Carlo Fuselli

“Come sta il barile?” is the question I have been asked at every family gathering for the past few years. The barrel is the funny name my family has for the detector I have been working with for my PhD: the XENONnT experiment. My usual answer is that the barrel is doing fine, sitting under a rock with nothing to do. The truth is not so far from that, but in this thesis I will try to give a more detailed answer. The barrel itself is a tank filled with liquid xenon, buried under a mountain at the underground Laboratori Nazionali del Gran Sasso in central Italy, and sitting inside a larger tank of water for further shielding. It was built to catch a particle of dark matter: the invisible substance that makes up most of the matter in the universe, felt through its gravity but never directly detected. Dark matter is thought to fill our galaxy and drift through ordinary matter almost undisturbed. Because it barely interacts with anything, it should easily pass through a mountain of rock, and so we hope it bumps into one of the heavy xenon atoms in our detector. At the same time, plenty of other things keep interacting with the xenon. Cosmic rays are one source, though the mountain above the laboratory blocks most of them before they arrive. Neutrinos from the Sun are another source: each one interacts only very rarely, but the Sun sends so many of them that some strike the xenon anyway, and when they do, they can look a lot like what we expect from dark matter. But the biggest source is in the detector’s own surroundings. Ordinary materials, rock, the detector’s own components, are slightly radioactive, and even the liquid xenon itself is not perfectly pure: trace amounts of radon, a radioactive gas, are dissolved in it. Their atoms are unstable and decay one by one, emitting alpha particles, electrons, or gamma rays, and occasionally neutrons. The water tank around the detector stops most of the neutrons, but the rest of this radiation reaches the xenon constantly. Keeping it under control took extreme care in choosing and cleaning every material in the detector, making XENONnT one of the cleanest, quietest places in the world. When a particle interacts inside the xenon, two things happen: electrons are liberated from xenon atoms, and a flash of light is produced. Sensors around the tank see this flash right away, and we call this first signal S1. The freed electrons then drift up through the liquid under an electric field. A few microseconds later they reach the gas above the liquid and make a second, bigger flash, which we call S2. Together, S1 and S2 tell us where the interaction happened and how much energy it deposited. This process, and the details of the detector that make it possible, from the photomultiplier tubes to the electronics that reconstruct S1 and S2, are the subject of Chapter 2. Interpreting it precisely also means trusting a simulation of the detector. Chapter 3 describes fuse, the simulation framework that reproduces the entire process from a particle interaction to a digitised waveform, producing fake data nearly indistinguishable from the real thing. XENONnT is the latest step in a hunt for dark matter that the XENON collaboration has pursued for almost 25 years, through a series of detectors, XENON10, XENON100, XENON1T, each one bigger and cleaner than the previous one. The current version holds 8.6 tonnes of xenon and has been running since 2020, without finding dark matter yet. While waiting to discover dark matter, the same detector is also good at other things. Its very low background and its ability to measure S1 and S2 with high precision make it useful well beyond the search it was built for. In this thesis, we use data from the detector to do three things: to model how liquid xenon converts energy into light and charge, to measure a nuclear decay’s energy spectrum precisely enough to test different theories of nuclear structure, and to search for dark matter through an alternative interaction, absorption rather than scattering. All three analyses also rely on the same trick to fight background. In each case, the process we are after produces not one signal, but two: a first signal followed, a short and predictable time later, by a second one. Ordinary background essentially never does this, so requiring both signals, in the right order and within the right time window, removes almost all of it at once. This is usually called a delayed coincidence. One of those decay chains is particularly useful. An atom of 214Bi decays by emitting an electron, turning into 214Po, which itself decays a fraction of a millisecond later by emitting an alpha particle. This is the signature we use: whenever we see this pair, we know we caught a 214Bi decay, cleanly separated from everything else happening in the detector. Chapter 4 uses this tagged sample to work out how a given amount of deposited energy splits between the light and charge signals in liquid xenon, a relationship that previous models got wrong by a wide margin at these energies. Getting it right matters for every other measurement described in this thesis, since it sets the ruler everything else is measured against. With that ruler calibrated, Chapter 5 goes a step further and asks what the beta decay itself looks like in detail: not just how much energy it releases on average, but the full spectrum of energies the electron can carry away. That spectrum is not a simple curve. Its exact shape depends on the internal structure of the atomic nucleus in a way that is hard to calculate, and different nuclear models disagree noticeably on what it should look like. Chapter 1 lays out the nuclear physics behind this: why the shape is hard to predict, and what a precise measurement of it can teach us about the nucleus. Using the same tagged sample of clean 214Bi decays, we measured that spectrum with enough precision to tell the competing models apart. Chapter 6 turns to dark matter again, but from a different angle. Most searches look for a dark matter particle bumping into a xenon nucleus and knocking it slightly, the way one billiard ball nudges another. For very light dark matter, that nudge is too gentle to see. An alternative is absorption: the dark matter particle is not deflected but destroyed, handing its entire mass over to the nucleus in one go. In xenon, this can leave behind an excited nucleus that settles down shortly afterwards by emitting an extra, delayed flash of light of its own, on top of the original signal. Requiring that second, delayed flash strips away almost all of the ordinary backgrounds that would otherwise swamp the search. It is the same trick that made the beta-decay measurement possible, now applied to a different physics question. No dark matter signal was found. But in part of the mass range, the limits this search places on how strongly it could interact are around ten thousand times tighter than what was known before. The thesis closes with an appendix on the software: the infrastructure that processes, stores, and distributes XENONnT’s data, built and kept running over several years by many people across the collaboration. It gives a short description of the main software and infrastructure, followed by some personal reflections on the practicalities of running and using it. The barrel, it turns out, had more to do than sit there and wait.