ORIGINAL PAPER

Vibrational spectroscopic analysis of an ambernecklace—a forensic historical study

Howell G. M. Edwards

Received: 9 December 2009 /Revised: 4 February 2010 /Accepted: 5 February 2010 /Published online: 8 April 2010# Springer-Verlag 2010

Abstract The vibrational infrared spectroscopic analysis ofan important historical necklace of 102 beads that arepurported to be made of amber indicated strong signalcharacteristics of cellulose nitrate with dark green-colouredareas of a naphthylamine dye. Confocal Raman depth-profiling spectroscopy using a 785-nm laser excitation, anovel application first applied here for the analysis ofinclusions in amber resin, confirmed that the beads wereamber resin and that residues of cellulose nitrate, camphorplasticiser and a naphthylamine dyestuff were present insurface cracks and inclusions in the bead matrix. The beadstringing material was confirmed as cellulose, which wasstained green in part with the dyestuff. Comparison of theRaman spectra of the amber beads with a resin databasesuggested that the amber was sourced from NorthernEngland. The scientific evidence supports the stylisticopinion that the necklace is an important example thatcould date from the 19th Century and that efforts had beenmade to coat it with a synthetic dyed polymer; this providesa rather unusual example of the chemical masking of agenuine article—a procedure that renders the article ofparticular interest.

Keywords Raman spectroscopy . Amber . Necklace .

Museum artefact . Disguise

Introduction

The application of vibrational Raman spectroscopic techni-ques to museum conservation science is relatively newcompared with analogous infrared spectroscopic applica-tions [1–3]; the first Raman spectroscopic studies wereconcerned with the characterisation of pigments on ancienthistoriated manuscripts [4, 5], and it was several years laterbefore the investigation of organic compounds in an art andarchaeological context was successfully attempted [6, 7].This was perhaps surprising in view of the specialadvantage of micro-Raman spectroscopy for the examina-tion of the interaction between pigments and organicsubstrates and for the identification of organic binders andvarnishes on substrates.

The diagnostic capability of vibrational spectroscopyfor the nondestructive discrimination between differentresins and biomaterials from several geographical sour-ces, and for the detection of thermal and environmentalspecimen degradation, is a powerful aid to the provisionof novel information for historical provenancing formuseum curators. In this context, the Raman spectra ofresin specimens from archaeological excavations hasrevealed some key vibrational features that facilitate thebotanical or generic classification of the specimen;examples include the important resins prized by ancientcultures, such as myrrh and frankincense [8–10]. Thedegradation suffered by resins in archaeological burialenvironments has also been studied [11], and the effectson the spectra were evaluated.

In particular, Raman spectroscopic studies of amberresins have been carried out to determine the geograph-ical provenancing of ambers and the discriminationbetween aged, mature resins up to 30 My and younger,immature resins such as copal [12–15]. In these studies,

H. G. M. Edwards (*)University Analytical Centre, Division of Chemical & ForensicSciences, University of Bradford,Bradford BD7 1DP, UKe-mail: h.g.m.edwards@bradford.ac.uk

Anal Bioanal Chem (2010) 397:2677–2683DOI 10.1007/s00216-010-3549-3

definitive information has been forthcoming, which hasestablished the role of Raman spectroscopy in a discrim-inatory context which had hitherto been the exclusivepreserve of infrared spectroscopy [16]. A recent study ofinsect inclusions in ancient amber [17] has been reportedin which the advantage of confocal Raman spectroscopyusing long wavelength laser excitation was manifest; fromthis study, the degradation of the insect keratin encased inits amber matrix could be determined using the transpar-ency of the surrounding amber to good effect. In this currentwork, we shall use this technique to even greater effect bydepth profiling through the inclusions in amber resins.

In the current study, we have extended the application ofconfocal Raman spectroscopy at a longer excitation wave-length to the characterisation of a suspected fake ambernecklace, which has some visible dark green inclusions andsurface deposits. The vibrational spectroscopic character-isation of the necklace materials has been determined, andnovel information has been provided about the authenticity ofthe resin beads comprising the necklace, which havestimulated suggestions forthcoming for its attribution.

Experimental

Specimen

The necklace comprised some 102 irregularly shaped,multi-facetted brownish-yellow resinous beads, each ofapproximately 10-mm dimension, threaded onto a fibrousknotted string, which was stained green in parts (Fig. 1).Many beads were shiny and translucent but mostcontained dark greenish-brown discolourations, whichunder optical microscopic examination proved to besurface deposits and near-surface inclusions. Signs ofresin degradation were evident on several beads by thepresence of surface cracking and areas of nontransparentsurface deterioration. The stringing appears white insidethe holes threading the beads but was stained greenelsewhere. The necklace is believed artistically to be inthe style of the 19th or early 20th Centuries and, if madeof amber, would be an important piece of jewellery fromthis period; however, some doubt had been cast uponthis artistic attribution from earlier preliminary andunpublished infrared spectroscopic measurements, whichindicated that the necklace was composed of a syntheticpolymer, namely cellulose nitrate.

Spectroscopic analysis

Preliminary, unpublished survey of FTIR infrared spectrahad been obtained hitherto from three specimens excisedfrom the necklace: a section of a small transparent bead,

a green powder taken from the surface of a clear beadand the stringing thread near the clasp. Transmissionspectra were obtained from an average of 64 spectralscans taken at 8-cm−1 resolution using a ×15 microscopeobjective. The infrared spectra of the bead, the greencolourant and the string thread clearly matched thereference spectra for cellulose nitrate, naphthylaminebrown F and a cellulose fibre, probably cotton. It wastherefore concluded [18–20] from literature comparisonsthat the necklace was made of cellulose nitrate, which hadbeen dyed to simulate amber along with appropriateinclusions and that the date of probable manufacture wasbetween 1860 and 1920, in conflict with the stylisticattribution. The green colouration in the stringing threadwas attributable to degradation products of cellulosenitrate and the dyestuff.

In the present work, Raman spectra were obtained using aBruker IFS 66/FRA 106 Fourier-transform instrument with aNd3+/YAG laser operating at 1,064 nm in the near-infraredregion; some 2,000 spectral accumulations were necessary to

Fig. 1 The amber(?) necklace of questionable authenticity with 102facetted beads studied in the present work

2678 A. Quye, H.G.M. Edwards

achieve suitable signal-to noise ratios at a spectral resolu-tion of 4 cm−1 over a scan period of typically 30 min persampling point. A selection of beads, comprising some40 beads in all, from the necklace was analysed, representinga range of specimens exhibiting clear translucency through toopaque beads; the sample illumination was effected bymounting each bead with a polished facet perpendicular tothe incident laser beam. The stringing thread was alsoexamined; in all cases, the necklace was not dismantled, andthe analyses were carried out on the beads and stringing threadin situ.

Secondly, a Renishaw InVia confocal Raman spectrom-eter operating at 785 nm with a ×50 microscope objectivewas used to analyse the composition of the dark-browninclusions in the resin beads by carrying out, non-

destructively, a confocal depth-profiling experiment fromthe surface through the resin up to and through the inclusionregions of interest; this procedure did not require thepolishing of the surface facets of the beads or any chemicalor mechanical pre-treatment of the sample. Several beadswere examined in this way.

Results and discussion

Hitherto unpublished infrared spectral data indicated thatthe suspected amber resin beads possessed absorptionbands characteristic of cellulose nitrate and a naphthyl-amine dyestuff with there being few or no evidence for thepresence of amber resin.

Fig. 2 FT Raman spectra ofcellulose fibres: (a) cotton, (b)linen with added jute, (c)stringing thread from beadnecklace. a Wavenumberregion,2,600–3,600 cm−1;b wavenumber region, 250–1,700 cm−1; 1,064-nm laserexcitation

Vibrational spectroscopic analysis of an amber necklace 2679

In contrast, Raman spectroscopy clearly demonstratedfrom the outset that the beads were composed of amberresin but also indicated that they contained featurescharacteristic of cellulose nitrate, camphor plasticiser anda naphthylamine dyestuff.

Figure 2a, b shows the FT Raman spectral stackplots ofthree spectra comprising (a) cotton, (b) linen with added jutefor fibre strengthening purposes and (c) the bead stringingmaterial in the necklace. Although cotton and linen havevery similar Raman spectra based on the cellulosic compo-sition of their individual fibres, namely cotton and flax, withcharacteristic bands at 1,478, 1,378, 1,337, 1,151, 1,096,985, 520, 436 and 378 cm−1, the analogous spectra of juteand flax have an additional feature at 1,601 cm−1 assignableto the stretching vibration of C═C unsaturation in the bastfibres [21, 22]. The wavenumber regions 2,600–3,200 and250–1,700 cm−1 in Fig. 2a, b, respectively, indicate that thebead stringing material is cellulosic and furthermore iscomposed of jute-free cotton or linen fibres.

Raman spectroscopic analyses of a selected number ofthe 102 resin beads in the necklace, some 40 in all,indicated that specimen heterogeneity was significant;although the primary Raman spectroscopic signals werethose of amber, as exemplified in the spectral stackplots inFig. 3 for selected beads numbered 2, 5 and 7 in the lowerthree plots, bead number 16 produced the uppermostspectrum in the stackplot; it can be inferred from thespectral data presented here that this bead seems to be madefrom a different material to the others. This conclusion wasmaintained over the cumulative analyses of some 40 beadsselected from the 102 beads comprising the necklace,selected on the basis of their appearance, which ranged

from beads that were composed of a translucent, clearbrown resin to those that exhibited “degradation” effects,which were manifest in areas of surface cloudiness and thepresence of areas of dark greenish-brown colour located atthe surface and in several subsurface inclusions. The amberRaman spectral signals obtained from the majority of thenecklace beads analysed here were compared with ourdatabase of published spectral features of a variety ofambers and copals of different maturity, age and geograph-ical distribution [12–15], although in all cases the genericRaman spectrum characteristic of amber resins is given,with bands at 2,930 cm−1 (CH stretching), 1,659 and 1,645(C═C stretching), 1,441 (CH2 scissors deformation), 1,362(CH2 deformation), 979 and 947 (CH rocking) and 746 and697 cm−1 (C–C stretching). It is also possible to identifyseveral combinations of bands whose presence serves tolocate the probable geographical origin of the amber resinin the necklace beads. In the case of the beads studied here,we can suggest that the amber resin came from the NorthSea, English coast. In particular, although the Ramanspectral bands of the “English” amber are very similar tothose of Baltic amber, the former has features at 3,080(unsaturated C═CH stretching), 2,984 (CH2 stretching),1,700 (C═O stretching) and 1,501 cm−1 (unassignedhitherto) that are absent from spectra of the Baltic amber.In addition, support for the English amber attribution isprovided by the Raman bands in the beads found at1,298 cm−1 and a distinctive triplet which occurs between700 and 800 cm−1, which have hitherto been reported tocharacterise verified samples of amber that were collectedon the North Sea coast of England. Figure 4a, b illustratesthese specific spectral differences.

Fig. 3 FT Raman spectra ofselected beads from a necklace,demonstrating their amber com-position; from bottom, beads 2,5, 7 and 16. Wavenumber range250–3,700 cm−1; 1,064-nmexcitation

2680 A. Quye, H.G.M. Edwards

The upper spectrum of Fig. 3, which was obtained frombead number 16, is clearly not assignable to amber bycomparison with the definitive amber beads also presentedin the same spectral stackplot. We can attribute this upperspectrum obtained from bead 16 to a beta-naphthylaminedyestuff [23]. Nevertheless, other regions of this bead, andof others studied here, do present the spectroscopicsignatures of amber, as do other beads from those selectedin this study. We can conclude, therefore, that several beadsin the necklace have been treated with a coloured dyestuffsometime after its manufacture and that possibly thistreatment has been abraded or worn off. In other beadsanalysed, the Raman spectra exhibited can be described ascompounded of the spectra of both the dyestuff and theunderlying amber resin, which are then superimposed uponeach other; this was quite a common occurrence, and the

upper spectrum in Fig. 5, which was obtained from a regionof bead number 22, is a good example of this.

A third scenario was manifest from the Raman spectralanalysis of the surface regions of other beads in that thesespectra were different from both those of the naphthylaminedyestuff and of the amber resin matrix; an example isshown in Fig. 6, seen in many beads of the 40 analysedhere, and this is clearly assignable to cellulose nitrate withcharacteristic Raman bands at 1,731 (C═O from camphorplasticizer in cellulose nitrate), 1,646 (NO2 asymmetricstretching in cellulose nitrate), 1,449 (CH deformation incamphor), 1,376 (ring vibration in cellulose nitrate), 1,283(NO2 symmetric stretching in cellulose nitrate), 862 (C–Cin camphor), 650 and 556 cm−1 (C–C ring deformations incamphor) in agreement with the recent definitive andelegant study of the Raman spectra of cellulose nitrates by

Fig. 5 FT Raman spectra oftwo regions of bead 22, show-ing the characteristic spectrumof amber in the lower spectrumand that of amber with thecoloured dye superimposedupon it in the upper spectrum;wavenumber range, 250–3,600 cm−1; 1,064-nmexcitation

Fig. 4 FT Raman spectra ofBaltic and northern Englishcoast ambers; bottom, Englishcoast; upper, Baltic. Wavenum-ber range 300–3,200 cm−1;1,064-nm excitation

Vibrational spectroscopic analysis of an amber necklace 2681

Paris and Coupry [24]. The presence of strong signals fromthe residual camphor plasticizer in cellulose nitrate isespecially interesting for attribution purposes since asignificant wavenumber shift is observed in the C═Cstretch in free camphor and that of the residual camphorplasticiser in cellulose nitrate, namely from 1,746 to1,731 cm−1, the latter being identical to the observed bandin the spectra recorded here, therefore confirming its originas a plasticiser for the cellulose nitrate.

Finally, a confocal depth-profiling Raman microspectro-scopic study of several beads from the necklace wasundertaken using a 785-nm excitation in an attempt toidentify the greenish-brown subsurface inclusions observedin many of the beads comprising the necklace. Thestackplotted spectra in Fig. 7 show the results of an exemplar

depth profile transect analysis of bead number 2 from thesurface into the bead and through an inclusion; over thewavenumber range 400–1,800 cm−1, the spectra, from the top,show the amber resin matrix above the inclusion, the interfacebetween the amber matrix and the inclusion, the inclusionitself and finally the amber matrix below the inclusion; thesespectra indicate that the transect through the bead and into theinclusion show amber, followed by an amber-rich regionsimilar to that shown in the uppermost spectrum in Fig. 5, theamber-deficient beta-naphthyl dyestuff in the inclusion and,finally, the amber matrix again. It is important to note herethat these spectral data were obtained non-destructively froma depth transect through a necklace bead and did not involvethe sectioning of the bead or chemical or mechanical pre-treatment of the specimen in any way.

Fig. 6 FT Raman spectrum ofcellulose nitrate with residualcamphor plasticiser; wavenum-ber range 250–4,000 cm−1;1,064 nm excitation

Fig. 7 Depth profile analysis ofan amber bead with dark-browninclusion; confocal Ramanmicrospectrometer with ×50microscope objective, 785-nmlaser excitation; wavenumberrange, 400–1,800 cm−1. Fromtop, surface of amber bead,junction between inclusionand amber matrix, inclusionand amber matrix belowinclusion

2682 A. Quye, H.G.M. Edwards

Conclusions

The vibrational spectroscopic study of a large 102-beadnecklace, which was originally believed to be made ofamber resin but now was considered to be of questionableauthenticity following an unpublished IR study, demon-strate the value of a nondestructive analytical approach;whereas infrared spectroscopy indicated predominantly abead composition of cellulose nitrate and a beta-naphthylamine dyestuff, Raman spectroscopy clearly indi-cates that the beads are made of amber, which showsremnants of both cellulose nitrate and of a naphthyldyestuff. Moreover, the Raman spectra of the amber beadsplace their geographical origin firmly on the North Seacoast of England. This amber necklace is believedstylistically to belong to the the 19th or 20th Centuries;however this attribution is compromised by the discoveryof the presence of a synthetic polymer, cellulose nitrate, anda synthetic dyestuff which both date from the later part ofthe 19th Century at the earliest. However, the analyticalinformation from the present Raman spectroscopic studystrongly suggests that a perfectly genuine amber resinnecklace had been “modified” by treatment with syntheticadditives; whatever the reason for this, which may havebeen undertaken as a protection, it could be hypothesisedthat the adoption by society of the new synthetic replace-ments for expensive natural materials, such as amber,tortoiseshell and ivory, that were much in vogue sociallyin the late nineteenth century generated the modifications tothis necklace, which have been detected analytically thatnow make it probably quite unique. Whereas there areexamples known of genuine and valuable artefacts beingcovered with commonplace materials in an attempt todisguise their true origins for the purpose of contrabandsmuggling and eventual illegal profiteering, the necklacesubject of the present study clearly does not belong to thiscategory. This makes the necklace probably unique andcertainly an example of forensic historical importance.

Acknowledgement The author is indebted to Dr Anita Quye of theDepartment of Conservation and Analytical Research, NationalMuseums of Scotland, Edinburgh, for background information andpersonal communication regarding the preliminary spectroscopic dataand investigations carried out on this unusual necklace.

References

1. Edwards HGM, Chalmers JM (eds) (2005) Raman spectroscopyin archaeology and art history. RSC Analytical ChemistrySpectroscopy Monographs Series. Royal Society of Chemistry,Cambridge, 476

2. Edwards HGM (2009) Raman spectroscopy in art and archae-ology: a new light on historical mysteries. In: Laane J (ed)Frontiers of molecular spectroscopy. Elsevier, Amsterdam,pp 133–173

3. Edwards HGM (2009) Raman spectroscopy of inorganicmaterials in art & archaeology: spectroscopic analysis ofhistorical mysteries. In J Yarwood, R Douthwaite, SB Duckett(eds) Specialist periodical reports, “Spectroscopic Properties ofInorganic and Organometallic Compounds: Materials andApplications”. Royal Society of Chemistry, Cambridge 40,pp 16–48

4. Guineau B (1989) Studs Conserv 34:385. Clark RJH (1995) Chem Revs 42:1876. Edwards HGM (2004) Analyst 129:8707. Vandenabeele P, Edwards HGM, Moens L (2007) Chem Revs

107:6758. Edwards HGM, Falk MJ (1997) J Raman Spectrosc 28:2119. Edwards HGM, David AR, Brody RH (2008) J Raman Spectrosc

39:96610. Edwards HGM, Sibley MG (1997) Spectrochim Acta, Part A

53:237311. de Faria DLA, Edwards HGM, Afonso MC, Brody RH, Morais JL

(2004) Spectrochim Acta, Part A 60:150512. Edwards HGM, Farwell DW (1996) Spectrochim Acta, Part A

52:111913. Brody RH, Edwards HGM, Pollard AM (2001) Spectrochim Acta,

Part A 57:132514. Brody RH, Edwards HGM, Pollard AM (2002) Biopolymers 67:12915. Vandenabeele P, Grimaldi DM, Edwards HGM, Moens L (2003)

Spectrochim Acta, Part A 59:222116. Beck CW (1986) Appl Spectrosc Revs 22:5717. Edwards HGM, Farwell DW, Jorge Villar SE (2007) Spectrochim

Acta, Part A 68:108918. Carlson J, Pretzel B, Price B (eds) (2000) Infrared and Raman

Users Group (IRUG) Spectral Database Edition 2000. IRUG,Philadelphia

19. Quye A, Williamson C (eds) (1999) Plastics: collecting andconserving. NMS, Edinburgh

20. Knecht E, Rawson C, Lowenthal R (1941) A manual of dyeing,volume II, 9th ed. London, pp. 556, 559

21. Edwards HGM, Farwell DW, Webster D (1997) SpectrochimActa, Part A 53:2383

22. Edwards HGM, Wyeth P, Farwell DW, Nikhassan NF, Garside P(2006) J Raman Spectrosc 37:1193

23. Vandenabeele P, Moens L, Edwards HGM, Dams R (2000) JRaman Spectrosc 31:509

24. Paris C, Coupry C (2005) J Raman Spectrosc 36:77

Vibrational spectroscopic analysis of an amber necklace 2683