Thread by @lzdarkmatter

Wondering how LZ works and what these new results mean? This is the thread for you!
#darkmatter 1/15

LZ has 3 detectors! The inner TPC, the dark matter (DM) target, is 7T of liquid xenon (LXe). The Skin, 2T of LXe, sits outside the TPC. The Outer Detector (OD) is 17T of Gd-loaded liquid scintillator. The Skin & OD veto backgrounds that could otherwise be mistaken for DM! 2/15

LZ started running in Dec '21. The first science run was 60 live days (live = detector is actively recording data) & results were released July '22 (WS2022). The new run (WS2024) was 220 live days. Given the 5.5T mass of LXe used in the analysis, that's 3.3 T-yrs of exposure 3/15

How does LZ detect particles? If a particle bounces off Xe, it excites it & light is emitted (S1 signal). It also knocks off some electrons. These e- are pushed to the top of the detector by an applied electric field, where in a gas layer they create more light (S2 signal). 4/15

Two types of "bounces" occur: on e- (electron recoil, ER), & on nuclei (nuclear recoil, NR). We can tell between them by the ratio of S1 to S2. The signals also pin-point the location of the recoil in 3D. This is helpful as most backgrounds are at the edge of the detector! 5/15

Most of what we see are electron recoils - these mostly come from radioactive decay - beta decays, and gamma-rays. Some are from neutrinos! Most of the time, these don’t look anything like dark matter, but a small fraction of them can - so we have to understand them well. 6/15

In WS2024, we used a new technique to help reduce Pb214 β-decays! We modelled the flow of the LXe, and used α-decays that precede Pb214 to predict where & when in the detector we might see the β-decay later on. This is one of our biggest backgrounds so this is a huge help! 7/15

We also studied and measured electron capture decays - where the nucleus absorbs an e- and then emits x-rays and Auger e-. Extra effort went into modelling them as we see that these decays are more NR-like (so have less S2) than β-decays, so can look more like dark matter. 8/15

We expect dark matter to cause NRs, but what else can? Neutrons do and can look just like DM, but the OD has a 92% efficiency at vetoing neutrons through their capture. Looking at what has been “tagged” by the OD (and the skin) gives us a strong handle on LZ’s backgrounds. 9/15

Another background is “accidentals”, where unrelated S1s & S2s occur close together & look like a real ER or NR. We target these with data selections, & study them by selecting those where the S2 is too far away from the S1 for it to be related ("unphysical drift time"). 10/15

As we are looking for a small signal on top of all these backgrounds, we must validate selections & mitigate bias. We do this via "salt" - fake NRs made from splicing ER S1s & S2s, injected into the data stream. Our analysis team can’t tell what is real and what is salt! 11/15

So, we’ve got our backgrounds, what now? We get our final dataset through a serious of “cuts”, selecting only the data that could possibly contain a dark matter signal. Once the cuts are finalised, the salt can be removed and in WS2024, 1220 individual events remain. 12/15

Now, it’s time for statistics. We feed in probability distribution functions in S1 vs S2 for our backgrounds & for dark matter signals. We can check the agreement of the data with the signal+background model. This is done several times over a range of dark matter masses. 13/15

Sadly, no signal was observed! So now, we set constraints on DM properties. This shows DM mass vs probability to interact with a nucleus; above the black line is ruled out, but DM can still be hiding below (& to the left, where constraints are weaker for lighter particles). 14/15

It might seem boring to see no signal, but learning about what dark matter isn’t helps tell us more about what it might be. For now, the mystery remains! We still have plenty more to learn from LZ, so stay tuned!✨ 15/15 🧵