Multivariate analysis of combined reflectance FT-NIR and micro-Raman spectra on oil-paint models

The potential of principal component analysis (PCA), on combined first derivative near infrared and Raman spectra, was explored and tested in order to exhaustively characterize different oil-paint models, prepared with linseed and poppy-seed oil, and to verify the behavior of these binders in the presence of a pigment. A series of oil-paint models, selecting pigments employed in artworks both in antiquity (like lead white, azurite, Afghanistan ultramarine) and contemporary art (such as phthalocyanine blue, zinc oxide and synthetic ultramarine blue), were prepared and analyzed by means of reflectance near infrared and micro-Raman spectroscopic techniques. At first, PCA was applied by taking into account all oil-paint models, pigments and drying oils. Subsequently, new statistical models were built focusing only on single pigment subsets; hence, offering a more realistic application of this method, once the pigment had been identified. The proposed procedure allowed us to recognize the pigments and binding media used to prepare each model. In addition, it proved to be even sensitive to the drying oil employed in the oil-paint mixture, enabling its identification. The application of PCA to a combination of different spectral regions (6000- 3900 cm-1 along with 1900-260 cm-1) allowed an enhanced level of spectral information to be extracted, with respect to its application on separated data-sets. It represents also a powerful tool to differentiate oil-paint models on the basis of their composition, including oils and pigments. Moreover, we verified that after a natural ageing of nine months, and in presence of a pigment, it is still possible to obtain information regarding the functional groups directly involved in the drying process (e.g. the 1 st overtones of methylenic stretching, 5800 cm-1 and 5698 cm-1, the C-H combination bands of methylenic stretching and bending modes, 4340 and 4261 cm-1, and the ν(C=C) stretching, 1654 cm-1).