Paper Explainer: Search for Gamma-ray Emission from Dark Matter Annihilation in the Small Magellanic Cloud with the Fermi Large Area Telescope

This is non-specialist description of my latest paper, written with Regina Caputo, Pierrick Martin, Eric Charles, Alyson Brooks, Alex Drlica-Wagner, Jennifer Gaskins, and Matt Wood. I want to thank all of them, and the Fermi-LAT organization (several of the authors are members of the collaboration).

That’s quite a laundry list of authors, but that’s because this was a pretty complicated work, and we needed a lot of people with specialized knowledge to be able to accomplish it. I’m pretty proud that we managed to assemble such a great team to get this physics done. Tackling questions of dark matter often ends up requiring such a diversity of skills that no one person can be an expert in all of them, so you need outside help.

Actually, this is a follow-up paper to a previous paper with a large intersection of the author list: “Search for Gamma-ray Emission from Dark Matter Annihilation in the Large Magellanic Cloud with the Fermi Large Area Telescope” After getting the gang together for that work, it seemed silly to let such an obvious follow up go uninvestigated, and lose all that institutional knowledge we had put together. The previous paper predates this blog, so I’ll try to explain both to some degree as I go along.


 Artists Rendition of Fermi-LAT (NASA)

Artists Rendition of Fermi-LAT (NASA)

These papers are using the experimental data from a NASA satellite, the Fermi Gamma-Ray Space Telescope. Fermi has a number of instruments on it, but the one of primary interest to particle physicists like me is that Large Area Telescope (LAT). That’s the big block on the top of Fermi in the picture. The LAT is about a meter by a meter square, and contains layers of metal and detector elements. If a gamma-ray enters the LAT, it leaves a track of energy deposition through the metal, which allows a measurement of the energy of the gamma-ray photon and the direct from which it came. The LAT rocks back and forth over the sky (unless something interesting happens, and then it briefly pivots to look at the event, before returning to scan-mode). Thus, over the years it’s been up, the Fermi-LAT has collected gamma rays from the entire sky, giving us a map of what the Universe looks like in this spectrum. Of course, the sky is a sphere, but if you take that sphere, and unwrap it to lay flat (like a projection of the Earth for a map) it looks like the figure seen here.

 Gamma Ray Sky, as Seen by Fermi-LAT (NASA, Fermi Collaboration)

Gamma Ray Sky, as Seen by Fermi-LAT (NASA, Fermi Collaboration)

Red are regions with lots of gamma-rays, blue and black have fewer. The plane you see across the sky is the Galactic Plane, the disk of stars that we picture when we picture “the Milky Way.” The Galactic Center is the region in the middle of picture, and you see some elegant arcs of gamma rays emerging from it. Pretty much all these gamma rays are coming from collisions of energetic particles and gas and dust in the Milky Way, that is, they are coming from “normal” physics. Then there is a background of gamma rays coming in from all directions (isotropic) that comes from the sea of high energy events all over the Universe, this is the extragalactic background.


I’m not interested in those gamma rays. I’m interested in gamma rays that might come from dark matter. Dark matter is… dark. And it doesn’t interact very much. But dark matter must have been produced in the early Universe somehow. We don’t know how, but one leading idea is that dark matter was produced when the Universe was very hot, and regular matter had lots of energy. At that point, it would be possible for regular matter to collide and make some new particle that was very heavy and didn’t interact with photons (if such a particle exists, of course). As the Universe cooled, regular matter stopped having enough energy to make these massive new particles, and they “froze out” of contact with the regular matter that went on to be the stuff we’re made out of. These heavy particles would be an ideal dark matter candidate.

Now, if that’s the case, in the Universe today occasionally, a particle of dark matter might smack into another particle of dark matter. If we assume that regular matter could collide to make pairs of dark matter, than pairs of dark matter could collide and turn into regular matter (again, this may not be how dark matter works, but it’s theoretically compelling for lots of reasons, so we should investigate). In most cases, the pair of normal matter produced in this collision would be unstable (as most of the Standard Model particles are unstable), and they would decay down to lighter, stable particles. In that decay, they would produce lots of photons, and these photons would be energetic enough to be called gamma rays.

So, if we have a gamma-ray telescope, we can look in the Universe to see if we see a bunch of extra gamma rays that might be coming from dark matter. Fermi is the ideal platform for such indirect detection.

Now, the darkness of dark matter means that dark matter-dark matter collisions must be very rare. Furthermore, I don’t know what energies of gamma rays dark matter might produce, since — newsflash — I don’t know what the hell dark matter is and therefore don’t know what it annihilates into, or with what available mass-energy. Also, there are a lot of astrophysical sources of gamma rays, so dark matter could be hiding behind them.

But the rate of dark matter hitting dark matter goes up if there is more dark matter around. In fact, it goes up as the square of the number of dark matter particles in a region (the square of the number density), since you need two particles to collide. Thus, you can win in the search for dark matter if you take the data from Fermi in regions where you know there’s lot of dark matter. This boosts the signal, and hopefully reduces the astrophysical backgrounds. You also need that region to be close, since the further away it is, the dimmer any gamma-ray signal would be (just like a light source appears dimmer as it moves away from you). The intrinsic “brightness” of a mass of dark matter is encompassed in a number called the $J$-factor, which encodes the boost from having lots of dark matter around, and how far away that dark matter is.


You may have noticed a problem here: I need to know where to look for dark matter to find dark matter. But I’ve never seen dark matter, that’s what this indirect detection is all about. But I do know about dark matter from it’s gravitational interactions. Dark matter is what pulls stars around, so I can use that to figure out where the dark matter is.

The most dark matter lives in the Galactic Center. The $J$-factor there is something like $10^{21}-10^{24}$ in the unhelpful unit of GeV$^2$/cm$^5$. This is the brightest source of dark matter in the sky. Unfortunately, the Galactic Center is also the brightest source of astrophysical background around. Nevertheless, people have looked there, and they’ve even seen something unusual: an excess of gamma rays peaking around 1 GeV in energy that isn’t explained by known sources. This was originally found by Dan Hooper and Lisa Goodenough in their analysis of the Fermi data (NASA requires all data to be public, so if you want, you can download the data set and go hunting) so I call this putative dark matter the Goodenough-Hooperon, Fermi has seen this signal too, so there clearly is something there. The problem is we can’t quite tell if its dark matter or something more pedestrian: like gamma rays emitted by electrons and positrons accelerated by pulsars.

As an aside, I feel like it’s important to remind people that the “boring” option for this signal is that antimatter is being produced by the rapidly spinning corpse of a dead star, accelerated to insane energies by magnetic fields so intense they'd kill you from AU away, shattering the atoms of interstellar gas to create a storm of super-X-rays, which are then thrown across ten thousand light years.

But that’s not dark matter, so its so boring, you know?


To try to disentangle the signal from the possible backgrounds, I wanted to look for dark matter indirect detection somewhere else. After all, if the Goodenough-Hooperon signal is dark matter, we’ll see it elsewhere. Typically, the next best place to look for dark matter after the Galactic Center are what are called dwarf galaxies. These are little satellite galaxies of the Milky Way, maybe 1/100000th the mass of the Milky Way. But they have a decent amount of dark matter, and very few stars, so their background is lower. The $J$-factor for these varies, but the very best dwarfs (not dwarves) have $J$ of $10^{19}$ — which implies anywhere from 100 to 100,000 times less dark matter signal than from the Galactic Center.

 Large and Small Magellanic clouds near the Milky Way. Photo from Miloslav Druckmüller

Large and Small Magellanic clouds near the Milky Way. Photo from Miloslav Druckmüller

People have looked in the dwarfs for dark matter, and their negative results have put some pressure on the Galactic Center signal, but its not ruled out yet in my opinion.

So, where else to look?

Well, if you live in the Southern Hemisphere, when you look up at the night sky you see the two things in the photo I'm showing here (not the comet). You see the Large and Small Magellanic Cloud.

 Rendition of the Milky way and large and small magellanic clouds. From NASA/CSC/M. Weiss.

Rendition of the Milky way and large and small magellanic clouds. From NASA/CSC/M. Weiss.

These are two satellite galaxies of the Milky Way. Though they are smaller than the Milky Way itself, they are not dwarf galaxies. The Large Magellanic Cloud (LMC) is 1/10th the mass of the Milky Way, and the Small Magellanic Cloud (SMC) is 1/100th the mass. From what we can tell, these two galaxies are in a complicated dance with each other, and are infalling towards the Milky Way for the first time. Which is great, because that means they haven’t lost their dark matter to the Milky Way yet. They are also not that far away (as these things go), so any dark matter signal would be fairly bright. I show an artists rendition of the Milky Way and Clouds relative size and location here. We on Earth are located about halfway along the Milky Way disk more or less in the side with the Clouds.

Interestingly, simulations suggest that the Milky Way is very unusual to have two such satellite galaxies around it. Something like 1 in 10000 unlikely.

So why had no one looked at these two targets yet? Well, they’re real galaxies. They have astrophysical backgrounds. So they’re complicated targets, and you need to work hard to subtract the background to get at any signal. But a back of the envelope calculation indicated that the Galactic Center signal might be visible in the LMC or SMC if we were lucky, so we forged ahead.


The first problem in both the LMC and SMC analysis was to know where the dark matter is in the Clouds. This was difficult, since the data was a bit sketchy. But Alyson Brooks is a simulator of dark matter in galaxies. So she could use her simulations to compare the LMC and SMC to computer generated galaxies which looked similar, and use those and the data that existed to fit the dark matter density in the Clouds. Here’s what that looked like for the SMC.

From this, we estimated that the $J$-factor for the LMC would be somewhere between $10^{19.5}$ and $10^{20.5}$, and the SMC was $10^{19}$. This is actually the very conservative assumption (which is what you need to do to set a limit). There could be a lot more dark matter in the center of these two galaxies, we just don’t know, though simulation suggests it is possible. If there is, then the type of analysis we did would become much more powerful. Unfortunately, we can’t tell with the current data. One thing I’m very excited about is the possibility of new astrometric surveys to tell us more about how stars are moving in the Clouds, and therefore how the dark matter is distributed. That would go far to answering these questions.

So we have the dark matter, now we need to know where the background gamma rays are coming from. I really want to thank Pierrick Martin, who is an astrophysicist working in Fermi and constructs gamma-ray maps of the Clouds, and who was willing to work with a theoretical physicist who keeps blabbering about dark matter. Using the Fermi data, he was able to construct a map of the LMC and SMC in gamma rays, and then build up a map of the gamma rays we expect from “normal” physics. Unfortunately, this is data-driven, so if the dark matter gamma rays happen to look like the background, we can miss something. Fortunately, from our simulations, we know what the distribution of dark matter should look like, and therefore we know what the shape of the dark matter signal should look like (another advantage over dwarf galaxies, which are point objects for Fermi. The Clouds are extended objects, allowing us to use morphology to distinguish signal and background). So both the spectrum and shape of the background would have to conspire against us. Which is still possible, but somewhat less likely.

Here’s the map of the SMC in gamma rays. Not too complicated. The LMC is much bigger and therefore more difficult to map out. The gold stars are known point sources of gamma rays, mostly supernovae remnants (I spent a lot of time complaining to one of my colleagues about a particular supernova remnant in the LMC, which was just in the wrong spot and could be obscuring our signal. He affected some degree of hurt: he’d written one of the early papers on that particular remnant. One person’s background is another’s fascinating physics.) The green star is a new point source we found in this study. The bright source in the center left is a star-forming region known as 47-Tuc. The blue circle is the kinematic center of the SMC. Any dark matter signal should be centered on this point.

 MAp of the SMC in gamma rays, fermi data. From Caputo et al.

MAp of the SMC in gamma rays, fermi data. From Caputo et al.

Now we have the signal expectation, we have the backgrounds, so we can look for any amount of signal over background. Since we don’t know what mass dark matter is, or what it annihilates into, we have to test a bunch of possibilities, as each predicts a different spectrum of dark matter gamma rays. This is really the meat of the paper and the work that went into it, and Regina, Eric, Alex, Jenny, and (other) Matt were all invaluable to this effort. I’m skimming over it a bit here, but doing this right is hard.

I’ll skip to the final results. We didn’t find dark matter (damn). Since we were being conservative, our limits are not as strong as the dwarf galaxy search, or the LMC search we did first. Again, I’d love to revisit this with more accurate maps of the LMC and SMC, because I think the possibility for a big improvements is there. But that data doesn’t exist yet, and we’d need to expand the collaboration to find the right people who are experts in astrometry.

Here’s the propaganda plot for one set of our limits. The vertical axis measures the rate at which dark matter would annihilate (moving up means more annihilation). The horizontal is the mass of the dark matter, and here we assumed that dark matter is annihilating into pairs of $b$-quarks.

The black line is the SMC limit, the red the LMC limit. Dark matter annihilating with larger rates than the lines is excluded, since we would have seen more gamma rays than we did. The region of interest for the Goodenough-Hooperon are the various ovals (from various searches done by non-Fermi teams). The blue line is the current best limit from dwarf galaxies. It appears to rule out the Galactic Center signal, though I think there’s still room for quibbling. You can see that our conservative limits aren’t competitive at the moment, but again, we’ve demonstrated that you can use complicated galaxies like the Magellanic Clouds to do precision searches for dark matter. We still don’t have the answer, but the search continues.

 Results plot from Caputo et al.

Results plot from Caputo et al.