Comparing post-acquisition methods for Raman-based mineral identification of particles in yellow earth pigments

Area of Science:

• Geochemistry
• Materials Science
• Spectroscopy

Background:

• Natural yellow earth pigments are complex mineral mixtures, primarily iron oxides.
• Identifying accessory minerals in these pigments is challenging due to spectral overlaps and sample heterogeneity.
• Accurate mineral identification is crucial for determining geological origins in cultural heritage artifacts.

Purpose of the Study:

• To evaluate post-acquisition data processing for robust mineral discrimination in yellow earth pigments using Raman spectroscopy.
• To compare different analytical strategies for mineral identification and pigment differentiation.
• To establish a reproducible workflow for pigment analysis in cultural heritage research.

Main Methods:

• Particle-resolved confocal Raman spectroscopy was employed on five yellow earth pigments (ochres and siennas).
• Statistical datasets (>200 spectra per sample) were generated for comprehensive analysis.
• Mineral identification utilized visual interpretation, automated spectral library fitting (3C model), and machine learning classification (XGBoost).

Main Results:

• Both the 3C model and XGBoost successfully achieved statistical separation of all five pigments.
• Automated methods enabled robust mineral discrimination, unlike K-means cluster analysis of visual interpretation.
• The study demonstrated the effectiveness of post-acquisition processing for Raman-based pigment analysis.

Conclusions:

• Systematic post-acquisition processing of Raman spectra enables reliable mineral discrimination in heterogeneous yellow earth pigments.
• Automated approaches (3C model, XGBoost) offer superior performance compared to visual interpretation for pigment differentiation.
• A reproducible Raman-only workflow is proposed for preliminary mineral assignment and provenance studies in cultural heritage.