Raman identification of pigments and opacifiers: Interest and limitation of multivariate analysis by comparison with solid state spectroscopical approach—I. Lead‐tin and Naples Yellow

Abstract

Abstract Lead‐tin/Naples Yellow provides a bright yellow color often used in combination with a blue‐colored glass to obtain a variety of green hues to decorate enameled glass, metalware, and ceramics. From the analysis of about hundreds of enameled objects from French and Chinese productions of the 18th century, we identify and compare the Raman signatures representative of these yellow pigments. For the first time, we propose several procedures that allow us to compare our data with those obtained from the literature on the spectra of synthesized pigments also characterized by XRD. A first approach is a reasoned analysis of the parameters obtained from the spectral decomposition carried out on spectra visually selected as the most different and the “purest.” A second approach uses multivariate analyses with principal component analysis (PCA) or hierarchical clustering of the spectra and of the band parameters originating from a spectral decomposition. It appears that the precise identification of the phase from simple observation of the sole Raman signature may be unreliable when only main peaks are detected due to the large non‐stoichiometry of the phases and the resulting similarity of the Raman signatures. PCA analysis of the total Raman spectra does not allow relevant classifications due to the strong contributions induced by the bands of the silicate matrix. However, the combination of a reasoned analysis of the Raman spectra with the PCA of spectroscopic parameters extracted from Raman spectral decomposition appears effective in categorizing the different pigments. From the results discussed, at least three types of Raman signatures are identified for each Sn‐ and Sb‐based pigment and correlate with the type, the period, and the location of enamel production.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 12 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.