Fresenius J Anal Chem (1994) 348:402-410 Freseniu$' Journal of © Springer-Verlag 1994 A critical evaluation of the environmental scanning electron microscope for the analysis of paint fragments in art conservation N. W. Bower 1, D. C. Stulik 2, and E. Doehne 2 t Chemistry Department, Colorado College, Colorado Springs, CO 80903-3294, USA 2 The Getty Conservation Institute, 4503 Glencoe Avenue, Marina del Rey, CA 90292-6537, USA Received March 29, 1993; revised July 5, 1993 Summary. Cross-sections from medieval paintings by Cenni di Francesco and Dosso Dossi were analyzed for the inor- ganic components as well as the binding media using an environmental scanning electron microscope (E-SEM). The advantages of this instrument compared to a normal SEM- EDX are illustrated and a number of optimization studies are reported. It was found that using a chamber gas pressure of 1.5 kPa and a tungsten source instead of the usual LaB6 source with the 38 kPa pressure normally used for imaging would significantly improve the x-ray analyses. Quantitative analyses for most of the common medieval pigments are also presented. Introduction The analysis of cross-sections of paint layers using microchemical and instrumental methods is important for the authentication and provenancing of works of art. The information obtained from these studies are useful not only for understanding the materials and techniques used by art- ists, but are also needed for developing an appropriate proto- col for the restoration or conservation of the work. As early as /800 wet microchemical tests were used to analyze artists' pigments in a study of medieval wall paint- ings [1]. Since then a variety of techniques have been employed, but light microscopy became the primary method after Laurie [2] demonstrated its use for the study of paint cross-sections at the beginning of this century. The appli- cation of electron microprobe analysis (EMPA) for microelemental analyses in conservation was demonstrated as early as 1963 [3, 4]. Scanning electron microscopy coupled with an energy dispersive detector (SEM-EDX) has ex- panded this approach, and both methods are widely applied in the field of art conservation today [5 - 7], although both methods require that non-conductive paint specimens be coated with a conductor and placed in a vacuum. Recent developments in SEM technology [8, 9] have removed these restrictions, opening the door to new applications and sim- pler sample preparations. Correspondence to: N. W. Bower Theory EMPA and SEM-EDX both use electron bombardment to excite x-rays in the sample [10, 11] that are characteristic of a sample's elemental composition. The degree of accuracy for analyses made with standardless ZAF corrections is approaching 5% for modern software packages from most of the detector manufacturers. This is good enough to correctly identify the common inorganic pigments used by artists. At atmospheric pressure an electron beam will be scat- tered and absorbed and the lower energy x-rays emitted from the sample may be attenuated. The heated filament normally used to produce the electron beam will also be destroyed. An additional problem is encountered with non-conductive samples such as paint films. The bombarding electrons can build up a local surface charge that is great enough to signifi- cantly deflect the beam away from the point of interest. The spreading of the incident electron beam from collisions with gas molecules and from sample charging will negatively impact the image quality of the electron micrograph and the area being analyzed spectrally. As a result of these problems, non-conductive specimens are typically coated with a conductive layer of gold, osmium, or carbon, The latter is best for x-ray analyses (though not for image quality) as it does not produce spectral lines at the energies typically detected with SEM-EDX instruments. However, this coating may interfere with further analyses as well as the image quality. For example, the identification of the binding media components using FT-IR microscopy [12] on the same sample would be difficult after the sample has been coated. The IR spectra are usually obtained from an adjacent cross-section obtained using a microtome. The IR analysis may also be done on the same cross-section that is used for the SEM-EDX analysis, if it is done before the sample is coated. However, there are times when it is helpful to reanalyze the same cross-section as more information is obtained or as new questions arise. The environmental scanning electron microscope (E- SEM) circumvents some of these problems. It uses a moder- ate pressure (0.75 to/50 kPa) of a gas such as water in the sample chamber instead of coating the sample to remove the excess charge [8, 9]. This is made feasible by a series of apertures and vacuum pumps between the sample chamber and the filament source.