Multi-Dimensional Machine Learning Analysis of Polyaniline Films Using Stitched Hyperspectral ToF-SIMS Data.

Area of Science:

• Materials Science
• Analytical Chemistry
• Machine Learning

Background:

• Self-organizing map with relational perspective mapping (SOM-RPM) is an unsupervised machine learning technique for visualizing high-dimensional hyperspectral data.
• Previous applications include analysis of time-of-flight secondary ion mass spectrometry (ToF-SIMS) hyperspectral images and 3D depth profiles.
• SOM-RPM aids in visualizing features, trends, molecular characteristics, and contaminant transport in complex datasets.

Purpose of the Study:

• To apply SOM-RPM to stitched ToF-SIMS datasets for direct 2D and 3D comparison.
• To analyze the effects of heat treatment on spin-coated polyaniline (PANI) films.
• To model PANI as a conformal coating for the aerospace industry.

Main Methods:

• Utilized SOM-RPM for analyzing stitched ToF-SIMS data from PANI films subjected to heat treatment.
• Trained a single SOM-RPM model on combined 2D and 3D datasets for comparative analysis.
• Performed quantitative assessment using peak ratios to evaluate chemical breakdown trends.

Main Results:

• Demonstrated precise equivalence between replicates in both spatial distribution and composition.
• Identified a clear annealing threshold for the PANI films.
• Highlighted subtle differences in peak intensity ratios, which spectral analysis alone struggled to quantify.

Conclusions:

• SOM-RPM provides insightful visualization and interpretation of hyperspectral data, even for complex, stitched datasets.
• The method is effective for characterizing material properties and identifying process thresholds, such as annealing in PANI films.
• SOM-RPM offers a robust approach for analyzing subtle chemical changes and ensuring quality control in material coatings for industries like aerospace.