A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis B. Kanngießer, W. Malzer, I. Mantouvalou, D. Sokaras & A. G. Karydas Applied Physics A Materials Science & Processing ISSN 0947-8396 Volume 106 Number 2 Appl. Phys. A (2012) 106:325-338 DOI 10.1007/s00339-011-6698-0

1 23

Your article is protected by copyright and all rights are held exclusively by SpringerVerlag. This e-offprint is for personal use only and shall not be self-archived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication.

1 23

Author's personal copy Appl Phys A (2012) 106:325–338 DOI 10.1007/s00339-011-6698-0

I N V I T E D PA P E R

A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis B. Kanngießer · W. Malzer · I. Mantouvalou · D. Sokaras · A.G. Karydas

Received: 3 April 2011 / Accepted: 7 November 2011 / Published online: 30 November 2011 © Springer-Verlag 2011

Abstract Quantitative X-ray fluorescence (XRF) and particle induced X-ray emission (PIXE) techniques have been developed mostly for the elemental analysis of homogeneous bulk or very simple layered materials. Further on, the microprobe version of both techniques is applied for 2D elemental mapping of surface heterogeneities. At typical XRF/PIXE fixed geometries and exciting energies (15– 25 keV and 2–3 MeV, respectively), the analytical signal (characteristic X-ray radiation) emanates from a variable but rather extended depth within the analyzed material, according to the exciting probe energy, set-up geometry, specimen matrix composition and analyte. Consequently, the in-depth resolution offered by XRF and PIXE techniques is rather limited for the characterization of materials with micrometer-scale stratigraphy or 3D heterogeneous structures. This difficulty has been over-passed to some extent in the case of an X-ray or charged particle microprobe by creating the so-called confocal geometry. The field of view B. Kanngießer () · W. Malzer · I. Mantouvalou Institut für Optik und Atomare Physik, Technical University of Berlin, Sekr. EW 3-2, Raum 346, Hardenbergstr. 36, 10623 Berlin, Germany e-mail: [email protected] D. Sokaras · A.G. Karydas Institute of Nuclear Physics, NCSR ‘Demokritos’, 153 10 Athens, Greece Present address: D. Sokaras Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, CA 94025, USA A.G. Karydas Nuclear Spectrometry and Applications Laboratory (NSAL), International Atomic Energy Agency, 2400 Seibersdorf, Austria

of the X-ray spectrometer is spatially restricted by a polycapillary X-ray lens within a sensitive microvolume formed by the two inter-sectioned focal regions. The precise scanning of the analyzed specimen through the confocal microvolume results in depth-sensitive measurements, whereas the additional 2D scanning microprobe possibilities render to element-specific 3D spatial resolution (3D micro-XRF and 3D micro-PIXE). These developments have contributed since 2003 to a variety of fields of applications in environmental, material and life sciences. In contrast to other elemental imaging methods, no size restriction of the objects investigated and the non-destructive character of analysis have been found indispensable for cultural heritage (CH) related applications. The review presents a summary of the experimental set-up developments at synchrotron radiation beamlines, particle accelerators and desktop spectrometers that have driven methodological developments and applications of confocal X-ray microscopy including depth profiling speciation studies by means of confocal X-ray absorption near edge structure (XANES) spectroscopy. The solid mathematical formulation developed for the quantitative in-depth elemental analysis of stratified materials is exemplified and depth profile reconstruction techniques are discussed. Selected CH applications related to the characterization of painted layers from paintings and decorated artifacts (enamels, glasses and ceramics), but also from the study of corrosion and patina layers in glass and metals, respectively, are presented. The analytical capabilities, limitations and future perspectives of the two variants of the confocal micro X-ray spectroscopy, 3D micro-XRF and 3D micro-PIXE, with respect to CH applications are critically assessed and discussed.

Author's personal copy 326

1 Introduction The development and assessment of non-destructive methods for investigations of art and archaeological objects has become more and more important in recent years. In this context X-ray emission techniques play a major role in various analytical investigations, for screening purposes towards a complete compositional analysis. Among them X-ray fluorescence analysis (XRF) is the technique mostly used as it is feasible at a laboratory scale or even on-site when operated with an X-ray tube. Particle induced X-ray emission (PIXE) and X-ray absorption fine structure spectroscopy (XAFS) are two other important techniques for investigations in the cultural heritage (CH) field; however, they do require largescale accelerators as sources. These three analytical methods were successfully used in the past for quantitative elemental analysis (XRF and PIXE) or for chemical speciation (XAFS) of homogeneous bulk or very simple layered materials [1–3]. Further on, the invention and the further development of efficient X-ray optics and magnetic lenses enabled the microprobe version of these techniques, facilitating twodimensional (2D) mappings in the sub-micrometer regime. As a next step the possibility to obtain depth resolved compositional information was aimed for. The conventional X-ray based elemental techniques (XRF/PIXE) have very limited capabilities to resolve a stratigraphy or in-depth concentration profiles. The introduction of a confocal geometry at first in a typical micro-XRF set-up [4–6] and later on at the micro-PIXE probe [7, 8], provided a novel tool to ‘see’ in a dynamic way within the interior of a material and to facilitate quantitative investigations of layered structures extended up to tens or even hundreds of micrometers. The basic idea of the confocal geometry is the restriction of the detector’s field of view by an X-ray optic in a microprobe set-up. When the focus of the detector’s optic is intersecting the focus of the excitation optic of the microprobe, a probing volume in the micrometer regime is formed. By moving the object to be investigated through this probing volume, element-specific intensity profiles versus specimen position can be recorded and depth-resolved information can be obtained. Together with the 2D scanning mode of the microprobe a three-dimensional (3D) spatial resolution can be achieved. Thus, three-dimensional micro-XRF (3D microXRF), micro-PIXE (3D micro-PIXE) and micro-XAFS (3D micro-XANES) became possible since the middle of the past decade. In the first part of this review experimental set-ups for confocal micro X-ray spectroscopy at synchrotron sources, at particle accelerators and at laboratory scale together with the methodological developments are summarized. This includes discussion of relevant analytical capabilities and limitations as well as a summary of the current status of quantification schemes.

B. Kanngießer et al.

The potential of confocal micro X-ray spectroscopy for cultural heritage (CH) related investigations was already perceived from the very beginning of its development. In fact, one of the first methodological demonstrations was carried out on paintings [4]. Since then, various CH objects have been investigated showing unique features of the technique. Especially the possibility of depth resolved elemental analysis by 3D micro-XRF has been extensively exploited. Up until now, two main CH investigation topics have been carried out. On the one hand, the elucidation of manufacturing techniques by analyzing multilayered structured materials, like paintings and decorated artifacts, is an important question. On the other hand, the identification of corrosion or surface alteration mechanisms for conservation/restoration purposes has been pursued. They will be demonstrated by selected CH applications in the second part of this review.

2 Depth resolved elemental analysis in conventional XRF and PIXE techniques Experimental and methodological developments in XRF/ PIXE techniques, using either fixed or multiple experimental set-up parameters, have supported elegant solutions in a broad range of analytical problems involving layered or in-depth inhomogeneous materials. In the simpler approach, i.e. without varying any experimental parameter, different emission lines are utilized and the variable information depth versus X-ray energy is exploited. For example, the XRF/PIXE intensity ratio among different emission lines of the same (K/L or L/M) or different elements can probe efficiently gilding layers of a few microns [9–12] or they can even resolve a near-surface stratigraphy [13, 14]. In general, the XRF/PIXE analytical information depth can be tuned by varying individually, or in a combined mode, certain experimental parameters such as the incident beam energy and the incident or detection angles. For example, low-energy (below 1 MeV) PIXE and variable detection angle is sensitive to near-surface regions at the µm scale for thin-film investigations [15, 16], whereas high-energy PIXE [17] offers valuable non-destructive bulk analytical information beyond extended surface layers (patina or corrosion layer, coating etc). On the other hand, energetic X-ray beams can investigate the spatial distribution of elements in subsurface paint layers [18, 19]. In the case of PIXE analysis, in particular, the incident proton energy variation method (the so-called ‘differential PIXE’) has found numerous applications in the field of CH, mostly in the study of paint layers [20–24] and plating of metallic objects [25, 27]. Differential PIXE takes advantage of the strong dependence of the atomic inner shell ionization probabilities with the proton energy; increasing the incident energy, a more extended layer of the

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

material under investigation contributes to the characteristic X-ray emission. Moreover, the gradient of the PIXE yield variation versus proton energy (the technique sensitivity) becomes larger for elements located deeper but within the incident beam range [28]. The main drawback of all the previous experimental approaches is that full quantification is limited and, if applicable, is confined to rather simplified layered structures. In principle, differential PIXE can reconstruct the paint layers and estimate, in some cases, the local paint layer thickness [24, 29, 30]. The application of advanced deconvolution algorithms further enabled the quantitative reconstruction of single or double layers on substrates (up to 20–40 µm), in terms of the metal content versus depth and incorporating an advance information for the density of the paint layers [23]. Similarly, tinning on bronze and gilding layers of silver and gold on brass, bronze and silver objects from a few up to 10 µm were reconstructed by means of differential PIXE for incident proton energies between 2.78 and 0.74 MeV [26, 27]. Integrated analytical approaches which combined different excitation probes and techniques have also aimed to resolve layered structures in various CH-related materials. In the MOLAB (the mobile facilities for in situ non-invasive measurements, co-funded by the European Commission) multitechnique approach, in-situ non-invasive investigation of layered matrices is assisted by combining XRF elemental information with molecular and structural insights from electronic and vibrational spectroscopies [31]. In a combined study for wood or canvas paintings, XRF analysis was used to reconstruct the detected layers’ thicknesses, whereas visible reflectance spectroscopy (vis-RS) reported the pigment concentration in the uppermost layer [32, 33]. Other efforts have tackled the stratigraphy problem of paint layers by the combined use of XRD and α-PIXE [34], in corrosion layers by combined XRF and LIBS (laser-induced breakdown spectroscopy) [35, 36], whereas the silver bulk composition (fineness) of Roman coins was investigated in a more elaborated way by incorporating deep proton activation analysis together with XRF and α-PIXE [37]. In general, again, complete quantification in a noninvasive manner relies on certain assumptions and a profound knowledge of the object investigated. One more recent example is the investigation of an easel painting with a portable XRF system by de Viguerie et al. [14]. With a simple differential approach they succeeded in estimating the layer thickness of varnish and of a lead-white-based painting layer of a Renaissance painting. The authors assumed a two-layered system even though the painting consists of at least three layers. This shows that it is often sufficient to work with simpler models and techniques to obtain reasonable data for understanding the structure of the investigated material. The combined and simultaneous application

327

Fig. 1 Principle optical arrangement for a confocal set-up

of RBS (Rutherford Back Scattering) and PIXE techniques to cross sections of paint layers, in contrast, can determine the absolute in-depth concentrations of the binder and of the inorganic element that is a major, minor or even a trace constituent of a pigment [14, 38]. However, in this case the method becomes destructive as cross sections have to be taken from the CH object.

3 Experimental set-ups for confocal X-ray spectroscopy analysis 3.1 General considerations A confocal X-ray spectroscopy set-up, irrespectively of the excitation mode, exhibits basic common characteristics and instrumental components: – An incident microprobe is formed by magnetic quadrupoles in the case of a charged particle beam or by an X-ray optical element, most often a polycapillary lens for X-ray tube excitation or a polycapillary half lens for synchrotron radiation. At synchrotron radiation facilities also other X-ray optics, like compound refractive lenses, are being used. Typical spot sizes for an incident photon microprobe created by polycapillary lenses range between 10 µm and 30 µm depending on the energy of the exciting radiation, whereas for the proton microprobe 1–2 µm in vacuum or about 14 µm in atmospheric pressure can be attained. – The second X-ray optical element placed in the detection channel is usually a polycapillary half lens attached and manually adjusted on the detector snout or placed independently onto a multidimensional motorized stage. The spot size apparent for the detector is dependent on the energy of the radiation being detected. The detector is

Author's personal copy 328

always a semiconductor detector registering an energydispersive spectrum. The overlap of the focal spots of the optics forms the probing volume enabling three-dimensionally resolved measurements. An alignment procedure in both modes is carried out to overlap the focal region of the incoming microprobe together with the detector’s lens focus. Figure 1 shows a sketch of the principle arrangement in a confocal set-up. – A specimen stage manipulator that allows at least three translations with less than one micron precision. For depth-profiling measurements, the investigated specimen is scanned within the probing volume with very few µm step size along the surface’s normal direction. Threedimensional measurements can be performed by translating the specimen in three directions through the probing volume. Other scanning possibilities exist with respect to the 3D micro-PIXE set-up. – Additionally, an optical microscope is used for the surveillance of the specimen. The short working distances of the X-ray optics of usually several millimeters only require an optical surveillance of the specimen to prevent damage. In addition, the optical microscope may facilitate the alignment procedure of the specimen when properly adjusted. 3.2 3D micro-XRF The 3D micro-XRF method has been realized almost at the same time at synchrotron radiation facilities and at laboratory scale with X-ray tubes. Today there exist reported confocal set-ups for 3D micro-XRF at seven synchrotron radiation facilities around the world. Namely, they can be found at BESSY [39], Hasylab [40] and ANKA [41] in Germany, at the ESRF in France [42, 43], at CHESS in the USA [44], at the Beijing synchrotron facility in China [45] and at the LNLS in Brazil [46]. In order to form the probing volume, all of the set-ups use a polycapillary half lens in the detection channel. It is mandatory for the confocal set-up as no other current X-ray optic is able to transport a broad energy range without the necessity to be moved. Slight differences occur in the choice of the detector but all of them are semiconductor detectors based on silicon. Likewise, differences for the used X-ray optics in the excitation channel can be found. The optics are polycapillary half lenses, monocapillaries, Kirkpatrick–Baez mirrors and compound refractive lenses (CRLs) producing a spot size in the succession of their listing with the smallest one for the CRLs. Nevertheless, for the spatial resolution attained with the probing volume the polycapillary lens in the detection channel is decisive. Roughly, spatial resolutions in terms of width of the probing volume down to around 10 µm are achieved for photon energies above 10 keV.

B. Kanngießer et al.

Fig. 2 Confocal set-up for 3D micro-XRF and 3D micro-XANES at the synchrotron radiation facility BESSY in Berlin

This holds for a confocal geometry in which the angle between the optical axes of the excitation X-ray optic and of the collecting X-ray optic is 90°: if this angle is larger a slightly better spatial resolution is obtained while the lateral spot size on the specimen’s surface becomes larger. Especially for stratified materials with homogeneous layers a deviation from the usual 90° arrangement of the X-ray optics may be beneficial for a higher spatial resolution. Figure 2 shows the confocal set-up at the synchrotron radiation facility BESSY in Berlin. The two polycapillary half lenses forming the probing volume, the seven-element Si(Li) detector and the specimen stage are pictured in top view. Laboratory confocal set-ups deploy X-ray tubes as source and a full polycapillary lens instead of a half lens to concentrate the tube radiation. The X-ray tubes being used have anode spot sizes of several tens of microns. The detection side remains the same as for set-ups at synchrotron radiation facilities. Various groups already published properties and characterization of their laboratory confocal set-ups [5, 6, 47–50]. The width of the probing volume achieved in these set-ups is in the range of micrometers larger than the ones at synchrotron radiation facilities. This can be mainly attributed to the anode spot size of the X-ray tube. But the most important difference lies in the flux provided by X-ray tubes, which results in lower elemental sensitivities by around one to two orders of magnitude. In comparison to conventional micro-XRF, the second lens reduces the sensitivity of confocal set-ups by around

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

329

Fig. 3 Sketch of the confocal set-up for 3D micro-PIXE presenting the equivalency of the 2D lateral scanning mode (left-hand side) to the specimen scanning mode (right-hand side)

two orders of magnitude. At synchrotron sources, detection limits below 100 ppm can be achieved with acquisition times of a few tens of seconds. However, for investigations of CH objects the sensitivity is sufficient. A more serious limitation is the shortening of the range of detected elements. The analysis of elements lighter than calcium is hampered as a consequence of the soft X-ray energy cut-off induced by the polycapillary lens placed in the detection channel. In principle, the creation of a probing volume allows not only for 3D elemental analysis but also for a combination of methods, like 3D micro-XRF combined with 3D micro X-ray absorption near edge spectroscopy (3D microXANES) for chemical speciation. First experiments have been reported using 3D micro-XANES for investigating geological samples by Denecke et al. [51] and by Silversmit et al. [42]. In order to perform these investigations without absorption distortion of the spectra, the samples had to be polished. For a broader use of 3D micro-XANES, especially for investigation of cultural heritage objects, we are currently developing an appropriate absorption correction. 3.3 3D micro-PIXE A proton microprobe features certain very interesting properties with respect to the implementation of a 3D microPIXE set-up: an excellent spatial resolution down to a few micrometers achieved by a sequence of magnetic quadrupoles, a fast and precise 2D scanning mode driven by magnetic deflection coils placed across the beam path and the potential to apply and combine together the analytical information from different ion beam analysis (IBA) spectroscopic techniques with complementary analytical range (organic elements) and elemental depth resolving capacity. 3D micro-PIXE can look into an extended depth of several tens of micrometers, whereas RBS spectroscopy is best suited for depth analysis below 10 µm (with 3-MeV protons). The 2D microprobe scanning mode mentioned above

offers important advantages in the 3D micro-PIXE experimental methodology. For example, if the investigated specimen has a layered structure, it is not required to perform a complete translation of the specimen along the beam direction in order to achieve in-depth-resolved elemental analysis. A lateral 2D scanning mode with the proton beam and with the specimen placed at a fixed position within the confocal volume (close to or at the focal point) is equivalent with the procedure where the specimen is translated in one direction within the probing volume. More specifically, any measurement that results from a microbeam incidence at the specimen’s lateral surface uniquely corresponds to an indepth translation value of the specimen within the probing volume. Sokaras et al. [52] have shown the equivalency of the two experimental modes as presented in Fig. 3. The lateral displacement of the proton microprobe from position A to position B (left-hand side) is equivalent with the respective specimen translation along the perpendicular to the surface axis (right-hand side). Further on, adding a translation movement to the specimen together with the 2D microprobe lateral scanning, a sequence of element-specific 2D intensity images versus depth can be generated. Contrary to the variety of existing 3D micro-XRF experimental arrangements, very few 3D micro-PIXE set-ups have been reported so far. At the Oxford nuclear microprobe endstation of the Jozef Stefan Institute (JSI), Ljubljana, Slovenia, a confocal PIXE set-up was realized within a spherical experimental vacuum chamber equipped with a five-axis goniometer for specimen positioning [7]. The specimen was placed with its surface vertical to the microbeam axis and by moving the specimen across that axis depth-resolved measurements were performed. In subsequent 3D micro-PIXE experiments the experimental methodology was further improved by utilizing the 2D lateral scanning mode for 3D element-specific analysis, whereas the geometry was also optimized by inclining the specimen with respect to the incident beam direction in order to bisect the angle formed

Author's personal copy 330

between the beam and the X-ray detector [53–56]. A typical size for the JSI microprobe, with 3 MeV of proton energy, was measured to be 1.6 µm (2.4 µm) in the horizontal (vertical) direction at the focal point, whereas it spreads up to 3 µm at the distance of ± 450 µm. The confocal geometry was formed after attaching respective polycapillary half lenses on the snout of two X-ray detectors, a high-purity germanium X-ray detector HPGe and a Si(Li) detector. Due to its thin Be window, the Si(Li) detector enabled the detection of X-rays with energies down to 800 eV. The proton dose accumulated over individual 3D micro-PIXE measurements was used for normalization purposes. The concept of confocal X-ray spectroscopy utilizing an incident charged particle microprobe was also transferred to the external nuclear microprobe of the AGLAE facility (C2RMF, Paris, France). A 3D micro-PIXE set-up with the specimen in air was developed, offering thus the possibility to investigate specimens or artefacts without any size restrictions [8]. The proton microbeam is extracted through a 100-nm Si3 N4 membrane into a helium atmosphere that minimizes the beam energy straggling, angular scattering and lateral broadening. In the case of the AGLAE set-up, the FWHM of the external 3.5-MeV proton microbeam was measured to be 14 µm through the derivative of a 2-µm copper foil knife edge scan performed at the focal point. It is estimated that the proton microbeam FWHM will increase up to about 24 µm at a position 500 µm away from the focal point. The confocal set-up at AGLAE included two Si(Li) X-ray detectors, one with a polycapillary half lens on its snout. The signal of the second Si(Li) X-ray detector was used as a monitor of the proton beam current. For the depth scans, the specimen was moved parallel to the normal of its surface. The depth-resolved capacity of the external microbeam confocal set-up was tested with a multilayered sample [8]. This self-manufactured sample was built up of three thin one-element foils (Ti 2 µm as upper layer, Ni 5 µm as intermediate layer and Cu 2 µm as bottom layer) with 8-µm-thick Kapton foils as spacer between the metal foils. This ‘sandwich’ sample was moved stepwise through the probing volume, whereas at each position individual fluorescence spectra were recorded. The centroids of the Gaussian-like intensity curves obtained from the principal fluorescence line of each metal foil represent actually their in-depth distribution. Further on, the FWHM of each intensity profile expresses the size of the probing volume for the particular X-ray energy. It is evident from this simple experiment that the depthresolving capacity of 3D micro-PIXE is better than the analytical potential of RBS spectroscopy.

B. Kanngießer et al.

4 Quantitative analysis in 3D micro-XRF and in 3D micro-PIXE 4.1 General considerations The experimental nucleus of the depth-resolving capabilities, the creation of a probing volume by X-ray optics, necessitates new quantification procedures. This is mainly due to the fact that the employed X-ray optics modify not only the spatial but also the intensity distribution of the source radiation, whereas both distributions are energy dependent. Especially the X-ray optic placed in front of the detector, which is a polycapillary half lens for all confocal set-ups, has a big impact on depth resolution and sensitivity. As generated by their working principle, polycapillary lenses produce focal spot sizes and exhibit a photon energy dependent transmittance. Furthermore, the recorded fluorescence radiation intensity that emanates from the created probing volume no longer expresses mass deposition of an analyte but its local elemental density [57]. Hence, conventional quantification procedures for micro-XRF or micro-PIXE cannot be employed. The main task for an adequate quantification procedure for both confocal methods is the description of the probing volume in terms of its energy-dependent shape and intensity distribution. 4.2 Quantitative analysis in 3D micro-XRF The first paper dealing with a description of the probing volume was published by Smit et al. [58]. The authors assumed a spherical probing volume and integrated its analytical description in the conventional fundamental parameter approach for XRF analysis assuming a specimen with a layered structure. With this first approximation to describe the influence of the probing volume in the measured X-ray intensities, concentrations and layer thicknesses of historic paint fragments and car paints could be estimated. Figures for the accuracy of this approach were not given. A refined model of the probing volume created by two polycapillary optics was proposed by Malzer and Kanngießer [57]. Assuming a two-dimensional function for the transmission function of each polycapillary lens, an ellipsoidal shape as an expression for the set-up’s sensitivity is obtained, see Fig. 4. The figure shows a principle sketch of both polycapillary lenses’ intensity distributions in the focal spot region. The width of the intensity distribution of a polycapillary lens in the focal spot region does not change over several hundreds of micrometers depending on the type of polycapillary lens. This is depicted by the two cylindrically shaped intensity distributions. The overlap of the intensity distributions results in an ellipsoidal shape representing the probing volume. The top view of the ellipsoidally shaped intensity distribution elucidates the change of sensitivity and

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

Fig. 4 Principle sketch of the intensity distributions of the two polycapillary lenses in the focal spot region creating the probing volume in the confocal set-up

spatial resolution when a specimen is moved perpendicular to its surface along the three directions indicated by the three angles. In contrast to conventional XRF analysis, the sensitivity is now converted to a spatially dependent function. This model is the heart of an expanded fundamental parameter approach that expresses the primary fluorescence intensity for a bulk material. Out of this expression a calibration procedure for thick reference material could be derived in which two relevant, energy-dependent parameters are obtained. The two relevant calibration parameters are the integral sensitivity and the width of the probing volume. For each fluorescence line both parameters have to be determined. Even if the properties of the two polycapillary lenses would have been determined independently before their use in the confocal geometry, a calibration procedure is still necessary. Slightly different illumination of the polycapillary lenses and divergence differences occurring by introducing this X-ray optic in a certain set-up can alter their transmission function and spot size, as has been shown by Wolff et al. [59]. Mantouvalou et al. [60] expanded the expression developed in [7] for bulk material to stratified material. The equation for the primary fluorescence intensity of stratified material delivers a quantitative reconstruction algorithm for the composition and thickness of layers. The intensity expression is used in a fitting algorithm to the measured depth profiles with the thicknesses of the layers and their compositions as free parameters. Therefore, the properties of the specimen obtained, i.e. the thickness of the layers and their compositions, are the results of this fitting procedure. For the quantification of unknown stratified material the number of layers and an initial guess for their composition must be given. This includes the ‘dark matrix’, which consists of the

331

non-detectable elements as well as the overall density of the layers. The validity and accuracy of the reconstruction procedure were tested with bulk and stratified reference materials. The latter was especially developed for this purpose as no adequate stratified reference material was available [61]. Compared to quantification procedures for conventional XRF the results are less accurate. Deviation of 20–30% to the reference values was found. Higher uncertainties may even occur for trace elements or for matrices with strong absorption. In comparison to classical micro-XRF, the quantification is more laborious and the results are less accurate. The reconstruction procedure has to be adapted to each specimen and judged by the user interactively. Different fitting strategies may be applied according to the nature of the specimen. The problem of the so-called ‘dark matrix’ is an important aspect for all fundamental parameter based quantification approaches. The ‘dark matrix’ is constituted by all elements that cannot be measured with XRF. These are light elements for which the sensitivity of the experimental set-up is not sufficient. In the case of 3D micro-XRF the problem is more pronounced. More elements constitute the ‘dark matrix’ as the sensitivity for light elements is reduced in comparison to XRF or micro-XRF due to the restricted transmission of the X-ray optics. In general, all the elements being lighter than Ca are hard to detect, whereas for state-ofthe-art micro-XRF analysis the measurable range starts with Na. Hence, for quantitative 3D micro-XRF it is even more important to gather information on the light-element constituents of the specimen. An easy to perform measure is the combination of micro-XRF and 3D micro-XRF. Experimental set-ups for 3D micro-XRF very often have the possibility to execute micro-XRF at the same time or consecutively. With its higher sensitivity for light elements, microXRF measurements may provide insights on more light elemental constituents of the ‘dark matrix’. Thus, it results in a better estimation of the matrix, which will facilitate an improved quantification for 3D micro-XRF, see for example [47]. For stratified material even a recursive procedure can be employed to improve the quantification. Here, the layer thickness determined with 3D micro-XRF is introduced in the micro-XRF quantification procedure to improve its accuracy. An example for this approach applied in the investigation of historical parchment has been recently published by Mantouvalou et al. [72]. The paper also exhibits the advantages of using scattered radiation in the quantification procedure. A different quantification approach for the 3D microXRF measurements was presented by Vekemans et al. [62]. The authors expanded the well-established multivariate data analysis in scanning micro-XRF experiments to 3D microXRF. Their approach was mainly motivated by the short acquisition time per measurement point that they encountered

Author's personal copy 332

during synchrotron radiation experiments. Therefore, elemental fluorescence intensities were grouped using a clustering technique. Elements with higher fluorescence energies for which absorption effects are not so pronounced were taken to locate specific regions with distinct specific elemental composition. Considering these regions as clusters, the whole fluorescence spectra were summed in order to obtain sufficient statistics for quantification. The following quantification of these sum spectra was carried out with a fundamental parameter procedure relying on a calibration with a reference material. Despite the fact that this approach has severe limitations due to the omission of the self-absorption correction, it presents considerable advantages when a vast amount of data obtained from only slightly changing matrices needs to be handled. The quantification procedures for 3D micro-XRF presented so far are suited for confocal set-ups at synchrotron radiation facilities using monochromatic X-rays. No quantification approaches or procedures have been published up to now for polychromatic excitation. The reason can be found in the very nature of the polycapillary optics; when polychromatic X-rays are transmitted through such an X-ray optic, the spot at the focus is actually a convolution sum of spots having different sizes for each transmitted X-ray energy. The flux contribution of each spot to the sum spot is determined by the spectral flux of the polychromatic source weighted by the transmission function of the polycapillary lens. This has to be taken into account by an adequate model for the quantification. Hence, current measurements with laboratory confocal set-ups yield qualitative elemental information only. Nevertheless, in certain cases also qualitative analysis of 3D micro-XRF measurements with polychromatic X-ray tube excitation is sufficient to solve specific, e.g. art historical, questions, as has been shown by Mantouvalou et al. [47]. 4.3 Quantitative analysis in 3D micro-PIXE A quantification procedure in 3D micro-PIXE analysis aims to determine the unknown element local densities for the general case of a 3D inhomogeneously distributed analyte. The crucial difference between 3D micro-PIXE and conventional PIXE analysis is the fact that instead of total analyte intensity at specific incident particle energy, an elementspecific intensity profile versus depth is obtained as the specimen is translated across the probing volume. An alternative procedure, relatively faster, more flexible and precise, is the 2D scanning of the specimen by the microprobe. In both 3D micro XRF and 3D micro PIXE techniques the probing volume does not represent an isotropic space in terms of excitation and detection efficiency of an analyte, even if matrix effects are neglected. At different regions of

B. Kanngießer et al.

the probing volume the incident flux varies and the generated X-rays do not possess the same probability to be captured by the X-ray lens placed in front of the detector. Therefore, at each specimen or/and microprobe position the contribution of a local elemental density concentration to the measured intensity is weighted/modulated by the probing volume ‘space’ probability. The first effort towards quantification in 3D micro-PIXE analysis would be to describe analytically the probing volume (probability) or, better expressed hereafter, the sensitivity function. The proton microprobe cross section can be fairly enough approximated by a rectangular or homogeneous flux density distribution. Hence, it can be expressed in terms of two Heaviside functions. The proton microprobe cross section does not change noticeably along its propagation direction for about 200–300 µm from the focal plane [7]. The acceptance function of the X-ray lens in the detection channel is well described by a Gaussian-like function. The convolution of the two distributions after proper transformation of coordinates will result in an analytical formula for the sensitivity function (assuming that the center of the microprobe direction is passing through the lens focus) that is expressed by the difference of two error functions. For a microprobe in vacuum (size below 2–3 µm) a Gaussian function represents quite accurately the depth sensitivity of the confocal micro-PIXE set-up. For a microprobe in atmospheric pressure (beam size > 12–20 µm), the shape of the sensitivity profile will start progressively and in particular at its tails to deviate from the Gaussian shape [52]. Sokaras et al. [52] have presented an analytical model that describes the PIXE intensity in confocal geometry of an analyte for the case of a layered structured material. The theoretical model has been further extended to include secondary inter-layer fluorescence enhancement [63]. Based on the proposed formalism, simulations of the 3D micro-PIXE analysis of various layered materials have highlighted the potential of the technique. For example, in the case of a composite paint layer with the first layer (20 µm) composed of 5% of azurite diluted in an organic binder and the second (30 µm) of 5% lead–tin yellow pigment onto 100% calcium carbonate substrate, direct analytical information can be extracted even from the raw data in terms of number of layers, the elemental composition per layer and some first approximation for the thicknesses of the first one or two layers. A more general 3D micro-PIXE quantitative formalism for 3D inhomogeneous materials has been proposed by Zitnik et al. [54]. In this approach the specimen is divided into microscopic voxels of a typical size 2 × 2 × 10 µm3 where the latter dimension is selected to be in accordance with the detector’s lens resolution. The formulation expresses the contribution of each cubic cell to the element-specific intensity recorded per pixel of sequential 2D images produced by the 2D microbeam scanning mode as the specimen is

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

333

Fig. 5 Virtual cross section across the interface between the armor and forge sections of the painting (left-hand side). The short solid line marked with red arrows in the reproduction of ‘The Armorer’s Shop’ indicates the location of the depth scans (adapted from [64])

translated within the probing volume. The authors have applied this formalism in the case of PM10 aerosol particles immersed in a very low density matrix (quartz filter) and determined with a micron precision positions and local concentrations of dust particles. In a subsequent application, Zitnik et al. [55] demonstrated a more advanced 3D compositional analysis and imaging of an extended irregular object (hematite) accounting properly for the proton microbeam energy loss and matrix X-ray self-absorption effects.

5 Applications of confocal micro X-ray analysis in cultural heritage This section gives an overview of confocal micro X-ray analyses of CH objects published so far. Among the first publications, emphasis is rather laid upon the principle usefulness of the method for CH investigations demonstrated on special CH objects. Others describe further developmental steps of the confocal set-up in combination with the work on concrete archaeometric questions. In latest publications, confocal micro X-ray analysis is used as a real analytical tool, often in combination with other methods, to answer archaeometric questions. CH applications related to the characterization of painted layers from paintings and decorated artefacts (enamels, glasses and ceramics), but also from the study of corrosion and patina layers in glass, are presented. Compared to the band width of applications described by Adriaens [64] in her constructive overview of research associated with the pan-European network COST Action G8, confocal micro X-ray analysis already covers a good part of it.

5.1 Paintings Kanngießer et al. [4] published the first investigations on CH objects and showed at the same time the proof-ofprinciple of 3D nicro-XRF at the synchrotron radiation facility BESSY II. Two Mughal miniatures (nos. 3 and 10) conserved in the album Inv. MIK I 5004 of the Museum of Indian Art, State Museums of Berlin, were analyzed for their paint layer composition. Questions of manufacturing technique and authentication were tackled. Successive paint layers could be distinguished with a depth resolution of about 10 µm. The elemental depth profiles showed a surprising succession of paint layers and the use of Zn for the ground layer in the case of the Mughal miniature no. 10. This was a first hint that this painting is a copy manufactured in the 19th century in contrast to the stylistic dating to the 17th century. The determination of elemental depth profiles in historic paintings was also the aim of Woll et al. [44]. They constructed a confocal set-up at the CHESS synchrotron radiation facility and characterized it. In addition, a semiempirical model for the evaluation of depth profiles of simple paint layer structures was presented. In two successive papers [65, 66] they used their development in 3D microXRF in combination with other methods to investigate a 17th century panel painting ‘The Armorer’s Shop’, which was attributed to David Teniers the Younger of Flanders. The study was motivated by the hypothesis that Jan Brueghel the Younger completed the depiction of the armor in this painting. It could be shown with 3D micro-XRF that a wooden plank on which the armor is depicted was painted before it was combined with the two other planks constituting the whole painting. A thorough analysis of the elemental depth profiles in combination with the findings of other methods, especially dendrochronology, strongly supports the art historical hypothesis.

Author's personal copy 334

Figure 5 shows a reproduction of ‘The Armorer’s Shop’ at which the short solid line marked with red arrows indicates the location of depth scans performed across the interface between the armor and forge sections of the painting. At the right-hand side elemental concentrations obtained from all depth scans are depicted, forming a virtual cross section across the interface. The lead and copper cross sections revealed the presence of a buried imprimatur layer, whereas iron, calcium and manganese are found only in the top layer. The buried layer is evidently a layer that was painted onto the armor plank before the plank was joined to the remainder of the panel. The succession of paint layers was also the objective of 3D micro-XRF measurements in the paper of Wei et al. [45]. The paper rather concentrates on the construction and characterization of the confocal set-up at the synchrotron radiation facility BSRF. The performance of the set-up has been evaluated and first investigations have been carried out on a faux bamboo painting of the Forbidden City in Beijing. Elemental depth profiling revealed a layered structure that gives a hint for restoration treatments of the painting. The latest publication on investigation of a painting was written by Faubel et al. [41]. This time a painting from the 20th century, Max Beckmann’s ‘Pierette und Clown’, was examined. Here the focus was laid upon the research on damage of paint layers caused by blisters and crater holes filled with metal soaps. These so-called protrusions could lead to the destruction of the ground layer. Understanding the damage processes may help to find adequate conservation techniques. The authors combined several analytical methods to elucidate the formation of the destructive metal soaps. For the measurements small samples of damaged areas were taken by the painting’s restorer. With 3D micro-XRF three-dimensional elemental distributions were recorded. They showed a highly inhomogeneous metal distribution where Zn plays an important role. The method was especially helpful for the measurements of closed blisters on which Raman spectroscopy could not be performed. The 3D micro-XRF measurements were carried out at the synchrotron radiation facility ANKA. All the investigations of paintings described above were carried out at synchrotron radiation facilities. This is mainly due to the fact that confocal set-ups there yield higher elemental sensitivities. Nevertheless, Kanngießer et al. [6] demonstrated with their first confocal laboratory set-up that 3D micro-XRF measurements on paint are also feasible with X-ray tubes. Elemental depth profiles of self-prepared structured paint layers and three-dimensional elemental distributions of inclusions of the paint were recorded, showing the principle suitability for CH investigations.

B. Kanngießer et al.

5.2 Glass A transition from research on paintings to glass is the investigation of reverse glass paintings by Kanngießer et al. [67]. In the center of this study was the question about damage phenomena of the paint layer reinforced by corrosion processes leading in the worst case to a complete loss of the glass decoration. Depth profiles of mobile elements were supposed to clarify the corrosion processes. 3D micro-XRF measurements carried out at the dedicated mySpot beamline at BESSY II could show that a corrosion process beginning at the glass/binding medium was initiated. In addition, with the use of an appropriate quantification procedure [60] the migration of relevant elements could be followed, thus reflecting the dissemination of the corrosion process. Another study on Medieval reverse glass paintings is presented by Mantouvalou et al. [47]. In this case the archaeometric question was directed towards the manufacturing technique. For medieval glass paintings a cold painting technique is characteristic. But in the preliminary XRF analysis black enamel was identified forming the black contour colour. Black enamel is typically employed for stained glass objects, i.e. it is fired onto the glass. In order to find out if a firing process has taken place, elemental depth profiles were recorded on several black enamel spots on a reverse glass painting of the 14th century, the ‘Lüneburger Meditationstafel’. The depth profiles were compared to those obtained from two self-manufactured reference samples with black enamel. One of them was fired, whereas the other one was just left to dry. The unfired sample showed a layered structure of paint and glass body. In contrast, the fired reference sample exhibited only one homogeneous layer. The depth profiles of the Medieval reverse glass painting revealed a clear layered structure. This is a strong indication that reverse painting on glass takes the glass substrate and the kind of contour painting from stained glass, but the further process of painting from panel painting. The mixing of two techniques might just be interpreted as a new art historian genre. The measurements were mainly carried out with a compact tabletop spectrometer for 3D micro-XRF operated with an X-ray tube. Even though quantitative investigations are not possible up to now for X-ray tube excitation in a confocal geometry, the study is a good example that qualitative 3D micro-XRF measurements also lead to illuminating results. Figure 6 shows the Medieval reverse glass painting ‘Lüneburger Meditationstafel’. The elemental depth profiles of a glass spot and of a black enamel spot, depicted at the righthand side, were obtained with the compact laboratory 3D micro-XRF spectrometer. The bottom profiles are normalized to 1 in order to exhibit the succession of fluorescence intensities [47].

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

335

Fig. 6 Elemental depth profiles of a glass spot and of a black enamel spot at the ‘Lüneburger Meditationstafel’ of the 14th century, National Museum Schwerin, Inv. no. G 2627 (adapted from [47])

5.3 Ceramics and glazes A 3D micro-XRF laboratory spectrometer was also employed by Nakano and Tsuji [68] in order to study Japanese lacquerware ‘Tamamushi-nuri’. The art objects serve as an example of a multilayered structure. The authors compare the performance of two different micro-XRF methods: micro grazing exit XRF (GE-XRF) and 3D microXRF. With both methods a two-layered paint structure could be detected. The difference between both methods in depth-resolved measurements was exemplified. Namely, with GE-XRF depth-selective measurements can be carried out which give integrated information over the footprint of the whole beam. In contrast, 3D micro-XRF enables a threedimensional scanning of the specimen. A special feature of the confocal set-up is the use of two X-ray tubes including the respective X-ray optics, improving thus the elemental sensitivity and expanding the analytical range. Recently, Guilherme et al. [69] published the first archaeometric investigations on Portuguese polychrome glazed ceramics. With a multimethodological approach pieces from two Portuguese production centers, namely Coimbra and Lisbon, were investigated. The main aim in this study was the elucidation of manufacturing techniques. Microchemical and microstructural data obtained by various methods revealed significant differences in the glazes produced in the two centers. In particular, the 3D microXRF measurements showed that the pigment elements are less disseminated throughout the glaze in the case of the pieces from Coimbra. Hence, the pigment elements are more concentrated at the surface of the ceramics, giving them a slightly different appearance.

The first example of applying 3D micro-XANES on CH objects can be found in the paper from Reiche et al. [70]. In their extensive study of Persian polychrome underglaze painted tiles from the 19th century, the authors employed a large variety of non-destructive spectroscopic techniques using microfocused beams. Among them 3D micro-XANES was used to confirm the chemical state of coloring agents in blue and blue-green zones. XANES spectra around the copper K-edge revealed the characteristic absorption bands for octahedral Cu2+ ions in an amorphous glassy phase. Further investigations with 3D micro-XANES were hampered during that time due to the lack of an appropriate absorption correction procedure. Likewise, the first investigations with 3D micro-PIXE at two different proton facilities present the new method’s applicability to CH research [7, 8]. Karydas et al. [7] presented the characterization of a layered structured material, i.e. of an Attic black gloss ceramic small fragment (TH/CLAKROP-34, 5 cent. BC, Makrygianni, Acropolis), that has been excavated in the period 2000–2003 at the Acropolis area. In principle, the fragment consists of two basic layers, one black gloss layer on the top with a typical thickness ranged between 20 and 25 µm fairly well adhered to a second layer, the ceramic porous body. According to AloupiSiotis [71], the Attic black gloss is produced by fine suspension of illitic clay, mainly consisting of polycrystalline magnetite particles of less than 0.2-µm size embedded in an amorphous vitreous matrix. The 3D micro-PIXE depth profiles have been described semi-quantitatively based on independent elemental concentrations measured by means of XRF and EPMA (electron probe micro analyses). This first

Author's personal copy 336

approach was evaluated to be very promising in terms of the analytical capacity of the technique to deduce the black gloss layer thickness and individual element concentrations per layer. The only exception was the depth profile of Fe, which can be taken as a sign for the existence of Fe-rich particles in the microprobe path. A simulation which takes into account a Fe-rich particle at a certain depth from the surface resulted in a better description of the shape and relative intensity of the Fe depth profile. In [8] 3D micro-PIXE measurements were carried out on an artificial prepared green patina layer, prepared onto the surface of a quaternary bronze specimen. The patina on the surface of the bronze specimen was produced by the art casting house Coubertin. The patina was treated in order to obtain a smooth surface, whereas a wax coating was applied for its future protection. The depth profiles showed a depletion of Zn and an enrichment of Cu in the patina layer. Hence, for the study of patinas 3D micro-PIXE is also a useful analytical tool.

6 Conclusions Overviewing the analytical features of confocal X-ray microscopy techniques, it should be quoted that their depth resolution is in principle tailored by the respective performance of the currently existing X-ray lenses and therefore its capabilities might be further extended in the future following the development of new-generation X-ray optics. Nevertheless, many synchrotron radiation beamlines dedicated to micro-XRF investigations offer the confocal technique as a standard option. Depth resolutions obtained go down to around ten micrometers. The experimental technique is also well established for XRF spectrometers with X-ray tubes, which may contribute substantially to a wider use and accessibility of the method. The sensitivity is considerably lower in comparison to the confocal set-ups at accelerators, but for many applications sufficient. A commercial spectrometer will soon be available. 3D micro-PIXE experiments have been reported for two beamlines up until now. Its analytical capabilities for many fields of applications have to be further explored. Amongst the two confocal microscopy variants, 3D micro-XRF is certainly more versatile considering the possibility to be also applicable on a small laboratory scale. But this explains only partly its wider applicability with respect to 3D micro-PIXE. The true non-invasive character of examination and the analytical range and sensitivity of 3D microXRF are in many cases advantageous for investigations in the field of cultural heritage. A proton microprobe, due to its superior spatial resolution, seems to be more suited for the investigation of 3D inhomogeneous materials with internal structures having a size at the very few micrometers

B. Kanngießer et al.

scale. In addition, the sensitivity of 3D micro-PIXE for light elements is better than for 3D micro-XRF. For the investigation of manuscripts and paintings, the depth profiles of the intensities of the characteristic X-ray lines are frequently qualitatively evaluated. Information on the depth distribution of the elements, on correlations etc can directly be obtained if the attenuation of the radiation is negligible. Quantitative evaluation is indispensable for objects made from glass or ceramics, as the intensity profiles are tremendously distorted by the attenuation of the radiation. Fortunately, methods for quantitative evaluation have been established for 3D micro-XRF and for 3D micro-PIXE. Quantitative evaluation has been applied to many types of CH objects, namely parchments, glasses and ceramics. The results contributed to research on e.g. corrosion, manufacturing or provenance. The solid mathematical foundation that has been developed to quantify confocal X-ray microscopy data adds to the technique, despite its complexity, the potential to reconstruct the stratigraphy of layered materials and the stoichiometry of buried layers. Unlike with conventional XRF, we would like to point out that currently quantification of depth profiles is a laborious task and requires considerable expertise, in particular if the attenuation is severe. A further emerging method of confocal micro X-ray spectroscopy is 3D micro-XANES. In principle, it offers the possibility to investigate the chemical state of a specimen in a depth-resolved and non-destructive manner. Currently, its application is hampered by particular absorption effects, which distort the XANES spectra. Correction algorithms are currently under development, but successful applications to CH objects are not yet published. In summary, confocal micro X-ray spectroscopy has reached a matured state as the latest developed analytical X-ray method. The course of its application for CH investigations shows all the phases of a new method towards its applicative implementation. First publications document the working principle, followed by examples of its first application on real CH objects. Now investigations using this method are published and are currently under way in which typical archaeometric questions, like manufacturing techniques or provenance, are being tackled. The investigations combine various analytical techniques where confocal micro X-ray spectroscopy plays a key role due to its nondestructive, depth-resolving capacity. These examples will hopefully constitute its widespread use on a routine basis in the field of CH investigations. The analytical characterization of cultural heritage materials and objects with rather complex structure will certainly benefit from the advances of X-ray microscopy techniques. The deep view into materials will certainly help to understand and preserve our cultural heritage in a better way.

Author's personal copy A deep view in cultural heritage—confocal micro X-ray spectroscopy for depth resolved elemental analysis

References 1. R.E. Van Grieken, A.A. Markowicz (eds.), Handbook of XRay Spectrometry, 2nd edn. (Dekker, New York, 2002). ISBN 0824706005 2. B. Beckhoff, B. Kanngießer, N. Langhoff, R. Wedell, H. Wolff (eds.), Handbook of Practical X-Ray Fluorescence Analysis (Springer, Berlin, 2006). ISBN 3-540-28603-9 3. S.A.E. Johansson, J.L. Campbell, PIXE: A Novel Technique for Elemental Analysis (Wiley, New York, 1988). ISBN 0471920118 4. B. Kanngießer, W. Malzer, I. Reiche, Nucl. Instrum. Methods Phys. Res. B 211–212, 259 (2003) 5. G.J. Havrilla, T. Miller, Powder Diffr. 19(2), 119 (2004) 6. B. Kanngießer, W. Malzer, A. Fuentes Rodriguez, I. Reiche, Spectrochim. Acta B 60, 41 (2005) 7. A.G. Karydas, D. Sokaras, Ch. Zarkadas, N. Grlj, P. Pelicon, M. Žitnik, R. Schütz, W. Malzer, B. Kanngießer, J. Anal. At. Spectrom. 22, 1260 (2007) 8. B. Kanngießer, A.G. Karydas, R. Schütz, D. Sokaras, I. Reiche, S. Röhrs, L. Pichon, J. Salomon, Nucl. Instrum. Methods Phys. Res. B 264(2), 383 (2007) 9. A. Gianoncelli, G. Kourousias, Appl. Phys. A 89, 857 (2007) 10. R. Cesareo, A. Brunetti, S. Ridolfi, X-Ray Spectrom. 37(4), 309 (2008) 11. A.G. Karydas, H. Brecoulaki, B. Bourgeois, Ph. Jockey, in Proceedings of 7th International Conference of Association for the Study of Marble and Other Stones in Antiquity. BCH Suppl., vol. 51 (2009), pp. 811–829 12. A.G. Karydas, Ann. Chim. 97(7), 419 (2007) 13. C. Neelmeijer, I. Brissaud, T. Calligaro, G. Demortier, A. Hautojärvi, M. Mäder, L. Martinot, M. Schreiner, T. Tuurnala, G. Weber, X-Ray Spectrom. 29(1), 101 (2000) 14. L. de Viguerie, L. Beck, J. Salomon, L. Pichon, Ph. Walter, Anal. Chem. 81, 7960 (2009) 15. B. Nsouli, M. Roumié, K. Zahraman, J.P. Thomas, M. Nasreddine, Nucl. Instrum. Methods Phys. Res. B 192, 311 (2002) 16. J. Miranda, J. Rickards, R. Trejo-Luna, Nucl. Instrum. Methods Phys. Res. B 249, 394 (2006) 17. A. Denker, J. Opitz-Coutureau, Nucl. Instrum. Methods Phys. Res. B 213, 677 (2004) 18. J. Dik, K. Janssens, G. Van Der Snickt, L. van der Loeff, K. Rickers, M. Cotte, Anal. Chem. 80(16), 6436 (2008) 19. K. Janssens, J. Dik, M. Cotte, J. Susini, Acc. Chem. Res. 43(6), 814 (2010) 20. C. Neelemeijer, W. Wagner, H.P. Schramm, Nucl. Instrum. Methods Phys. Res. B 118, 338 (1996) 21. C. Neelemeijer, M. Mäder, Nucl. Instrum. Methods Phys. Res. B 189, 293 (2002) 22. P.A. Mandó, M.E. Fedi, N. Grassi, A. Migliori, Nucl. Instrum. Methods Phys. Res. B 239, 71 (2005) 23. Ž. Šmit, M. Uršiˇc, P. Pelicon, T. Trˇcek-Peˇcak, B. Seme, A. Šmrekar, I. Langus, I. Nemec, K. Kavkler, Nucl. Instrum. Methods Phys. Res. B 266, 2047 (2008) 24. N. Grassi, Nucl. Instrum. Methods Phys. Res. B 267, 825 (2009) 25. G. Demortier, J.L. Ruvalcaba-Sil, Nucl. Instrum. Methods Phys. Res. B 118, 352 (1996) 26. Ž. Šmit, M. Holc, Nucl. Instrum. Methods Phys. Res. B 219–220, 524 (2004) 27. Ž. Šmit, J. Isteniˇc, T. Knific, Nucl. Instrum. Methods Phys. Res. B 266, 2329 (2008) 28. G. Lagarde, P. Midy, I. Brissaud, Nucl. Instrum. Methods Phys. Res. B 132, 521 (1997) 29. I. Brissaud, G. Lagarde, P. Midy, Nucl. Instrum. Methods Phys. Res. B 117, 179 (1996) 30. I. Brissaud, A. Guilló, G. Lagarde, P. Midy, T. Calligaro, J. Salomon, Nucl. Instrum. Methods Phys. Res. B 155, 447 (1999)

337

31. C. Miliani, F. Rosi, B.G. Brunetti, A. Sgamellotti, Acc. Chem. Res. 43(6), 728 (2010) 32. L. Bonizzoni, A. Galli, G. Poldi, M. Milazzo, X-Ray Spectrom. 36(2), 55 (2007) 33. L. Bonizzoni, S. Caglio, A. Galli, Appl. Phys. A 92, 203 (2008) 34. L. Pappalardo, G. Pappalardo, F. Amorini, M.G. Branciforti, F.P. Romano, J. de Sanoit, F. Rizzo, E. Scafiri, A. Taormina, G. Gatto Rotondo, X-Ray Spectrom. 37(4), 370 (2008) 35. J.A. Pérez-Serradilla, A. Jurado-López, M.D. Luque de Castro, Talanta 71, 97 (2007) 36. V. Kantarelou, Ch. Zarkadas, A. Giakoumaki, M. Giannoulaki, A.G. Karydas, D. Anglos, V. Argyropoulos, Innovative investigation of metal artifacts, in Proceedings of METAL-07, vol. 2 (2007), pp. 35–41 37. F. Rizzo, G.P. Cirrone, G. Cuttone, A. Esposito, S. Garraffo, G. Pappalardo, L. Pappalardo, F.P. Romano, S. Russo, Microchem. J. 97(2), 286 (2011) 38. L. Beck, L. de Viguerie, Ph. Walter, L. Pichon, P.C. Gutierrez, J. Salomon, M. Menu, S. Sorieul, Nucl. Instrum. Methods Phys. Res. B 268, 2086 (2010) 39. B. Kanngießer, I. Mantouvalou, W. Malzer, T. Wolff, O. Hahn, J. Anal. At. Spectrom. 23(6), 814 (2008) 40. B. De Samber, G. Silversmit, K. De Schamphelaere, R. Evens, T. Schoonjans, B. Vekemans, C. Janssen, B. Masschaele, L. Van Hoorebeke, I. Szaloki, F. Vanhaecke, K. Rickers, G. Falkenberg, L. Vincze, J. Anal. At. Spectrom. 25, 544 (2010) 41. W. Faubel, R. Simon, S. Heissler, F. Friedrich, P.G. Weidler, H. Becker, W. Schmidt, J. Anal. At. Spectrom. (2011). doi:10.1039/c0ja00178c. 42. G. Silversmit, B. Vekemans, S. Nikitenko, S. Schmitz, T. Schoonjans, F.E. Brenker, L. Vincze, Phys. Chem. Chem. Phys. 12, 5653 (2010) 43. L. Vincze, B. Vekemans, F.E. Brenker, G. Falkenberg, K. Rickers, A. Somogyi, M. Kersten, F. Adams, Anal. Chem. 76, 6786 (2004) 44. A. Woll, J. Mass, C. Bisulca, R. Huang, D.H. Bilderback, S. Gruner, N. Gao, Appl. Phys. A 83, 235 (2006) 45. X. Wei, Y. Lei, T. Sun, X. Lin, Q. Xu, D. Chen, Y. Zou, Z. Jiang, Y. Huang, X. Yu, X. Ding, H. Xu, X-Ray Spectrom. 37, 595 (2008) 46. R.D. Perez, H.J. Sanchez, M. Rubio, C.A. Perez, X-Ray Spectrom. 40(1), 19 (2011) 47. I. Mantouvalou, K. Lange, T. Wolff, D. Grötzsch, L. Lühl, M. Haschke, O. Hahn, B. Kanngießer, J. Anal. At. Spectrom. 25, 554 (2010) 48. X. Lin, Z. Wang, T. Sun, Q. Pan, X. Ding, Nucl. Instrum. Methods Phys. Res. B 266, 2638 (2008) 49. K. Nakano, K. Tsuji, J. Anal. At. Spectrom. 25, 562 (2010) 50. D. Wegrzynek, R. Mroczka, A. Markowicz, E. Chinea-Cano, S. Bamford, X-Ray Spectrom. 37, 635 (2008) 51. M.A. Denecke, B. Brendebach, W. De Nolf, G. Falkenberg, K. Janssens, R. Simon, Spectrochim. Acta B 64, 791 (2009) 52. D. Sokaras, A.G. Karydas, W. Malzer, R. Schütz, B. Kanngießer, N. Grlj, P. Pelicon, M. Zitnik, J. Anal. At. Spectrom. 24, 611 (2009) 53. M. Zitnik, P. Pelicon, N. Grlj, A.G. Karydas, D. Sokaras, R. Schutz, B. Kanngießer, Appl. Phys. Lett. 93, 094104 (2008) 54. M. Zitnik, P. Pelicon, K. Bucar, N. Grlj, A.G. Karydas, D. Sokaras, R. Schutz, B. Kanngiesser, X-Ray Spectrom. 38(6), 526 (2009) 55. M. Zitnik, N. Grlj, P. Vaupetic, P. Pelicon, K. Bucar, D. Sokaras, A.G. Karydas, B. Kanngiesser, J. Anal. At. Spectrom. 25, 28 (2010) 56. N. Grlj, P. Pelicon, M. Žitnik, P. Vavpetiˇc, D. Sokaras, A.G. Karydas, B. Kanngießer, Nucl. Instrum. Methods Phys. Res. B (2011). doi:10.1016/j.nimb.2011.02.072 57. W. Malzer, B. Kanngießer, Spectrochim. Acta B 60, 1334 (2005) 58. Z. Smit, K. Janssens, K. Proost, I. Langus, Nucl. Instrum. Methods Phys. Res. B 219–220, 35 (2004)

Author's personal copy 338 59. T. Wolff, I. Mantouvalou, W. Malzer, J. Nissen, D. Berger, I. Zizak, D. Sokaras, A.G. Karydas, N. Grlj, P. Pelicon, R. Schütz, M. Žitnik, B. Kanngießer, J. Anal. At. Spectrom. 24, 669 (2009) 60. I. Mantouvalou, W. Malzer, I. Schaumann, L. Lühl, R. Dargel, C. Vogt, B. Kanngießer, Anal. Chem. 80, 819 (2008) 61. I. Schaumann, W. Malzer, I. Mantouvalou, L. Lühl, B. Kanngießer, R. Dargel, U. Giese, C. Vogt, Spectrochim. Acta B 64, 334 (2009) 62. B. Vekemans, L. Vincze, F.E. Brenker, F. Adams, J. Anal. At. Spectrom. 19, 1302 (2004) 63. D. Sokaras, A.G. Karydas, Anal. Chem. 81, 4946 (2009) 64. A. Adriaens, Spectrochim. Acta B 60, 1503 (2005) 65. A. Woll, J. Mass, Ch. Bisulca, M. Cushman, C. Griggs, T. Wazny, N. Ocon, Stud. Conserv. 53, 93 (2008) 66. Ch. Bisulca, A. Woll, J. Mass, N. Ocon, C. Griggs, T. Wazny, M. Cushman, in Proceedings of 9th Int. Conf. NDT of Art, Jerusalem, Israel, 25–30 May, 2008

B. Kanngießer et al. 67. B. Kanngießer, I. Mantouvalou, W. Malzer, T. Wolff, O. Hahn, J. Anal. At. Spectrom. 23, 814 (2008) 68. K. Nakano, K. Tsuji, X-Ray Spectrom. 38, 446 (2009) 69. A. Guilherme, J. Coroado, J.M.F. dos Santos, L. Lühl, T. Wolff, B. Kanngießer, M.L. Carvalho, Spectrochim. Acta B 66(5), 297 (2011) 70. I. Reiche, S. Röhrs, J. Salomon, B. Kanngießer, Y. Höhn, W. Malzer, F. Voigt, Anal. Bioanal. Chem. 393, 1025 (2009) 71. E. Aloupi-Siotis, Recovery and revival of attic vase-decoration techniques, what can they offer archaeological research, in Papers on Special Techniques in Athenian Vases (J. Paul Getty Museum, Los Angeles, 2008), pp. 113–128 72. I. Mantouvalou, T. Wolff, O. Hahn, I. Rabin, L. Lühl, M. Pagels, W. Malzer, B. Kanngießer, Anal. Chem. 83(16), 6308 (2011)