X-ray fluorescence in Member States (Italy): Portable EDXRF in a multi-technique approach for the analyses of large paintings

Energy-dispersive X-ray fluorescence (EDXRF) with its portable capability, generally characterized by a small Xray tube and a Si-PIN or Si-drift detector, is particularly useful to analyze works of art. The main aspect that characterizes the EDXRF technique is its non-invasive character. This characteristic that makes the technique so powerful and appealing is on the other hand the main source of uncertainty in XRF measurements on Cultural Heritage. This problem is even more evident when we analyze paintings because of their intrinsic stratigraphic essence. As a matter of fact a painting is made of several layers: the support, which can be mainly of wood, canvas, paper; the preparation layer, mainly gypsums, white lead or ochre; pigment layers and at the end the protective varnish layer. The penetrating power of X rays allows that most of the times the information of all the layers reaches the detector. Most of the information that is in the spectrum arrives from deep layers of which we have no clue. In order to better understand this concept, let us use the equation of A. Markowicz. in which the various uncertainties that influence the analyses with portable EDXRF are reported. Let us adjust this equation for non invasive portable EDXRF analysis. The second, the third and the fourth term do not exist, for obvious reasons. Only the first and the last term influence the total uncertainty of an EDXRF analysis. The ways to reduce the influence of the fifth term is known by any scientist: good stability of the system, long measuring time, correct standard samples, good energy resolution etc. But what about the first term when we are executing a non invasive analysis? An example that shows the influence of the sample representation in the increasing of the uncertainty of a XRF analysis is the case in which we are asked to determine the original pigments used in a painting. If we have no clue of where restoration areas are dislocated on the painting, the probability of analyzing a "wrong" pigment is surely high and therefore high is the probability of having a high uncertainty in our results. Without a good representation of the painting to analyze, the total uncertainty could be very high even applying correctly all the analytical procedures for XRF analysis. The only way to reduce the influence of the sample representation in the EDXRF technique is to backup the EDXRF analysis with different kinds of techniques that help reduce the total uncertainty. These techniques, totally non invasive such as EDXRF, can be the imaging techniques such as UV light, IR reflectography and X radiography. These techniques have already an independent application in Cultural Heritage with their own proto - their results to the fact that: UV light is able to detect surfaces restorations, IR reflectography intercepts restorations in deeper layers while X rays define deeper problems. These imaging techniques gain new importance from a scientific point of view when they help choosing the right spot of analysis and therefore reduce the total uncertainty of EDXRF analyses. Surely the sample representation of a specific surface of the painting is greatly increased when, prior to any EDXRF analysis, we are able to fulfill non invasive imaging techniques. In a few words, we can state that Markowicz's equation demonstrates that portable non invasive EDXRF in Cultural Heritage needs a multi-technique approach to reduce the total uncertainty of its results