Analysis of Atomic Charge State and Atomic Number for VAMOS++ Magnetic Spectrometer using Deep Neural Networks and Fractionally Labelled Events

A. Lemasson · 2025

Read Analysis of Atomic Charge State and Atomic Number for VAMOS++ Magnetic Spectrometer using Deep Neural Networks and Fractionally Labelled Events by A. Lemasson (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-07109.txt.

Introduction

Rejmund and Lemasson hand a tedious nuclear physics chore to deep neural networks. Their spectrometer sorts atomic nuclei by charge and number, but a flawed detector window blurs the readings. Hand analysis takes months of meticulous work. The team labeled only a small, clean slice of events precisely.

Networks trained on that slice learned to classify the rest on their own. Analysis time collapsed from months to hours. Dropping human judgment also made results standard and repeatable. The method delivers sharp charge and number spectra from messy real data. The authors see it as a template for other instruments with imperfect detectors.

Catalog record

CREATORA. Lemasson
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-07109

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Analysis of Atomic Charge State and Atomic Number for VAMOS++ Magnetic Spectrometer using Deep Neural Networks and Fractionally Labelled Events A. Lemasson Abstract The VAMOS++ magnetic spectrometer is a multi-parametric system that integrates ion optical magnetic elements with a multi-detector stack. The magnetic elements, along with the tracking and timing detectors and the trajectory reconstruction method, provide the analysis of the magnetic rigidity, the trajectory length between the beam interaction point and the focal plane of the spectrometer, and the related velocity and mass-over-charge ratio. The segmented ionization chamber provides the energy measurements necessary to analyze the atomic charge state and atomic number. However, this analysis critically suffers from inherent limitations due to the variable thickness and non-uniformity of the entrance window of the ionization chamber and other detector imperfections. Conventionally, this meticulous, detailed analysis is exceptionally tedious, often requiring several months to complete. We present a novel method utilizing deep neural networks, trained on an experimental dataset with only a small fraction of precisely labeled events for the lowest and best-resolved atomic charge states or numbers. This innovative approach enables the networks to autonomously and accurately classify the remaining events. This method drastically accelerates the acquisition of high-resolution atomic charge state and atomic number spectra, reducing analysis time from months to mere hours. Crucially, by discarding human bias, this approach ensures standardized, optimal, and reproducible results with unprecedented efficiency. 1 Introduction Magnetic spectrometers are indispensable tools in nuclear structure and reaction mechanism studies, enabling the precise identification and characterization of nuclei produced in nuclear collisions. They provide event-by-event isotopic identification of reaction products, crucially ensuring clean selection for subsequent analysis. The spectrometers like VAMOS++ [ 1 , 2 ] at GANIL and PRISMA [ 3 ] at LNL Legnaro are particularly suited for reactions occurring near the Coulomb barrier, where a large angular and momentum acceptance is necessary for efficient collection of the nuclear reaction products of interest. However, this large angular and momentum acceptance is associated with the need for large optical elements in the spectrometer, which often results in a design that exhibits highly non-linear optics. To overcome this challenge, ray-tracing techniques are employed to reconstruct the trajectories of the detected heavy ions [ 4 , 5 ] . This process allows for the determination of the magnetic rigidity and related observables for these ions. The VAMOS++ spectrometer, with its large angular and momentum acceptance, is specifically designed to provide isotopic identification of incoming ions, resolving them by atomic mass number ( A ), atomic charge state ( q ), and atomic number ( Z ). The magnetic rigidity ( B\rho ), the trajectory length ( l ) between the beam interaction point and the focal plane of the spectrometer, the ion’s velocity ( v ), and the related mass-over-charge ratio ( A/q ), are obtained using the tracking and timing detectors and the trajectory reconstruction methods [ 4 , 5 ] . However, achieving accurate isotopic identification critically depends on the precise analysis of the atomic charge state ( q ) and atomic number ( Z ), both derived from energy measurements within the segmented ionization chamber [ 4 ] . This work introduces a groundbreaking method to significantly simplify, accelerate, and enhance the reliability of this crucial analysis step. 2 Motivation VAMOS++ operates with heavy ions ranging from carbon ( Z=6 ) to uranium ( Z=92 ) at energies between approximately 2 MeV/u and 10 MeV/u. However, achieving accurate energy measurements from the segmented ionization chamber data [ 4 ] is severely complicated by the highly variable thickness and non-uniformity of the material preceding the active volume. This is particularly challenging due to the unpredictable shape and amplitude of large-area window deformations, which are highly susceptible to pressure changes. To maximize the number of identifiable ions in terms of atomic number by utilizing the correlations between energy loss and residual energy, the thickness of the windows is minimized to reduce the population of the unresolved low-energy phase space associated with the Bragg peak. The VAMOS++ focal plane detectors cover an active area of 1000\times 150~\text{mm}^{2} . Within this large area, the deformation of the ionization chamber windows alone can introduce a significant variation of approximately 3~\% in the measured total energy. Figure 1 vividly illustrates these large amplitude deformations in both vertical (a) and horizontal (b) views, highlighting the extent of this challenge. Additionally, any imperfections of the detector, such as charge collection or non-linearities in the electronics, must also be considered during the energy measurement analysis process. Figure 1 : Deformation of the ionization chamber windows: This figure presents the vertical (a) and horizontal (b) views of the ionization chamber window after a test conducted on the test bench. Achieving a precise absolute energy calibration of the ionization chamber before each experiment is impractical. This is due to the unavailability of radioactive sources covering the required heavy ion and energy ranges. Moreover detailed pre-experiment mapping of the active region of the ionization chamber via elastic scattering of specific ions is prohibitively costly and time-consuming. Therefore, our analysis strategically leverages the inherent self-consistency of each experiment’s data, augmented by fundamental physics principles and approximate online calibration coefficients for raw energies from each ionization chamber row. 3 Experimental methods Figure 2 : Non-uniformity of the material: Two-dimensional correlation plot between atomic charge state ( q ) and (a) the horizontal position ( x_{ic} ) and (b) vertical position ( y_{ic} ) on the entrance window of the ionization chamber. The red arrows in panel (a) indicate the positions of the vertical window holding wires, spaced by 30 mm, shown in Figure 1 . The charts (a) and (b) were obtained using approximate calibration coefficients ( C_{i} ) for the raw energies ( E_{ic_{i}} ) measured by each row of the ionization chamber. Figure 3 : Initial identification charts: (a) Two-dimensional correlation plot between atomic charge state ( q ) and mass-over-charge, ( A/q ). Each accumulation for a given atomic charge state corresponds to a specific atomic mass number. (b) Two-dimensional energy correlation plot between the energy loss ( \Delta E ) and residual energy ( E_{res} ) resulting in correlated bands for each atomic number. For the lowest E_{res} , the unresolved part of the spectra, associated with the Bragg peak, was omitted. The charts (a) and (b) were obtained using approximate calibration coefficients ( C_{i} ) for the raw energies ( E_{ic_{i}} ) measured by each row of the ionization chamber. The experimental data employed in this study was acquired from during the E826 GANIL experiment [ 6 ] designed to detect and identify the fission fragments generated in fusion-fission and transfer-fission reactions. These reactions were induced by a 238 U beam with an energy of 5.88 MeV/u on a 0.5 mg/cm 2 thick 9 Be target. The VAMOS++ spectrometer was positioned at an angle of 20^{\circ} relative to the beam axis. The fission fragments with atomic numbers spanning approximately 30 to 60 were detected by the VAMOS++ spectrometer, with energies ranging from approximately 4 MeV/u to 10 MeV/u. This resulted in a relatively broad distribution of atomic charge states from 25 to 50 . The ion tracking and timing were ensured by the

Read the complete text