Direct Detection of Dark Matter

This podcast script comes from the episode first released on June 30, 2020.

Direct Detection of Dark Matter

Hi all, and welcome to this episode of Physical Attraction.

A little bit of background here. I’m currently working on some scripts for our next series, but it will take a while for them to be ready. However, one of the topics that we were going to touch on has been in the news lately, and I got a listener request to discuss it in more detail, and I saw a very good lecture on the topic at Oxford a year or so ago by Professor Elena Aprile, which a lot of this discussion is going to be based on — so this is going to be a one-off episode which will deal with this specific topic, slightly out of context, and when the series comes out we’ll deal with it in a lot more detail later on.

So what I want to talk about, specifically, is some of the experiments that are going on around the world right now to try and indirectly detect dark matter. But to do that, we need to quickly explain what dark matter is.

What is dark matter? Dark matter is matter that you can’t see because it doesn’t interact with electromagnetic radiation. Consequently it does not absorb, reflect, or emit electromagnetic radiation of any frequency, which is why we can’t see it with our telescopes, and can only infer that it is there based on how its gravitational pull affects normal matter. But, of course, that barely answers the question.

Essentially, dark matter remains something of a mystery. Astrophysical explanations suggest that perhaps up to 85% of the actual mass in the Universe is simply unaccounted for. We know that it has to exist, because if it didn’t, galaxies would fly apart; the cosmos would be expanding at a different rate; and we would be unable to explain our observations of how galaxies rotate within our currently-accepted — and highly verified — theories of gravity.

If you allow yourself to introduce a large amount of matter that can’t be seen, though, the mathematical models work, and we can explain our observations more easily.

I’m not going to go into too much detail here, but it does remain one of the rather crazy facts about astrophysics. We think only around 1% of the matter in the Universe consists of stars. Perhaps another 7% is visible gas that forms in haloes around galaxies. Another 7% is in the intergalactic medium — that vast, sparse plasma of atoms and electrons that fills the space between galaxies. But the overwhelming majority, 85%, of the mass in the Universe is dark matter.

Lots of people have found this philosophically unsatisfying — it’s almost as if we’re just filling in all of the gaps in our existing theory with some convenient ingredient that we can’t see but the fact is that most alternative theories — such as modified theories of gravity — require things that are just as arbitrary and even more unverifiable to be true for everything to hold together.

Without going into too much detail, though, there are several different lines of evidence that have lead the consensus that there must be some kind of dark matter to be pretty strong.

We have precision cosmology → COBE, WMAP, Planck — these satellites that have looked into the cosmic microwave background radiation, the leftover radiation from early on in the Universe’s history. This allows us to infer our how much of the universe’s energy must be gravitating matter, how much must be dark energy pushing the expansion of the Universe, and so on.

We also have the work by Vera Rubin, who noticed a discrepancy in how galaxies were rotating. Typically if you have a rotating galaxy, you can figure out how fast the rotation should be based on how much mass is towards the centre of the galaxy. Of course, this is because the gravitational pull acts as the centripetal force for the rotation. Typically you would expect the rotational velocity to change as you go further out in the galaxy, due to the different amount of mass that’s enclosed by the rotating stars towards the edge.

Vera Rubin (1928–2016)
=> studies of rotational velocities of galaxies

=> rotational velocity profile not at all understood from the visible disk
=> instead, rotational velocity remains constant

=> appears to be 10x more mass that’s not visible than visible mass to explain rotational profile
=> Similar things discovered in earlier discoveries of superclusters.

An even more obvious point of difference is the gravitational lensing; general relativity. Due to the way that general relativity works — with matter bending spacetime, and hence bending light — we can see distortions in far-off galaxies that we know are down to this effect of gravitational lensing, and the strength of the lens depends on the amount of mass between us and the object we’re looking at. So, again, this gives us an independent way of determining how much mass is between the distant object and us, and yet another confirmation that there must be more mass in the galaxy and in the universe than we can actually see.

All of this means that we’re fairly confident that some kind of dark matter must exist — that it probably clumps together in haloes, and that it exerts gravitational influence on stars, galaxies, and the fabric of space-time in general to explain our astronomical and cosmological observations. But obviously, it would be nice if we could actually detect this dark matter and learn a little bit more about what 85% of the actual matter in the Universe is made up of.

There are plenty of different theories as to what dark matter could be, but one theory that has been popularly explored amongst physicists are the WIMPs, or Weakly Interacting Massive Particles.
Why is that?

Well, there are lots of supersymmetric and theoretical extensions to the standard model for the particles that exist in the Universe where you naturally predict that Weakly Interacting Massive Particles exist. [For more, listen back to our series on particle physics, Concealing a Hadron.]

I think the main reason to hope that dark matter is made up of WIMPs is because it would be pretty helpful if it was. There are four fundamental forces; the strong nuclear interaction, which applies to baryons: electromagnetism, gravity, and the weak nuclear force. We know that whatever dark matter is, it can’t interact electromagnetically, or we could see it. We suspect that it can’t be made of baryons, and it won’t interact strongly, because of what we know about baryons. We know that it interacts gravitationally; that’s how we can detect it. So you have to hope that it might also interact via the weak force — and then we have another way to study it and understand it, and possibly detect it directly.

If WIMPs are really the source of dark matter, then we might have some reasonable hopes for detecting them. Depending on the mass of the WIMP, we know that there would have to be a lot of them about.

If you take some fairly reasonable estimates for what the WIMPs might be like, then you might expect that billions of them could be passing through your body every year. Of those, most will pass straight through, but perhaps ten in a year could interact with a nucleus somewhere in your body. You don’t notice, of course, but if these figures are true, then we could put a detector in place which might possibly be able to detect some of these interactions directly, and determine if WIMPs truly are dark matter.

One of the first papers that talked about this was the 1985 paper called “Detectability of certain Dark-Matter Candidates”, by Witten and Goodman. The fundamental idea here was pretty simple; hope that the WIMPs are relatively massive; and as the WIMP scatters off a nucleus, you may not be able to measure the WIMP itself, but you can measure the recoil in the nucleus.

One issue here is really nailing down exactly how many of these particles you might be expecting to see. That way, you know how sensitive you’re going to need to build your detector, how big it will have to be, and what the signal is that you’re really expecting to find with this direct detector.

And that depends on all kinds of things. If the WIMP is heavier, you obviously need fewer of them to make up the dark matter that we know is in the Universe. We also need to know how much of that dark matter is in the milky way.

And finally, we need to know something called the interaction cross-section. In essence, this is really a probability. It has units of area, so you can think about it as the “surface area” that a particle has to hit for a particular interaction to take place. But it simply measures how likely the particle is to interact with the target.

So the expected number of WIMPs that you’ll find will depend on the number of particles in the Milky Way halo, their velocity distribution (integrate over that, like a Maxwellian), the number of nuclei in the target, the mass of the WIMP, and the theoretical interaction cross-section.

The issue is that the theory doesn’t constrain this interaction cross-section much — because we don’t know all that much about what WIMPs are like! So, if you’re not detecting them, then it simply might just be the case that WIMPs interact less often than you hope they do. As dark matter detectors have got more and more sensitive, they have effectively ruled out types of WIMP that are happy to interact more and more regularly, and constrained this interaction cross-section. We now know that if WIMPs exist at all, they have to be pretty unlikely to interact with ordinary matter.

If you make a few assumptions — for example, assume the mass of the WIMP is 50GeV [GeV are units of energy, but since E=mc² for particle physics we can use energy and mass interchangably. A proton is around 1 GeV in mass, so this would be about 50x heavier than a proton.] If you assume a cross-section, sigma at 1e-46 m², which is in the sort of range that we haven’t necessarily probed yet, then you expect there would be 15 events per tonne of detector per year that will have more than 10keV energy transfer — i.e. in the range that we might detect them.

And now you can see why the search for WIMP dark matter has been so difficult. Not only do we not really know what we’re expecting to find, and so there’s not really a characteristic signal to look for that we can unambiguously say would be dark matter, but if we have a target made of Xenon [we’ll talk more about types of detector later] then even if you have 1,000kg of Xenon target, with these parameters for the WIMP which are considered quite reasonable, you’d only expect to see 15 events in a year! That’s an extremely low rate. And observing just one event isn’t going to be enough; you’d want plenty to build up significance and assure yourself that you’re really seeing something that genuinely could be a dark matter signal.

And, as with many physics experiments these days, the real challenge of the experiment — and the real triumph if signals are detected — is correcting for background noise. To say we’re looking for a needle in a haystack is unfair. We’re looking for an atom of the iron that makes up the needle in a haystack. There are huge amounts of background noise which could potentially produce similar recoil signals.

For example, there are 30,000 radioactive decays in our lungs every hour, 15 million potassium atoms decaying inside you every hour from bananas and food, natural radioactivity in the Earth and in the buildings that surround us sends 200 million gamma rays through us every hour. Intrinsically, the background of events that can potentially interfere with your detector and mimic an unusual signal.

Consequently, the process of making these detectors is very difficult. They must be very massive and able to distinguish the very weak WIMP signal from a huge background noise.

To reduce this noise as much as possible, every element in the detector process is screened as much as possible to reduce the radioactivity! It can take months to screen them. This can make it extremely difficult to arrange things with manufacturers — if you want some new wiring, a new metal case for the lab or something, you need to have it made to incredible standards of low radioactivity which is not something people usually consider. Trying to identify the lowest-activity material appropriate to build the detector is extremely difficult, which is why many different types of detector have been tried.

Neutrinos and neutrons are both quite large problems as they can behave exactly like WIMPs in the detector. You’ll remember that neutrinos are these ghostly, neutral particles which, by and large, pass entirely through the Earth without interacting with ordinary matter. The process for detecting a neutrino is quite similar to the dark-matter detection methods: you generally have an extremely large detector, weighing many tonnes, which aims to find the rare interaction between the neutrino and conventional matter. But filtering out a neutrino background is not particularly easy; a hundred trillion neutrinos pass through your body every second, mostly produced in interactions that are occuring in the Sun and streaming towards Earth alongside photons of light — and almost all of them travel directly through the Earth and pass through the other side. There’s not much you can do to shield against that.

So essentially, the race to detect WIMP dark matter directly with one of these detectors has always had this innate limit: the “neutrino floor”. There’s a certain limit beneath which you’re looking at the neutrino flux background and not WIMPs, and you won’t be able to distinguish any WIMP signal from the neutrino interactions in your detector.

For quite a while, this was a theoretical limit. But as the direct dark matter detectors have grown more and more sensitive, there has been a deep concern that they may well reach the neutrino floor without finding anything. As the detectors grow more and more complicated, and new generations of device are constructed, they can rule out lots of different properties for dark matter. We now know that, if it is exists at all in the form of WIMPs, it’s going to have to be a kind that interacts with ordinary matter very rarely.

You may be asking yourself: how do we know that the dark matter signal is above the neutrino floor? The answer is that we simply don’t. There’s no real reason to expect that it might be. We build our detectors in hope, rather than expectation. It’s perfectly possible that dark matter does exist as WIMPs, but it interacts so rarely that we’ll never be able to distinguish it from this background noise of neutrinos — at least, not with the kind of experiments we can concieve of today. This was what struck me when I watched the talk — realising how close the experiments have come to the neutrino floor in recent years. It’s an incredible achievement on behalf of the experimenters. But it seemed, at least a few years ago, that it might be that they had all spent decades producing brilliant experiments that merely conclusively ruled out finding detectable dark matter. And that would not be particularly satisfying at all.

So it’s worth saying that a VAST ARRAY OF DETECTORS TRIED!

Bubble chambers → charged particles forming tracks of bubbles
Noble liquids → not much radiation, looking for recoil
Cryogenic bolometers → incredibly cold detectors which heat up marginally due to the interactions
Scintillating crystals → the idea being that any interaction sets off a big oscillation in the crystal, which can be detected more easily

Directional detectors → to see if there’s any difference in the directionality of the detection. One of the reasons you might want this is to point the detector in the direction where the solar neutrino flux is the weakest, so your background noise effect is as weak as possible.

These detectors have names like SIMPLE


etc, etc., etc.

So when you imagine these detectors, you have to essentially imagine some kind of unusual setup. It might be in an underground laboratory a long distance from anything else to cut down on the radioactive background signal. It might consist of many tonnes of cryogenically cooled material, or a carefully-arranged crystal lattice, or a huge liquid xenon gas chamber. Deep underground, the detector waits, silently, passively, for years — hoping for that rare, fleeting blip which could be the signal of a ghostly dark-matter particle alerting us to its presence by a very uncommon interaction with visible matter.

So for many years, the field has extremely competitive, but over recent years (from 2010–2020), liquid xenon detectors are getting better and better. To illustrate what this means, we think back to that interaction cross-section for a 50GeV WIMP. Over recent years, since 2000, these have been probed from 10^-40 /cm² to 10^-47 cm/2. In other words, they have got ten million times more sensitive over the last 20 years, and they are approaching the neutrino floor at around 10^-49. The detectors are now so sensitive that they can detect one event per tonne of detector material per year, which is just a crazy statistic that illustrates just how these detectors are working.

What happens if we hit this neutrino floor? Well, technically, we still can’t rule out the idea that WIMPs are the solution to the dark matter conundrum. All we will know for sure is that we can’t detect them directly in this way, because this coherent neutrino scattering signal will get in the way. Then the direct hunt for WIMPs is basically over.

So Professor Aprile described her work on the DARWIN, the dark matter wimp search with liquid xenon, which will look for the next-generation, explore entire region for heavy wimps to Neutrino Floor. In the next decade or so, she expected to either find detectable WIMPs, or rule them out altogether. [This talk was a couple of years ago.]

So, why are Xenon detectors leading the way over everything else? Professor Aprile explained this to us — obviously she is a little biased because she actually invented this method for detecting dark matter! Essentially, as a reminder, xenon is a noble gas. It’s the heaviest if you exclude radon. Radon is radioactive so it’s obviously a terrible material for this kind of detector. It has a large nucleus, which is good for heavy WIMPs, as your scattering will be detected more easily. When liquified, it’s a dense liquid, which is good for a heavy, but compact and homogenous target — remember, you want to try and get many tonnes of this material together to maximise your chances of observing the scattering event. The cryogenics are easy for xenon; there are mature technologies can liquefy it and keep it very cold for a long time, so you don’t need to worry about that. And finally, if you did have a WIMP interaction, you might expect two different kinds of signal.
The xenon might be ionized, i.e. an electron is knocked off the xenon atom. That would produce a light signal when another electron recombines with the xenon, which you can detect. And there’s also scintillation. In simple terms, you can imagine the WIMP hits in this interaction, the nucleus recoils, this can ionise the nucleus (charge signal), or it can send the atom into an excited state and cause photons to be emitted. There would also be a signal of heat from the elastic scattering of nucleus. The heat itself can’t be detected yet in this kind of detector, but that’s the sort of thing that some of the crystalline detectors are looking for. It just so happens that xenon as a target would produce a really large number of photons per interaction, so you get a particularly strong light signal when you use xenon if one of these interations take place.

The type of photon you might get if xenon was excited, and then de-excited, in such an interaction is a particular line at 178nm. That’s in the high-UV spectrum. We know how to detect this pretty well, but manufacturing the photo-tube which can detect photons at this wavelength is again a challenge because it can’t be radioactive.

Some other advantages to the xenon detector are that the process of producing it generally doesn’t have much background radioactivity. Once you remove trace amounts of krypton, there’s no radioactivity in the xenon sample. And it only costs around $2000/kg to produce, which is substantially cheaper than gold, which allows you to make a slightly bigger detector on a budget. Finally, the mean free path of Gamma Ray is only a few cm. In other words, a gamma ray won’t make it too far into the material without being absorbed.

So considering the evolution of this particular set of detectors, the first was the Xenon10, which was 5kg of xenon with 1000 background events / tonne keV day. Then the Xenon100 which was 34kg with 5.3 events/tonne/keV/day. The newest detector was finished in 2018 — the Xenon 1 tonne device, which is 1000kg with just 0.2 events / tonne kev day in the background. So it’s not just about building endlessly bigger detectors, but instead reducing the background that arises from the materials as much as you possibly can. And they’ve reduced it to an incredibly small rate.

There is now some work on building the Xenon N Tonne detector, which would take us right down to the neutrino floor, basically. It will be built in the Gran Sasso lab, which you may remember was the one that gave that faster-than-light neutrino signal some years ago. It’s not as deep as Snolab or the Jinping Lab in China, but deep under the mountains, which makes it a good place to finish these things.

Describing this collaboration a little bit, there are 170 scientists in the XENON collaboration across various different universities, lots of postdocs and grad students, from all around the world. One advantage that Professor Aprile pointed out about this is that the size of the detector is relatively small compared to something like the LHC, which allows you to contribute to virtually every aspect of it — something like CERN is a much larger collaboration, where you will only ever work on your tiny bit and will never know all of the other people working on the same project.

Back in 2018, when Professor Aprile gave this talk, they didn’t have an awful lot of exciting results to show from Xenon 1 Tonne. None of the events they had detected could be a WIMP. What they were looking for was a three-sigma excess — in other words, something that’s very unlikely to have occurred simply by chance. But at that time, the results were just consistent with the hypothesis that WIMPs don’t exist — so all they had done is rule out a bunch of different possible WIMPs, in terms of their potential mass and cross-section to interact.

But there were some cheering results. They had found that their device was better than competing devices, meaning it remains a state-of-the-art dark matter direct detection unit. And they also did observe the rarest decay on Earth ever seen in a detector. This decay is called the two-neutrino double-electron capture in 124 — Xenon.

So what happens in this incredibly rare interaction is that two protons in the xenon atom spontaneously, and simultaneously, absorb electrons — in electron capture. Those protons become neutrons, and so you have a new element formed — tellerium.

Now this is pretty amazing here. The half-life for this interaction is 10²² years. In other words, a trillion times the age of the Universe. If you had an atom of Xenon sitting around, you would have to wait a trillion times the age of the Universe for it just to have a 50/50 chance of this interaction taking place. [Obviously they have many more than a trillion atoms in place, which is how they are able to observe this in a few years.]

So what this showed, at the time, was that the detector was capable of seeing very rare events, even though it was not necessarily designed to observe them. Unfortunately, though, they had actually seen too many of these events to be convinced that they were from WIMPs — the signal they’re actually looking for is much smaller than that!

So that was the story when I saw this talk a couple years back. The collaboration has built this incredibly impressive detector, capable of detecting these incredibly rare events. But they hadn’t found what they were looking for, which was evidence of dark matter.

Now, though, that may have changed, and this is the story that caused me to dig out my notes from the talk and write this episode! Because Xenon1T has found something that looks like a signal — but if it is a signal, it’s not from WIMPs, but instead possibly from axions, a competing dark-matter candidate particle.

Quoting from the brilliant Natalie Wolchover at Quanta Magazine:

“The physicists who run the world’s most sensitive experimental search for dark matter have seen something strange. They have uncovered an unexpected excess of events inside their detector that could fit the profile of a hypothetical dark matter particle called an axion. Alternately, the data could be explained by novel properties of neutrinos.

More mundanely, the signal could come from contamination inside the experiment,” due to tritium atoms that have not been discovered yet. The collaboration wants to finish work on the Xenon N tonnes experiment which should start operating later this year to be sure that the signal is not from this contamination.

“If this turns out to be a new particle, then it’s a breakthrough we have been waiting for for the last 40 years,” said Adam Falkowski told Quanta, a particle physicist at Paris-Saclay University in France who was not involved in the experiment. “You cannot overstate the importance of the discovery, if this is real.”

So we’ve been discussing how Xenon1T was hoping to find WIMPs uisng these nuclear recoils, but as they built successively more sensitive detectors, they weren’t seeing any excess of these events. Professor Aprile is actually quoted in this article describing the whole process as a “desperate saga.” But instead what they’ve actually detected is an excess of electron recoils. From Quanta again:

“As the WIMP search kept coming up empty, XENON scientists realized several years ago that they could use their experiment to search for other kinds of unknown particles that might pass through the detector: particles that bang into an electron rather than a xenon nucleus.

They used to treat these “electronic recoils” as background noise, and indeed many of these events are caused by mundane sources such as radioactive lead and krypton isotopes. But after making improvements to dramatically reduce their background contaminations over the years, the researchers found that they could look for signals in the low-level noise.

In their new analysis, the physicists examined electronic recoils in the first year’s worth of XENON1T data. They expected to see roughly 232 of these recoils, caused by known sources of background contamination. But the experiment saw 285 — a surplus of 53 that signifies an unaccounted-for source.

After rejecting all possible sources of error they could think of, the researchers came up with three explanations that would fit the size and shape of the bump in their data plots.

First and perhaps most exciting is the “solar axion,” a hypothetical particle produced inside the sun that would be similar to a photon but with a tiny amount of mass.

Any axions produced recently in the sun couldn’t be the dark matter that has shaped the cosmos since primordial times. But if the experiment has detected solar axions, then it means axions exist. “Such an axion could also be produced in the early universe and then would make up some component of dark matter,” said Peter Graham, a particle physicist at Stanford University who has theorized about axions and ways to detect them.

Researchers said the energy of solar axions inferred from XENON1T’s bump doesn’t fit with the simplest models of axion dark matter, but more complicated models can probably reconcile them.

Another possibility is that neutrinos — the most mysterious of the known particles of nature — might have large magnetic moments, meaning they’re like little bar magnets. Such a property would allow them to scatter with electrons at an enhanced rate, explaining the surplus of electronic recoils. Graham said neutrinos possessing a magnetic moment “would also be very exciting since it indicates new physics beyond the Standard Model.”

But it’s also possible that trace amounts of tritium, a rare hydrogen isotope, are present in the xenon tank, and that their radioactive decays generate electronic recoils. This possibility “can be neither confirmed nor excluded,” the XENON1T team wrote in their paper.”

So, unfortunately, I don’t want to rain on anyone’s parade when it comes to this story, but as we generally find with these experimental physics stories that are really pushing the boundaries of our ability to detect things… when you get a unusual or unexplained signal, it’s quite often down to some experimental error or background that you didn’t account for, and that’s what you should assume it is until you’re very convinced otherwise. Quanta again points out that the axion theory, and the neutrino theory, both seem to be contradicted by our measurements from astrophysics:

“Outside researchers say there are “not red, but orange flags,” as Falkowski put it, that point to the boring answer. Most importantly, if the sun creates axions, then all stars do. These axions pull a small amount of energy away from the star, like steam carrying away the energy of a boiling kettle. In very hot stars like red giants and white dwarves, where axion production should be greatest, this energy loss would be enough to cool the stars down. “A white dwarf would produce so many axions that we wouldn’t see hot white dwarves around today like we do,” said Zurek.

Neutrinos with large magnetic moments have been similarly disfavored: In comparison to standard neutrinos, more of them would be spontaneously produced inside stars, sapping away more of the stars’ energy and cooling down hot stars more than is observed.”

So I think most people, if you asked them to bet, would suggest that this new result is most likely down to tritium contamination and background radiation in the detector which is not yet accounted for. The alternative theories seem unlikely, and of course, this detector was built for WIMPs and is now being observed to try and detect these other possible phenomena, so you can’t be sure that they have cancelled out or fully considered these background events, when what they are looking for is WIMPs.

The story of the Xenon detectors in the last few decades is still fascinating and very much worth telling in my mind, though, because in the next few years, we will know for sure whether our leading dark matter candidate is going to be possible to detect or not. If it’s undetectable, at least in this way, that is still a result — and will probably motivate a whole heap of alternative theories as to what dark matter might be made up of, hopefully with some testable predictions that will allow the hunt for this mysterious 85% of the Universe’s matter to continue.

The mystery of dark matter might still be outstanding, but the mystery of this particular signal, luckily, is due to be solved pretty soon. When the Xenon N tonne detector goes online later this year, the scientists involved in the collaboration suggest that they will only need a few months of data and analysis to determine whether there’s really something there for this unexplained signal — and, at the same time, it will also essentially either find WIMPs or rule out the idea of ever finding them within a few years. So, in some ways, this is the last-chance saloon for direct dark matter detection. And it’s a reminder, as much as we know about the Universe, there’s still so much we don’t know, cannot say, and which may remain unexplained for generations to come.

Thanks for listening etc.