Dylan Smith, Robert H. Smith Research Conservator, National Gallery of Art, spoke as part of the Theory and Practice Ring at the “WCG’s 3-Ring Circus.” For conservators and conservation scientists, one of the most challenging aspects of quantitative x-ray fluorescence spectroscopy (XRF) analysis is reliable calibration to ensure repeatability and permit data comparison across institutions using the same instrument. Dylan Smith’s presentation Portable XRF Analyses of Copper Alloys and Recent Thoughts on Calibration reviewed the factors that make qualitative calibration difficult; a method for improved calibration with standards developed specifically for the cultural heritage sector; and a case study that demonstrates the level of accuracy that can be achieved when a calibration created with dependable standards is applied to XRF data collected with a handheld spectrometer.
Smith began with a discussion of the benefits of empirically derived calibrations over factory calibrations provided by spectrometer manufacturers. Empirical calibrations are created by analyzing known standards and plotting the information received from the detector against the known proportions of the standards tested to create a calibration curve. Due to complicated energy interactions between metallic elements within alloys, it is necessary to create calibration curves using alloys rather than pure metal standards. For this reason, factory calibrations that are truly meant for industrial use are not typically adequate for analysis of cultural heritage objects. Further, minor and trace elements are not typically well represented in factory calibrations. To address these issues and the specific needs of conservators, the British firm MBH Analytical, Ltd. worked with conservators and scientists at the J. Paul Getty Museum, the National Gallery of Art (NGA), and a group of other institutions to develop the Cultural Heritage Alloy Reference Material (CHARM) standards for quantitative elemental analysis. The CHARM standard set has significantly expanded the elements of interest and greatly improved the fit of calibration curves.
To determine the level of interchangeability of calibrations, NGA created calibrations for three different Bruker Tracer portable XRF spectrometers for comparison. To even consider using a calibration from one instrument with another, the spectrometers must be operated with the same filters and kilovoltage (kV) to create the empirical calibrations. Comparisons were then made of the quantified results calculated by applying the calibration created for one instrument to the spectral data collected on another XRF. The general agreement of the results was very good, however, some differences in the measured value where observed. Although the exact cause is not clear, it appears to be associated with the sensitivity at different energies of the individual detectors, not small variations in the operating amperage.
The question of whether or not an a calibration created for one instrument using the CHARM standards could be applied to an existing data set collected by another comparable instrument was raised as the Courtauld Gallery of Art prepared for the exhibition Court and Craft: A Masterpiece from Northern Iraq (February 20 – May 18, 2014). This exhibition featured one of the most extraordinary objects in the Courtauld’s permanent collection: an intricate, 14th-century metalwork bag created for the Mongol court. Referred to as The Courtauld Bag (Mosul, 1300-1330), it is extravagantly inlaid with gold and silver and is the only one of its kind known.
A technical study, which included quantitative elemental analysis of the alloys that comprise the bag and a number of other objects, was undertaken in conjunction with the close of the exhibition. The Courtauld, however, was concerned about the accuracy of their portable XRF spectrometer and reached out to NGA for assistance in performing the analyses and interpreting the data collected on the study objects. To this data, NGA applied a calibration created with the CHARM standards using the same model of spectrometer and instrumental conditions that were employed to analyze the objects. Later, the British Museum lent the Courtauld a set of CHARM standards to calibrate its instrument and apply this calibration to the data already collected. This allowed for a custom calibration specific to the Courtauld instrument to be created. When results from this calibration were compared to the quantitative interpretation with NGA’s calibration, again close agreement was achieved.
Smith concluded his presentation with the following takeaways from the comparative spectrometer study and analysis of The Courtauld Bag: It is always best practice to calibrate your spectrometer with known standards, but in a pinch, it is possible to use another institution’s empirically derived calibration for the same instrument. A broader range of alloys and trace elements found in art objects will be accounted for by applying a calibration from a cultural heritage institution than with those supplied by the manufacturer for industrial applications. Further, with increased use of highly accurate alloy standards designed for the analysis of cultural heritage, elemental data interpretation will become more consistent, reliable, and shareable across institutions performing similar technical investigations.
Summarized by: Michelle Sullivan, Graduate Intern in Paper Conservation, National Gallery of Art
Graduate Fellow in Paper Conservation, Winterthur/University of Delaware Program in Art Conservation