Development of soft X-ray and neutron diagnostics for space-resolved spectroscopic measurements at RFX-mod2

The upgrade of RFX-mod (RFX-mod2) requires advanced diagnostics capable of providing space-resolved and energy-resolved measurements during Magnetic Reconnection (MR) events, which are fast transient phenomena characterized by intense bursts in particle flux at the detectors. In this context, this thesis focused on the development of soft X-ray (SXR) and neutron/gamma diagnostics for RFX-mod2, together with dedicated software for simulations and data analysis. For the SXR diagnostic, the design of an energy-resolved detection system based on Gas Electron Multiplier (GEM) detectors was presented, accounting for the main challenges imposed by the RFX-mod2 environment, including operation in strong magnetic fields and possible detector gain variations. A custom simulation software, named REVOLT-U, has been developed to model the energy-resolved response of SXR pixelated detectors in nuclear fusion experiments. Its application to the development of an energy-resolved 1D tomographic reconstruction algorithm for RFX-mod2, and to the design of the SXR diagnostic system, supported the realization of a three-dimensional Computer-Aided Design (CAD) model of the diagnostic. The diagnostic is characterized by significant flexibility, as its parameters can be adapted to a wide range of operating conditions, depending on the experimental target and the plasma scenario. As a neutron/gamma diagnostic for RFX-mod2, a system combining EJ-276D plastic scintillators and LaCl$_3$:Ce inorganic scintillators has been proposed. The three-dimensional CAD model of the diagnostic system was presented, with the detectors positioned as close as possible to the RFX-mod2 device in order to maximize the neutron flux, while ensuring adequate detector shielding for background mitigation and the definition of three lines of sight (allowing a basic level of spatial resolution). The main challenge was identified in the high magnetic field at the detectors location ($\sim$ hundreds of mT), which led to the adoption of silicon photomultipliers (SiPMs) as photodetectors due to their insensitivity to magnetic fields. To address the potential challenges posed by pile-up, complex radiation fields, and the use of SiPMs coupled with LaCl$_3$:Ce scintillators, a neural network-based algorithm was developed for automated waveform processing, including PSD and pile-up recovery. This algorithm enables operator-independent particle identification and the recovery of events affected by pile-up. This can lead to a significant increase in the number of usable events compared to ideal pile-up rejection schemes, depending on the incident fluxes at the detectors and the observed pile-up probability. The proposed neutron/gamma and SXR diagnostics are installed in the same poloidal sector of RFX-mod2, providing simultaneous information on ion and electron acceleration during MR events, resolved in space, energy, and time, with a potential sub-millisecond temporal resolution limited primarily by the available radiation fluxes. Gamma-ray spectroscopy with spatial information could also contribute to the study of runaway electrons in future RFX-mod2 tokamak discharges. Finally, although presented for RFX-mod2, the methodologies and software tools developed in this PhD work are potentially extensible to other fusion experiments and diagnostic systems, potentially representing a broader contribution to the field of nuclear fusion diagnostics.