DOI: 10.1007/s00339-004-2510-8

Appl. Phys. A 79, 171–176 (2004)

Materials Science & ProcessingApplied Physics A

g.m. ingo1,�

e. angelini2

t. de caro1

g. bultrini3

Combined use of surface and micro-analyticaltechniques for the study of ancient coins1 CNR-Istituto per lo Studio dei Materiali Nanostrutturati, CP 10, 00016 Monterotondo Stazione, Roma, Italy2 Dipartimento Scienza dei Materiali ed Ingegneria Chimica, Politecnico di Torino,

corso Duca degli Abruzzi n. 24, 10129 Torino, Italy3 Dipartimento di Scienze Chimiche, Università di Catania, v. le A. Doria n. 6, 95125 Catania, Italy

Received: 11 June 2003/Accepted: 18 December 2003Published online: 19 May 2004 • © Springer-Verlag 2004

ABSTRACT By means of the combined use of surface andmicro-analytical techniques the surface chemical compositionof ancient coins and some aspects of their manufacturing tech-niques and of degradation mechanisms have been elucidated.Two case histories are described concerning silver RomanRepublican coins and some coins plated with thin films ofsilver and gold. In particular, the coinage methods, the sil-vering and gilding techniques and the origin of the embrit-tlement of these selected Roman coins have been studied bymeans of the combined use of selected-area X-ray photoelectronspectroscopy (SA-XPS) and scanning electron microscopy andenergy-dispersive spectrometry (SEM+EDS). This innovativeapproach has been utilised in order to gain further insight intothe microchemical structure of the external regions of the coinsas well as of the bulk features. The results show the use of mer-cury to coat a copper or silver core with a thin film of preciousmetals that could be considered the most important advancein the technology of gilding to be made in antiquity. Further-more, the microchemical investigation of brittle Roman silvercoins has allowed us to identify the origin of this troublesomeproblem. The microchemical results indicate that brittleness isinduced by the presence of a low amount of lead that is retainedin supersaturated solution when the cast blank was produced.This latter element segregates at the grain boundaries during thecoin production and the subsequent long-term ageing at roomtemperature, thus inducing the alloy fracturing along the weak-ened grain boundaries.

PACS 68.55.Jk; 68.35.Dv; 68.37.Hk; 68.55.Nq; 81.05.Bx

1 Introduction

Most people who are concerned with the metal-lurgical or chemical investigation of ancient coins will befamiliar with the large number of publications on the subjectproduced by scientists in the last decades. But these scientistswill be the first to admit that in spite of these relevant effortsthere are many aspects about ancient coins that are poorlyunderstood, such as coin-production details or the plating

� Fax: +39-06/90672714, E-mail: ingo2@mlib.cnr.it

methods [1–6]. Other interesting aspects relate to the prove-nance of metals or the chemical composition and its variationas well as the elemental local composition changes which areof great value to understand the manner in which the coin wasmade, processed or degraded. Other interesting aspects relateto the overstruck coins (i.e. coins scratched slightly off andrestruck) or the subaerate coins that appear to be struck onsilver but have a copper core. In Roman times, these lattercoins were generally minted by the official issuing authoritiesunder a lack of precious metals and the production of coinsonly coated by a noble-metal thin film as well as the contentof Ag or Au of the coinage reflect periods of severe economicdifficulties.

The case histories that are presented in this paper havebeen accumulated over a short period as the result of exami-nations finalised to the restoration and conservation of coinsproduced in Roman times in the Mediterranean basin. Theinvestigations were carried out to save these important wit-nesses of our ancient economy as well as to obtain informationon the technological competence reached by Romans on thealloying and minting processes.

The first case relates to the surface microchemical inves-tigation of some silver- and gold-plated Roman coins coatedby using the sophisticated technique of the mercury amalgam.The second case relates to one of the most interesting and curi-ous problems concerning the silver Roman Republican coinscalled serrati. These coins are different with respect to thecommonly used and most diffused regular silver coins calleddenarii, being characterised by a notched edge.

2 Experimental

The XPS investigation has been carried out on anEscalab Mk II spectrometer using both unmonochromatedAl Kα1,2 and Mg Kα1,2 radiations as excitation sources (hν =1486.6 eV and hν = 1253.6 eV, respectively). The size ofthe analysed area for the selected-area X-ray photoelectronspectroscopy (SA-XPS) measurements was about 0.15 mmin diameter; for the large-area XPS characterisation the areawas about 1 cm2. Argon-ion etching was performed usinga rastering differential pumping ion gun VG A61 operated ata pressure of 5 ×10−6 mbar and at 2 keV, thus producing an

172 Applied Physics A – Materials Science & Processing

ion-current density of about 0.3 µA/mm2, as measured usinga Faraday cup near the target surface.

Both scanning electron microscopy (SEM) and energy-dispersive spectrometry (EDS) characterisations were carriedout by using a Cambridge 360 scanning electron microscopeequipped with a LaB6 filament and a backscattered electrondetector with four sectors.

To analyse the sectioned microchemical structure of thecoins a sample was removed with a jeweller’s diamond saw,embedded in a resin and metallographically polished with car-borundum papers and diamond pastes up to 0.25 µm.

X-ray diffraction (XRD) patterns were recorded directlyon the samples, by multiple scanning using an automatedSeifert XRD-3000 diffractometer. The identification of thespecies was carried out by using a Seifert XDAL 3000 Soft-ware Index I.

The fractured surface of the serrati for SA-XPS andSEM+EDS analyses was obtained by fracturing the coin aftercooling at liquid-nitrogen temperature.

3 Results and discussion

3.1 Silver and gold Roman coins plated using mercuryamalgam

Hereafter we report the results of an investigationcarried out on plated Roman coins recently found during ar-chaeological excavation in Rome and therefore not subjectedto later addition and restoration procedures. In particular, theplated coins have been examined in as-received condition offinding and we can exclude that the gilding was repaired or re-furbished in recent times, as could be not ruled out for otherancient objects. In particular in Fig. 1 a gold-plated Romancoin (on the left) and a silver-coated Roman denarius (on theright) are shown.

These images evidence that only little of the plating re-mains in situ but that thin layers of silver and gold are yetvisible in several large areas of the coins.

In Fig. 2, backscattered electron images and ED spectrafor these coins are reported. These latter spectra have beentaken on the areas where the precious metals are still presentand clearly show the presence of mercury in the thin metallayer, thus demonstrating the use of an amalgam [7, 8].

FIGURE 1 Images of a gold-plated Ro-man coin (left) and a silver-coated Romandenarius (right) produced by using mer-cury amalgam. Both coins have been foundduring archaeological excavation in recenttimes in Rome. It is worth noting that thedenarius was minted in Rome in 62 BC andis the first precisely dated example of theuse of mercury in Rome and likely in Eu-rope

According to this method a mixture of gold or silver withmercury, i.e. the so-called amalgam, was first spread ontothe surface to coat it and then a thermal treatment at about250–300 ◦C was carried out for a few minutes, thus causingthe mercury to evaporate. For these reasons, the fire gild-ing is very easy to detect because the Hg is never driven offcompletely by the heating process and the residual content ofHg can be detected by EDS or other surface analytical tech-niques. The EDS spectra shown in Fig. 2 disclose that theHg amount in the silver-coated coin is higher with respect tothe gold-coated coin because silver has a much higher solidsolubility for mercury than copper and it can absorb its ownweight of mercury in solid solution [8]. Furthermore, the im-ages shown in Fig. 2 evidence that corrosion products haveerupted through the gilding in several places and have forcedthe precious thin metal layer outwards.

In order to confirm the presence of Hg on the surface ex-ternal layer of the gold-coated coin SA-XPS was used and theSA-XPS photoemission signals in the energy region of Au 4fand Hg 4f peaks are reported in Fig. 3 as a function of Ar-ion sputtering (2 keV, 0.3 µA/mm2, each etching time 600 s)starting from the surface in as-received state (first spectrum).

These SA-XPS results confirm the presence of mercuryon the thin gold film on the coin. The SEM+EDS resultsindicate some differences in the production and minting pro-cesses because it seems that the gold coin was first struck andthen gilded, while the silver-plated coin was first coated witha silver–mercury layer and then struck.

Concerning the practice of fire gilding, as pointed out byOddy [7], the evidence for the classical world is unclear aboutthe date of the introduction of this method in Europe. Onlyone or two undoubtedly authentic objects are known in Eu-rope which clearly indicate the use of mercury and which canbe precisely dated before the 2nd century AD. Furthermore,unfortunately, the origin and use of mercury gilding for coat-ing copper or silver artefacts have been episodically exploredin the last decades by archaeologists and ancient technol-ogy experts only via visual inspection without any scientificinvestigation.

For this reason the results of this study are relevant becausethe examined silver-coated coin was minted in Rome in 62 BCand is the first precisely dated example of the use of mercury

INGO et al. Combined use of surface and micro-analytical techniques for the study of ancient coins 173

FIGURE 2 Backscattered electron imagesand ED spectra for the gold-plated Romancoin (top) and for the silver-plated Romandenarius (bottom). The ED spectra havebeen taken on the areas where the preciousmetals are yet present and clearly show thepresence of mercury in the thin metal layer,thus demonstrating the use of an amalgam

FIGURE 3 XPS photoemission signals in the energy region of Au 4fand Hg 4f peaks, 80–90 eV and 90–100 eV, respectively, for the surfaceof the gold-plated Roman coin as a function of Ar-ion sputtering (2 keV,0.3 µA/mm2, each etching time 600 s) starting from the surface in as-received state (first spectrum)

in Rome and possibly in Europe [6, 7]. Indeed, the method forcoating copper artefacts by using mercury was discovered bythe Chinese in the 4th century BC and until now it was felt thatthis sophisticated technique reached the Roman world in the2nd century AD [7]. On the contrary, Anheuser [8] has shownthat in the time range from the 1st century BC to the 1st cen-tury AD mercury gilding was also known in the western worldand that the earliest European artefacts coated with a thin filmby using mercury originated from Celtic contexts. These lat-ter suggestions are reinforced by the contemporary historicalsources given by Vitruvius and Pliny the Elder. However, theabove-shown results indicate that in the Roman world the dateof mercury application can be shifted about 100–200 years to62 BC, because the use of Hg was also known during the Ro-man Republican period.

174 Applied Physics A – Materials Science & Processing

3.2 Serrati silver Republican Roman coins

In Fig. 4, the image of a silver Roman Republicancoin of the series serrati is shown.

The typical sawtooth pattern of a serratus is characterisedby a number of notches ranging from 25 to 30 and a depthof 0.6–1.3 mm. These notches were obtained by mintinga blank that was previously notched around the coin edge witha chisel. These coins appear clearly very different with respectto the commonly used and most diffused regular silver coinscalled denarii that have a regular circular shape.

The serrati and denarii coin production was probably car-ried out adopting the same procedures consisting of hammer-ing of a heated blank between an immobile and a mobile die.The practice of serration was confined to specific issues andwas especially used intermittently from the late 2nd centuryBC to the early decades of the 1st century BC. The reasonfor the contemporaneous minting at the same time and in thesame place of serrati and regular denarii remains at presentuncertain [9]. It is also unclear why this complex labour-consuming method was applied considering the large scaleof silver coinage required by the widespread trade within theRoman world. Furthermore, the serrati are often found inan extremely brittle condition even though they do not showsigns of heavy external corrosion and silver must have beenquite ductile when the coins were struck. Indeed, the serraticoins are easily broken with little applied force and with onlya small deformation. This mechanical feature is not typicalonly of the serrati but has been noted also in silver coins ofancient Greece, in Saxon and English coins and in late me-dieval coins of India as well as in Roman and Egyptian silver

FIGURE 4 Image of a serratus (diameter≈ 20 mm)

Coin Ag Cu Sn Zn Pb Fe(wt %)

S 1 97.12 0.37 0.22 0.36 1.36 0.08S 2 96.88 0.42 0.56 0.47 1.22 0.13S 3 96.52 0.78 0.31 0.43 1.72 0.12S 4 97.08 0.34 0.61 0.38 1.44 0.09D 1 90.22 8.42 0.33 0.15 0.96 n.d.D 2 94.43 4.42 0.17 n.d. 0.39 0.07D 3 94.39 4.54 0.02 n.d. 0.23 n.d.D 4 92.49 7.12 0.04 n.d. 0.25 n.d.

n.d.: not determined

TABLE 1 Chemical compositionof coeval silver Roman Republicanserrati (S) and denarii (D) coins ofeight different series

artefacts [10–12]. The production of a notched edge and thebrittleness, therefore, might be due to the chemical compo-sition and microchemical structure as well as to the ageingprocess inducing a drastic change in the metallurgical and mi-crochemical structure of the coins. Furthermore, it is worthnoting that the surface of the serrati is generally characterisedby the presence of a network of fine cracks as shown in Fig. 4.

In Table 1, the chemical compositions of coeval silver Ro-man Republican serrati and denarii coins are reported.

In all the brittle coins, without exception, lead has beenfound in appreciable quantity while the denarii are charac-terised by a lower amount of lead and a higher content ofcopper ranging from 4.4 weight percent (hereafter wt %) to8.4 wt %. These data seem to suggest a qualitative correla-tion between the amount of lead, the practice of serrationand the embrittlement that could be enhanced by a long-termageing process in the silver–lead system [12, 13]. This lattersystem has been extensively investigated and the probabil-ity of an ageing process related to thermal treatment effectshas been demonstrated [12–14] and related to the very lowsolid-solubility of lead in silver at room temperature, whilethe complete solubility is obtained when a silver–lead alloyis molten. In particular, with a lead amount of 0.8 wt %, theage-embrittlement can be induced after a thermal treatmentof a few hours at a temperature of 100 ◦C and the ageingproceeds more and more quickly with increasing tempera-ture [12–14].

Dealing with the presence of an appreciable lead concen-tration in the serrati coins as evidenced by the data reported inTable 1, this occurrence is surely due to the direct use of silverobtained via cuppelation and not completely refined. In order

INGO et al. Combined use of surface and micro-analytical techniques for the study of ancient coins 175

FIGURE 5 Fracture surface of a serratus andits chemical composition obtained via EDS (top).The coin was fractured in air after cooling atliquid-nitrogen temperature. SEM image of thesurface obtained by polishing to mirror-like fin-ish a serratus (bottom). The micrograph dis-closes the presence of a network of microcracksat the notch

to confirm the relationship suggested above, a serratus coinwas fractured in air after cooling at liquid-nitrogen tempera-ture. The morphology of the fracture surface is shown in Fig. 5(top) by a SEM micrograph and the chemical composition ofthis surface is described by the EDS spectrum reported in thislatter figure.

The SEM image reveals an intergranular fracture alongthe boundaries of large equiaxed grains (300–500 µm diam-eter) typical of a brittle fracture caused by impurities in silveror silver alloys [15]. In Fig. 5, a SEM image for the sur-face obtained by polishing to mirror-like finish a serratus isalso reported. This latter micrograph discloses the presence ofa network of microcracks at the notch. It is worth noting thatthe EDS spectrum reported in Fig. 5 shows a very small peakof lead. Since this information does not allow us to prove thecontribution of surface phenomena in inducing the brittleness

of the serrati, SA-XPS was performed to identify the chem-ical nature of the fracture surface. The results of the SA-XPSsurvey are shown in Fig. 6.

They disclose the presence of an appreciable amount ofmetallic lead and a low amount of copper that segregate tothe surface of the grains, thus inducing the alloy fractur-ing along the weakened grain boundaries. On this basis, itis possible to suggest that lead remained in supersaturatedsolution when the cast blank was produced and began to seg-regate to the grain boundaries during the cooling of the castblank.

This process also occurred during the heating prior to hot-minting at 200–250 ◦C and continued during the subsequentlong-term ageing at room temperature. Furthermore, in orderto identify the process of an ageing embrittlement that inducesan enrichment of lead along the grain boundaries, a SEM

176 Applied Physics A – Materials Science & Processing

FIGURE 6 SA-XPS survey of the fracture surface of a serratus

FIGURE 7 SEM micrograph of a cross-sectioned serratus after etchingwith acidified potassium dichromate carried out for evidencing the wigglygrain boundaries

micrograph of a cross-sectioned serratus near the centre isshown in Fig. 7.

This picture reveals the microstructure of the grain bound-aries after an attack with acidified potassium dichromate andconfirms the occurrence of a long-term enrichment processthat takes place at low temperature from lead–silver supersat-urated solution [10]. The lead enrichment during twenty cen-turies deforms the grain boundaries and the slip lines and thuscreates double-edged and dark fine lines along their length.

With these considerations in mind, we suggest that the ser-ration of the cast blanks was performed in order to reducethe probability of fracture during the hot minting of nearlypure silver blanks intrinsically brittle due to the presence oflead. Indeed, the minting process of Roman Republican silvercoins induced a radial elongation of the blank of about 20%and a thickness reduction of about 30% due to the compres-sion. Under these conditions, the presence of a large numberof notches could help to reduce the effects of the radial forcesthat could cause unwanted deep cracks in the coins. This issimply a possible explanation of what might have happened.It is unfortunate that so many interesting technological pro-cesses learned in antiquity were lost and are now unknown

because of the extreme secretiveness of the craftsmen or forthe lack of written sources.

4 Conclusions

By means of the combined use of surface andmicroanalytical techniques the surface chemical compo-sition and some aspects of the manufacturing techniquesand of degradation mechanisms of ancient coins have beenelucidated.

Two case histories are described concerning Roman Re-publican silver coins and some coins plated with thin films ofsilver and gold.

The results show the early use of mercury in Europe to coata copper or silver core with a thin film of precious metals. Thiscan be considered the most important advance in the technol-ogy of gilding to be made in antiquity.

Furthermore, the microchemical investigation of brittleRoman silver coins has allowed us to identify the origin ofthis troublesome problem. The microchemical results indicatethat brittleness is induced by the presence of a low amountof lead that is retained in supersaturated solution when thecast blank was produced. This latter element segregates to thegrain boundaries during the coin production and the subse-quent long-term ageing at room temperature, thus inducingthe alloy to fracture along the weakened grain boundaries.

ACKNOWLEDGEMENTS This work was carried out in theframework of a project financially supported by the European Commis-sion (INCOMED project, acronym EFESTUS, Contract No. ICA3-CT-2002-10030). The authors would like to thank Dott.ssa Silvana Balbi de Caro(Director of the Roman National Museum of Rome) for the samples of theRoman coins, Dr. Alessio Mezzi and C. Veroli (CNR-ISMN) for XPS andXRD analyses and, further, the SEM laboratory (Universita Tor Vergata,Roma – Area della Ricerca di Montelibretti del CNR).

REFERENCES

1 G.M. Ingo, G. Bultrini, E. Lo Piccolo, L.-I. Manfredi: Archaeometry 39,59 (1997)

2 D.A. Scott: J. Am. Inst. Conserv. 29, 193 (1990)3 D.A. Scott: Metallography and Microstructure of Ancient and Historic

Metals (The Getty Conservation Institute, Paul Getty Museum, Malibu,CA 1991)

4 G.M. Ingo, G. Bultrini, D. Attanasio: Archaeometry 43, 529 (2001)5 G.M. Ingo, L.-I. Manfredi, G. Bultrini, G. Chiozzini, S. Caroli: in Proc.

XII Int. Numismatischer Kongr., ed. by B. Kluge, B. Weisser (StaatlicheMuseen zu Berlin, Berlin 2000) p. 306

6 P.A. Lins, W.A. Oddy: J. Archaeolog. Sci. 2, 365 (1975)7 A. Oddy: ‘Gilding of Metals in the Old World’. In: Metal Plating and

Patination, ed. by S. La Niece, P.T. Craddock (Butterworth-Heinemann,Oxford 1993)

8 K. Anheuser: Bull. Met. Mus. 26, 48 (1996)9 M.H. Crawford: The Roman Republican Coinage (Cambridge University

Press 1975) p. 56910 R.J.H. Wanhill, J.P.H.M. Steijart, R. Leenheer, J.F.W. Koens: Archaeom-

etry 40, 123 (1998)11 I.G. Ravich: in Proc. Tenth Trienn. Meet. Int. Council of Museums Com-

mittee for Conservation, Vol. II (Reinhold Publishing Co., Washington,DC 1993) p. 792

12 F.C. Thompson, A.K. Chatterjee: Stud. Conserv. 1, 155 (1954)13 L. Addicks: Silver in Industry (Reinhold, New York 1940)14 A. Butts: Copper, the Science and Technology of the Metal. Its Alloys and

Compounds (Am. Chem. Soc. Monogr. Ser. 12) (Reinhold PublishingCo., New York 1954) p. 702

15 D. Ott: Handbook on Casting and Other Defects in Gold Jewellery Man-ufacture (World Gold Council, Industrial Division, London 1997)