A Shift-Invariant Deep Learning Framework for Automated Analysis of XPS Spectra

TL;DR

This project introduces a machine learning solution using a Spatial Transformer Network (STN), a type of neural network that implicitly learns to align spectra and demonstrates that neural networks can effectively learn the underlying relationships between spectral features and chemical composition when they are able to intrinsically account for variable shifts.

Abstract

X-ray Photoelectron Spectroscopy (XPS) is a crucial technique for material surface analysis, yet interpreting its spectra is often challenging for both human analysts and automated methods due to the prevalence of variable spectral shifts and overlapping peaks. This project introduces a machine learning solution using a Spatial Transformer Network (STN), a type of neural network that implicitly learns to align spectra. An STN model was designed to classify the chemical environments present in an input spectrum, using functional groups as a proxy. The model was trained and tested on a large synthetic dataset of 100,000 spectra, created by linearly combining real experimental data from a library of 104 polymers. \cite{RN22} To simulate experimental variability, random uniform shifts and broadening were applied to the data. The STN was found to effectively correct for random electrostatic shifts (up to 3.0 eV) and achieved relatively high accuracy ($\sim$ 82\%) in identifying functional groups, despite utilizing a much simpler architecture than previous work. These findings demonstrate that neural networks can effectively learn the underlying relationships between spectral features and chemical composition when they are able to intrinsically account for variable shifts. This work advances the development of more reliable automated XPS analysis, offering potential as an assistive tool for researchers and as a core component in future autonomous systems like self-driving laboratories.

Figures (9)

• Figure 1: High-resolution C 1s and O 1s core level XPS spectra of poly(methyl methacrylate) (PMMA).RN22 Peak assignments for the primary chemical environments are annotated on the experimental data (blue line). A duplicate spectrum with a uniform +1.6 eV shift (red, dotted line) is overlaid for comparison.
• Figure 2: Diagram comparing architectures of a Multilayer Perceptron (MLP), a Convolutional Neural Network (CNN) and a Spatial Transformer (STN)-based NN, and how they process a spectrum (S) into predicted label (L).
• Figure 3: Model test classification prediction accuracies for models (multilayer Perceptron (MLP), Spatial Transformer Network-Neural Network (STN-NN) and convolutional neural network (CNN, kernel size = 3.0eV)) trained and tested at different levels of maximum shift.
• Figure 4: Comparison of C 1s and O 1s peaks for both epoxide (blue) and aliphatic ether (red) functional groups. Peaks were isolated by fitting from experimental spectra of PGMA and PEG respectively.
• Figure 5: Process of applying random uniform shifts to a subset of 10 synthetic test spectra (a $\rightarrow$ b), and their subsequent alignment by the STN layer (b $\rightarrow$ c). All plots are cropped to highlight the C 1s region (280–292 eV).