The Very Interesting Gas That Doesn't Do Anything

"Atomic hydrogen is the reservoir of fuel for star formation." "Neutral hydrogen is a galaxy's fuel tank." These stock phrases are a mantra I fall back on for beginning both public outreach and journal articles. But are they accurate ? A recent paper has got me wondering if we've got something terribly, terribly wrong. It seems that when galaxies merge, although star formation rates increase, the gas really doesn't do a lot at all. It's a bit like pulling the plug out and not seeing the water level drop.


We Need To Talk About Hydrogen - Well, I Do, Because They Pay Me Money To Look For This Stuff, But You Can Just Listen If You Like.

Atomic hydrogen (HI, pronounced, "H one" since it's a Roman numeral 1, not an I) is the simplest element there is. One electron orbiting one proton. That's it. Electrically neutral overall, since the proton and electron have equal but opposite charges, it should be as simple as you can get.

In reality things can be a great deal more complicated if you want to bring quantum into it, which I don't. Anyway this simple picture is good enough for what I'm going to discuss here.

That picture is a little over-simplified for what we'll need though. For example, two hydrogen atoms can bond to form an H₂ molecule :

The electrons from each atom are now associated with both protons, forming a covalent bond.

Or the hydrogen can have its electron stripped away by an energetic photon (i.e. light, often near hot young stars) to become HII (confusingly also pronounced "H two"), which is really just a cloud of protons and unbound electrons. It can even gain an electron to become the negatively charged hydride ion, though this is not thought to be important in astronomy.

The current working model (I would not call it a consensus by any means) is that it's usually only the H₂ that's important in forming stars. Atomic hydrogen is generally so hot (1,500 - 10,000 Kelvin) that its own internal pressure prevents it from collapsing into stars. Molecular hydrogen is colder, which means it's less able to hold itself up and more prone to collapsing. That view is by no means certain, and there are some hints that in some circumstances, HI can indeed collapse into stars without forming molecular hydrogen first.

When we go looking for atomic hydrogen, we mostly find it in blue, star-forming galaxies (they look blue because the most massive stars are blue and don't live long).

Random sample of galaxies detected by their hydrogen emission from a large survey. True, many are red, but when you look at them more closely they usually have blue star-forming regions as well.

Looking in more detail, we see "holes" in the atomic hydrogen, where it looks as though the gas has cooled to become molecular and is forming stars. We also see a few cases where there's atomic hydrogen but no star formation. And that's the (simplified) version of why we think atomic hydrogen is usually just a sort of reservoir : ultimately it will cool into molecular hydrogen and form stars, but it doesn't normally form stars directly.

OK, That's Enough About Hydrogen, Please Tell Me About These New Results Now Or I Will Become Cross.

When galaxies merge, the gas collides and things get seriously messy. Not for nothing was a Hubble press release entitled, "Galaxies Gone Wild !"

Needless to say, galaxy collisions and mergers are complicated. But generally if the galaxies both contain gas, collisions result in a massive increase in star formation as the gas compresses and cools. So the atomic hydrogen becomes molecular hydrogen which becomes stars, and everyone's happy, right ?

Oh, would that 'twere that simple. The new paper shows that when you stick two galaxies together, the atomic hydrogen does sod all. Well, it might get splashed about a bit, but its mass doesn't decrease and in fact it might even increase slightly. Which as far as the "fuel reservoir" idea is concerned is like being poked in the eye with a sharp stick.


How Do We Know This ?

Galaxy collisions are slow, grand affairs that can last for billions of years - so we can't just watch galaxies merge and see what their HI does. The metaphor that's usually used is that if you want to learn about how trees grow, you look at many different trees. So it is with galaxies. By finding enough isolated galaxies and galaxies which have already merged, the authors try to look at what happens to the gas statistically.

That's not any easy process. After the merger happens, a lot of information about the original galaxies gets lost. So unfortunately there's just no way to know what sort of galaxies were involved in the original collision. But most mergers are thought to occur between spiral galaxies (because these are far more common, except in galaxy clusters), and spiral galaxy gas content doesn't vary very strongly depending on their precise morphology.

What the authors do is define a sample of post-merger galaxies, then for each of these they find isolated galaxies which have the same mass in stars. Then they measure the gas content of all the galaxies in both samples using existing data from the ALFALFA survey and their own Arecibo observations (They've got me to thank for that since I supervised Derik Fertig (second author on the paper) at an Arecibo summer school. The fact that the other authors are far more experienced senior astronomers has, obviously, absolutely nothing to do with it whatsoever).

What they find is that the amount of gas is the same in both samples. That is, a galaxy with a billion stars that's quietly minding its own business has the same amount of gas as a post-merger of a billion stars. Even though new stars are forming*, somehow the gas just nonchalantly sticks its hands in its pockets and goes, "meh".

* Before the galaxies merged they would have had less than a billion stars in total.

It's not quite as simple as that though. It's not terribly likely that all of the mergers are formed from equal-mass galaxies. And less massive galaxies tend to have higher gas fractions - that is, more gas relative to their stars. So if you stick two unequal-mass galaxies together, and none of the gas gets turned into stars, you'd expect the gas fraction of the post-merger object to be a bit higher that a normal galaxy of the same stellar mass.

There isn't really a handy analogy for this, so let's make one up. And let's bring Worf back into it, why not. Suppose Worf goes to a party at Starfleet headquarters and brings a bottle of strong Klingon blood wine. Captain Picard is making do with regular human wine. When his glass is half-empty, a waiter asks if he wants a top-up. Worf, however, is honour-bound to offer the captain a refill from his blood wine instead. If Picard chooses the blood wine, he'll end up with a stronger drink than if he accepts the regular wine, even though the same amount would have been added.

It's the same with gas fractions. Smaller galaxies are more "potent", they contain more gas per star, so merging them with a larger galaxy is like topping up orange juice with vodka. You'll get a lot more drunk that way.

As long as one bottle is of Klingon blood wine,
obviously.

When reading the paper I thought to myself, "man I could sure use a glass of wine right about now !". But then I thought, "well, maybe you could account for the mergers of unequal-mass galaxies by choosing random galaxies from the isolated sample and adding up their stellar and gas masses". Lo and behold, the authors did exactly that in the next paragraph ! What they found was that the difference in gas fractions of the initial galaxies would increase the gas fraction in the post-mergers quite considerably, not by a very dramatic amount but easily enough that they should have been able to measure it in their sample.

So, does that mean that some of the gas is being consumed after all ? The gas fraction may not have changed, but it is less than if you stuck two unequal-mass galaxies together. Well, maybe, though it seems a bit suspicious that all of those tremendously complex processes that happen during the merger just bring the gas fraction down to that of a normal, isolated galaxy. We had a saying during the undergraduate course on general relativity : it all cancels and equals nought - an awful lot of work needed to find out that nothing's happening.

At this point the team turn to numerical simulations to figure out how much gas should be consumed by star formation. This is perhaps the weakest aspect of the paper. Ideally, they'd set up their own simulations and track the evolution of the atomic and molecular gas and the stars, but this is tremendously difficult to do. We're talking about many months (or more) of extra work, so it's perfectly understandable that they don't do this.

Instead they use an existing set of simulations which are (it must be said) vaguely-described here (again this is understandable since that publication is only a letter, not a full article). What they do is track the total mass in stars, then use the known gas fraction relation (from observations of isolated galaxies) to calculate how much atomic hydrogen there is during the whole merging process (presumably because the simulation itself doesn't distinguish between the different forms of hydrogen we discussed earlier).

What they found by doing this is that the star formation process shouldn't change the gas fraction much at all. So the fact that the post-mergers don't have as much gas as expected can't be due to star formation.

But that assumes that there really is a decrease in the gas content as the galaxies merge. Now I mentioned earlier that there was some suggestion that the gas fraction actually increases a little bit from the mergers. That's a little more tentative. The thing is, not every galaxy in their sample had detectable atomic hydrogen at all, but the detected fraction of the post-mergers was double that of the isolated galaxies of the same stellar mass. That is, if you randomly choose a post-merger and an isolated galaxy of the same mass, you're much more likely to detect gas in the post-merger than the isolated galaxy. Which suggests that post-mergers actually have higher gas fractions than their parent galaxies did.

Another important factor is that the ALFALFA survey isn't as sensitive as we might like. That means it's only detecting particularly gas-rich objects which, say the authors, reduces the expected difference in gas fractions between the isolated and post-merger galaxies - so their calculated differences are probably too large. Many of their isolated galaxies that have no detected gas from ALFALFA probably do contain some gas, just not enough to be detectable. When you run the numbers, say the authors, that means that it's very possible that smashing galaxies together increases both their star formation rate and their gas content.

Which is a lot like pulling the plug out and seeing the water level rise. It's weird.

Believe me, if you Google image search "weird" you'll get far stranger stuff than this.

What ? How Could This Happen ?!? What Does This MEAN ?!?!?!?!

To summarise, it seems that when galaxies merge their atomic gas content remains (at best) unchanged, even though part of their gas is turned into stars. Simulations suggest that the fraction that forms stars is very small, but it looks plausible that the atomic gas content actually increases, somehow.

One thing this study doesn't look at directly is the molecular gas, which is what we think is more directly responsible for star formation. Could it be that the stars which form as a result of the merger do so from the existing H_2, perhaps due to the shock of the collision increasing the density ? Unfortunately, say the authors, previous studies have found that the molecular gas content also increases during mergers, so, nope.

It just seemed really wrong not to include a picture of the famous merging "Antenna" galaxies in this post somewhere, so here we go.

But before we go saying, "a wizard did it !", the authors suggest a possible explanation for where this extra gas comes from. Galaxies, it's thought, may be surrounded by large halos of ionized hydrogen. Just how much is not known. Normally it could be slowly trickling down, keeping the reservoir's topped up, but during a merger it might cool more rapidly. Simulations say that's possible, so, maybe. Whether this is the correct explanation or if it's due to something else entirely, we just don't know.

If it's true, then atomic hydrogen is just the middle step in the process - a complex system of dams rather than just one reservoir. It's not a totally unprecedented idea, but it will mean quite a rethink. We'll have to wait and see what other observations and simulations have to say : this one study is important, but not enough by itself to prove what's going on.

Then there are elliptical galaxies. They're thought to be formed by mergers, as this rather nice simulation shows :

But they usually don't have any atomic hydrogen. Where's it gone ? And yet, sometimes they do - occasionally they have very large amounts of it. It's a mysterious mystery, right enough. As usual, we need more data. But the more we learn about galaxies, the more complicated they become. Eventually, perhaps, we'll have enough data and good enough theories to have a real explanation. For now, we're still learning. And that's fun.