Improving ultracold neutron statistics for the n2EDM experiment

Abstract

The n2EDM experiment, currently being commissioned at the UCN source at the Paul Scherrer Institute, aims to search for the neutron electric dipole moment (nEDM) with a sensitivity of 1.1 × 10^−27 e ⋅ cm. This requires improving both the statistical and the systematic uncertainties, compared to the previous experiment. The n2EDM experiment is a Ramsey experiment with ultracold neutrons (UCN). The statistical sensitivity depends on the number of UCN counted in the detectors at the end of each Ramsey cycle. Therefore, it is important to maximise the UCN yield of the UCN source and the UCN transport and storage in the n2EDM experiment. In this thesis, I will present work on both of these aspects. The UCN source at the PSI is a pulsed, spallation-driven, solid-deuterium-moderated, superthermal UCN source. UCN are generated by the downscattering of cold neutrons via the excitation of phonon states in the solid deuterium lattice. Using Raman spectroscopy, we determined that the hydrogen deuteride (HD) concentration in the UCN source solid deuterium is (0.210 ± 0.002 sys ± 0.016 stat) %. At this concentration, neutron capture in hydrogen is not the dominant loss channel for UCN. It is not recommended to replace the UCN source deuterium with even higher-purity deuterium. During UCN source operation, rapid temperature cycling causes a rough layer of ‘frost’ to build up on the solid deuterium surface, which reduces the UCN extraction rate and thus the UCN yield. The UCN yield is recovered by a gradual thermal cycling procedure, called ‘conditioning’. We tested a new method of conditioning that relies on radiation-induced heating from proton beam pulses instead of electric heaters. We verified that conditioning with proton beam pulses is reliable enough to be automated, and determined the optimal conditioning frequency. In 2022, we measured a record UCN yield during production pulses, of 5.71 × 10^7 normalised to 2.2 mA of beam current, with a solid deuterium mass in the moderator vessel of 5.28 kg. With 5.68 kg of solid deuterium in the moderator vessel, we measured a lower UCN yield. We suspect that in the latter case, the deuterium surface was not sufficiently cooled, because the filling level exceeded the height of the cooling channels in the moderator vessel, reducing UCN output. We expect we can achieve even larger UCN yields with a moderator vessel with adapted cooling channels. We compared the UCN intensities at the UCN source beamports South and West-1. For storable UCN, the UCN intensity ratio at these beamports has a small relative standard deviation of 0.14 %. While the UCN intensity ratio of West-2 to South/West-1 is less stable, we can use a detector on West-2 to monitor the UCN intensity in an experiment on South or West-1 with an accuracy of 0.4 %, when we apply drift correction and limit the number of pulses to 50. We determined that the UCN intensity at beamport South was (81.6±0.2) % of the intensity at West-1. We measured the UCN transmission of the UCN guides for n2EDM using a direct transmission setup. The best guides had a transmission of more than 95 %, normalised to a guide length of 1 m. Some UCN guides had bad UCN transmission due to wavy irregularities on the interior of the guides. We made new guides from tubes without these wrinkles to provide more UCN to n2EDM. The UCN guides that are installed closest to the precession chamber are not allowed to contain magnetic dipoles larger than 15 nA m^2. This would disrupt the magnetic field gradient reconstruction by the caesium magnetometer array, and cause systematic errors. We searched for dipoles using our magnetic gradiometer and found that the guides satisfy this limit. We investigated the shift in resonance frequency in n2EDM caused by the effect of the π/2 -pulse for the mercury comagnetometer on the UCN spin state, as well as a possible shift induced by a tilt between the static magnetic field and the rotating magnetic field axis during the Ramsey sequence. We designed window functions to limit these frequency shifts to a value well below the systematics budget.