Tuesday, 27 April 2010

More Trouble with DAMA

I haven't blogged about dark matter for almost 2 months, and already there is a pile of long overdue dark news. This post is about a couple of recent unpublished results that mean trouble for theorists trying interpret the DAMA signal.

Recall that the DAMA experiment has observed a few percent annual modulation of the recoil rate registered by their sodium-iodide crystal detector. This modulation could be due to a change of dark matter flux as the Earth moves around the Sun. However, other dark matter detection experiments (maybe except for CoGeNT) do not observe any signal, which puts strong constraint on the properties of the dark matter particle that could explain all available data. Vanilla-flavor models are by far excluded, however until recently two slightly more involved yet still plausible scenarios appeared marginally allowed:
  1. Weak scale inelastic dark matter. In this scenario a dark matter particle with mass of order 100 GeV scatters to an excited state with order 100 keV mass splitting. The inelastic scenario favors heavy targets (such as DAMA's iodine, A = 127), and enhances the modulation rate (only dark matter particles from the tail of the velocity distribution can scatter, so that small changes of Earth velocity can significantly change the available phase space).
  2. Light (5-10 GeV) elastic dark matter. This scenario favors very light targets (such as DAMA's sodium, A = 23) and experiments with low detection thresholds (such as DAMA's 2 keV), as light dark matter particles cannot give a large push to heavier target nuclei.
A few weeks ago, the former possibility was blasted by CRESST - yet another dark matter experiment under Gran Sasso mountain. CRESST uses CaWO4 crystals as the target, and detects scintillation and phonons to discriminate nuclear recoils (expected from dark matter particle) from alpha, beta, and gamma recoils (induced by ubiquitous backgrounds). The presence of tungsten (A = 184) in their crystal makes it very sensitive to the inelastic scenario. But the latest preliminary results presented at the Wonder2010 conference do not show a clear signal. Although CRESST has a handful of (most likely) background events in the signal band, the number of hits is much smaller than that predicted by the inelastic scenario consistent with the DAMA signal. The collaboration claims that the DAMA region is excluded by more than 3 sigma. This should be treated with a grain of sodium chloride as the CRESST data are not yet public and the assumptions that enter the derivation of the limits are not clearly spelled out in the slides. But most likely, the inelastic window is getting closed.

The explanation of DAMA via a 5-10 GeV dark matter particle is also facing problems. The (marginal) consistency of this scenario with null results from other experiments hinges on the so-called channeling effect in sodium-iodide crystals. Normally, an incoming particle recoiling against the crystal nuclei deposits most of the recoil energy in the form of lattice excitations (not observed by DAMA) while only a small fraction goes into scintillation (observed by DAMA). Channeling refers to the situation when an incoming particle gets caught along the symmetry plane of the crystal undergoing a series of small-angle scatterings and losing most of its energy via scintillation. Since a fraction of less energetic recoils can be detected thanks to channeling, the detection threshold of the experiment is effectively lowered. The effect is especially important for light dark matter because in this case the recoil spectrum is very sharply peaked toward lower energies. The channeling probability reported by DAMA is very large, of order 30 percent in the interesting range of recoil energies, which would greatly increase their sensitivity to light dark matter.

Given its importance you may expect that channeling in sodium-iodide crystals has been carefully studied by the DAMA collaboration. However, DAMA would not be herself if she dwelled on such trivialities. Instead the collaboration estimated the channeling probability using monte carlo simulations based on a theoretical model not applicable for the actual problem. Recently I came across slides from the Snowpac2010 workshop describing an independent attempt to estimate the channeling fraction using more reliable theoretical assumptions. The preliminary results contradict the conclusion of the DAMA collaboration: the channeling probability in sodium-iodine is negligible. If this is right, simple models of light dark matter cannot consistently explain the DAMA oscillation results.

Assuming that both of these preliminary results are true, we are confronted with an embarrassing situation: there is no plausible theoretical interpretation of the DAMA results. What remains on the market are rather exotic models (e.g. resonant dark matter) or Frankenstein models that patch up several non-trivial effects (inelastic+form factor, inelastic+streams, and so on). So theorists need to think harder. At the same time, the need to independently verify the DAMA experimental results becomes even more acute. Maybe a socially sensitive hacker could upload DAMA's raw data on WikiLeaks ;-)

14 comments:

Javier said...

It is wonderful that you make this post precisely today.

Yesterday in arxiv blog there was an entry about mirror dark matter: http://www.technologyreview.com/blog/arxiv/25100/#comments

There they discussed two papers:

http://arxiv.org/abs/1004.1424

and

http://arxiv.org/abs/1003.0014

I have read the first one and I find somewhat surprising the proposal. I thought that it was ruled out the possibility that dark matter could have an intricate structure (specially an standard model like one).

But I am newby to the business of very modern and detailed cosmology and dark matter in particular (your blog is my reference on the subject). Consequently I ask you about the reliability of those "mirror dark matter" models.

Javier said...

Ok, if it is not t much too ask I have another question.

If we discard DAMA we still have the CoGeNET results and the 2 sigma CDMS announcement.

I was wondering if some Kaluza-klein like dark matter (for universal extra dimensions) could fit that results.

My main motivation for it is that as far as I now that could be the easiest way to modify the F-theory GUTs of Vafa to fit that results.

Jester said...

I've never studied mirror dark matter models in details. I'd be surprised too if this was allowed. In any case, as far as i understand, in the context of DAMA this just an example of the light dark matter explanation, and suffers similar problems if there is no channeling.

As for KK dark matter, this is what I call vanilla-flavor WIMP. It cannot be used to explain CoGeNT (as to fit the CoGeNT data you need a 10 GeVish dark matter particle) and the 2 CDMS events also prefer a lighter particle.

zupan said...

Nice post! To the list of Frankenstein models you can probably add also the spin-dependent inelastic scattering that we found to work for DAMA and is not excluded by others in 0912.4264. It can originate from tensor x tensor interactions, but cooking up a realistic model that does it is really nontrivial.

Anonymous said...

I think it's a bit premature to take anything from CRESST too seriously:

"Unfortunately we were forced due to the wonder conference to show some data before our analysis was finished. We are still working on it and try to
understand what is going on. To deal with this problem we decided within the collaboration not to quote some important numbers needed for a data
interpretation( as exposures). Therefore I cannot answer any of your questions.
Sorry for that"

CRESST's 40 keV upper bound on scattering means that they are incredibly sensitive to DAMA's Iodine quenching factor. Drop QI from 0.085 (which is on the high side) to 0.06 and DAMA's signal goes flying out of the CRESST's signal region. If they cannot use the signal above the nuclear form factor zero (which has dozens of events), then they will not be able to test Iodine scattering in DAMA.

Anonymous said...

Nice post, but you may have missed one transparency in the CRESST Wonder stack (look carefully): the rumor in the street is that they see a signal consistent with DAMA and CoGeNT, precisely in the 5-10 GeV region, and the beginning of a modulation. No channeling invoked.

Jester said...

I didnt miss it. But other rumours say that the rate varies between different detectors.

Anonymous said...

Not what I heard, not what they say in that presentation... "Rate in all detectors equal within statistics"...

Jester said...

Then it's getting curiouser and curiouser...

Anonymous said...

All these experiments will have tremendous trouble demonstrating these are not backgrounds, but curioser indeed. Maybe CRESST will make an official noise soon?

holden said...

Your comment about the channeling effect
in NaI(Tl) seem not so honest. In the slides
from Gelmini and Gondolo
you link it is written that DAMA was the first Dark Matter
experiment that stress the importance of this effect
and take it into account.

The channeling calculation of DAMA have been
obtained in the framework of Lindhard's theory
and it is described in details in the paper arXiv:0710.0288
where also the implication are reported.
Gelmini and Gondolo have considered another model
and start some analytical calculation but they
said "Analytic results may not be enough, and we may
collaborate with other groups to carry out sophisticated
Monte Carlo simulations to confirm our results."

You jump to conclusion faster than the authors...
I understand the channeling is an argument interesting and open
and completely neglected for Dark Matter before DAMA paper.

All this is useful to understand that the comparison
of results obtained by different experiments is not so trivial
as one con think.

When you speak about DAMA seems resentful.
Inciting someone to commit an offence is off colour.
It is the 2nd time you write something about DAMA that
can sound slanderous.

Jester said...

My understanding is that DAMA's 0710.0288 assumes that channeling occurs when the recoil angle wrt to the crystal plane is less than some critical angle. This seems to be the right assumption when you shoot ions at the crystal. However, when the atom is kicked from a crystal lattice site, there is the issue of blocking which has not been taken into account.

Of course the results of Bozorgnia et al. are preliminary and it's hard to conclude anything before the paper is out. But put this together with the fact that channeling has never been observed experimentally at relevant energies, and that CDMS does not see any indication of channeling in their Ge crystals, and you get a hint that the effect might not be real. Given the importance of the effect for dark matter model building, I guess it's worth mentioning on a blog.

Anonymous said...

Holden, DAMA & co. quite simply messed up the treatment of channeling: what Gondolo and collaborators are doing these days is to look at it from precisely the point of view of Lindhard's theory, with all physical processes properly modeled, rather than departing from the wrong set of assumptions. Jester is just presenting the cold facts.
(Jester, according to Gondolo & co. CDMS would be blind to channeling, however small, because of its operating temperature)

Blake said...

LOL, oooh you better watch out Jester, you're "slandering" DAMA now with your wild insinuations! What is this group's deal anyway? What a bunch of drama queen flakes. Since we know you're here listening DAMA, why don't you REPLACE YOUR SCINTILLATOR to see if your signal disappears, hmmmm? I won't hold my breath.