Extension study of a statistical age prediction model for acrylic paints

Abstract

In this work, the robustness and potential applicability of statistical age prediction models applied to the dating of different acrylic paints were studied. The FTIR-ATR analysis of three acrylic colours (Hansa yellow, phthalocyanine green and ultramarine blue) from two manufacturers (Liquitex® and Vallejo®) subjected to accelerated ageing was carried out. The acrylic paints were characterised and the modifications of their ATR spectra throughout ageing were studied. Predictive models developed with the Liquitex® brand containing phthalocyanine green pigment were then applied to other colour and brands of acrylic paints and their robustness and feasibility were studied based on calculated accuracy error values. The influence of the pigment on the ageing of the paint components, the type and quantity of additives present in the acrylic paint as well as the ageing conditions to which it was subjected were decisive in the short-term predictive model, which explains the low accuracy values obtained for all the acrylic paints analysed. However, the slower degradation processes taking place in the longer term and the stabilisation of the acrylic paints at higher stages of ageing made them fit successfully into the long-term model, obtaining an error of between 14 and 23%. Thus, the predictive statistical model is robust and feasible to be used for different colours of the same brand of acrylic paint as well as for acrylic paints of different brands that have been long-term aged under slightly different conditions of accelerated ageing. In conclusion, this methodology could be a promising tool in the field of dating contemporary artworks of a certain age.