JFAP 2007-5-1 Embrittlement of Ancient Silver

41Journal of Failure Analysis and Prevention Volume 5(1) February 2005

Ancient silver may become brittle and damaged owing to long-term corrosion and changes in themicrostructure. Recognition and determination of corrosion-induced and microstructurally-inducedembrittlement, and also their synergy, are important for restoration and conservation of ancient and historicsilver. The types of embrittlement are described and illustrated, using examples of ancient and historicsilver artefacts, including the famous Gundestrup Cauldron, a masterpiece of European Iron Agesilverwork. In particular, the use of automated Electron BackScatter Diffraction (EBSD) enables improvedanalysis and assessment of corrosion-induced embrittlement. The knowledge obtained from detailedinvestigations is helpful not only in determining the best ways to restore and conserve embrittled silverobjects, but also in defining the possible extent of the embrittlement problem. This is illustrated by astraightforward statistical analysis.

Keywords:

JFAPBC (2005) 1:41-54 ASM InternationalDOI: 10.1361/15477020522294 1547-7029 / $19.00

1. IntroductionAncient metallic objects give rise to many ques-

tions. The basic questions of identity, provenanceand authenticity are generally answered by ClassicalArchaeology, though uncertainties can remain. Moredetailed questions often require Archaeometallurgy,a discipline that has expanded greatly over the lastdecades. Table 1 lists many of the topics and ques-

R.J.H. Wanhill, National Aerospace Laboratory NLR, Amsterdam, The Netherlands. Contact e-mail: [email protected]. Copyright Institute of Materials Engineering Australasia. Reprinted with permission from International Conference of Failure Analysis and MaintenanceTechnologies (ICFAMT), April 29-30, 2004, Brisbane.

ancient silver, cracks, corrosion, embrittlement, grain boundaries

Embrittlement of Ancient SilverR.J.H. Wanhill

tions that archaeometallurgists may be called uponto consider and answer.

The present paper considers a specific problem,the embrittlement of ancient silver. This might seemvery restricted, but a comprehensive treatment ofthis problem involves all of the topics and most ofthe questions in Table 1.

Ancient silver can be embrittled by long-termcorrosion and microstructural changes.[1-10] Thereare two basic types of embrittlement, corrosion-induced and microstructurally-induced. These areable to act synergistically, resulting in a derivativethird type.

The types of embrittlement have recently beendiscussed at length,[8-10] and will be reviewed insection 2. This is followed by three case histories,widely separated in provenance: the famousGundestrup Cauldron, a masterpiece of EuropeanIron Age silverwork; an Egyptian silver vase fromthe Ptolemaic period; and a Byzantine paten circa600 A.D.

The knowledge gained from detailed examinationof such ancient and historic artefacts can be impor-tant in determining the best ways to restore andconserve embrittled silver objects. The results of thecase histories can also be combined with a straight-forward statistical analysis of the chemical compo-

Table 1 Archaeometallurgical Topicsand Questions

Further Identification and ProvenanceMetal/alloy compositionsOrigins of metals and ores

Manufacturing and CraftsmanshipFabrication methodsChoices of methodsDamage AssessmentDeformationCorrosionEmbrittlement (cracks)Previous restorationsRestoration and ConservationMethodsAuthenticity ChecksGenuineModern fakeAncient fake

42 Journal of Failure Analysis and PreventionVolume 5(1) February 2005

Embrittlement of Ancient Silver (continued)

sitions of many ancient silver objects to enable somegeneral statements about the extent of silverembrittlement.

2. Types of Embrittlement

2.1 Corrosion-Induced EmbrittlementCorrosion-induced embrittlement is due to several

forms of selective corrosion. Intergranular corrosionis the most commonly reported. This can occur inmechanically worked and annealed objects, whichconstitute the majority. Interdendritic corrosion canoccur in castings, which are uncommon, especiallyin the Old World.

Corrosion along slip lines and deformation twinboundaries can occur in objects that have not beenannealed after (final) mechanical working, whichincludes striking a coin[1] and decorating by chasingand stamping.[7] Inside the metal these kinds ofcorrosion can lead to additional corrosion alongsegregation bands. These bands are the remains,

modified by working and annealing, of soluteelement segregation (coring) and interdendriticsegregation that occurred during solidification of aningot or cupelled button.

Figure 1 illustrates the kinds of selective corrosion.The examples are eclectic: a Roman cup,[3] a Sicantumi,[11] and an Egyptian vase.[7] Intergranularcorrosion is attributed partly to low-temperaturesegregation of copper.[2,3,6] This segregation, calleddiscontinuous or cellular precipitation, sometimescauses the grain boundaries to appear meandering,as will be shown in section 3. Interdendritic andsegregation band corrosion are consequences of high-temperature segregation of copper during metalsolidification. Corrosion along slip lines anddeformation twin boundaries is due to locally highstrains and possible long-term segregation of soluteor impurity elements to the highly strained regions.

2.2 Microstructurally-Induced EmbrittlementMicrostructurally-induced embrittlement is seen

Fig. 1 Selective corrosion of high-silver-content ancient silver: (a) intergranular corrosion (Source: Ref 3); (b) interdendritic corrosion(Source: Ref 11); (c) corrosion along slip lines, deformation twin boundaries and segregation bands (Source: Ref 7); and (d)crystallographic fracture due to corrosion along slip lines and deformation twins (Source: Ref 7)


to be characterized by brittle intergranular fracture,with sharply defined cracks and grain boundaryfacets, as shown in Fig. 2. Note the bodily displacedgrain: this is a characteristic of severe embrittlement.

The embrittlement is most probably a conse-quence of long-term low-temperature ageing, where-by an impurity element, or elements, segregates tograin boundaries. The available evidence indicateslead to be the most likely perpetrator.[1,8] thoughthis has yet to be verified directly. Other impurityelements might be involved, notably bismuth.[8]

2.3 Synergistic EmbrittlementFigure 3 gives examples of synergistic embrittle-

ment. Corrosion along slip lines, deformation twinboundaries and segregation bands can result incracks. These cracks can then initiate fracture along

Fig. 2 Brittle intergranular fracture (microstructurally-inducedembrittlement) in an Egyptian vase. Source: Ref 7

Fig. 3 Examples of synergistic embrittlement: (a) corrosion (causing fracture) along slip lines intersecting grain boundary facets, (b) corrosionalong deformation twin boundaries intersecting a grain boundary, and (c) corrosion along segregation bands intersecting grainboundary facets. Source: Ref 7



microstructurally embrittled grain boundarieswhich may fracture anyway, though less easilyunder the action of external loads. In turn, the grainboundary fractures expose more slip lines, defor-mation twins and segregation bands to the environ-ment and increase the opportunities for corrosion.

3. Case Histories

3.1 The Gundestrup CauldronFigure 4 shows the reassembled Cauldron, which

is the largest surviving silverwork from the EuropeanIron Age, dating to the 2nd or 1st century B.C.E.Owing to its size, high quality workmanship andiconographic variety, the Cauldron has been thesubject of many studies, particularly of its origin,which is still controversial.

3.1.1 Analysis technique: EBSD. Four smallmetallographic samples from different parts of theGundestrup Cauldron were lent to the NLR byPeter Northover, Oxford University. The sampleswere examined using a Field Emission Gun Scan-ning Electron Microscope (FEG-SEM) combinedwith automated Electron BackScatter Diffraction(EBSD) equipment.

EBSD is a powerful technique for microstructuralanalysis, providing many options. For the Gunde-strup Cauldron samples the following options werefound to be of most use:[12]

Inverse pole figure (IPF) colour-coded maps

Boundary rotation angle maps

Coincidence site lattice (CSL) maps

Figures 5 to 9 illustrate the EBSD analysis results,which fall into two categories:

1. Sample 366. This sample was essentiallyannealed and virtually free of corrosion. The mostsignificant results were a random microtexture, Fig.5(a), and extensive grain boundary precipitation ofcopper, giving the grain boundaries a meanderingappearance, as in Fig. 5(b). The details in Fig. 6prove it to be discontinuous precipitation.[13-15] Theprecipitate was finely mottled and had widths up

Fig. 4 The Gundestrup Cauldron: The Cauldron consists oftwelve plates and a bowl, all of 95-98% silver. Chemicalanalyses show copper as the main alloying (or impurity)element.

Fig. 5 Inverse pole figure (IPF) colour-coded map and boundaryrotation angle map for Gundestrup Cauldron sample366. The yellow-coded boundaries are mainly annealingtwins.


to 7 m. These characteristics are important, as willbe discussed in section 5.

2. Samples 361, 363, 365. These samples containedincreasing amounts of remanent cold-deformation

Fig. 6 Details of precipitate nucleation and growth inGundestrup Cauldron sample 366

and corrosion damage. In Fig. 7 to 9 the remanentcold-deformation is visible as red regions (slip) andirregular yellow boundaries (deformation twins). Thecorrosion damage is visible as black regions, repre-senting mainly intergranular cracks but also trans-crystalline cracks (best seen in Fig. 9a). However,there was no evidence of discontinuous precipitation.

3.1.2 Discussion of EBSD results. The differ-ences between sample 366 and samples 361, 363and 365, with respect to remanent cold-deformation,corrosion and discontinuous precipitation, areremarkable for two reasons:

Fig. 7 IPF colour-coded map and boundary rotation angle mapfor Gundestrup Cauldron sample 361

Precipitate behavior: Nucleation at original green grain boundary Growth into green grain, changing its lattice orientation

to that of the contiguous grain having the purple matrixand reddish-brown annealing twin

Precipitate behavior: Separate nucleations at original grain boundary between

purple and pink grains Growth in opposing senses: one nucleation into the

pink grain, changing its lattice orientation to that of thecontiguous purple grain; one nucleation into thepurple grain, changing its lattice orientation to that ofthe contiguous pink grain



1. Although the link between remanent cold-deformation and corrosion damage was noted pre-viously,[7] the present results indicate that cold-deformation was primarily responsible for corrosion,while extensive discontinuous precipitation wasinnocuous. This is notable because the eminentmetallurgist Cyril Stanley Smith took the view thatgrain boundaries along which discontinuous precip-itation had occurred seem to be highly susceptibleto corrosion.[2] On the other hand, the present resultsare consistent with the experience of Peter North-over,[16] who observed intergranular corrosion andcracking in ancient Bactrian silver despite coppercontents less than 1%, which is almost certainly too

low for discontinuous precipitation to occur.[8]

2. There is a possible, or probable, link betweenremanent cold-deformation and the occurrence ofdiscontinuous precipitation. Experiments haveshown that cold-deformation can reduce the earlygrowth rate of discontinuous precipitation in silver-copper alloys at elevated temperatures, and this couldbe due to deformation-induced continuous precipi-tation within the grains.[17] A similar effect may haveoccurred in samples 361, 363 and 365, even to theextent that discontinuous precipitation was pre-vented. However, verification of this will requiretransmission electron microscopy (TEM).




3.2 The Egyptian VaseFigure 10 shows the restored vase, which is a rare

survivor from the Ptolemaic period, being dated tobetween 300 and 200 B.C.E. The chased andstamped decorations represent flowers and lotus andacanthus leaves. The designs and form blenddifferent cultural traditions, including Egyptian andPersian.

3.2.1 Analysis techniques: SEM, EDX. The vasewas investigated by SEM metallography andfractography in 1995, taking two small samples fromthe lip and one from the lower wall.[7] EDX analysesgave the following average chemical composition, inwt.%: Ag 97.1; Au 0.8; Cu 0.9; Pb 0.7; Bi 0; Sn0.2; Sb 0.3.

Figures 1(c), 1(d), 2, 3, and 11 illustrate the SEMresults. The vase has proven to be archetypal for:

1. Corrosion along slip lines, deformation twinboundaries and segregation bands

2. Microstructurally-induced embrittlement,whereby the vases chemical composition (0.7% Pb,no Bi) indicates lead to be the most likely perpetrator

3. Synergistic embrittlement

4. The link between remanent cold-deformationand corrosion damage

Figure 11 illustrates the link between cold-deformation and corrosion in an especially significantway. The SEM metallograph used backscatteredelectron imaging, which revealed a local deformationpattern. The adjacent schematic interprets thedeformation pattern from the slip-line field theoryof indentation.[18]

This theory predicts that when ti/w = 4.4, atension zone forms at the surface opposite the

Fig. 10 The Egyptian vase

Fig. 11 Through-thickness backscattered electron SEMmetallograph and schematic of an external chaseddecorating groove in the Egyptian vase shown in Fig. 10(Source: Ref 7,8). The sample is from the vase lower wall.



indented (chased) groove, as well as the always-present compression zone directly under the groove.The actual value of ti/w is 4.2, which is close enoughto justify the interpretation. The tension zonepromoted corrosion and intergranular fracture at andnear the internal surface of the vase, i.e., it causedhidden damage.

3.2.2 Discussion of the results. As mentioned insubsection 3.2.1, investigation of the Egyptian vaseprovided archetypal information on several aspectsof ancient silver embrittlement. Two points partic-ularly relevant to restoration and conservation arementioned here:

1. Though microstructurally-induced embrittle-ment is probably much less common than corrosion-induced embrittlement (see section 6), the combina-tion synergy is very detrimental, rendering an objectnot only frangible but friable. Special care shouldbe taken to conserve such objects.[7]

2. The interpretation of Fig. 11 leads to a conclu-sion of general significance: all thin-walled artefactswith chased decorations should be examined fordamage at and near the corresponding internal orrear surface locations. This conclusion is reinforcedby the Byzantine paten case history, discussed next.

3.3 The Byzantine PatenFigure 12 shows the Byzantine paten, which has

been dated to circa 600 A.D. The paten is a rare andhigh-quality liturgical altar object. The important

central tableau is well preserved. However, there hasbeen extensive breakage along annular decoratinggrooves.

3.3.1 Analysis techniques: SEM, EDX, EBSD.A sample from the paten has been examined bySEM metallography and EDX in a preliminaryinvestigation at the Netherlands Institute for Cul-tural Heritage (ICN) in Amsterdam. It is intendedto continue this investigation using FEG-SEM +EBSD equipment at The Open University in theU.K.

Figure 13 illustrates the SEM + EDX results. The

Fig. 12 The Byzantine paten. Photograph courtesy of The MenilCollection, Houston

Fig. 13 Examples of (a) intergranular corrosion and (b)discontinuous precipitation of copper at grain boundariesin the Byzantine paten. EDX analyses showed primarilysilver at location 1, and silver and copper at locations 2-4. SEM metallographs and analyses courtesy of InekeJoosten, ICN


sample showed surficial intergranular corrosion, Fig.13(a), and at higher magnifications abundantevidence of discontinuous precipitation of copperat the grain boundaries, Fig. 13(b). This precipita-tion gave the grain boundaries a meanderingappearance, as in Fig. 5(b).

3.3.2 Discussion. The preliminary status of thisinvestigation allows concluding only that the patenhas undergone limited intergranular corrosion, andthat discontinuous precipitation of copper hasoccurred. However, by analogy with the Egyptianvase, see subsections 3.2.1 and 3.2.2, the damagepattern seen in Fig. 12 strongly suggests thatbreakage was due to preferential corrosion along theannular decorating grooves, and that the corrosionwas due to remanent cold-deformation under thegrooves. In turn, this suggests that the grooves onthe intact part of the paten should be assessed fordamage and use of a possible remedial measure, suchas applying a protective coating to the rear side.

4. Remedial Measures4.1 Introduction

Modern restorations and conservation are con-cerned (or should be concerned) with both technicaland ethical aspects. Essentially, this means respectingan objects integrity and using reversible remedialmeasures. However, reversibility is a controversialtopic and is not always practicable.[4,19]

4.2 Potential Remedial MeasuresBearing the remarks in subsection 4.1 in mind,

Table 2 summarises how the basic condition andtype of embrittlement of ancient silver point topotentially sanctionable remedies. This table shows

that corrosion protection can be a generally appli-cable measure, but heat treatment is not.

4.2.1 Corrosion protection. The most likely andfeasible corrosion protection measure is cleaning,outgassing to dry crack surfaces and any entrappedcorrosion products, and application of a protectivecoating.

The choice of cleaning methods and protectivecoatings requires much forethought and care.[4,20]

An innovative cleaning method is hydrogen plasmareduction.[21] This requires no more than an hour,at temperatures of 40 to 100 C, which minimisesor avoids significant alterations of an objects micro-structure, especially at the lower end of the temper-ature range. The hydrogen plasma reduces surfacecorrosion products to metallic silver, and offers apossible alternative to heat treatment of objectsseverely embrittled by corrosion. Such heat treat-ments require temperatures of 700 C or more.[6]

Recent coating developments suggest thatParylene-type coatings would be ideal for finalprotection. These coatings can be tailored to require-ments, they are very thin and uniform, and theycan be applied by vapour condensation under greatlyreduced atmospheric pressure.[22] This latter featuremeans their application can be directly preceded byconsiderable outgassing and drying of cracks andentrapped corrosion products in the objects.

The need to remove or exclude moisture is alsowhy it may be sanctionable to disassemble oldrestorations and reassemble with modern non-hygroscopic adhesives and fillers, followed by out-gassing and a protective coating.

4.2.2 Heat treatment. Nominally intact artefacts

Table 2 Potential Remedial Measures for Restoration and Conservation of Ancient Silver

Nominally Intact Artefacts and Coins

Undeformed Corrosion protectionDeformed Corrosion protection; heat treatment of coins to remove microstructural embrittlement,

followed by corrosion protection

Restored ArtefactsOld restoration Corrosion protection; disassembly, reassembly, and corrosion protectionModern restoration Corrosion protection

Fragmented Artefacts and CoinsAssembly and corrosion protectionHeat treatment, assembly, and corrosion protection



almost certainly would not be heat treated. Coinsare possible exceptions if microstructurally em-brittled, see Table 2, since they are small, whichmakes it easier to heat treat them. Coins are alsooften less rare, so that heat treatments may beperceived as less risky.

At the other extreme, if it is decided to restoreseverely embrittled and fragmented objects, then heattreatment may be essential.[4,6] As mentioned insubsection 4.2.1, heat treatment to remove severecorrosion-induced embrittlement requires a temper-ature of at least 700 C. The objects need to be inan inert environment or under charcoal. Time attemperature can be short, 5 to 10 minutes, but thetemperature is so high that it excludes artefacts withsoldered joints, niello inlay and gilding. There isalso a risk of incipient melting in silver containingseveral weight percent of copper and other alloyingelements.

Heat treatment to remove microstructurally-induced embrittlement can be done at lower temper-atures. However, the available data suggest temper-atures of at least 200 C, and possibly as high as500 C.[1]

Heat treatments are obviously irreversible remedialmeasures, and information about an objects manu-factured condition and subsequent history is or canbe lost. There is also the possibility that futureinvestigators could be misled by the changed micro-structure. These considerations, together with therisk of further damage, show that heat treatmentsshould be allowed only if preceded by thoroughdiagnostic investigations and if judged feasible byexpert technical staff. A classic example is therestoration of the Hockwold Treasure.[19]

5. AuthenticitySome 25 years ago Schweizer and Meyers made

the interesting proposal that the widths and detailedmorphology of discontinuous precipitation of copperin silver could enable authentication of ancient silverobjects.[23,24] This proposal has been reconsidered,taking into account the Gundestrup Cauldron resultspresented in subsection 3.1.1.[25] A summary isgiven in the next two subsections.

5.1 Discontinuous Precipitate WidthsDiscontinuous precipitate growth rates in binary

silver-copper alloys have been determined in the

temperature range 200 to 375C.[17,26,27] Lowertemperatures are unfeasible,[17] requiring imprac-ticably long heat treatment times.

Assuming Arrhenius-type reaction kinetics,extrapolation of the data indicates a maximumgrowth rate of 103 m/year at ambient temp-eratures.[8,25] For the Gundestrup Cauldron thiswould mean a maximum precipitate width of 2.1to 2.2 m. However, sample 366 had precipitatewidths up to 7 m, see Fig. 5(b) and 6(a). This is sofar beyond the predicted maximum as to discountusing precipitate widths for authentication.

5.2 Discontinuous Precipitate MorphologyBy analogy with other alloy systems, for example

pearlite in iron-carbon alloys, Schweizer and Meyerssuggested that discontinuous precipitation of copperin silver might be characterized by regular lamellae,whose spacing would depend on the ageingtemperature such that one could distinguish betweengenuine long-term precipitation at ambienttemperatures and short-term precipitation atelevated temperatures.

However, all the evidence,[25] exemplified by Fig.6 and higher SEM magnifications, shows that theprecipitate is finely mottled, not lamellar. As before,one must conclude that the precipitate cannot aidauthentication.

6. Extent of the EmbrittlementProblem

6.1 BackgroundOpinions differ on the seriousness of ancient silver

embrittlement. Schweizer and Meyers[24] andKallfass et al.[5] aver that silver is often extremelybrittle. However, Peter Northover has examinedhundreds of silver artefacts and coins, finding onlya small proportion to be badly embrittled.[16] Onneutral ground, Thompson and Chatterjee[1] andWerner[3] state that it is well known that certainsilver objects have become very brittle.

In view of the evidence, see sections 2 and 3 ofthe present paper, there appear to be three primaryfactors in the embrittlement of ancient silver. Theseare the copper and lead contents and remanent cold-deformation. By considering these factors it ispossible to make some general statements on theextent of the embrittlement problem, but first it is


necessary to discuss the chemical compositions ofancient silver and statistically analyse the data formany objects of high silver content.

6.2 Compositions of Ancient Silver6.2.1 General remarks. Native silver alloys may

or may not have been used for Old World ancientobjects.[28-30] However, the scarcity of native silvercompared to silver-containing minerals, mostly leadores, and the early development of lead cupellation,resulted in pyrometallurgy becoming the mainsource of silver.[29,31-34]

Cupellation is very effective in producing silverabove 95 wt.% purity,[33,35] though it usuallycontains minor-to-trace amounts of gold, copper,lead and bismuth, and traces of antimony, arsenic,tellurium, zinc and nickel.[29,34,35] Gold, copper,lead and bismuth contents aregenerally below 1 wt.% for eachelement.

Cupellation experiments byMcKerrell and Stevenson[35]

showed that copper was reduced to0.3 to 0.5 wt.%, with lead andbismuth remaining between 0.5 to1 wt.%. From these results itappears that copper contents above0.5 to 1 wt.% indicate deliberateadditions to increase the strength,and also the wear resistance of coins.Remarkably, this practice seems tohave been followed from as far backas 3000 B.C.E.[29]

6.2.2 Statistics of copper andlead in objects of high silvercontent. A compilation was madeof the copper and lead percentagesfor many Old World artefacts andcoins containing at least 95 wt.%silver.[8,36] Figure 14 presents themedian ranked data on probabilityplots. Figure 14(a) indicates thatthe cumulative occurrence ofcopper percentages fits a normaldistribution, apart from the tails.The 2 test for goodness of fitconfirmed that the distribution wasnormal, including the tails, at asignificance level of 5%. Figure

14(b) shows that the cumulative occurrence of leadpercentages appears to follow a log-normaldistribution. The 2 test for goodness of fit rejectedthis, since there were too many data in the range0.4 to 0.7 wt.% lead.

Of course, there is no intrinsic reason why any setof data should be perfectly described by simplemathematical distributions. However, whenadequate descriptions are obtained, the statisticalresults can be most useful. From this admittedlysubjective viewpoint, Fig. 14 leads to the followingthree conclusions:

1. The data show a high degree of homogeneity,even though taken from a wide variety of sources.[8,36]

This is most probably because cupellation was thesource of the silver.

Fig. 14 Probability plots of copper and lead contents in ancient artefacts and coins containingat least 95 wt.% silver



2. Since the copper percentages are normallydistributed, this implies that any additions of copperto improve the strength and wear resistance wereunsystematic.

3. Although the lead percentages did not satisfythe 2 test for goodness of fit to a log-normaldistribution, the straight line fit in Fig. 14(b) isadequate. The practical significance of this is that alog-normal or approximately log-normal distribu-tion pertains to the concentrations of chemicalprocess residues.[37] In other words, the adequatecorrelation in Fig. 14(b) suggests that the leadpercentages represent the residual lead contents aftercupellation.

6.2.3 Embrittlement-composition relationships.Figure 14 includes data for the GundestrupCauldron samples and the Egyptian vase. These dataand the case histories in section 3 enable the follow-ing remarks on corrosion-induced, microstructurally-induced and synergistic embrittlement:

1. Corrosion-induced embrittlement: see Fig. 14(a).The Gundestrup Cauldron samples 363 and 365and the Egyptian vase contained remanent cold-deformation and had undergone extensive corrosion-induced embrittlement. In contrast, the GundestrupCauldron sample 366, with much higher coppercontent, was essentially annealed and free ofcorrosion. These results demonstrate that remanentcold-deformation is more important for corrosion-induced embrittlement than copper content. Also,Fig. 14(a) shows that half the objects had coppercontents less than 2 wt.%. Like the Egyptian vase,many of these objects would contain remanent cold-deformation to provide sufficient strength foreveryday use. In other words, corrosion-inducedembrittlement could be a widespread problem inhigh-purity ancient silver. Of course, there areadditional factors, including an objects burial timeand the average temperature, moisture content, pHand chemical composition of the burial environment,especially the salt, nitrate and nitrite contents.[31]

However, the foregoing hypothesis is consistent withPeter Northovers observations about Bactriansilver,[16] mentioned in subsection 3.1.2.

2. Microstructurally-induced embrittlement: see Fig.14(b). Mechanical testing of age-hardened silver-lead and silver-lead-copper alloys[1] and observationsabout the Egyptian vase indicate that 0.7 to 0.8

wt.% lead is sufficient to embrittle silver. The pub-lished evidence for microstructurally-inducedembrittlement of ancient silver is very limited atpresent,[1,7] but Fig. 14(b) shows that about a quarterof the objects had lead contents more than 0.7 wt.%.This is a significant fraction, and so detailed exam-ination of other embrittled objects should revealfurther instances of microstructurally-inducedembrittlement: however, see point 3 immediatelyfollowing.

3. Synergistic embrittlement. Since corrosion-induced embrittlement is more in evidence thanmicrostructurally-induced embrittlement, it seemsmore likely that synergistic embrittlement wouldoccur rather than microstructurally-inducedembrittlement on its own. As mentioned in sub-section 3.2.2, synergistic embrittlement is verydetrimental. It is important to recognise synergis-tically embrittled objects because their conservationrequires special care.[7]

7. Concluding RemarksCurrent knowledge enables identifying and

providing explanations of the types of embrittlementof ancient silver, namely corrosion-induced,microstructurally-induced and synergistic embrittle-ment. Two case histories, the Gundestrup Cauldronand an Egyptian vase, have given important andarchetypal information. Automated ElectronBackScatter Diffraction (EBSD) has improved theanalysis and assessment of corrosion-inducedembrittlement by demonstrating the primary detri-mental effect of remanent cold-deformation in anobject. This is also attested to by a third case history,a Byzantine paten.

The information from detailed case histories isimportant for determining the best ways to restoreand conserve embrittled silver objects. The remedialmeasures fall into two main categories: corrosionprotection and heat treatment. Corrosion protectionis a generally applicable measure, though the choiceof method requires much prior consideration. Heattreatments are difficult to justify, but may beessential for some severely embrittled and fragmentedobjects.

Case histories can also contribute to assessing theextent of ancient silver embrittlement. This wasshown by combining results for the GundestrupCauldron and the Egyptian vase with a straight-


forward statistical treatment of the copper and leadcontents of many ancient objects containing at least95 wt.% silver. The indications are that corrosion-induced embrittlement could be a widespread prob-lem in high-purity ancient silver, and that detailedexamination of other embrittled objects is likely toreveal further examples of synergistic embrittlement.It is also possible, but less likely, that micro-structurally-induced embrittlement will be foundon its own.

AcknowledgmentsSeveral colleagues assisted with this work: Peter

Northover, Oxford University; Bart Ankersmit andIneke Joosten, Netherlands Institute for CulturalHeritage; Ron Leenheer, Allard Pierson Museum,Amsterdam; Joanna Cook, The Menil Collection,Houston; Jean-Paul Steijaert and Tim Hattenberg,NLR.

References1. F.C. Thompson and A.K. Chatterjee: Stud. Conserv., 1954,

1, pp. 115-26.

2. C.S. Smith: The Interpretation of Microstructures ofMetallic Artefacts, Application of Science in Examination ofWorks of Art, W.J. Young, ed., Boston Museum of FineArts, Boston, 1965, pp. 20-52.

3. A.E. Werner: Two Problems in the Conservation ofAntiquities: Corroded Lead and Brittle Silver, Applicationof Science in Examination of Works of Art, W.J. Young, ed.,Boston Museum of Fine Arts, Boston, 1965, pp. 96-104.

4. R.M. Organ: The Current Status of the Treatment ofCorroded Metal Artefacts, Corrosion and Metal Artefacts,NBS Special Publication 479, National Bureau of Standards/U.S. Department of Commerce, Washington, 1977, pp.107-42.

5. M. Kallfass, J. Paul, and H. Jehn: Prakt. Metallogr.-Pr. M.,1985, 22, pp. 317-23.

6. I.G. Ravich: Annealing of Brittle Archaeological Silver:Microstructural and Technological Study, 10th TriennialMeeting of the International Council of Museums Committeefor Conservation, Preprints of the Seminar: August 22/27,1993, II, Washington, 1993, pp. 792-95.

7. R.J.H. Wanhill, J.P.H.M. Steijaert, R. Leenheer, and J.F.W.Koens: Archaeometry, 1998, 40, pp. 123-37.

8. R.J.H. Wanhill: Archaeological Silver Embrittlement: aMetallurgical Inquiry, NLR-TP-2002-224, April 2002,National Aerospace Laboratory NLR, Amsterdam.

9. R.J.H. Wanhill: JOM-J. Min. Met. Mat. S., 2003, 55(10),pp. 16-19.

10. R.J.H. Wanhill: Archaeometry, 2003, 45, pp. 625-36.

11. D.A. Scott: Archaeometry, 1996, 38, pp. 305-11.

12. R.J.H. Wanhill, T. Hattenberg, and J.P. Northover: ElectronBackScatter Diffraction (EBSD) of Corrosion, Deforma-tion and Precipitation in the Gundestrup Cauldron, NLR-TP-2003-490, Oct 2003, National Aerospace LaboratoryNLR, Amsterdam.

13. D.B. Williams and J.W. Edington: Acta Metall., 1976, 24,pp. 323-32.

14. W. Gust: Discontinuous Precipitation in Binary MetallicSystems, Phase Transformations, The Institution ofMetallurgists, London, 1979, pp. II-27-68.

15. R.D. Doherty: Diffusive Phase Transformations in the SolidState, Physical Metallurgy, R.W. Cahn and P. Haasen, ed.,Elsevier Science B.V., Amsterdam, 1996, vol. II, pp. 1456-58.

16. J.P. Northover: personal communication, Department ofMaterials, Oxford University, 1999.

17. W. Scharfenberger, G. Schmitt, and H. Borchers: Z.Metallkd., 1972, 63, pp. 553-60.

18. W. Johnson and P.B. Mellor: Plasticity for MechanicalEngineers, D. van Nostrand Company Ltd., London, 1962,pp. 333-34.

19. W.A. Oddy and R. Holmes: The Hockwold Treasure, TheArt of the Conservator, W.A. Oddy, ed., British MuseumPress, London, 1992, pp. 137-50.

20. J. van Reekum and E. Moll: Coating Silverware: fromDaily Use to Museum Object, Zeven IJzersterke Verhalenover Metalen (in Dutch), H.A. Ankersmit and J.A. Mosk,ed., Netherlands Institute for Cultural Heritage,Amsterdam, 2000, pp. 74-79.

21. K. Schmidt-Ott: Plasma Reduction of Silver Surfaces (inGerman), EXPOSURE 2001: Corrosion, Conservation &Study of Historic Metals in Situ, on Display & in Storage, tobe published by Archetype, London, 2004.

22. R. Wood: Mater. World, 2000, 8(6), pp. 30-32.

23. F. Schweizer and P. Meyers: MASCA Journal, 1978, 1, pp.9-10.

24. F. Schweizer and P. Meyers: A New Approach to theAuthenticity of Ancient Silver Objects: the DiscontinuousPrecipitation of Copper from a Silver-Copper Alloy,Proceedings of the 18th International Symposium onArchaeometry and Archaeological Prospection, Rheinland-Verlag GmbH, Cologne, 1979, pp. 287-98.

25. R.J.H. Wanhill, J.P. Northover, and T. Hattenberg: On theSignificance of Discontinuous Precipitation of Copper inAncient Silver, NLR-TP-2003-628, Dec 2003, NationalAerospace Laboratory NLR, Amsterdam.

26. B. Predel and H. Ruge: Z. Metallkd., 1968, 59, pp. 777-81.

27. W. Gust, B. Predel, and K. Diekstall: Z. Metallkd., 1978,69, pp. 75-80.

28. A. Lucas: J. Egypt. Archaeol., 1928, 14, pp. 313-19.

29. N.H. Gale and Z.A. Stos-Gale: J. Egypt. Archaeol., 1981,67, pp. 103-15.



30. G. Philip and T. Rehren: Oxford J. Archaeol., 1996, 15, pp.129-50.

31. W. Gowland: Archaeologia, 1918, 69, pp. 121-60.

32. N.H. Gale and Z.A. Stos-Gale: Sci. Am., 1981, 244(6), pp.142-52.

33. R.F. Tylecote: The Prehistory of Metallurgy in the BritishIsles, The Institute of Metals, London, 1986, pp. 54-61.

34. Ch.J. Raub: The Metallurgy of Gold and Silver inPrehistoric Times, Prehistoric Gold in Europe: Mines, Metal-lurgy and Manufacture, G. Morteani and J.P. Northover, ed.,Kluwer Academic Publishers, Dordrecht, 1995, pp. 243-59.

35. H. McKerrell and R.B.K. Stevenson: Some Analyses ofAnglo-Saxon and Associated Oriental Silver Coinage,Methods of Chemical and Metallurgical Investigation of AncientCoinage, E.T. Hall and D.M. Metcalf, ed., Royal Numis-matic Society, London, 1972, pp. 195-209.

36. R.J.H. Wanhill: Ancient Silver Embrittlement: Signifi-cances of Copper, Lead and Cold-Deformation, NLR-TP-2003-617, December 2003, National Aerospace Lab-oratory NLR, Amsterdam.

37. C. Lipson and N.J. Sheth: Statistical Design and Analysis ofEngineering Experiments, McGraw-Hill Book Company,New York, 1973, p. 60.

JFAP 2007-5-1 Embrittlement of Ancient Silver

Documents

Transcript of JFAP 2007-5-1 Embrittlement of Ancient Silver