Space station’s detector has not found dark matter, despite what some media reports say

On April 5, 2013

AMS detector on the ISS

Let me start with this: the AMS-02 detector on the International Space Station has not detected dark matter. It hasn’t found “indications” of dark matter, or even “hints.” It certainly is not providing the “best evidence yet” of dark matter’s existence.

What AMS has done is measure, to very high accuracy, the amount of antimatter the galaxy is bombarding us with. It might be coming from dark matter… but really there’s no compelling evidence that it is. And it’s certainly too soon to break out the champagne.

So what is AMS really telling us? The AMS experiment detects cosmic rays — protons, electrons, and the antimatter counterparts of each, antiprotons and positrons. Before the experiment ran, we had predictions of how the matter/antimatter fraction should vary with the energy of the particles.

AMS tells us our predictions were wrong. The antiprotons look about right, but there’s a huge excess of high-energy positrons over what astrophysical models predict, and a bump in the electron flux at high energies. All of these results were actually seen by earlier experiments PAMELA and Fermi, but AMS confirms them to higher precision and higher energies. There’s more antimatter than we thought; now we have to figure out why.

Dark matter

Here’s where dark matter comes in. One thing several models for the dark matter particle have in common is the idea that it might be its own antiparticle and thus be able to annihilate with other particles of its kind. (It’s a quirk of particle physics that a particle can be its own antiparticle, which always strikes me as a bit weird and kind of lonely, but there you go.) This annihilation produces particle-antiparticle pairs, which then interact with other particles through a particle cascade, and at the end you can get electron-positron pairs. Voila! There’s your positron excess! Problem solved!

Only, maybe not. Dark matter annihilation isn’t the only thing that makes positrons. And to make an excess of positrons without an accompanying excess of antiprotons, your dark matter model has to be… rather specific.

Basically all we can say is that there’s some kind of high-energy process out there that we haven’t accounted for. But if it’s not dark matter, what can it be?


The most popular non-dark-matter interpretation of the positron excess is that it comes from pulsars, which can use their extreme magnetic fields to accelerate particles and create electron-positron pairs. The fact that pulsars do this is solidly in the realm of known physics, and theoretical models can easily fit the signals seen in the cosmic ray experiments.

So why consider dark matter at all? Why not just give up and say, “well, it’s probably pulsars, we can all go home”?

What if…?

While it’s tempting, accepting the pulsar hypothesis would be premature. Pulsar models are highly uncertain, and there could even be some other astrophysical process contributing to the problem. The data can’t yet tell the difference.

So what if it really is dark matter? What kind of dark matter would that be?

There are two major conditions for the dark matter model to fit the data.

1) It has to reproduce the positron flux. This requires the annihilation cross-section — which determines how direct a collision needs to be for annihilation — to be much larger locally than in the early universe. There are models that can get both annihilation rates right, but they need a boost: more and denser clumps of nearby dark matter, or a so-called Sommerfeld enhancement to increase the annihilation rate, or some kind of resonance.

2) While providing the positron excess, it has to ignore antiprotons. Most theories would produce excesses in each, so we have to make the model discriminate. One way is to suppose the particle only annihilates into electron-like particles. Another is to use a short-lived intermediary particle that is itself lighter than a proton. The dark matter particles annihilate into the intermediary, and since you can’t decay into something heavier than yourself, the intermediary can only decay into light things like electrons and positrons.

There are models that meet these criteria, and it’s basically a matter of taste whether you prefer them to more “traditional” models such as the supersymmetric neutralino, since the data can’t really tell the difference. On the one hand, you might have to invent a new force in the dark sector to explain the positron excess. On the other hand, sticking with the bog-standard neutralino might not be so great either — supersymmetry is under heavy pressure from the fact that the LHC so far hasn’t seen it. Maybe it’s a good time to think outside the box. Really, we just need more data, using different kinds of physics (gamma rays, direct detection, colliders) to give us some experimental handle on the theoretical possibilities.


As pointed out by the Resonaances Blog, the AMS press release mentioned dark matter nine times. They said the signal was consistent with dark matter, but “not yet sufficiently conclusive to rule out other explanations.” Unsurprisingly, several media outlets ran with this, announcing that hints of dark matter have been found, or that this is the first evidence for dark matter that we’ve seen. Neither of these statements is true. While intriguing, the AMS signal is not a “hint” of dark matter — it’s an anomaly that needs to be explained. And it’s certainly not the best evidence for dark matter we have. Check out the Bullet Cluster, or any number of extremely compelling astrophysical results that, however indirectly, make it abundantly clear that dark matter is real, and abundant.

Another misconception making the rounds (and encouraged by the press release) is that we are really close to confirming the signal is actually dark matter. The story goes that all we need is a sharp cut-off in the positron flux at energies just above what we’ve measured, and that will be a “smoking gun”; pulsar emission would only have a gradual drop. This is wrong on two counts. One, it’s not necessarily true that the dark matter cut-off would be sharp. Models that use an intermediary in the annihilation, such as described above, can have a much softer drop at high energies. And two, some models of pulsar emission also have a very sharp drop. The shape of a cut-off at higher energies will be interesting and useful, but will by no means definitively answer the question.

Another thing being held up as a sure sign of dark matter is the fact that the signal appears isotropic, coming equally from all parts of the sky. Contrary to what some media outlets are saying, dark matter isn’t distributed evenly throughout the universe — it traces regular matter, and in our galaxy is most dense in the center. But charged particles lose so much energy interacting with matter and radiation in their paths there that none of the positrons from the galactic center would make it to us. Even worse, charged particles are strongly deflected by the galaxy’s complex magnetic field, which makes it impossible to know with much accuracy which direction they came from initially. So while it’s true that the local dark matter density is close to constant, we’d expect the signal to be basically isotropic anyway. And the magnetic field means positrons from pulsars — even just one or two pulsars very nearby — would also appear close to isotropic. The directional data isn’t good enough to tell the difference yet, and might not be for the duration of the experiment.


We don’t yet know what the AMS signal comes from, but we know it’s something interesting. I wouldn’t say it’s necessarily new physics or astrophysics, but it’s intriguing nonetheless. Personally, I think it’s probably pulsars (partially because that’s a more conservative view), but I’d be extremely happy to find out it was dark matter all along. We just need more data. And in the meantime, theorists like me are going to have a lot of fun with the mystery.

Image: The International Space Station, space shuttle Endeavour remains docked with the station and the Alpha Magnetic Spectrometer-2 when it was newly installed. Credit: NASA

About Katie Mack

Katherine (Katie) Mack is a theoretical astrophysicist working at the University of Melbourne as a postdoctoral fellow on an ARC Discovery Early Career Researcher Award. She studies dark matter and the growth of cosmological structure in the universe, developing new ways to shed light on the early universe and fundamental physics using astronomical observations. She is also an active science communicator, participating in a range of science outreach programs, and has written for Sky & Telescope,, and the Economist tech blog, among others. She blogs about cosmology at The Universe, in Theory.

12 Responses to Space station’s detector has not found dark matter, despite what some media reports say

  1. Great article. It’s good to hear somebody who knows what they’re talking about countering some of the press-release hype and nonsense that finds its way into our newspapers unchallenged.

  2. Thank you, very clear and above all informative.

    It is clear, I think, that the AMS result is not even an observation, even less the extraordinary evidence needed for an extraordinary claim of DM annihilation observation. And it is probably not what theorists would consider ´”hint” (somewhat confident observation).

    On the other hand I think the current hype of “no DM” goes too much in the other direction. The source model seems to be compatible with a DM annihilation source, it (the model, not the data) certainly has the required cutoff that I don’t think has been sen before. As long as it is a competitive model (and as source model it was robust, predicted with different energy range data and also other experimental data, perhaps even Fermi-LAT if you allow adjusting the p/e ratio) it is an outsider’s “hint” (somewhat competitive models).

    As for supersymmetry being “under pressure”, that is only correct if you want it to be TOE “natural” and/or solve the hierarchy problem, right? Unless LHC surpasses Planck energies (we wish), I assume you can squeeze supersymmetry into the gap. (Disclaimer: My interest is astrobiology, hence secondarily cosmology. Not particle physics as such.)

  3. Thanks for your comments!

    Torbjörn: Supersymmetry isn’t an area I’m really working in, but my understanding is that the space of models currently allowed is starting to get more restricted, and those more in the know than me are starting to look around at alternatives. Here’s a nice article discussing all the ways supersymmetry could still be out there: and here’s another (slightly more detailed) post, that goes into more detail about some of the models people are thinking about:

    There’s certainly a lot of room for supersymmetry to still work out, and it may be that we’ll never see it with the LHC even if it’s the right model. As a theorist, though, I’m always keen for reasons to start exploring new kinds of models and solutions to long-standing problems, so the lack of any evidence for supersymmetry seems like as good an excuse as any to think a bit outside the box. Most importantly, though, I think we need to make sure that any new theories we come up with are straightforwardly falsifiable (and we need to come up with new ways to test existing theories), so we don’t end up with a huge number of new models, all consistent with the data, and all distinguishable only by theoretical aesthetics. I think this happens sometimes with inflation models, and it doesn’t seem to do much to advance the field.

  4. Katie:

    Re out-of-the-box thinking and inflation, I look to relativity. Hence I favour something akin to infinite gravitational time dilation where the early universe expands at its own sweet pace, but to observers within the universe, either on the scene or at a later epoch, the early expansion appears to be almost instantaneous. IMHO it’s worth checking out papers by David Wiltshire in Canterbury NZ and David Wands in Portsmouth:


    Also see “Reality check at the LHC” on physicsworld along with Peter Woit’s blog:

    Personally I think the issue with supersymmetry is that it proposes a superpartner for say the electron without a clear understanding of the electron itself. I don’t know if you know anything about topological quantum field theory, but there’s papers out there that describe the electron as an electromagnetic standing-wave “structure” something like the blue DNA torus on this biological physics web page: . The spin ½ is represented by the moebius-like twist. With an electron model like this, it’s difficult to see how the electron could possibly have a spin 0 superpartner. Overall, I’d say the right approach is to complete the Standard Model before proposing “beyond the Standard Model” hypotheses.

  5. Nice discussion, Katie!

    The first report of the AMS team seems to indicate more missing data than ‘missing mass’! As I understand, while the AMS team only reported results for for positron detections at energies up to 350 GeV (72 @ 260-350 GeV), ostensibly due to concerns over the number of detections at higher energies and their statistical relevance. However, since the assertion is that the nature of the fall-off in detections will be the characteristic that indicates whether the positron emission source is dark matter annihilations, one wonders who is being protected by this omission… Please see Table I,

    If the nature and distribution of all background detections sources cannot be definitively determined, how can the fall-off in high energy detections be evaluated to indicate then number of detections attributable to DM annihilations? I presume that if the fall-off of detections at high energies might be steep if all detections were produced by DM annihilations, but how steep will the fall-off be if only 10% of high energy detections are from DM annihilations?

    As often happens, the identification of dark matter seems to be dependent of the results of models constrained by the results of models constrained by… underlying assumptions. Better stated:
    “There are significant modeling uncertainties in all of these scenarios, though, such as how particles propagate through the Galaxy, the mass of the dark matter particles, and their interaction rates. But the ideas are attractive and remain fodder for lively speculation.”

    Finally, you mention that “… dark matter isn’t distributed evenly throughout the universe – it traces regular matter, and in our galaxy is most dense in the center.” Isn’t that the conclusion of particle physicists considering that an exotic dark matter particle must effectively interact with baryonic matter only through gravitation?

    It seems there is a pervasive, fundamental disconnect between particle physicists and astrophysicists here since, as I understand, the primary requirement for galactic dark matter is that it produce observed rotation curves using established methods of gravitational evaluation.

    As I understand, the general distribution of DM required to produce observed spiral galaxy rotation curves is ‘cuspy halo’ that extends to very large radii, well beyond the galactic disk. Please see

    On the other hand, I’m not aware of any explanation of any property of exotic DM could allow it to maintain the necessary ‘cuspy halo’ configuration…

  6. Katie Mack, “And to make an excess of positrons without an accompanying excess of antiprotons, your dark matter model has to be… rather specific. …
    Maybe it’s a good time to think outside the box. …
    While intriguing, the AMS signal is not a “hint” of dark matter…”

    Your statement above is still the most accurate one after 6 months went by. Today, the dark matter and the dark energy mysteries are no more. They can be easily resolved with half page calculations, as we do have Planck data (dark energy = 69.2; dark matter = 25.8; and visible matter = 4.82) and the AMS02 data now. Since these are simple numbers, the *predictions* of a model for these numbers are very straightforward, no debate can come about.

    For dark matter, with the Pimple model (that is, every particles carry the same mass-charge, see ), there are 48 matter particles (matter + anti-matter) while only 7 of them [the first generation matter (not anti-matter)] gives out lights (excluding e-neutrino). Thus, the dark mass/visible mass ratio = [41 (100 – w)% / 7] . The *w* is the percentage of the dark matter which does give out lights. According to the AMS02 data, it is between 8 to 10%. By choosing w = 9, the d/v ratio = 5.33 (while the Planck data shows d/v ratio = 25.8/4.82 = 5.3526). Details, (see ).

    For dark energy, it uses an iceberg model (see ). That is, the Time, Space and Mass (dark + visible) form an iceberg system, while the mass is the iceberg. And, they three take the *equal* share. So, the dark mass = [(33.3 – 4.82) x (100 -9)%] = 25.91 (while the Planck data is 25.8), with d/v ratio = 5.37. The 9% here is the melting ratio from the dark matter. Thus, the dark energy = 66.66 + [(33.3 – 4.82) x 9%] = 66.66 + 2.56 = 69.22 (while the Planck data is 69.2).

    One interesting thing here, the dark/visible ratio was calculated with two different pathways. Yet, the average [(5.33 + 5.375)/2] = 5.3527, exactly the same as the Planck data.

    With these calculations, the Nature is much simpler than we can ever imagine. Yet, numbers are numbers, and there is no debate-point for these calculations.

    • “… there are 48 matter particles (matter + anti-matter) while only 7 of them [the first generation matter (not anti-matter)] gives out lights (excluding e-neutrino). Thus, the dark mass/visible mass ratio = [41 (100 – w)% / 7] . The *w* is the percentage of the dark matter which does give out lights.”

      If I follow this correctly, you’re suggesting that it is your proposed number of different types of particles that determines their representation within the universe (i.e. ratio of dark matter:ordinary matter)?

  7. @James T. Dwyer, “…(i.e. ratio of dark matter : ordinary matter) …”

    I did not use the term *ordinary matter* in my previous comment but with *dark mass/visible mass*, as the dark mass arises also from the *ordinary matter*.

    Now, I would like to make one more point on this issue. The Planck data (dark energy = 69.2; dark matter = 25.8; and visible matter = 4.82) consists of *three* numbers, with an very interesting relationship (d/v = 5.3526). Yet, in my previous comment, this d/v ratio was calculated with two different ways.
    a. The pimple model (every SM particle carries the same mass-charge while its measured mass is only the pimple on that mass-charge) — the d/v ratio = 5.33
    b. The iceberg model (the total mass-land takes only 1/3 = 33.33 % of the total energy of this universe) — the d/v ration = 5.3755 and the dark energy is 69.22

    There are 4 interesting points on the above.
    i. There is a 0.85 % difference between the two d/v ratio.
    ii. The dark mass melting ratio is chosen with the same number (9%) for both calculations, based on the AMS 02 data.
    iii. The pimple model is in fact embedded in the iceberg model, that is, they are not disjointed.
    iv. These two models are deterministic models, no quantum consideration at all. This is another strong evidence on the *quantum algebra ( ). By adding the quantum effect, the 0.85% difference could be understood.

  8. We will make a new approach for an effect known as “Dark Energy” by an effect on gravitational field.

    In an accelerated rocket, the dimensions of space towards movement due to ‘Lorentz Contraction’ are on continuous reduction.

    Using the equivalence principle, we presume that in the gravitational field, the same thing would happen.

    In this implicates in ‘dark energy effect’. The calculi show that in a 7%-contraction for each billion years would explain our observation of galaxies in accelerated separation.

    Lorentz Contraction

    If we suppose that gravitational field contracts the space around it (including everything within), we can explain the accelerated separation from galaxy through this contraction without postulating ‘dark energy’.

    The contraction of space made by gravity would cause a kind of ‘illusion of optic’, seem like, as presented below, that galaxies depart fastly.

    The contraction of space would be equivalent to relativistic effect which occurs in a special nave in high-speed L.M.: With regard to an observer in an inertial referential stopped compared to a nave, the observer and everything is on it, including own nave, has its dimension contracted towards to movement of nave compared to a stopped observer (Lorentz Contraction).

    This means that the ‘rule’ (measuring instruments) within the nave is smaller than the observer outside of moving nave.

    The consequence is, with this ‘reduced rule’, this moving observer would measure things bigger than the observer would measure out of nave.

    An accelerated rocket and its continuous contraction

    In the same way, if we think of an accelerated increasing speed rocket, its length towards movement – compared to an inertial reference – will be smaller, and ‘rule’ within the nave will decrease continuously compared to this observer.

    We would think of ‘equivalence principle’ to justify that gravitational field would have the same effect on ‘rules’ (measuring instruments) as an accelerated rocket would do within the nave, but, now, towards all gravitational field and not, in the case of rocket, only at acceleration speed.

    I.e., the gravitational field would make that all rules within this field would be continuously smaller regarded to an observer outside of gravitational field and this would make, as we can see, these observers see things out of field be away fastly.

    Anyway, even if “equivalence principle” can’t be applied into a gravitational field to show that the space is contracting around it, we can take it as a new effect on gravitational fields and this would explain the ‘dark energy effect’.

    The “dark energy” through gravitational contraction:

    Let’s think what would happen if a light emitted by a star from a distant galaxy would arrive into our planet:

    Our galaxy, as well as distant galaxies, would be in continuous contraction, as seen before, due to gravity.

    A photon emitted by a star from this distant galaxy, after living its galaxy, would go through by an “empty” big space, without so much gravitational influence, until finally arrives into our galaxy and, lastly, to our planet.

    During this long coursed way (sometimes billion years), this photon would suffer few gravitational effect and its wavelength would be little affected.

    However, during this period, our system (our rules) would still decreasing due to gravitational field, and when this photon finally arrives here, we would measure its wavelength with a reduced ‘rule’ compared to what we had had at the moment when this photon was emitted from galaxy.

    So, in our measurement would verify if this photon had suffered Redshift because, with reduced rule, we would measure a wavelength longer than those was measured. The traditional explanation is “Shift for Red” happened due to Doppler Effect compared to galaxy separation speed!

    End of Dark Energy

    Farthest a galaxy is from viewpoint, more time this light will take to arrive us and more shrunken our ‘rule’ will be to measure this photon since it had been emitted; so it would be bigger than wavelength, which would induce us to think of faster galaxy separation speed.

    This acceleration (this new explanation, only visible) from distant galaxies took astronomers to postulate the existence of a “Dark Energy” would have a repulsive effect, seems like they are getting away faster.

    But if acceleration is due to our own scale reduction, this dark energy wouldn’t be necessary anymore, because what makes this separation accelerated is, actually, our own special contraction. This would be the end of dark energy.