Methodology for processing backscattered electron images. Application to Aguada archaeological paints

Scanning electron microscopy is a powerful technique in several fields of science and technology. In particular it is an important complement in the characterization of materials for which X-ray analysis is not possible. Such is the case of thin paint layers on ceramic pots, in which, even for low incident energies, the electron interaction volume can be greater than the paint thickness—in addition to the problem arising from similar compositions. With the aim of complementing other common techniques used in compositional materials characterization, in this work, an image-processing software has been developed, which implements a new methodology for the treatment of backscattered electron (BSE) images in order to bring to evidence small mean atomic number contrasts, usually imperceptible to human eye. The program was used to study black and white pigments of ceramic pieces belonging to the Ambato style of “Aguada” culture (Catamarca province, Argentina, IV–XII centuries AD). Although the BSE images acquired for these samples showed no apparent contrast between sherd and black and white pigments, through image-processing algorithms using different space filters, chemical contrast between regions has been brought to evidence with a minor detail loss. This has been accomplished by applying a smoothing filter, after which the main routine for contrast enhancement reveals details in the grey-level region of interest; finally, a filter for edge enhancement permits to recover some details lost in the previous steps, achieving satisfactory results for the painted sherd samples analyzed. In order to validate the mean atomic number differences found between each pigment and the ceramic body, X-ray diffraction diagrams have been refined with Rietveld method using the software DIFFRACplus Topas®, arriving to mineralogical differences which agree with the results obtained. As a consequence of this study, the program developed has proven to be a suitable tool for routine analysis of samples with slight chemical contrast.