Annual Modulation at CoGeNT

When talking about dark matter (DM), there's a standard line that gets used from popular talks through to journal papers: we only know about its gravitational interactions.  That is, we've measured its existence and abundance from how it affects galaxy rotation curves, or the structure of the cosmic microwave background; but we have no direct information about any other types of coupling it might have to the ordinary stuff we are made of¹.

Of course, there are a lot of searches of various types looking for those interactions.  One of the most basic is direct detection, building a very sensitive and low-background experiment and looking for dark matter scattering off the atoms in your apparatus.  It is here that one of the more enticing, puzzling and long-standing mysteries of dark matter is to be found; the fact that several experiments claim signals, that seem to be ruled out by other searches that found nothing².

My attention was drawn by the publication on the arXiv today of another paper in the signal column.

The CoGeNT experiment published their results from three years of data.  CoGeNT has published hints at a signal before.  Particularly interesting is that CoGeNT is one of two experiments to claim evidence for an annual modulation in their signal.  That is, the strength of the signal varies with time with a period of about one year.

The origin of this modulation is straightforward.  Because dark matter and stuff like us interact only weakly, the Earth, Sun and entire Solar System can be thought of as moving through a stationary background of dark matter particles.  The Earth, of course, moves around the Sun.  In June, the Earth is moving in the same direction as the Sun:

and in December, the opposite direction:

This means that the speed we hit dark matter particles is higher in June and lower in December.  There is more energy in the collisions in June, so we are more likely to see something going on in our equipment.

The CoGeNT data is quite compelling.  Even at the raw, unprocessed level we can see a definite variation in signal with time in one of the four channels:

While after removing certain known backgrounds, we can see this in all of them:

Particularly impressive is the top one, where we see arrows show the expected peaks.  They line up very well with the observed peaks, a promising sign!

There are, however, a number of concerns.  The fact that the other channels do not give DM-like signals is not a problem; the second plot may well be background, while the bottom two plots are at different energies, which potentially tells us something about the energy and mass of the DM.  However, the fact the other four channels do show some variation in their signal is a reminder that annual variations are very common on Earth.  We really need to be sure that there isn't some unknown background.  And the actual statistical significance of this result is not great; 2.2 σ, which can be produced by chance a few percent of the time.

The authors of this paper are well aware of this fact, and make the important point that we need to compare with other data.  This is the point of the arrows in the plot above: they are the expected peaks based on the only other DM experiment to see annual modulation.  That is definitely a good sign.

Against this is the mystery mentioned above: other experiments would seem to rule out the DM interpretation of these signals.  The CoGeNT team address this by considering that the distribution of DM in the galaxy might not be the simplest one usually assumed.  This question has been considered before, but never quite seems to work.  And something similar happens here:

This plot has the dark matter mass on the horizontal axis, and the coupling strength on the vertical.  The various closed curves are the preferred regions for this result (red) and other possible signals.  The other experiment that claims annual modulation is both dark grey and pink; this shows the effect of changing the DM density (and some other tweaks I'm not going into here).  With this change, the different loops nearly all overlap, another hint that they might have a common origin.

But those exclusions from experiments who found nothing are also shown on this plot, as the dotted and dashed lines.  The regions above and to the right of these lines are ruled out.  The LUX and XENON 100 lines exclude all of the regions preferred by signals, a serious challenge to the DM interpretation of these signals.

There are possible explanations for this discrepancy, beyond the obvious one that these signals are really an unknown background.  To go into that would take me beyond the scope of this blog post.  Let me just finish by saying that, despite the more negative tone I adopted at times, I find this result intriguing.  With the lack of anything unexpected from the LHC, signals like this are all the more hopeful; and the disagreement between experiments may point to surprises on the horizon in DM physics.

---

1 Given that dark matter is roughly five times as abundant as normal matter, why might call it the ordinary and us the rare stuff!^↩
2 One experiment, DAMA, claims a statistical significance of 8.9.  To put that into perspective, the probability of such a signal by chance is more than a millions times smaller than the Higgs discovery signal.^↩