Educational x-ray experiments and XRF measurements with a portable setup adapted for the

characterization of cultural heritage objects

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2010 Eur. J. Phys. 31 419

(http://iopscience.iop.org/0143-0807/31/3/001)

Download details:

IP Address: 195.130.109.152

The article was downloaded on 30/11/2012 at 09:56

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING EUROPEAN JOURNAL OF PHYSICS

Eur. J. Phys. 31 (2010) 419–431 doi:10.1088/0143-0807/31/3/001

Educational x-ray experiments and XRFmeasurements with a portable setupadapted for the characterization ofcultural heritage objects

I Sianoudis1, E Drakaki2 and A Hein3

1 Department of Physics Chemistry & Material Technology, Technological Educational Institute(TEI) of Athens, Ag. Spyridonos 12210 Egaleo, Greece2 Department of Physics, NTUA, Athens 15780, Greece3 Institute of Materials Science, N.C.S.R. Demokritos, 15 310 Aghia Paraskevi, Greece

E-mail: jansian@teiath.gr, edrakaki@gmail.com and hein@ims.demokritos.gr

Received 16 July 2009, in final form 4 January 2010Published 10 March 2010Online at stacks.iop.org/EJP/31/419

AbstractIt is common to modify valuable, sophisticated equipment, originally acquiredfor other purposes, to adapt it for the needs of educational experiments, withgreat didactic effectiveness. The present project concerns a setup developedfrom components of a portable system for energy dispersive x-ray fluorescencespectroscopy (EDXRF). Two educational modules have been developed onthe basis of this setup. Module 1 comprises a series of x-ray laboratoryexercises investigating basic principles, such as the verification of Moseley’slaw, Compton’s law and the Lambert–Beer law. Module 2 concerns thecalibration of the XRF with reference materials, aiming to get quantitativemeasurements of the elemental composition of objects of cultural interest.The application of the calibrated experimental setup is demonstrated withindicative measurements of metal objects and pigments of wall paintings, inorder to discuss their spectra, and their qualitative and quantitative analyses.The setup and the applied experiments are designed as an educational packageof laboratory exercises on the one hand for students in natural sciences, and onthe other for the education of students who will work in the field of culturalheritage, such as conservation science or archaeological science.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

X-ray fluorescence analysis (XRF) is one of the most common methods for the determination ofthe elemental composition of materials. There are basically two different kinds of techniques,

0143-0807/10/030419+13$30.00 c© 2010 IOP Publishing Ltd Printed in the UK & the USA 419

420 I Sianoudis et al

wavelength dispersive XRF (WDXRF) and energy dispersive XRF (EDXRF). Due to thebetter spectral resolution, WDXRF provides advantages particularly in terms of the numberof elements in the sample matrix to be analysed and their detection limits. EDXRF, however,in general allows for faster measurement. In cultural heritage, for example, EDXRF hasclearly gained importance in the last 15 years as an easy and fast tool for non-destructivedetermination of the chemical composition of artifacts. This development is based essentiallyon two technical advancements of the method. First, due to the development of Peltier cooledPIN detectors with sufficient efficiency and energy resolution to replace the nitrogen-cooled Sidetectors, it became possible to assemble small portable systems [1]. Second, developmentsin x-ray optics increased the spatial resolution [2]. However, even without additional x-rayoptics a recent portable XRF can be used to determine the major element composition of anartifact with sufficient precision and accuracy using a spot size in the range of a few mm2. Inthis way, within several minutes, for example, specific pigments in a painting can be examinedor particular areas on the surface of a metal object to investigate possible corrosion processes[3].

Even though the analytical method and its applications are common, the physical andspectrometric basics behind those are still complex, and imparting this knowledge willcontribute to basic understanding of atomic physics. Therefore, the components of a portableEDXRF setup were assembled for a series of educational experiments, in order to teach someof the basic principles of the interactions of x-rays and matter, such as the characteristicsof fluorescence lines, attenuation and the Compton effect. This first part of the experimentsconstitutes an educational module suitable for students in natural science. Based on the firstmodule, a second module was designed which focused more on the application of the techniquewith regard to cultural heritage. In general, as far as archaeological science or conservationscience are concerned, there are numerous applications of analytical systems and in particularportable XRF systems, in terms of materials to be analysed and questions to be answered.Study of materials in the field of cultural heritage therefore requires a basic knowledge ofthe characteristics of particular analytical techniques. Educational experiments enhance thecomprehension of the potential and of the limitations, of a technique like XRF. Both modulescan be implemented either in conjunction or independently.

In the present case, the educational objectives are the origin of an XRF spectrum,the relation between spectral energy and a particular element and the differences betweenqualitative and quantitative analysis. The students are supposed to learn how an XRF spectrumis generated and in which way it can be interpreted, in terms of qualitative and quantitativepeak evaluation. Basic understanding of terms such as detection limit, precision and accuracywill be achieved. An essential educational objective is the assessment of the potentials ofthe techniques applied to different materials. This concerns energy-dependent absorptionof x-rays in air, as well as matrix effects. Therefore, a set of sample objects, representingdifferent materials, are measured by the students. This set comprises metal objects composedof different alloys, a group of ceramic and soil samples, glass samples and a series of pigments.In this paper, the analysis of metal objects and pigments in wall paintings is demonstrated.

2. Analytical nomenclature

For the terms accuracy, precision and detection limit in the context of quantitative analysis,the nomenclature suggested by the International Union of Pure and Applied Chemistry(IUPAC) is followed [4]. While accuracy is the closeness of the analytical results to thetrue values, which can be tested for example with standard reference materials, precision isthe closeness of independent analyses, expressed in most cases as estimation of the statistical

Educational x-ray experiments and XRF measurements on cultural heritage objects 421

(a) (b)

Figure 1. (a) Overview of the standard XRF setup for the x-ray experiments. (b) A graphicalrepresentation showing a sample excited by the beam of an x-ray source with an Ag-anode withthe detector placed at a 90◦ angle. A collimator was used in order to control the beam size, whilea series of Al foils has been placed between the standard sample and the detector, in order toattenuate sequentially the intensity of the x-ray beam. The white arrow on the case of the x-raysource (a) indicates the path of the x-ray beam.

experimental error. The detection limit is the minimum detectable quantity, i.e. the quantitywhich corresponds to the minimum statistically significant signal. The detection limit dependson the signal-to-noise ratio, which can be controlled for example by the measurement period,the detector geometry or the intensity of the x-ray source.

3. Experimental approach

The detector used is an Amptek XR-100CR, semiconductor diode (Si-PIN), optimized todetect x-ray photons directly in the energy range of 1 to 30 keV. The basic principle of asemiconductor x-ray detector is that each photon produces a number of free electrons andholes, proportional to its energy, which are collected by electrodes in an external electricfield. Incomplete collection of electron–hole pairs and interferences due to the limited timeresolution can affect the signals and are source of electronic noise. The pre-amplified signalsare processed with an end-amplifier and a multichannel analyzer (MCA of Phywe or 8000A ofAmptek), sorting them in 4 k addresses (channels), according to their energy. The multichannelanalyzer is connected to a computer on which the data are presented in the form of spectra andgraphs. The spectral lines are assigned to the specific elements by using the software ADMCAand the quantitative analysis software XRF-FP, both of Amptek, while the data acquisitionuses the software Measure of Phywe and Origin of OriginLab Co. The x-ray beam is generatedwith a mini mobile x-ray source with Ag-anode (Amptek Laser-X) providing a low current(10 μA) with a maximum energy of 35 keV. The system is provided with two mini diodelasers, mounted under the constructive board, in order to define the exact excitation point ofthe sample. In this way, the distance of the sample to the x-ray source and the detector couldbe kept constant.

The experimental setup is an in-house construction, designed for the presented XRFmeasurements of different objects and standard reference materials (figure 1), mainly foreducational purposes.

For the basic x-ray experiments it was reassembled, like for the verification of the Comptoneffect (figure 2). For the calibration procedure and the verification of Moseley’s law, a seriesof metal reference standards and a multi-element standard in form of a pellet are used. Forthe verification of the validity of the Lambert–Beer law a simple aluminum foil is used as

422 I Sianoudis et al

(a) (b)

Figure 2. The x-ray setup reassembled for the Compton effect experiment (a). A Plexiglas wasused as a scattering target placed between the source and detector, which was positioned at differentscattering angles (b).

Figure 3. Characteristic XRF spectrum of the standard (BAM 376) sample used for the energycalibration of the system (the vertical scale is logarithmic). Recording spectrum time: 100 s.

an absorber material, appropriately folded, in order to achieve different values of surfacedensities. The thickness of each aluminum foil was approximately 12 μm.

4. Module 1: Experiments for the demonstration of the basic principles of XRF

4.1. Calibration

The first step in the series of x-ray experiments was the energy calibration of the spectrometerwith a commercial standard (BAM 376). For this reason, a sample in form of a metal discwas used, comprising components of Mn, Fe, Ni and Cu. Figure 3 shows the acquired XRFspectrum with the identified Kα-lines of the respective elements. A linear fit was applied basedon the reference values for the energies of the Kα emission lines and the channel numbers ofthe respective peak maxima in the spectrum.

The resulting calibration line is shown in figure 4, indicating a sufficiently linear relationbetween channel number and reference energy in the examined range of the spectrum. In thecase of the example presented, one channel corresponds to approximately 14 eV.

Educational x-ray experiments and XRF measurements on cultural heritage objects 423

Figure 4. Fitted calibration line, based on the data of the components’ energies contained in thestandard sample and the channel number of the corresponding x-ray lines in the XRF spectrum(figure 3).

4.2. Moseley’s law

The correlation of emission energy and atomic number was investigated in the second step.The energy Emn of each transition from atomic level m to atomic level n is proportional to thesquare of their atomic number. The relation between emission energy and atomic number isdescribed by Moseley’s law:

Emn = RH · (Z − σ)2 ·(

1

n2− 1

m2

)(1)

with RH being the Rydberg constant and σ being the shield constant of the atom. In the caseof the Kα emission line with m = 2 and n = 1, the energy is

E21 = 34 · RH · (Z − σ)2 (2)

or √E21 = B · Z − B · σ (3)

with

B =√

0, 75 · RH . (4)

Moseley’s law is an empirical solution for the characteristic x-rays. It indicates that as onegoes from the lighter elements to the heavier elements, the energy of the characteristic x-raysemitted from a sample increases in a regular manner and is approximately proportional to(Z − σ )2, from which the atomic number Z of a sample can be identified [5, 6].

The validity of Moseley’s law, within the range (22 > Z > 35) for atomic number, isshown in the plot of figure 5 and after appropriate fitting to the measured data, the linearrelationship obtained is in accordance with equation (3). The Rydberg constant as well asthe shield constant of the atom can also be determined. Table 1 shows sufficient agreementbetween the experimental and the theoretical values.

424 I Sianoudis et al

Figure 5. Plot of the square energy of the Kα and Kβ emission lines of the XRF spectrum of themulti-component standard sample, containing the elements Ca, Ti, Cr, Fe, Ni, Zn and Br, as afunction of atomic number.

Table 1. Experimental and theoretical values of the Rydberg constant and shield constant of theatom for the Ka and Kβ lines.

Theoretical value Experimental value

RHKα (eV) 13.6 13.90 ± 0.20RHKβ (eV) 13.6 13.53 ± 0.10σ Kα 1.0 1.23 ± 0.10σ Kβ 1.0 1.78 ± 0.10

4.3. Lambert–Beer law

The intensity of x-rays after being transmitted through a material is attenuated. This is a resultof interactions with the material depending on its thickness, which is commonly expressedwith the surface density α and the absorption coefficient μ (E), which is a material constant[7]. This correlation is described in the Lambert–Beer law:

N = N0 · e−μ(E)·σ ⇒ ln

(N

N0

)= −μ(E) · σ (5)

where N0 is the original number of counts and N the number of counts after passing thematerial.

The absorption coefficient is energy dependant. For the experiment, the fluorescenceradiation of the multi-element standard was attenuated with aluminum foils of varioussurface densities, and the relative intensity of four specific Kα emission lines was determined(figure 6). The standard was a multi-element standard, which has Fe, Ni, Zn and Br aselements. Those elements have characteristic Kα energies at 6.4, 7.5, 8.6 and 11.92 keV,respectively. As shown in figure 1, the multi-element standard was used as a target sample toproduce secondary x-ray fluorescent radiation with given and different energy of the photons.The foils, mounted vertically in front of the detector, lead to attenuation of the beam as shownin the plot (figure 6) by reducing the number of photons reaching the detector.

Educational x-ray experiments and XRF measurements on cultural heritage objects 425

Figure 6. Plot of the absorption of four characteristic Kα energies of the standard obtained withAl foils of various surface densities. From the slope of the fitted lines, the absorption coefficientwas directly derived.

Table 2. Experimental and theoretical values of absorption coefficients at different energies.

E (keV) μtheo (cm2 g−1) μexp (cm2 g−1)

6.40 91.7 104 ± 147.47 57.5 68 ± 78.63 38.9 51 ± 411.92 15.1 17 ± 2

In table 2, the experimental values of the absorption coefficient for aluminium at theseenergies are listed in comparison with theoretical values taken from the literature. Theexperimental values of the absorption coefficient agreed with theoretical values to someextent, taking into account, however, that the foils used in this experiment were not pure Albut contained impurities (mainly Fe and less Ti and Pb), while various uncertainties in thevalues of the surface density, due to the folding of the Al foils, could not be avoided. Inaddition, there is a small possibility of nonlinear increase in the intensity of the fluorescentradiation as a function of the layer thickness [8].

The thicknesses of Al foils ranged approximately up to 72 μm, corresponding to surfacedensities of up to 0.0192 g cm−2.

4.4. Compton scattering

The last experiment of the first module concerns the Compton effect. X-ray interactionby inelastic scattering with electrons in matter results in a decrease in energy (increase inwavelength) of the x-ray. Part of the energy of the x-ray is transferred to a scattering electron,which recoils and is ejected from its atom, and the rest of the energy is taken by the scatteredphoton. This ‘degraded’ energy of the x-ray photon, observed in its wavelength, is measuredas a spectral shift �λ, which depends on the scattering angle α:

�λ = λ′ − λ0 = λc · (1 − cos α) (6)

where λ′ and λ are the wavelengths of the photon after and prior to inelastic scattering,respectively. The Compton wavelength λC is a physical constant, having a value of

426 I Sianoudis et al

Figure 7. Series of XRF energy spectra, which were measured at different angles in relation tothe x-ray source, after scattering on the Plexiglas plate. The wavelength was calculated from thecorresponded energies, after an appropriate calibration. A blue spectral shift of the wavelength forAg-Ka (inelastic scattering) is shown, while the elastic scattering is just observable.

Figure 8. Plot of a spectral shift of the characteristic line (Ag-Kα), of the anode material of thex-ray source, versus the cos(α). The obtained data from the spectra (see figure 7), showed an α

satisfactory approach to the theoretically expected results.

2.426 × 10−12 m according to the literature. For the experiment, the Kα line of the Agsource was used, with a wavelength of λ0 = 56.2 × 10−12 m. The radiation of the x-ray tubewas directed to a Plexiglas disc of ∼1 cm thickness (figure 2). The scattered x-ray radiationwas recorded at different angles relative to the x-ray beam of the tube (figure 7).

The spectral shift of the Kα line is presented in figure 8 against cos(α) and a linear fit wasapplied. The experimental value of λc = (2.26 ± 0.06) × 10−12 m, which is the slope of thefitting line, as well as the corresponding intercept ((2.30 ± 0.04) × 10−12 m) shown in figure 8agrees sufficiently with the theoretical value. The small difference between them, withinthe experimental error, is mainly caused by the non-accurate determination of each value ofwavelength corresponding to the Kα line, due to the noise of the spectra.

Educational x-ray experiments and XRF measurements on cultural heritage objects 427

Figure 9. XRF application on an outdoor bronze monument (Victoria Square, Athens, Greece).The two photos show an overview and the detail of the in situ XRF measurement (right). The XRFspectrum (left) corresponds to a spot on the arm (top right).

5. Module 2: XRF applications on cultural heritage objects

In recent years, considerable research has been focused on non-destructive techniques for thecharacterization of cultural heritage objects [9]. An introduction to these analytical approachesis a subject included in the curriculum of the students at the Department of Conservation at theTEI of Athens. Apart from a theoretical background in different non-destructive techniques,the students are also practically trained during their laboratory courses. The present XRF setuphas already been used in laboratory courses for metal conservation and in the framework ofstudent projects. The suggested module 2 combines measurements of different materials on amore systematic basis, imparting at the same time the theoretical background of instrumentalanalysis.

5.1. In situ measurements of metal objects

The first example presented shows the analysis of a large outdoor bronze monument in VictoriaSquare in Athens, Greece (figure 9). The portable XRF was applied on site [10]. The spectrumpresented shows a brass composition with Cu and Zn being the main components and Cr, Feand Pb being minor components. The Cl and Ca content identified may be related to thepresence of corrosion products, such as copper chlorides, or environmental contamination,such as bird dung.

428 I Sianoudis et al

(a) (b)

Figure 10. XRF application on a fragment of pipe recovered from the ‘Patris’ shipwreck. Thephoto (a) shows the in situ measurement in the laboratory. In the XRF spectrum (b) the mainelement peaks are identified.

The second example shows the analysis of a fragment of metal pipe belonging to anassemblage of metal finds recovered from the ‘Patris’ shipwreck. The conservation treatmentof this assemblage was the object of a recent laboratory course at the Department ofConservation. The steamboat ‘Patris’ sank in 1868 close to the Greek island Kea whereit was discovered in 2007. A large assemblage of metal objects was brought to the surfacehaving been at the bottom of the sea for almost 140 years. The saltwater environment withits high ionic conductivity requires special and in most cases immediate treatment of theobjects in terms of passivating specific corrosion processes. Therefore, in this case study,even qualitative examination of alloys and corrosion products provides valuable informationin relation to the conservation of the metal finds (figure 10). In this regard, the identificationof contact areas between different metals or alloys was of particular interest.

5.2. Quantitative analysis of modern coins

In order to demonstrate the quantitative analysis of XRF spectra and the accuracy of themeasurements, a set of Euro coins was examined, the reference composition of which canbe found on the Web (http://www.copperinfo.co.uk/coins/) [11]. For calibration of the XRS-FP software, a combination of two standard reference materials was used, the BAM-367,providing calibration factors for Cu, Ni, Fe and Mn concentrations, and the BCR-A, providingadditional calibration factors for Zn, Sn and Pb concentrations. Even though the live time of100 s for the measurements was comparitively small, the results are in sufficient accordancewith the reference concentrations (table 3). However, due to the setup used—performingthe measurements in air—it was not possible to determine aluminium. Therefore, the Alconcentration was set fixed at 5 wt% for the 10- and 50-cent coins consisting of Nordic gold.Furthermore, the tin concentrations in the analysed coins were in the range of the lower limit ofdetection. Even though Sn was detected at least in the 50-cent coin, the relative measurementerror was close to 100%. In order to improve the signal-to-noise ratio, either the lifetime ofthe measurements has to be increased or the parameters of the x-ray tube have to be adjusted,such as the current or the voltage.

Educationalx-ray

experiments

andX

RF

measurem

entson

culturalheritageobjects

429

Table 3. Quantitative analysis of modern coins. The concentrations presented were determined with the software Amptek XRS-FP (3.3.0). Two reference alloys were used for calibration.

1 Cent 2 Cent 5 Cent 10 Cent 50 Cent1 Euro(inner)

2 Euro(inner)

2 Euro(outer)

wt% err wt% err wt% err wt% err wt% err wt% err wt% err wt% err

Al 5 (fixed) 5 (fixed)Mn 0.3 0.1Fe 0.5 0.1 0.5 0.1 0.4 0.1 0.3 0.1 0.3 0.1 0.7 0.2Ni 0.2 0.1 0.2 0.1 0.2 0.1 0.6 0.2 0.8 0.2 23.1 1.0 6.0 0.4 17.5 1.0Cu 98.4 2.3 98.6 2.1 99.0 2.1 88.9 2.0 87.2 2.0 74.8 2.0 76.6 1.6 75.1 2.3Zn 0.5 0.1 0.4 0.1 0.4 0.1 5.0 0.5 5.4 0.5 1.6 0.3 17.0 0.8 6.4 0.7Sn 1.3 1.2Pb 0.5 0.3

Reference values

Al 5.0 5.0MnFeNi 25.0 5.0 25.0Cu 100.0 (surface) 100.0 (surface) 100.0 (surface) 89.0 89.0 75.0 75.0 75.0Zn 5.0 5.0 20.0Sn 1.0 1.0Pb

430 I Sianoudis et al

(a) (b)

Figure 11. XRF spectra of fragments from wall paintings. (a) The spectrum presents, apart fromcalcium and iron, clear L-lines of lead, indicating the use of minium (red lead). (b) Apart fromcalcium and iron, mercury is apparently present in the paint layer, indicating the use of cinnabar,which was mixed with red ochre.

5.3. Identification of pigments in wall paintings

A third set of samples comprised fragments of Byzantine wall paintings. The sampleshad already been examined beforehand using scanning electron microscopy in combinationwith energy dispersive spectroscopy (SEM-EDS) [12]. The XRF examination was tested onthese samples in order to examine the feasibility of in situ measurements for future studies.Therefore, a live time of 300 s was selected. Because the measurements were performed inair, it was not possible to determine light elements (Z < 20) due to the absorption of theirlow x-ray fluorescence energies. Possible ways to suppress this absorption would be eitherto enrich the air with helium or to reduce the atmospheric air pressure with an appropriatevacuum pump. The spectra in general showed the presence of calcium, related to the limeplaster. Furthermore, most of the red pigments could be identified as red ochre due to theiriron content. In two cases, however, different red pigments could be identified (figure 11).A dark red streak was obviously applied by using a lead pigment, probably minium, and ina bright red paint layer the red ochre was mixed with cinnabar (HgS). In order to investigatethe entire palette of pigments used in wall paintings, the setup will have to be modified. Forexample, by the application of a helium flow in front of the x-ray source and detector, theabsorption of the fluorescence radiation of light elements can be reduced.

6. Conclusions

A setup has been developed, assembled with components of a portable system for XRFspectroscopy, in order to be used for educational purposes. Using this setup, a series of familiarand conventional laboratory exercises, such as the verification of Moseley’s law, Compton’slaw and the Lambert–Beer law, was realized with satisfactory success. Together with theconception of an instructive procedure for XRF calibration, these experiments constitute auseful educational package of laboratory exercises for students in natural sciences. Apartfrom this first module, a second module was developed for the education of students who willwork in the field of cultural heritage, such as conservation and archaeological scientists. Inthis second module, students learn more about XRF applications and about the potential and

Educational x-ray experiments and XRF measurements on cultural heritage objects 431

limitations of XRF in comparison with other analytical methods both in terms of measurementconditions and of elements which can be detected and quantified, respectively. The twomodules can be implemented independently or in conjunction.

Acknowledgments

We would like to thank A Karydas and C Zarkadas from the Institute of Nuclear Physics, NCSR‘Demokritos’ in Athens and V Argyropoylos and D Charalambous from the Department ofConservation of Antiquities & Works of Art at the TEI of Athens for their collaboration in theoutdoor monuments case study, in the framework of the EC-funded programs ‘Archimedes’and PROMET.

References

[1] Bronk H, Roehrs S, Bjeoumikhov A, Langhoff N, Schmalz J, Wedell R, Gorny H E, Herold A andWaldschlager U 2001 ArtTAX—a new mobile spectrometer for energy-dispersive micro x-ray fluorescencespectrometry on art and archaeological objects Fre. J. Anal. Chem. 371 307–16

[2] Zarkadas C and Karydas A G 2004 A portable semi-micro-x-ray fluorescence spectrometer for archaeometricalstudies Spectrochim. Acta B 59 1611–8

[3] Guilherme A, Cavaco A, Pessanha S, Costa M and Carvalho M L 2008 Comparison of portable and stationaryx-ray fluorescence spectrometers in the study of ancient metallic artefacts X-Ray Spectrom. 37 444–9

[4] Currie L A and Svehla G 1994 Nomenclature for the presentation of results of chemical analysis Pure Appl.Chem. 66 595–608

[5] Conover C W S and Dudek J 1996 An undergraduate experiment on x-ray spectra and Moseley’s law using ascanning electron microscope Am. J. Phys. 64 335–8

[6] Gudennavar S B, Badiger N M, Thonfadarya S R and Hanumaiah B 2003 Verification of Bohr’s frequencycondition and Moseley’s law: an undergraduate laboratory experiment Am. J. Phys. 71 822–5

[7] Sianoudis I A, Karydas A G, Zarkadas C and Drakaki E 2006 X-ray fluorescence spectroscopy(XRF): suggestions for development of lab-exercises and research applications e-J. Sci. Technol. 2 86http://e-jst.teiath.gr/deutero teuxos.htm

[8] Linke R, Schreiner M, Demortier G and Alram M 2003 Determination of the provenance of medieval silver coins:potential and limitations of x-ray analysis using photons, electrons or protons X-Ray Spectrom. 32 373–80

[9] Adriaens A 2005 Non-destructive analysis and testing of museum objects: an overview of 5 years of researchSpectrochim. Acta B 60 1503–16

[10] Argyropoulos V, Charalambous D and Vossou-Domi A et al 2007 A conservation strategy for documenting thecorrosion of outdoor bronze monuments in Greece Proc. of the 7th European Conf. ‘SAUVEUR’, SafeguardedCultural Heritage, Understanding & Viability for the Enlarged Europe, Prague vol 2 pp 887–9

[11] Peralta L, Farinha A C and Rego F 2008 What are the 50 cent Euro coins made of? Eur. J. Phys. 29 901–9[12] Hein A, Karatasios I and Mourelatos D 2009 The Byzantine wall paintings from Mani (Greece): microanalytical

investigation of pigments and plasters Anal. Bioanal. Chem. 395 2061–971