Characterisation and classification of binders used in art materials at the class and the subclass level

摘要

SIMCA pattern recognition is used with amino acid chromatographic profiles in a large homemade collection of natural protein binders obtained following old recipes traditionally used by painters and considered here as the standard of classification. An initial cluster analysis of the full dataset made it possible to distinguish three main classes of protein binders: albumin, casein and collagen-like substances. An additional iterative study of each class revealed a new subclass, i.e., glair, yolk and whole egg for the albumin class; goat, sheep and cow for the casein class; and mammals and fish for the collagen class. Optimized SIMCA models for each class and subclass were obtained with good results in terms of sensitivity (90a€“100%), specificity (73a€“100%) and interclass distance (1.4), providing identification of the protein binder present in a set of samples of different origins such as natural products, commercial binders and works of art considered cultural heritage.