Estimation of Electron Screening Potential in the 6Li(d,{\alpha})4He Reaction Using Multi-Layer Perceptron Neural Network

TL;DR

This paper employs a multi-layer perceptron neural network to analyze experimental data and accurately estimate the electron screening potential in the 6Li(d,α)4He nuclear reaction, improving understanding of sub-Coulomb energy reactions in astrophysics.

Contribution

It introduces an ANN-based method to extract electron screening potentials from experimental data, providing a more robust alternative to traditional extrapolation techniques.

Findings

Screening potential estimated at 147.95 eV

ANN approach effectively models energy-dependent S-factors

Method enhances accuracy of low-energy nuclear reaction analysis

Abstract

Reactions between light charged nuclei at sub-Coulomb energies are crucial in astrophysical environments, but accurate cross-section measurements are hindered by electron screening. Traditional methods, including polynomial extrapolation and the Trojan Horse Method, often yield screening potentials exceeding adiabatic predictions. Building on the success of an MLP-based Artificial Neural Network (ANN) for the 6Li(p, {\alpha})3He reaction [1], this work applies the same approach to the 6Li(d, {\alpha})4He reaction. Experimental astrophysical S-factor data from literature are reanalyzed using the ANN to model the energy-dependent S-factor. The bare S-factor is extracted from data above 70 keV, where screening effects are minimal, and the screening potential is obtained by comparing with the low-energy region. The resulting screening potential is 147.95 eV, demonstrating the robustness of ANN-based methods for evaluating electron screening in low-energy nuclear reactions involving light nuclei.