Reflections - Abstract - Josef Hormes

The corrosion of stained glass:
a chemical analysis using synchrotron radiation based X-ray absorption
and X-ray fluorescence spectroscopy

Josef Hormes ^(1,2)*, Wantana Klysubun⁽³⁾, Markus Kleine⁽⁴⁾

The corrosion of glass is a very complex process as it depends on external conditions (micro-climate, pollution) as well as on the composition of the glass (color, content of potassium etc.). While the basic mechanisms of the degradation process seem to be understood, there are still several open questions that should be answered for finding the "best" method for restoration (e.g. cleaning) and conservation of historic glass.

Because of its special properties (high intensity, collimation, continuous spectrum from the Infra-red to the hard X-ray region) synchrotron radiation based techniques allow the detailed and (at least in principle) non-destructive chemical characterization of the weathering crust but also of the "pristine glass": synchrotron-radiation excited X-ray fluorescence (SR-XRF) allows the (semi-quantitative) determination of the elemental composition of the samples incl. minor and trace elements and also "low-Z-elements" such as S, Si, Mg and Na. X-ray absorption near edge structure (XANES) spectroscopy provides detailed information about the chemical speciation of the elements of interest and thus direct information about corrosion processes. In most cases, for example, the valence state of the atomic species of interest can be determined and conclusions can be drawn on the chemical environment of these atoms by comparing the XANES spectra of the "unknown compound" to those of appropriate reference compounds (fingerprinting method). By focussing the X-ray beam, both techniques can be applied 2d-spatially resolved and by focussing also the fluorescence radiation on the detector side also depth profiles can be measured.

In this contribution, we present several examples for the applications of SR-XRF and XANES experiments for the analysis of corrosion products from glasses from different production times and different origins: some glasses from the Cathedral in Paderborn (~ 12th century) several glasses of different color from window 116 of the Cathedral in Chartres (~ 1230 A.D.) and glass from one window from the Cathedral in Sevilla (16th century). For the glasses from Chartres only the mechanically removed corrosion products were measured and for the glasses from Paderborn the glasses with the corrosion layer. Parts of the glass from the Seville Cathedral were cleaned using three different techniques:
1. mechanically using a spatula,
2. "chemically" using EDTA,
3. mechanically using a diamond file.
The corrosion products that incur using techniques 1 and 3 were collected and included into the investigation so that the corroded glass, the glass cleaned with one of the 3 techniques, the "pristine glass" and the corrosion products could be measured separately.

Experiments were carried out at the beamline 8 at the Synchrotron Light Research Institute (SLRI) in Nakhon Ratchasima (Thailand) and at the DCM beamline of the Center for Advanced Microstructure and Devices (CAMD) at Louisiana State University in Baton Rouge (USA). Fluorescence for XRF measurements was excited using monochromatic radiation of 10 keV and detected using either a 13-element Ge-detector or a Si-drift detector. For the assignment of elements and a rough calibration of fluorescence intensities NIST standard glass sample NIST 610 was measured. XANES spectra were recorded using double-crystal X-ray monochromators equipped with crystals suitable for the corresponding X-ray range - in most cases InSb(111) for lower and Ge(220) for higher energies. Most XANES experiments were carried out in fluorescence mode with the set-up used for the SR-XRF experiments.

As an example for the SR-XRF experiments, Figure 1 shows the SR-XRF spectra of parts from the front side of the "Seville glass" cleaned with the 3 different techniques. By looking, for example, at the Zn K_alpha line it is obvious that the different techniques remove significantly different amounts from this element that is obviously not part of the original glass, whereas an effective removal of the corrosion layer increases the intensity e.g. for the Ca K_alpha indicating that Ca is a part of the original "pristine" glass.

Figure 2 shows as an example for the XANES spectra, the Cu-K-XANES spectrum of a powdery corrosion product from a glass sample from the Cathedral in Chartres the in comparison with the spectra of two reference compounds (CuO and CuSO₄). Just by the visual comparison of the spectra it is obvious that Cu in the sample is not a sulfate as often discussed in the literature.

Authors' institutions:
1 Institute of Physics, Bonn University, Nussallee 12, D-53115 Bonn, Germany;
2 Center for Advanced Microstructures and Devices, Louisiana State University, Baton Rouge, USA;
3 Synchrotron Light Research Institute, Nakhon Ratchasima, Thailand;
4Glasmalerei Peters GmbH, Paderborn, Germany;
*Hormes@LSU.edu