A controversial and unconfirmed observation of dark matter made by the DAMA group in Italy may have an even stranger source than previously thought, according to physicists in the UK. Their research suggests that the signal seen by DAMA is neither from dark matter nor from background radiation. Instead, they say that the signal could be the result of a fault in the DAMA detector's data-collecting apparatus.

Starting in 1998, the DAMA-LIBRA experiment – nestled deep underground at the Gran Sasso National Laboratory in Italy – has reported an annual oscillation in the signal from its dark-matter detector. Some physicists believe that this variation is the first direct detection of dark matter and is a result of the Earth moving throughout the galaxy's halo of dark matter. Further data collected by the collaboration over the past 17 years has given the measurement a statistical significance at 9.3σ – well beyond the 5σ that usually signifies a discovery in particle physics. But apart from the CoGENT dark-matter experiment in the US, no other dark-matter searches across the globe have detected a similar effect, calling the claim of the first direct detection of dark matter into question.

Mimicking muons?

Last year, physicist Jonathan Davis, who was then at Durham University in the UK and is now at the Institut d'Astrophysique de Paris in France, developed a new model to explain the signal, suggesting that neutrons scattering in the detector could easily mimic the annual signal. According to Davis, these neutrons would be released when solar neutrinos and atmospheric muons scatter in the shielding material or the rock that envelops the experimental set-up. He pointed out that the rate of muons from cosmic rays decaying in the atmosphere varies across the year, peaking around 21 June; while the rate of solar neutrinos, which also varies annually, peaks around 4 January. Taken in conjunction, neutrons from both of these sources also have a rate that varies annually but peaks somewhere in between the two and can match the DAMA peak, which occurs in late May. While the idea of muons mimicking the DAMA signal is not new, the timing of muons in isolation does not match the DAMA data, and so the idea was previously dismissed. Davis's model solved this problem by adding the effect of solar neutrinos.

Now, Vitaly Kudryavtsev and Joel Klinger from the University of Sheffield in the UK have, for the first time, carried out the complete 3D modelling of muons and muon-induced neutrons at DAMA, taking into consideration the Gran Sasso mountain profile and the detector configuration, including all of the layers of its shielding. Kudryavtsev, who has been working in dark-matter searches for many years and has been involved in designing upcoming projects such as the LUX-ZEPLIN, told physicsworld.com that he and Klinger are well placed to carry out a full assessment of the background radiation in the NaI detectors used by DAMA. This has allowed them to establish unambiguously whether the modulated rate of events observed could arise out of any kind of background radiation. The researchers say that their results show conclusively that such a neutron flux induced by muons at Gran Sasso is too low by several orders of magnitude to mimic the DAMA modulation.

Minimal effect

Kudryavtsev says that while the DAMA collaboration itself has provided estimates of various backgrounds in their papers, no full modelling of the muon-induced background has been done so far. Davis's paper from last year also only provides a rough model, which was already refuted by another team of researchers last November, who said that the real effect of muon-induced neutrons and neutrino-induced neutrons is well below Davis's estimates and the DAMA signal.

Kudryavtsev and Klinger's results also agree, and indeed the two results taken together "indicate that it is very difficult to build any model to explain the DAMA signal. This conclusion is based on the analysis of the energy spectrum of events as measured by DAMA and calculated from [background] radioactivity", says Kudryavtsev. "The measured spectrum can be fitted reasonably well with the [background] radioactivity model leaving very little room to any additional signal, whether from dark matter, background or any instrumental effect," he adds.

Notoriously complicated

Davis has responded to the claims, saying that while the software package – which the duo has used to simulate how the neutrons get from where they are first produced, in the rock or shielding around the detector, to the detector itself – is commonly used in the community, such simulations are "notoriously complicated". Davis says that the current dark-matter models, which explain DAMA, are quite complicated and unnatural, so it is getting to the point where mundane models are the only option. "My opinion is that people are not currently creative enough with these models. The way these neutrons actually interact in the DAMA detector is not well understood, although this itself is a contentious issue, which some will disagree with. In these simulations the neutrons will just scatter in the DAMA detector, but it's possible that there are further interactions such as neutron capture (followed by the emission of a photon), which some have suggested in the past can enhance the rate of neutron events by a large factor. Essentially, things are not as well-understood as they appear," says Davis.

Davis also says that a key point in his paper is that "since the DAMA detector can't distinguish what is causing its scattering events, one doesn't really know what the events are. I suggested neutrons as a simple plausible model and somewhere to start off, but there are lots of alternatives". He says that the neutrons could be captured rather than scattered in the DAMA detector, or there could be another temperature-related source for the signal or it could even be something scattering with the electrons in DAMA instead of nuclei.

Basic requirements

"We do not think it is fruitful to discuss the period or phase of modulation of a background source without also discussing the amplitude of this source. There may be several potential sources of modulated signals and this is, of course, a disadvantage of the technique used by DAMA," says Kudryavtsev. He explains that all of the experiments currently at the forefront of dark-matter research "rely on a powerful discrimination between potential signal and backgrounds. DAMA does not have this option, and measures the total event rate and its time variations. However, any model that claims to explain the DAMA signal should be consistent with the measured amplitude of the signal". In addition, the duo says that such a model should satisfy some other requirements – that the amplitude of the effect must be very small compared with the DAMA event rate, that the modulation amplitude of the effect must not be much smaller than the average amplitude of the effect, and that the phase and period of the modulation must be predicted simultaneously.

"In particular, any signal summed with the predicted radioactive background should give the measured rate for all energies of interest. At the moment we are not aware of any model for an observed modulated signal that would satisfy these requirements," says Kudryavtsev. "Bear in mind that we know quite a lot about radioactivity and muons...certainly much more than we know about dark matter...and so any explanation of the DAMA signal will be very difficult, if we fully trust the statements made by DAMA about full control of temperature variations etc."

While Kudryavtsev and Klinger's model does not provide a plausible explanation of the DAMA signal, the pair are clear on the fact that they do not think it arises from dark matter either, because other experiments have already excluded this possibility and in any case the dark-matter signal is difficult to fit into the measured rate. "We believe that it is very difficult to explain the DAMA signal by any model, whether this is dark matter or a background. There is simply not much room in the measured event rate to accommodate the reasonably well-known radioactive background and the signal," says Kudryavtsev, adding that it may be "an artificial effect caused by the rejection of photomultiplier tube noise as implemented in the data analysis by DAMA. If the cut used to separate noise from real events somehow varies with time, then more events may be accepted as real events during certain periods, causing modulation in the event rate. This is, of course, just another speculation, and there is no proof of this being happening in the data analysis."

Davis himself believes that our best hope of understanding the DAMA modulation is other experiments, like the upcoming DM-Ice, which are looking to replicate DAMA. "Fundamentally, we can simulate all we like, but what we really need is more data," he says.

A preprint of the work is available on the arXiv server.