Non-destructive characterisation and classification of ceramic artefacts using pEDXRF and statistical pattern...

Area of Science:

• Archaeological Science
• Materials Science
• Analytical Chemistry

Background:

• Analysis of Neolithic ceramic artifacts from three Serbian sites using portable energy dispersive X-ray fluorescence (pEDXRF) spectrometry.
• Application of dimension reduction techniques, including principal component analysis (PCA) and scattering matrices, to identify key elemental features.

Purpose of the Study:

• To develop a classification procedure for archaeological ceramics based on elemental composition.
• To evaluate the efficacy of dimension reduction methods for distinguishing ceramic provenances.

Main Methods:

• Utilized pEDXRF spectrometry for elemental analysis of ceramic samples.
• Employed PCA and scattering matrices-based dimension reduction for data analysis.
• Developed linear classifiers for sample classification.

Main Results:

• Scattering matrices-based dimension reduction demonstrated higher efficiency compared to PCA.
• A decision-making procedure successfully classified 7 out of 8 unknown ceramic samples.
• Elemental composition data from pEDXRF is viable for provenience studies despite technique limitations.

Conclusions:

• Elemental composition analysis via pEDXRF is a valuable tool for archaeological ceramic provenience studies.
• A robust classification procedure can accurately assign ceramic artifacts to their origin based on elemental signatures.