Data fusion of Raman hyperspectral imaging data using multi-block chemometric methods for the characterization of copper preservation

Details

Supervisors

Delcorte, Arnaud ; Neff, Delphine ; Ammari, Faten

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en chimie et science des matériaux, à finalité spécialisée

Abstract

enThe copper corrosion patina is an intricate chemical matrix consisting primarily of copper, cuprite (Cu2O), and brochantite (Cu4SO4(OH)6); along with several other minority copper salt constituents. The outermost layer, brochantite, is the distinctive blue-green patina associated with copper corrosion. Contemporary preservation research is focused on non-toxic, eco-friendly, sol-gel coatings for copper corrosion protection. The coatings for cultural heritage artifacts present a unique challenge, it must preserve the underlying metal, the current corrosion layers, and prevent modification of the aesthetics. Specifically, alkoxysilane sol-gels doped with carboxylic acid compounds have been shown to inhibit the corrosion process. This study intends to evaluate the copper corrosion inhibition treatment of tetraethyl orthosilicate (TEOS) doped with octanoic acid applied to the copper patina surface. Treated copper cross-sections were analyzed by Raman spectroscopy using two different excitation wavelengths (473 nm and 532 nm) to analyze the penetration depth and interactions of the inhibitor within the corrosion layers. The corrosion products and inhibitor produce distinct responses from the two wavelengths to create unique, but complementary “blocks” of data. Machine learning and multi-block techniques exploit this collective relationship to enhance the identification of the corrosion and coating-related phases. The multi- block fusion techniques revealed that the octanoic acid penetrated the corrosion layer and reacted with the brochantite to form copper octanoate. But, this reaction was concentrated on the outer boundary of the layer rather than distributed uniformly throughout the brochantite layer. Comparing the different multi-block methods showed that they produced largely convergent results for this two-block dataset. These findings support the use of dual-excitation Raman imaging combined with multi-block fusion as an accessible, non-destructive strategy for evaluating corrosion inhibitor coatings on cultural heritage copper artifacts.