Extracting thin film structures of energy materials using transformers

Abstract

Neutron-Transformer Reflectometry and Advanced Computation Engine (N-TRACE ), a neural network model using transformer architecture, is introduced for neutron reflectometry data analysis. It offers fast, accurate initial parameter estimations and efficient refinements, improving efficiency and precision for real-time data analysis of lithium-mediated nitrogen reduction for electrochemical ammonia synthesis, with relevance to other chemical transformations and batteries. Despite limitations in generalizing across systems, it shows promises for the use of transformers as the basis for models that could replace trial-and-error approaches to modeling reflectometry data.

Figures (9)

• Figure 1: The workflow chart depicts two paths: the conventional offline iterative parameter extraction (left), and the proposed N-TRACE based on auto extraction. The proposed method replaces the labor-intensive iterative search process with a neural network, thus transferring the search work to the training process.
• Figure 2: Workflow diagram illustrating N-TRACE's data preprocessing steps, converting raw measurements $R(q)$ into latent embedding $\Phi$.
• Figure 3: Schematics of the N-TRACE model, which consists of a pre-processing unit (top), transformer encoder based translation unit (middle) and a single fully connected layer (FC) for post-processing.
• Figure 4: Schematic of the transformer-based model architecture for predicting the scattering length density (SLD) profile from reflectometry (NR) data.
• Figure 5: Training and testing loss.