Raman spectroscopy of artists' materials: Advances in characterisation and analysis
AffiliationSchool of Physics
Document TypePhD thesis
Access StatusOpen Access
© 2020 Deborah Lau
Cultural heritage research and conservation practice seek to preserve our cultural heritage. Understanding the composition of the materials that comprise the artwork or cultural object is critical to inform collections management and preservation treatments; however methods of analysis are constrained to those that are either non-destructive or can obtain the desired information from micro-samples in order to retain the integrity of the object. Raman spectroscopy is an ideal technique for characterising cultural materials as it is non-destructive, requires relatively little sample preparation and utilises short measurement times. Micro-Raman is especially useful for examining micro-samples and painting cross sections due to its spatial resolution being sufficiently high to target individual pigment grains. Despite the non-destructive, micro-sample analysis advantages, there are limitations to the use of Raman spectroscopy in the cultural heritage conservation context. Firstly, fluorescence is frequently observed in cultural materials. Secondly, with the large number of compositionally complex and heterogenous materials encountered in conservation, there is a need for advanced methods that can deal with large sample datasets. These methods are needed to facilitate examination of both non-spatial data and spatial (imaging) data, to extract the maximum amount of information possible from the limited sample available. It is the aim of this thesis to demonstrate how Raman micro-spectroscopy can provide new and useful information about paintings using the following strategies. (1) Creation of a reliable pigment database supported by X-ray diffraction data to confirm the structural identity of each pigment. At the time this work was started, a spectral library for pigment identification was necessary to have a comprehensive set of pigment reference spectra against which to compare unknowns. A Raman spectral pigment reference library was developed comprising over 180 samples from the National Gallery of Victoria’s pigment collection. Samples were validated using X-ray Powder Diffraction (XRPD) prior to Raman analysis. This Raman spectral pigment library can also support future identification of materials and artworks, alongside other Raman spectral databases that are now available. (2) Utilisation of the database in conjunction with longer wavelengths to mitigate fluorescence effects. In the presence of fluorescent components, Raman analysis is hampered. To mitigate the impact of fluorescence observed at 514nm, the efficiency of 830nm wavelength incident irradiation was examined. It was found effective and used to answer five research questions (case studies) regarding authenticity and art historical practice and to inform attribution, provenance studies and conservation treatments: a) Mock-up paintings were prepared to trial the experimental methodology. The overpaint in a 16th century portrait miniature was identified as Zinc white pigment, indicating the overpaint was applied after the mid-19th century. b) An Australian Impressionists artist’s catalogue from 1889 was examined and the inks found to contain Prussian blue and vermilion as the main pigments, with a minor addition of minium and perhaps a lake pigment, providing insight into the artist’s technical methods. c) Overpainted cracks in Tom Roberts’ iconic painting Shearing the Rams were suspected to have been due to the type of pigment used. The main pigment was vermilion, which is not known to cause cracking, so the cause of cracking is now believed to be due to the high ratio of binding medium in the paint. d) The Finding of Moses (1712) was reattributed to Giambattista Tiepolo (1696-1770), after Prussian blue was identified as a key component of the paint layer. e) The organic blue colour in an Indian Palampore was dissimilar to indigo but matched a published spectrum of indigo on silk, highlighting the importance of local structure and bonding on subtle features in Raman spectra. (3) Identified a practical method of Surface Enhanced Raman Spectroscopy for conservation to increase the Raman signal and make it visible over the photoluminescent background. This work reviewed several SERS substrate configurations, then prepared and evaluated SERS substrates prepared by a) colloidal Ag nanoparticles, b) Ag coated nanospheres, c) Ag foil etching and d) electroless deposition of Ag on a Cu coupon. The ease of production and reproducibility were used to select the most practical substrate for SERS analysis in conservation being SERS substrates prepared using an electroless deposition method. The selected substrate was used to identify dammar as the varnish used on an important Italian Renaissance painting by Tiepolo with the outcomes published in 2008. (4) Developing new methods of data analysis for managing complexity in spectra and large data sets. Multivariate analysis techniques have been used to analyse spectral datasets in numerous fields and provide an excellent opportunity to enhance the analysis of large Raman spectra datasets in conservation. Principal components analysis (PCA) and hierarchical clustering analysis (HCA) methods were used to visualise the data structural relationships amongst Raman spectra of natural and synthetic resins. It has been demonstrated that the two most utilised natural resins, dammar and mastic, are able to be distinguished from one another by PCA and HCA of their Raman spectra, irrespective of their supplier and the naturally occurring sample variance. This work also shows, using PCA and HCA, that the synthetic cyclohexanones resins Ketone N and Laropal K 80 are indistinguishable whilst the other synthetic painting varnish cyclohexanone, MS2A, is easily separable. The synthetic resins were found to be quite homogeneous in composition with little variability in their Raman spectral response compared to a very much greater degree of variance was observed within the natural resins: amber, copal, colophony and sandarac. Finally, a multivariate image analysis method, assembling the data into a 3D data-cube and using PCA and clustering techniques, was developed. The method for assembling and analysing the spectral 3D data cubes was achieved using prepared samples of known pigments in binder. The technique was used in the analysis of an Italian renaissance painting. PCA and clustering methods were applied to SEM-EDS elemental maps of Ti, Sn, Si, Pb, Mn, Mg, K, Fe, Cu, Ca, Al, S (corrected for Pb) and O, to develop a compositional map of the materials used and indicate their sequence in the layered construction of the painting. Secondly, using Raman maps of spectral intensity collected at 830 nm to mitigate fluorescence, and using the spectral database, vermilion, lead-tin yellow type 1 and a blue-green pigment consistent with terre-verte or another green silicate pigment were found in the paint layer. The ground layer was found to contain anhydrite with large gypsum inclusions. The identification of these components has led to the attribution of a previously anonymous painting to Dosso Dossi with the outcomes published in 2008, receiving 70 citations until 2019. Multivariate methods developed here have been further applied in published research in both conservation and non-conservation applications, which is noted in this thesis.
KeywordsRaman; Raman spectroscopy; Micro-Raman spectroscopy; Conservation science; Cultural heritage; Painting analysis; Pigment analysis
- Click on "Export Reference in RIS Format" and choose "open with... Endnote".
- Click on "Export Reference in RIS Format". Login to Refworks, go to References => Import References