Deep beneath the mountains of Gran Sasso in Italy, the world's most sensitive dark matter experiment has made a surprise detection. No, it's not dark matter. Instead, the experiment has detected significantly more particle interaction events than predicted by the standard model of particle physics.

Instead of the 232 ± 15 low-energy events expected in a year's worth of data, from February 2017 to February 2018, the XENON1T Dark Matter Experiment detected 285 - a whopping 53 more than the prediction, and well outside the error margin.

Excitingly, the large international team of physicists involved in the collaboration doesn't know what's causing the excess, even though they have been working on the results since 2018.

After careful consideration, they have boiled their options down to three possibilities: one fairly mundane… and two others that would have a huge impact on our understanding of fundamental physics.

The researchers presented their findings in an online seminar on June 17, and have prepared a paper that is currently in pre-print ahead of peer review.

"We observe an excess that's greater than three sigma, and we don't know what it is," said physicist Evan Shockley of the University of Chicago.

XENON1T is a tank filled with 3.2 metric tons of ultra-pure liquid xenon, and fitted with arrays of photomultiplier tubes. It's completely sealed and completely dark, in order to detect the scintillation and electroluminescence produced when two particles interact with each other, producing tiny flashes of light and a tiny shower of electrons ejected from a xenon atom - what is known as electron recoil.

Since most of these interactions occur from known particles, it's a relatively straightforward matter to estimate the number of background events that should be occurring. This is how the number 232 for low-energy electron recoil events was derived.

So, "whence the additional 53 events" is the big question.

The first, and most mundane, of the three scenarios that could have produced additional particle interactions is a previously unconsidered source of background events, caused by very small amounts of a rare radioactive isotope of hydrogen called tritium.

Tritium, the researchers noted, could have been introduced into the detector through the cosmogenic activation of xenon, and hydrogen in the detector materials themselves. It would only take a minute amount of tritium - just a few atoms for every 1025 atoms of xenon, way too small to be detected. Attempts to detect tritium by other means were fruitless, so the tritium hypothesis could be neither confirmed nor ruled out.

The second, more intriguing possibility is that the signal could be caused by neutrinos. These particles are similar to electrons but have almost no mass and no charge, and they interact with other particles very infrequently. This is just as well, since neutrinos are the most abundant particle in the Universe.

According to the team's calculations, neutrinos could be responsible for the excess signal if they had a stronger magnetic moment - that is, magnetic strength and orientation - than we thought. If these stronger magnetic moment neutrinos are responsible for the signal, we would very possibly need new physics to explain how they can exist.

The big problem with this scenario is that these neutrinos are produced in stellar cores (among other places), and would be produced in greater numbers in very hot stars, such as white dwarfs, from which they would extract energy, reducing the star's heat. We have yet to observe such heat loss consistent with the energy extraction of neutrinos with strong magnetic moments. This creates what the researchers call a "strong tension" with their own results.

The third scenario is a type of hypothetical particle called a solar axion. This is the best fit to the data, with a confidence level of 3.5 sigma - that is, a 2 in 10,000 chance that the signal is a random fluctuation. (The other two scenarios have a confidence level of 3.2 sigma.)

This would actually be huge, since to date we have not detected axions of any kind. Axions are a type of particle hypothesised in the 1970s to resolve the question of why strong atomic forces follow something called charge-parity symmetry, when most models say they don't need to.

Axions of a specific mass are a strong dark matter candidate. Solar axions, hypothetically streaming from the Sun, are not the same as the dark matter candidate axions, but would be a strong hint to their existence - if solar axions exist, other axions should also exist.

The problem with this scenario is very similar to the problem with the neutrinos. If the Sun can produce axions, so should all stars; and, once again, the observed heat loss in very hot stars places stringent limits on axion interactions with subatomic particles.

So, we're left with a dilly of a pickle, and one that's only going to be solvable with - you guessed it - more experiments. Since XENON1T is upgrading to its next phase, XENONnT, we'll just have to hold onto our hats for now.

"The signals discussed here can be further explored in the next-generation detectors," the researchers wrote in their paper.

"XENONnT, featuring a target mass of 5.9 tonnes and a factor of ∼6 reduction in ER background, will enable us to study the excess in much more detail if it persists. Preliminary studies based on the best-fit results of this work suggest that a solar axion signal could be differentiated from a tritium background at the 5 σ level after only a few months of data from XENONnT."

The pre-print paper has been uploaded to the XENON1T website.