Image: the event which has sent the particle physics community into overdrive, indicated by the yellow arrows. Plot as presented by Sam Eriksen at the TeVPA 2026 conference on behalf of the LZ collaboration.
On 16 June 2023 at 3:22pm and 39 seconds local time, at the bottom of a disused mine deep in South Dakota, something extraordinary happened.
That something was so unlikely, so unexpected that it set the particle physics community aflame with speculation, when it was publicly disclosed by the LUX-ZEPLIN collaboration on 1 September this year, 29 days ago, in a talk at the TeVPA conference. In that time, hundreds of pages of scientific papers have been written on the topic, special national-level meetings have been organised, presentations hurriedly prepared, frantic emails and WhatsApp messages composed. And I dare say, hundreds of arguments already have been had. All for one event, three and a half years ago, which may turn out to be of no significance at all; or may be the first hint of one of the greatest discoveries in the history of particle physics. This post is dedicated to explaining my current understanding of what is going on.
The LUZ-ZEPLIN experiment.
Images: (left) the LZ detector in real life. (right) A visualisation of the flashes of light produced when a Dark Matter particle interacts with the Xenon within. Images courtesy of LZ UK https://lz.ac.uk/lz-experiment/how-it-works/
LUZ-ZEPLIN (affectionately known as LZ to its friends) is a direct Dark Matter detection experiment, one of the most sensitive ever created. It is basically an enormous tub of multiple tonnes of incredibly dense matter (liquid Xenon with a gas layer at the top), in a strong electric field, surrounded by light detectors, at the bottom of a deep, abandoned mine in South Dakota.
The idea is the following: if a Dark Matter particle collides with one of the Xenon atoms, it will give it a little bit of its energy… Enough to excite the Xenon atom: this means its orbiting electrons will go up one energy level. It will then de-excite, meaning that it will release a photon: a tiny flash of light, which is measured by the light-sensors around the experiment. We can this flash S1.
The same DM interaction will also typically knock one electron out of the Xenon atom. That electron will travel upwards in the electric field, and when it hits the gas phase it will create a second flash of light. That’s S2.
We can locate the position of the interaction in the Xenon tub by the pattern of light recorded from. S2 (basically it will be directly above where the interaction took place). We can locate the depth at which the interaction took place by measuring the time between S1 and S2 (we known how fast the electron travels in the electric field). And most importantly, we can tell what energy was deposited in the interaction (the Recoil Energy) by the ratio of the number of photons measured in S1 and S2.
So what’s all the fuss about?
Image: the plot of S1 versus S2 (in number of photons) for all the events recorded in a year of data taking between spring 2022 and spring 2023.
The event in question can be seen in the plot above, the little black dot indicated by the yellow arrows. All the other black dots (other recoded events) lie on the blue curve which represents the expected background for this experiment: basically residual radioactivity from the cavern and instruments, which follow a known pattern.
The event of interest is nowhere near those backgrounds. It lies in a region where no known experimental backgrounds could reasonably populate. In the paper, the LZ authors go to grain pains to explain all the potential “conventional” explanations for such a measurement. None fit. No known source of background could have caused such an event.
Once all other explanations have been discounted, the remaining hypothesis is that this event was caused by a Dark Matter particle colliding with the Xenon.
Now you can probably see why this single flash of light (well, ok, two flashes S1 and S2) have completely bewitched the community. As I often write in my research proposals… Understanding Dark Matter and its relationship with the Standard Model is one of the most urgent questions of fundamental physics today. This is a key goal of several fields, including high energy physics, cosmology and astrophysics. Dark Matter’s effects have only ever been observed gravitationally, and a direct detection would have a transformational effect on the field and point the way to further discoveries. It would be, in simple terms, an incredible breakthrough.
And an obvious question arises: if this event really is Dark Matter, could we observe at ATLAS or CMS too, providing an irrefutable slam-dunk and earn someone a Nobel Prize ? We will get to that in a second, but first…
But hang on a second, how significant is it, statistically ?
A measurement in physics is only ever as good as its uncertainties. This case is no different. There are two things to consider: systematic uncertainties from imperfections in understanding of the instrument, background or methods; and statistical uncertainties which are to do with the amount of data collected.
The systematic uncertainties translate into the question: how many background events did we actually expect with that combination of S1 and S2 ? the answer is “basically zero”, but with a small uncertainty. Nevertheless, even accounting for the uncertainty, the expected background is still significantly less than one, by orders of magnitude.
The statistical uncertainty is another matter: one event even in a zero-background experiment is not enough to guarantee a discovery. There is still a chance that it could be a statistical fluke. Put together, the significance of this observation is about 3.4σ, reduced to about 2.5σ once the Look Elsewhere Effect is accounted for. That means, roughly speaking, and if there were no mistakes in the LZ analysis or missed backgrounds, that there is a roughly one in forty chance that this event could be a statistical fluctuation.
That may sound significant but, in particle physics, we usually set the threshold at less than one three hundred for “evidence” of a signal, and less than one in three and a half million for a discovery. So we are not there yet.
But LZ have another batch of data left to analyse and relate, of roughly the same size. And other competition experiments could also probe this region with comparable datasets. It would probably only takes three events in a tiny background experiment to reach the discovery threshold.
What do we actually learn about what Dark Matter could be from that event?
This question is of paramount importance to the collider community, us at ATLAS and CMS. If we can make an observation ourselves, this would be a “smoking gun” (incontrovertible evidence), and would give us a complementary handle to study any discovery that is made.
So, what can we learn from that one event about what sort of Dark Matter we might be faced with, and what future discoveries it could lead to ? Actually, quite a lot.
The first thing is that the measured recoil energy is, well, unexpected even for simple Dark Matter theories. The recoil energy for the LZ event is about 250 keV. keV means kilo-electron volts: that’s the energy needed to move one electron by one meter in a 1000-volt electric field. For a collider physicists like myself, It’s really small, since in collider physics we normally work in GeV or TeV (billions or trillions of electron-volts). But for a direct dark matter experiment it’s a lot. That’s because we normally assume there is a single Dark Matter particle interacting with the Xenon, and if you figure out the typical recoil energy for that process, you realise that 250 keV is way on the upper end of what is kinematically possible. In other words, if you saw a recoil energy that high, then statistically you should have seen a lot more already with lower recoil energies, basically populating the rest of the S1/S2 plane along the red line in the plot above. Clearly we don’t see that.
But there is a way out. Let’s imagine there are two dark matter particles, one slightly heavier than the other. Then you can imagine a lighter dark matter particle flying in, interacting with the Xenon, and being turned into the heavier dark matter particle in the process. Since you need to borrow some of the energy to create the mass of the heavier dark matter particle, the effect is to shift the recoil energy window upwards, which would explain why we don’t see any lower- energy events.
Ok, so now we have two dark matter particles in our model. What else can we learn ? Well, two things: first, that the difference in the mass of the two particles cannot be that large, actually, probably not much more than the recoil energy. So the two dark matter particles would have only a difference of a few hundred keV in mass (oh yeah, we measure energies and masses in the same units in particle physics, see Natural Units).
The other thing we can learn is the absolute mass scale of these particles: we can figure this out by determining what the most likely speed of the dark matter particle was when it collided with the Xenon. If you assume that the Dark Matter in the Milky Way is roughly stationary (which is the picture we seem to get from gravitational measurements), then you can say that the speed at which the solar system is rotating around the galactic center is a sort of minimum value. It’s like the solar system (and hence the earth) are in a “wind” of Dark Matter. We know that value, it’s about 200-300 km/s. The upper limit is given by the fact that if Dark Matter travels too fast, it escapes the Milky Way entirely (a bit like there is a speed at which rockets can leave earth’s orbit). That’s about 500 km/s. So, in short, we know Dark Matter should be colliding the Xenon at most at something like 700-800 km/s. We can then work out what Dark Matter masses can give us the observed recoil energy assuming that collision speed. It turns out that really light Dark Matter can’t do it. In fact, it seems like the most likely value would be something like 1 TeV.
At this stage, that’s basically all we can extract from that one event, but it’s already not bad. And another thing… TeV may sound familiar to you. LHC beams collide at 13 TeV.
Enough suspense already, can we see it in ATLAS or CMS???
Image: a hard choice to make… with my apologies to Jeff Shahanian.
Ok, so we have two pieces of information: two (presumably neutral) particles, each around 1 TeV in mass and with a mass splitting of a few hundred keV. Of course, other ways to read the event are possible, but this one is the simplest. It’s also been widely discussed, because Supersymmetry can naturally accommodate such a configuration. And hence many theory papers came out with Higgsino interpretations early on. However, basic TeV scale Higgsino models are already sort of ruled out by cosmic ray experiments like IceCube… But there are several ways to get this sort of particle content, including dark QCD and others. Regardless of the specific model interpretation, we can consider the phenomenology we might expect at the LHC.
If you’ve been reading my blog long enough, or followed my particle physics lectures, you might know that small mass differences mean suppressed decay rates. In other words, the heavier dark matter particle would be a Long Lived Particle. My first reaction: Holy Sh*t I was right all along! Not so fast. A tiny (keV) mass splitting means a REALLY long lifetime. Even if you produced the heavy dark matter particle in LHC collisions, it would travel all the way out of ATLAS or CMS before decaying. All you’d see is the missing energy. And even if it did decay in the detector, the amount of energy released would be minuscule: the resulting photon or lepton or gluon would not be reconstructible (insufficient energy by an order of magnitude for our trackers or calorimeters to measure).
Ok, but who said that the dark matter had to be electrically neutral ? Well, this is unlikely because otherwise you’d probably see more activity in the LZ Xenon on the way in and the way out. But ok, there are theoretical ways out. A generic prediction of many new physics models is the existence of five Higgs-like bosons: three of which are neutral and two electrically charged. Assuming the known Higgs is the lightest, perhaps you could say that the LZ ones are the other two neutrals. And maybe we could spot the charged ones at the collider.
What you’d see then is one of two things: a charged particle track being created at the collision point and then “disappearing” as it decays into another neutral Dark Matter particle. Or conversely, a track which seems to appear out of nowhere.
If all else fails, we still have the missing energy strategy: even if you can’t see the Dark Matter directly because it decays outside of ATLAS, it would carry away a lot of energy. You can spot the “Missing Energy” out by trying to balance the energy recorded in the rest of the ATLAS detector. You still need have something to balance against, a typical choice being a jet or photon from initial state radiation. This has been a generic search strategy in ATLAS for years (the famous “mono-jet” searches), and yours truly even made a measurement the missing energy plus jets spectrum a few years back.
So now what? We wait
In summary, what should we look for in ATLAS? well, pretty much the same stuff we were already looking for, but with renewed vim and vigour. We have ongoing displaced and disappearing track analyses. We have ongoing mono-jet-plus-missing energy searches. In our analyses so far, we did not spot any major excesses. But perhaps with the oodles of data we are analysing now for the full LHC Run 3, or even the larger datasets we will have in the next decade, we could make a corroborating observation. In the meantime, LZ will be analysing the rest of their data in the next couple of years, and we should get responses from the other direct dark matter experiments.
So, that’s my understanding so far. It’s exciting, but it’s one event. It’s for sure set our community ablaze, but we need to be patient. If nothing else shows up with the release of more data from LZ and similar, it’s probably game over. However, if another event pops up in future data releases at consistent recoil energies, or if ATLAS/CMS spot something, it’s game on. It could go either way at this point.
On a personal note, it feels like a shame that this pops up just as my convenership of the Exotics search group ends (my last day is tomorrow…more about that in the next post), but on the plus side I get to take part in the adventure as an analyst, in the trenches doing the real work !
My apologies to my friends in the direct dark matter detection community if I have misrepresented anything about your work. Corrections gladly implemented on request 🙂