·Table of Contents
·Conservation and Restoration in Art and Architecture
XRF Spectrometers for Non-Destructive Investigations in Art and Archaeology: the Cost of Portability
Dip. Ing. Chimica e dei Materiali, Università di Roma La Sapienza, ITALY.
Marco Ferretti ,
Istituto per le Tecnologie Applicate ai Beni Culturali, Consiglio Nazionale delle Ricerche, Rome, ITALY
X-ray fluorescence (XRF) has been used for the investigation of archaeological and historical materials for some fifty years; though a great variety of spectrometers has been successfully employed, it is only by portable systems that non-destructivity - one of the most attractive features of this technique - can be fully exploited; thanks to the possibility of working in situ, portable spectrometers have virtually extended the range of use of XRF to any type of object. This paper considers the most significant applications published so far with a special focus on the relationship between portability and other instrument characteristics such as, for instance, detection limits. It also discusses whether spectrometers used to investigate archaeological and historical materials are actually suitable for this purpose or, one could say, whether the designer has clear ideas on the analytical needs in this field. Finally it considers the possibility of improving the spectrometers' characteristics and establishing new analytical approaches thanks to recent advances in detector technology.
The importance of scientific support for the study and conservation of cultural property has long been recognized: the first laboratory was established in the Königliche Museum in Berlin in 1888, followed in the first decades of the 20th century by the British Museum's Research Laboratory (1920) and the Laboratoire de Recherche des Musées de France (1931). Since then the analysis of constituent materials remains one of the main concerns not only to improve the knowledge of the object and its context, but also to understand the deterioration processes and identify proper conservation treatments.
The fact that archaeological and historical objects are often unique, artistically relevant and not easily movable clearly encourage the use of non-destructive techniques and portable instruments. On this ground X-ray fluorescence (XRF), which is an elemental analytical technique, is one of the most widely explored. Though early applications date back to the 50's and 60's [1-11], portable or, at least transportable, XRF spectrometers started being effectively used in early 70's, when semiconductor detectors became available [12, 13]. Carrying the object to the laboratory was no more necessary: thanks to the feasibility of in situ investigations the application of XRF was virtually extended to any type of object. In recent times a significant improvement in spectrometers' portability has been produced by both thermoelectrically-cooled detectors [14-16] and miniaturized X-ray tubes.
Starting from the recent literature, this paper discusses how the choice of the system components affects the performance of the spectrometer, with special regard to the above mentioned thermoelectrically-cooled detectors and miniaturized X-ray tubes; some considerations will also concern the purpose of the analyses on art materials and whether the presently available portable spectrometers are suitable for real analytical needs.
PHYSICAL PRINCIPLES AND INSTRUMENTATION
The physical principles of X-ray fluorescence are simple and well known: electronic transitions can be induced in the inner shells of the atoms by electromagnetic radiation - or charged particles - of suitable energy; such transitions result in the emission of X-rays whose energy and intensity are related to the type and abundancy of the atoms concerned by the interaction. In practice there are limitations in detecting light elements in air: the low fluorescence yeld (the probability that ionization in a given shell results in the emission of a characteristic X-ray) and the attenuation of X-rays by air and by the detector window generally prevent elements lighter than sulphur (Z = 16) from being detected.
Due to attenuation in the matter, only the X-rays emitted in the first layers under the surface can reach the detector; Fig. 1 shows, for a typical bronze alloy and for each element, the part of the sample concerned by the analysis.
Fig 1: Depth of material concerned by XRF analysis for a typical bronze alloy for the analytes Cu, Pb and Sn; x(50%) and x(90%) are the depths from which 50% and 90% of the emitted X-rays come.
If one considers that the surface of archaeological and historical materials is in most cases deteriorated and differs in composition from the bulk, it is apparent that quantitative measurements may require surface abrasion or even sampling, depending on the material; conversely if a measurement has to be strictly non-destructive, one cannot also expect reliable quantitative data. For each experimental situation the choice of one approach over the other depends on:
Fig 2: Typical portable ED spectrometer using a cryogenically-cooled HPGe detector.
- the type and intrinsic value of the object,
- the aim of the investigation,
- the instruments available and, last but not least,
- the scientist's or conservator's personal idea of acceptable damage.
According to whether the wavelength or the energy of fluorescence radiation is analyzed, one can have wavelength-dispersive (WD) or energy-dispersive (ED) spectrometers, respectively; for the purposes of this paper only the ED ones will be considered; Fig. 2 shows a typical portable ED spectrometer equipped with a cryogenically-cooled HPGe detector.
it provides the exciting radiation and is characterized in terms of energy and intensity, which both affect detection limits; though in priciple radioisotopic sources can be used as well, aspects related to radiation hazard and system performance lead to prefer X-ray tubes. Low-power, air cooled miniaturized tubes are extremely attractive for their low-weight; on the other hand the low voltage and current provide poor detection limits, regarding in particular Ag, Sn, Sb whose K-lines are unefficiently, if not at all, excited. Reversely high-power tubes provide a wider excitation range and higher photon output, but the shielding and the cooling system make them heavy and complex.
when connected to the counting chain, it allows to measure the energy and intensity of outcoming fluorescence radiation; it is characterized by energy resolution and efficiency, the latter having a strong effect on detection limits. Within the familiy of solid state detectors one has two main groups:
Group a) includes Si(Li) and HPGe detectors; they are provided with good energy resolution (about 150 eV at 5.9 keV), thick depletion layers - which means intrinsic efficiencies close to 100% for all the eneries of practical interest  - and large surfaces, that mean high detection efficiency; typical figures for planar HPGe's are 13 mm and 200 mm2 for thickness and surface, respectively. On the other hand need for liquid N2 strongly reduces portability and autonomy, furthermore, due to its large dimensions, the detector stays relatively far from the measurement point. All the detectors of group b) do not need liquid N2 and have small dimensions, which allow the detector to be put very close to the measurement point; however they have, with the exception of HgI2 detectors, thin depletion layers - about 300 mm, corresponding to about 5% intrinsic efficiency for the Sn K-line - and small surfaces (less than 10 mm2), that cause really poor detection efficiencies; the group includes:
- cryogenically (liquid N2) -cooled detectors and
- thermoelectrically (Peltier effect) -cooled detectors.
One of the most relevant features of any analytical system is the detection limit; for XRF spectrometers it can be defined as follows:
- HgI2 detectors, with good both energy resolution (about 200 eV at 5.9 keV) and intrinsic efficiency,
- SiPIN detectors with poor both energy resolution (about 250 eV at 5.9 keV) and detection efficiency and
- Si Drift (SD) detectors, with good energy resolution (about 160 eV at 5.9 keV, even for high count-rates) and poor detection efficiency.
where: D.L. is the detection limit, s(NB) is the standard deviation of the background count-rate, the figure at the denominator is the slope of the function N(c) in the neighborhood of the detection limit, N(c) is the net count-rate as a function of the element concentration c.
The above equation clearly shows that the detection limit is an inverse function of the local slope of the calibration curve or, which is the same, of the count-rate produced by a given concentration c. The latter can be increased by:
In achieving the optimum - and subjective - compromise between portability and system performance one has to take all the above into account. Fig. 3 shows the calibration curves of Pb and Sn obtained with a HPGe and with a SiPIN, red and blue lines respectively, on bronze alloys at 60 kV [18, 19].
- exciting K- instead of L-lines,
- increasing the tube photon output,
- increasing the detector efficiency and
- reducing the object-to-detector distance.
Fig 3: Pb and Sn calibration curves for HPGe (red) and SiPIN (blue) detectors.
The tube-sample-detector relative position and the collimation are optimized separately for each detector. Despite a straight line is a very rough approximation for a calibration curve, yet it can be used to evaluate the slope and, consequently the detection limit. Ferretti et al.  show that, on actual bronze archaeological objects, their spectrometer reaches detection limits of 0.7 weight % for Pb and 0.06 weight % for Sn; it is the same instrument used to derive the red calibration curves of Fig. 3. The figures given show that for Pb the SiPin detection limit is comparable to that of HPGe, whereas for Sn it is almost one order of magnitude worst.
ACTUAL PORTABLE XRF SPECTROMETERS
This paragraph is aimed at comparing, whenever possible, different constructive solutions and the concerned performances for some portable XRF spectrometers described in the literature; Tab. 1 reports the relevant features of the systems considered.
1 CC = cryogenically-cooled; TC = thermoelrctrically-cooled
||Estimated weight [kg] 2
|Hall et al. 
||Air-cooled X-ray tube (Mo target, 30 kV, 1 mA max.) + 241Am radioisotopic source (20 mCi)
||Criogenically-cooled Si(Li) detector (res. 160 eV at 5.9 keV)
||60 (including electronics)
||The line at 60 keV of 241Am is used to excite the k-lines of Sn and Sb
|Lutz et al. 
||Air-cooled X-ray tube (Rh target, 50 kV, 0.35 mA max.)
||Thermoelectrically-cooled Si(Li) detector (res. 170 eV at 5.9 keV)
|Ferretti et al. 
||Water-cooled X-ray tube (W target, 60 kV, 4 mA max)
||Cryogenically-cooled HPGe detector (res. 195 eV at 5.9 keV)
||20 (only the mobile part)
||The tube has a closed-circuit cooling system
|Cesareo et al. 
||Air-cooled X-ray tube (W target, 50 kV, 1 mA max.)
||Thermoelectrically-cooled HgI2 detector (res. 200 eV at 5.9 keV) or SiPIN detector (res. 250 eV at 5.9 keV)
||The heat exchange inside the tube shielding is provided by a non-conductive liquid in natural convection
|Longoni et al. 
||Air-cooled X-ray tube (W target, 30 kV, 0.1 mA max.)
||Thermoelectrically-cooled Si Drift detector (res. 155 eV at 5.9 keV)
|Caneva et al. 
||Water-cooled X-ray tube (W target, 60 kV, 4 mA max)
||Thermoelectrically-cooled SiPIN detector (res. 250 eV at 5.9 keV)
||25 (only the mobile part)
||The detector produces a spurious Ag peak
The tube can be equipped with a closed-circuit cooling system
|Table 1: Relevant features of some portable XRF spectrometers described in the literature.|
2 Weights estimated by the authors
After the constructive features it is interesting to consider the detection limits of the spectrometers, which are summarized in Tab. 2.
|Hall et al. 
||0.03 (AgK), 0.01 (AgL), 0.1 (SnK), 1.5 (SnL)
||Matrix: Cu alloys
Sample: presumably calibration std.
|Lutz et al. 
||Matrix: Cu alloys|
Sample: drill shavings
|Ferretti et al. 
||Matrix: Cu alloys
Sample: surface of the object, with no abrasion
|Cesareo et al. 
||No detection limits are given
|Longoni et al. 
||Not worst than 0.15
|Not worst than 0.20 (SnL)
||Matrix: Cu alloy|
Sample: calibration std.
|Caneva et al. 
||Matrix: Cu alloy
Figures estimated by comparison with those of ref.  (see text)
|Table 3: Detection limits in weight % for some portable XRF spectrometers described in the literature.|
Tabs. 1 and 2 are worth further considerations:
- as regards the possibility of exciting efficiently the K-lines of heavy elements, the X-ray tubes have not improved much with respect to that used by Hall et al. in 1973; if one relies on the L-lines, Ag, Sn and Sb cannot be resolved from one another: this means that the figure given by Longoni et al. makes sense only in absence of Ag and Sb; the tube described by Cesareo et al. seems promising, the question is whether the natural convection of the inside liquid would be as much effective in non-vertical positions;
- some thermoelectrically-cooled detectors have reached the same energy resolution than the cryogenically-cooled ones, moreover they accept much higher count-rates; surface and thickness still have to be increased to have reasonably good efficiencies at "high" energies; the manufacturers have to pay more attention in the choice of materials: for instance the SiPIN detector used by Caneva et al. produces a spurious Ag peak of about 4 cps with the tube working at 60 kV, 4 mA;
- as regards the flexibility of the system, a support and positioning device is as much important as the spectrometer itself, particularly when large objects such as paintings and statues are investigated: the shorter is the time needed to position the spectrometer, the larger the number of measurements; more efforts should be spent in developing such devices;
- it is surprising that scientists who are among the world leaders in portable systems design seem to neglect the aspects related to detection limits; in our opinion this is a further - if ever necessary - indication of the lack of real co-operation between designers and field users.
USE OF ELEMENTAL COMPOSITIONAL DATA IN ART AND ARCHAEOLOGY
Elemental compositional data are frequently used to study archaeological and historical materials with regard to provenance, fabrication technology and manufacturing technique; they can help to distinguish non-original material and, in lucky cases, to spot fakes. Less frequent applications can be found in conservation, where the bare elemental information is rarely sufficient and much has to be known about the aggregation of the elements. Some materials of archaeological and historical interest are discussed here in deeper detail as regards the information provided by elemental analysis.
the close chemical connection between the object and the raw material, as well as the homogeneity of the trace element composition within a single deposit, allow in many cases a precise identification of clay sources [22-26]; this can be used to study provenance and trade routes.
obsidian sources can be very effectively identified by elemental composition: this is one of the most successful applications of elemental analysis in the history of archaeometry .
opposite to ceramics, metals do not keep a close chemical connection between the raw materials and the alloy, moreover they can be re-melted; this precludes in general the possibility of compositionally-based provenancing, nevertheless differences in elemental composition can be significant of different fabrication contexts [28-34]; for instance non-original parts can be identified in statues, doors and other complex objects ; used in combination with microstructural techniques, elemental analysis can provide information on the fabrication technology, with special regard to casting, soldering and repairing.
similarly to metals, the composition of glass cannot be univocally related to provenance because it can be - and actually was - recycled; however elemental analysis can provide important information on fabrication technology, with special regard to the use of different fluxes - which can be related to different geographical areas and historical periods - decolorants, colorants and opacifiers [35-40].
a relevant part of painting materials, such as canvas, wood, glue, binding media, some pigments, varnishes, etc., is organic and, therefore out of the elemental analysis concern, except perhaps for trace elements; conversely another relevant part, mainly grounds and pigments, is inorganic and can be characterized by elemental analysis; as regards the sense and significance of quantitative determinations, one should consider that a) the dilution in other colours precludes the possibility of quantitatively measuring the composition of the base colours and b) it is however difficult to sample selected layers in such etherogeneous materials. Nontheless "simple" qualitative analyses are often sufficient to identify pigments [41, 42]; Tab. 3  shows examples in which the pigment is identified by the presence of one or two characteristic elements.
||Class of pigments
|Yellow Brown Red Green
||Minerals composed of iron oxides and silicates
||Natural compounds (azurite, malachite, atacamite) and artificial compounds (acetates, resinates, chlorides, silicates, sulfates)
||Basic lead carbonate 2PbCO3·Pb(OH)2
||Glass SiO2 (65-70%), K20 (10-20%), Al203 (0-8%), CoO (1-18%)
||Mercuric sulfide (cinnabar) HgS
||Lead tetroxide Pb304
||Arsenic trisulfide As203
||Manganese dioxide MnO2 (pyrolusite)
||Cadmium sulfide CdS
||Titanium dioxide TiO2
||Pb + Sn
||Lead-tin yellow, giallolino
||Lead-tin oxide Pb2SnO4 and PbSnO3
||Pb + Sb
||Lead antimonate yellow, antimony y., Naples y.
||Lead antimonate Pb3(SbO4)2
||Fe + Mn
||Manganese dioxide (8-10%) + iron hydroxide (45-55%)
||Pb + Cr
||Lead chromate PbCrO4
||Pb + Cr
||Basic lead chromate PbCrO4·Pb(OH)2
||As + Cu
||Emerald green, Scweinfurt g.
||Copper acetoarsenite Cu(CH3COO)2·3Cu(AsO2)2
||Zn + Ba
||Zinc sulfide + barium sulfate ZnS+ BaSO4
|Table 3: Qualitative identification of pigments by the presence of one or two characteristic elements.|
For each of the above materials (except paintings) the range of concentration of typically determined elements is given by Lahanier et al. , who show that the majority of significant elements often have concentrations below 1%; as an example of the sensitivity required in painting analysis the work by Ferretti et al.  shows that the maximum Sb count-rate (K-line) found in one of the paintings - the Martirio di Quattro Santi - equals that of 0.2% Sb in a copper alloy. It is clear that, if one wants to distinguish among similar materials by elemental analysis, he has to rely on minor and trace elements that, with respect to major components, have a much wider variation range. Whatever of these materials one may want to analyze, significant results are related to the capabiltiy of detecting tens or hundreds of ppm's.
The features of different portable XRF spectrometers have been discussed with special regard to the choice of X-ray tubes and detectors; this choice affects both portability and analytical performances. Moreover it has been shown that good detection limits are essential to investigate archaeological and historical materials since in most cases trace elements provide more information than the major ones.
XRF offers a unique opportunity for the non-destructive study of ancient materials and portability is essential in making this technique effective. On this ground big steps forward have been made: spectrometers weighting few kg have been built thanks to the combined use of miniaturized X-ray tubes and thermoelectrically-cooled detectors. However, though it is accepted that portable spectrometers be less performant than the huge laboratory ones, there are minimal sensitivity requirements below which the analysis is almost useless.
Recent literature promotes the use of XRF spectrometers whose features are not adequate to obtain really important information, in particular detection limits are too poor. End users should push the designers - who are seemingly not reactive on this ground yet - to produce spectrometers suitable for the real analytical needs of the field, needs that in most cases go much further than knowing whether a given objetc is made of bronze or brass.
The authors wish to thank Prof. G. Pappalardo of the LANDIS Laboratory, at the Istituto Nazionale di Fisica Nucleare LNS, Catania, for the use of his SiPIN detector.
- E.T. Hall Some Uses of Physics in Archaeology Year Book of the Physical Society (1958) 22-34.
- C.M. Kraay The Composition of electrum coinage Archaeometry 1 (1958), 21-23.
- C.H.V. Sutherland and M.R. Harold The silver content of Diocletian's early post-reform copper coins Archaeometry 4 (1961) 56-61.
- E.T. Hall and G. Roberts Analysis of the Moulsford Torc Archaeometry 5 (1962) 28-32.
- M. Banks, N. Elphinstone and E.T. Hall Bristol Blue Glass Archaeometry 6 (1963) 26-30.
- E.T. Hall, M.S. Banks and J.M. Stern Uses of X-ray Fluorescence Analysis in Archaeology Archaeometry 7 (1964) 84-89.
- R.J. Gettens and E. West Fitzhugh Azurite and Blue Verditer Studies in Conservation 11 (1966) 54-61.
- J. Plesters Ultramarine Blue, natural and Artificial Studies in Conservation 11 (1966) 62.
- R.J. Gettens, H. Kuhn and W.T. Chase Lead White Studies in Conservation 12 (1967) 125-139.
- H. Kühn Lead-Tin Yellow Studies in Conservation 13 (1968) 7-33.
- B. Mühlethaler and J. Thissen Smalt Studies in Conservation 14 (1969) 47-61.
- R. Cesareo, F.V. Frazzoli, C. Mancini, S. Sciuti, M. Marabelli, P. Mora, P. Rotondi and G. Urbani Non-Destructive Analysis of Chemical Elements in Paintings and Enamels Archaeometry 14 (1972) 65-78.
- E.T. Hall, F. Schweizer and P.A.Toller X-ray fluorescence analysis of museum objects: a new instrument Archaeometry 15 (1973) 53-78.
- R. Cesareo, G.E. Gigante, P. Canegallo, A. Castellano, J.S. Iwanczyk and A. Dabrowski Applications of non-cryogenic portable EDXRF systems in archaeometry Nucl. Instr. and Meth. A 380 (1996) 440-445.
- J. Lutz and E. Pernicka Energy dispersive X-ray analysis of ancient copper alloys: empirical values for precision and accuracy Archaeometry 38, 2 (1996) 313-323.
- A. Longoni, C. Fiorini, P. Leutenegger, S. Sciuti, G. Fonterotta, L. Strüder and P. Lechner A portable XRF spectrometer for non-destructive analyses in archaeometry Nucl. Instr. and Meth. A 409 (1998) 407-409.
- R. Cesareo, G.E. Gigante and A. Castellano Thermoelectrically cooled semiconductor detectors for non-destructive analysis of works of art by means of energy dispersive X-ray fluorescence Nucl. Instr. and Meth. A 428 (1999) 171-181.
- M. Ferretti and P. Moioli Calibration of a transportable XRF system with standard copper alloys unpublished (1996).
- C. Caneva and M. Ferretti Design and calibration of a transportable XRF system using a SiPIN detector unpublished (1999).
- M. Ferretti and P. Moioli The use of portable XRF systems for preliminary compositional surveys on large bronze objects. A critical review after some years' experience in Proceedings of the International Conference Metal 98, Draguignan 27-29 May 1998, Eds. W. Mourey and L. Robiola (1998) 39-44.
- M. Ferretti, L. Miazzo and P. Moioli The application of a non-destructive XRF method to identify different alloys in the bronze statue of the Capitoline Horse Studies in Conservation 42 (1997) 241-246.
- M. Picon, M. Vichy and E. Meille Composition of the Lezoux, Lyon and Arezzo Samian Ware Archaeometry 13 (1971) 191-208.
- M. Picon, C. Carre, M.L. Cordoliani, M. Vichy, J.A. Hernandez and J.L. Mignard Composition of the La Graufesenque, Banassac and Montans terra sigillata Archaeometry 17 (1975) 191-199.
- P.J. Ballié and W.B. Stern Non-destructive surface analysls of Roman terra sigillata: a possible tool in provenance studies? Archaeometry 26 (1984) 62-68.
- G. Schneider, B. Hoffmann and E. Wirz Significance and dependability of reference groups for chemical determinations of provenance of ceramic artifacts in Proceedings of the 18\th international symposium on archaeometry and archaeological prospection, Bonn, 1978, Archaeo-Physika 10 (1980) 269-283.
- G. Schneider Bronze casting at Olympia in Classical times MASCA Res. Pap. Sci. Archaeol. 6 (1989) 17-24.
- O. Williams-Thorpe Obsidian in the Mediterranean and Near East: a provenancing success story Archaeometry 37 (1995) 217-248.
- P.T. Craddock The composition of copper alloys used by the Greek, Etruscan and Roman civilizations. 1 - The Greeks before the Archaic Period Journal of Archaeological Science, vol. 3, n. 2 (1976).
- P.T. Craddock The composition of copper alloys used by the Greek, Etruscan and Roman civilizations. 2 - The Archaic, Classical and Hellenistic Greeks Journal of Archaeological Science, vol. 4, n. 2 (1977).
- P.T. Craddock The composition of copper alloys used by the Greek, Etruscan and Roman civilizations. 3 - The origins and early use of brass Journal of Archaeological Science, vol. 5, n. 1 (1978).
- R. Cesareo, M. Ferretti, G. Guida and M. Marabelli Le leghe: i risultati in: Marco Aurelio - Mostra di Cantiere, Istituto Centrale per il Restauro - Comune di Roma (1984) 43.
- Relazione sulle analisi in: Due Bronzi da Riace, Bollettino d'Arte, Serie Speciale n. 3, parte I (1984) 85-106.
- P.T. Craddock Three thousand years of copper alloys: from the Bronze Age to the Industrial Revolution in: Application of Science in Examination of Works of Art (eds. P.A. England and L. Van Zelst) Museum of Fine Arts, Boston (1985) 59-67.
- J. Riederer Die Metallanalyse in: Ein Depot zerschlagener Grossbronzen aus Augusta Raurica (eds. B. Janietz Schwartz and D. Rouiller) Römermuseum Augst (1996) 120-143.
- G.A.Cox and K.J.S. Gillies The x-ray fluorescence analysis of medieval durable blue soda glass from York Minster Archaeometry 28 (1986) 57-68.
- J. Henderson The scientific analysis of ancient glass and its archaeological interpretation in: Scientific Analysis in Archaeology, Oxford University Committee for Archaeology (1989) 30-62.
- I.C. Freestone Looking into glass in: Science and the Past (ed. S. Bowman), the Trustees of the British Museum (1991) 37-56.
- I.C. Freestone Theophilus and the composition of medieval glass in: Materials Issues in Art and Archaeology III (eds. P.B. Vandiver, J.R. Druzik, G.S. Wheeler and I.C. Freestone) MRS vol. 267 (1993) 739-746.
- R.H. Brill Chemical analyses of early glasses (1999).
- M. Ferretti, P. Moioli and C. Seccaroni The apsidal stained-glass window of Ortvieto cathedral: characterization of the materials in course of publication.
- M. Ferretti, G. Guidi, P. Moioli, R. Scafe and C. Seccaroni The presence of antimony in some grey colours of three paintings by Correggio Studies in Conservation 36 (1991) 235-239.
- C. Seccaroni, R. Scafè, P. Moioli, G.F. Guidi and M. Ferretti L'impiego della fluorescenza X per indagini non-distruttive sui dipinti, I Beni Culturali tutela e valorizzazione, anno II n.6 (1994) 38-44.
- P. Moioli and C. Seccaroni, personal communication (1998).
- Ch. Lahanier, F.D. Preusser and L. Van Zelst Study and conservation of museum objects: use of classical analytical techniques Nucl. Instr. and Meth. B 14 (1986) 1-9.