exist ? And how come their globular cluster populations appear to be very different to other galaxies ?

We still don't know. We do know that UDGs are weird, but as to whether they're pointing to a deep flaw in our models or just some incompleteness or other... the jury's out. Or more accurately, the trial hasn't even started yet : we need more evidence before we can even really begin. But at least what this paper adds is what evidence we should go after. If UDGs are found to have chonks of molecular gas, that would falsify their model straight away. If they're not, we keep investigating.