Towards the formation of a forward-boosted beam of neutral antihydrogen for gravitational studies

The matter/antimatter asymmetry is one of the greatest mysteries in modern physics. In fact, the observable universe’s lack of antimatter poses a considerable problem for physicists, especially because the Standard Model predicts that the Big Bang should have produced identical amounts of matter and antimatter. The deviations observed experimentally in the production of matter with respect to antimatter, both in the leptonic and the hadronic sectors, are still insufficient to account for the matter’s domination of the universe. Therefore, one possible explanation for the discrepancy is that antimatter could interact differently than matter with gravity, resulting in a violation of the (Weak) Equivalence Principle. The Equivalence Principle, one of the pillars of general relativity, postulates that the gravitational charge of antimatter should be equal to its inertial mass, as it is for matter. However, some physics beyond the Standard Model (BSM) theories admit a difference in the gravitational properties of matter and antimatter [8]. Measuring antimatter’s gravitational force is a challenging task, both because of its scarcity and the complexities involved in trapping and handling antimatter particles. Indirect limits based on astronomical data have been posed [9], and lately, indirect measurements have progressed significantly [10]. Nonetheless, these indirect methods depend on theoretical models involving additional hypotheses: therefore they cannot be as conclusive as direct measurements. To date, the only direct measurement of the matter-antimatter gravitational interaction is from ALPHA [11]: while eliminating the possibility of “pure-repulsion”, its precision still lacks towards what is considered theoretically intriguing. This PhD thesis has been developed in the context of the AE¯gIS experiment (Antimatter Experiment: Gravity, Interferometry, Spectroscopy), which aims at measuring directly the acceleration exerted on antimatter by the gravitational field of the Earth and potentially by other gravity-like interactions. Specifically, antihydrogen has been selected as the test particle, as it is the simplest system of a neutral antimatter that can be synthesized1. To perform such a measurement, the methodology chosen is to create a pulsed beam of antihydrogen atoms, well defined in time, and let it pass through the grids of a moir´e deflectometer [13] while accelerating because of the influence of the Earth’s gravitational field. By the pattern created by the particles passing through the grids, the displacement due to gravity can be determined. For the formation of antihydrogen, AE¯gIS relies on the charge-exchange reaction between a positronium atom (Ps) (which is the bound state of an electron and a positron) and an antiproton: in the reaction, the electron is swapped with the antiproton, and the antihydrogen atom is thus formed. The AE¯gIS experiment is located at the European Organisation for Nuclear Research (CERN), in the Antimatter Factory (AD, from Antiproton Decelerator). The AD is the sole source of bunched trappable antiprotons in the world: they are a key ingredient for the formation of antihydrogen. The AE¯gIS experimental apparatus consists of multiple subsystems linked together, used in different combinations depending on the experiment’s needs (e.g. antihydrogen formation instead of positronium cooling). The principal piece of hardware is the main vacuum chamber, hosting the cryostat, the magnets system and the traps. The magnet system divides the traps into two zones, one at 5T and one at 1 T: in the first region, the capture trap is present, with electrodes tunable in the ±200V range and three dedicated electrodes going up to 15 kV, and it is used to capture the antiprotons coming from the decelerators, after being moderated by a material degrader. The high magnetic field is used to provide stronger radial confinement to the more energetic particles and to ensure a more efficient ¯p cooling via electron sympathetic cooling (which is proportional to B2). In the 1T region, instead, is present the formation trap: here the cold antiprotons are transferred from the capture trap, and the antihydrogen production takes place. A lower magnetic field ensures a higher rate of formation: in fact, the antihydrogen charge-exchange cross-section is proportional to the fourth power of the Rydberg level of the positronium involved (σ ¯H ∝ n4 Ps). But the maximum Rydberg level achievable by Ps is limited by the magnetic field, since it is bounded by the motional Stark effect induced field ionisation (nmax ∝ B−1/4) [14]. Therefore, a lower magnetic field in the formation region can enhance greatly the production rate of antihydrogen. To the main vacuum chamber, the positrons line is connected, which is used to inject positrons in the formation trap, to form the positronium atoms needed for the ¯H formation. It consists of a 22Na source emitting e+, moderated by a solid Neon moderator and stored in a Surko trap. To form positronium, a bunch of positrons are extracted and collided into the e+→Ps converter, consisting of a nanochanneled porous silica plate. A series of two lasers, then, is used to excite the so-formed Ps from the ground state to a high Rydberg state (1S →3 P → 17÷32). The cloud of Ps then expands towards the ¯p plasma, and ¯H formation can take place. The number of antihydrogens thus created is determined by observing, using the scintillators posed around the cryostat, the difference in annihilation rates arising when both positrons, antiprotons, and lasers are present, in opposition to the lack of (at least) one of them. AE¯gIS successfully produced cold antihydrogen in pulsed mode in 2018 [15], marking the end of its Phase 1. The formation rate was determined to be approximately 0.05 ¯Hper decelerator cycle, which lasts ∼ 110 s. This result showed that to arrive at a gravity measurement with a precision in the order of 1 %, antihydrogen formation needs to be improved significantly: in order to gather enough statistical data, a rate of approximately 1 ÷ 10 ¯H per decelerator cycle is necessary, which is more than two orders of magnitude greater than what was obtained in Phase 1. A substantial reduction in temperature is also required, from the previously reached level of around 400K to a few tenths of kelvin. Therefore, AE¯gIS entered in 2019 (together with CERN LS2), its Phase 2 (which is going to last until CERN LS3, in 2025), with four goals: the two just aforementioned aims (2-3 order of magnitude higher rate and one order of magnitude colder ¯H, with respect to Phase 1), together with the formation of a forward-boosted beam of antihydrogen, and the development of a moir´e deflectometer prototype for inertial measurement. To achieve these objectives, major upgrades of the apparatus have been deemed necessary, starting from the formation scheme. The Ps target has been positioned on the axis of the trap, to illuminate the ¯p plasma collinearly, and not perpendicularly as before: this has raised the maximum Ps Rydberg level from ∼ 19 to above 32 by reducing the Ps ionisation due to the motional Stark effect (nmax Ps ∝ θ−1/4 Ps|B). The updated formation scheme necessitated a redesign of the formation trap, which was built and installed in 2022. Additionally, the e+→Ps target has been optimised by fine-tuning the morphology of the nano-channel, so to increase up to five times the efficiency of positronium generation. In the context of the upgrades to the AE¯gIS apparatus, my main contribution has been the development of the new control system controlling the entire experiment. In fact, AE¯gIS has been operated using multiple independent control systems for each subsystem, coordinating them by a grown-up program and by the manual labour of multiple experts simultaneously. This already resulted demanding from the scientists at the end of Phase 1, when antihydrogen production was successfully attempted. But with the introduction, in 2021, of ELENA (Extra Low Energy viii Antiproton ring) [16], the modality of antiprotons delivery changed from 8 h daily shifts to continuous cycles: keeping the experiment up day and night would have become completely unfeasible with the previous system, ultimately leading to beam time loss and team overwork. Therefore, it was decided to completely rewrite the control system, creating a distributed system with stability and automation as the main targets. The new control system is called CIRCUS (Computer Interface for Reliably Controlling, in an Unsupervised manner, Scientific experiments) [1], whose main component is the TALOS framework (Total Automation of Labview Operations for Science) [2]. TALOS creates an environment where all individual control programs are subdivided into atomic modules, called MicroServices, which are integrated into a single, coordinated, distributed system: these features are the base that enables the complete automation of experimental procedures since high-level decisions often rely on parameters residing on multiple computers. TALOS is built upon the distributed system architecture provided by the Actor Model [17], which facilitates the modular structure. In particular, it is realised by dividing the code into standalone units, called MicroServices, each with a precise scope and function, which operate in parallel, communicating through non-blocking messages, to ensure complete asynchronous functionality. The unification of the distributed system is realised by a common process, called Guardian, an instance of which is executing on each machine. Every Guardian monitors the status of the other Guardian in the experiment’s computer network and oversees the local MicroServices, thanks to a series of three distributed watchdog systems. This system ensures that no component becoming unresponsive can pass unnoticed: it significantly strengthens the reliability and safety of the system, ultimately leading to unsupervised operation for extended periods. Furthermore, the automation of the system was augmented by the introduction of an optimiser, that leverages the feedback given by the online analysis of the data acquired, to autonomously find the best parameters that optimise a series of predefined observable CIRCUS, per se, is experiment-agnostic, being general enough to be usable by other experiments than AE¯gIS, in particular nuclear, atomic and quantum ones. Therefore, to be able to adopt it in AE¯gIS, I also coded the majority of the MicroServices necessary to manage most of the detectors and actuators, plus the interface with the decelerators and the data acquisition interface, among others. The implementation of CIRCUS would not have been possible without upgrading the control electronics from the previous custom-made hardware (with a very limiting software interface) to a new one based on ARTIQ/Sinara [18], an open hardware & software ecosystem expressly created for quantum physics experiment. The hardware modularity has enabled progressive migration and guarantees future-proofness; its ns synchronisation (internal and w.r.t. an external clock) capability has further improved the timing accuracy of the apparatus. Conversely, the programming interface done in ARTIQ, a Python-based real-time language, has greatly simplified the generation of experimental procedures, enabling both a library-based approach (which minimises debugging time and code duplication) and the integration with TALOS, to ensure the automation of the full system. In this respect, during this thesis, I contributed to the installation and integration of the Sinara hardware, and I wrote part of the libraries used to operate it and, in particular, to interface the FPGA seamlessly with the CIRCUS control system. During the three antiprotons campaigns (2021, 2022, 2023) that occurred during the development of this thesis, the system was tested with the particles, and the various operations needed to form antihydrogen were re-developed using the new control system. In this thesis, the physics motivations that brought these procedures to the form that we found most performant, which is also presented, are explained. The main operations tested and optimised are: antiprotons capture, trap closure time optimisation, antiproton beam alignment, traps voltage reshape, electrons loading, ¯p sympathetic cooling, ¯p plasma compression with rotating-wall technique, electrons removal before antiprotons transfer, antiprotons transfer and recapture in the formation trap, ¯p ballistic transfer to the formation trap, electrons recycling, antiprotons partial recycling, positronium formation optimisation, laser synchronisation for Ps excitation, antihydrogen formation. ix Chapter 0. Abstract These operations were implemented sequentially, by consolidating each one before passing to the subsequent one. A key ingredient was the extensive use of custom libraries for the Python code of the various experiments: every time an atomic operation was tested and optimised, it was defined as a function in the AE¯gIS libraries, so to be able to recall it in all the subsequent operations scripts. This greatly facilitated the development, since the new code was progressively added to the already consolidated one, minimising code duplication, script proliferation, and debugging time. To guide the success and performance of the operations, two detectors were mainly used: the scintillating slabs surrounding the main apparatus, acquired via photomultipliers (PMTs) and digitised, and a combination of a multichannel plate (MCP), a phosphor screen and a camera, placed downstream with respect to the formation trap. The firsts are fundamental to monitoring and understanding the lifetime and the quality of the operations with the antiprotons, by understanding the time and position of the various annihilations. The second is mainly used to understand radial plasma profiles and to perform time-of-flight (ToF) analyses. In the end, to determine the formation rate of antihydrogen with the new apparatus, I took part in the data analysis, by developing one of the three parallel analyses that have been performed. One analysis, analogue to the one already performed in 2018, used the scintillators counts to discriminate the difference in ¯p annihilation profiles in case that both ¯p, positrons and laser were employed together, or that one of them was missing: by the difference, ¯H production can be inferred. My data analysis, instead, used different techniques to look at the MCP data to understand not only the formation rate but also, by comparison with the scintillators’ data, the possible forward-boosted antihydrogen formation. The images were algorithmically selected, background corrected, and binarised; then, clusters were extracted. A Bayesian test was finally performed between runs where ¯H production was attempted, versus runs where it was suppressed by omitting one of the lasers. The analyses hint toward a successful production of antihydrogen with the upgraded apparatus, at a higher rate than the one seen at the end of Phase 1: nevertheless, it has been difficult to determine the direction of production with the MCP analyses, and deeper analyses and/or more data is needed to fully conclude. Overall, the work accomplished during this PhD thesis has been fundamental to achieving the formation of antihydrogen using the ballistic antiprotons transfer, which leads to the creation of a forward-boosted beam of neutral antihydrogen. This is an important milestone towards measuring directly the gravitational interaction between matter and antimatter, which could lead to a confirmation of the WEP on antimatter or could find a violation and, so, hint at new physics.