Decoding Molecular Geometries in Coulomb Explosion Imaging via Physics-Informed Deep Neural Network

TL;DR

A deep learning framework is introduced that bridges the gap by directly recovering position-space molecular structures from Coulomb explosion momentum patterns by combining CEI simulations with a neural network trained to establish the mapping between momentum-space Newton plots and real-space geometries.

Abstract

Determining the absolute configuration of gas-phase molecules in position-space has long been a fundamental challenge in molecular physics. While strong-field-induced Coulomb explosion imaging (CEI) has emerged as a powerful tool for probing molecular stereochemistry in momentum-space, reconstructing the original three-dimensional structure of polyatomic molecules remains a long-standing challenge due to the inherent complexity of multidimensional inversion. Here, we introduce a deep learning framework that bridges this gap by directly recovering position-space molecular structures from Coulomb explosion momentum patterns. Our approach combines CEI simulations with a neural network trained to establish the mapping between momentum-space Newton plots and real-space geometries. The trained model demonstrates high fidelity in reconstructing the structure of CHF$_3$ from experimental CEI data. This generalizable framework can not only be extended to other molecular systems but also opens avenues for time-resolved structural analysis of molecular dynamics.

Figures (4)

• Figure 1: Schematic diagram of the deep neural network.
• Figure 2: (a) The sampled geometries of CHF$_3$ from Wigner distribution. (b) Three-dimsnsional momentum-space Newton diagram by the CEI simulation. (c) The reconstructed position-space image of CHF$_3$ from the simulated momentum-space Newton diagram. (d) Comparsion bewteen the angular distribution of the sampled and reconstructed $\measuredangle$HCF and $\measuredangle$FCF angles of CHF$_3$. All of the Newton diagrams are presented in normalized scale and the angular distributions are normalized by their peaks.
• Figure 3: (a) Schematic of the experimental setup. (b) Experimental momentum-space Newton plot of CHF$_3$ measuring all 5 ions in coincidence. The the gray scatter points represent the 3D molecular configuration and the corresponding 2D projections onto three planes are illustrated by the colorful plots. Comparing the experimental (c) and simulated (d) Newton plots on the $p_{yz}$ plane.
• Figure 4: (a) The position-space goemetry distribution of CHF$_3$ retrieved from the momentum-space normalized Newton plot. (b) $\measuredangle$HCF and $\measuredangle$FCF angular distributions. Restructed (c) and sampled (d) position distributions of three fluorine atoms.